ChemWiki - User contributions [en]

Rep:Efr114MgOwiki

2017-03-24T11:49:38Z

Efr114: /* Quasi-Harmonic Approximation Method */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

The purpose of this experiment was to simulate the thermal expansion of MgO and to determine the thermal expansion coefficient for MgO using both the Quasi-Harmonic Approximation method and the Molecular Dynamics method.

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature.

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Fig. 4 a-e. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon.
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. 5a, this grid corresponds to one k-point sampled for the calculation. Fig. 5a has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 2 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|'''Fig. 5a''' Phonon DOS with a 1x1 grid ]]
|[[File:5x5efr114.PNG|thumb|200px|'''Fig. 5b''' Phonon DOS with a 5x5 grid ]]
|[[File:10x10newefr114.PNG|thumb|200px|'''Fig. 5c''' Phonon DOS with a 10x10 grid]]
|[[File:20x20efr114.PNG|thumb|200px|'''Fig. 5d''' Phonon DOS with a 20x20 grid]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|'''Fig. 5e''' Phonon DOS with a 30x30 grid ]]
|[[File:35x35efr114.PNG|thumb|200px|'''Fig. 5f''' Phonon DOS with a 35x35 grid ]]
|[[File:50x50.PNG|thumb|200px|'''Fig. 5g''' Phonon DOS with a 50x50 grid]]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure 5g, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation at 300K. The shinking factor was varied between 1 and 30 in steps of 5 and the optimum grid size was found to be between shrinking factor 15 and 20. The shrinking factor was then varied between 15 and 20 in steps of 1 to determine the optimal grid size. Table 2 shows the values of HFE and the difference from the most accurate value of HFE (<math> \Delta </math> HFE) for a range of shrinking factors.
{| class="wikitable"
!Shrinking factor
!HFE / eV
!<math> \Delta </math> HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|-
| colspan="3" |Table 2: the values of HFE and the difference from the
most accurate value of HFE for a range of

shrinking factors
|}

The accuracy of the HFE increases with increasing shrinking factor.
To obtain an accuracy of 1 meV, a grid of 2x2x2 should be used. For an accuracy of 0.5 meV, a grid of 2x2x2 should be used and to obtain an accuracy of 0.1 meV, a grid of 3x3x3 should be used.
The data in this simulation suggests that an 18x18x18 grid is the optimum grid size as it is the lowest grid size that still gives the energy to an accuracy of 6 decimal places.
Considering the qualitative visual inspection method as well as the energy calculations, an 18x18x18 grid was chosen for further calculations.

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The variance of Helmholtz free energy with temperature can be seen in Fig. 6a. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. 6a can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>component becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity.

{|align="center"
|[[File:Free energyefr114.PNG|thumb|250px|'''Fig. 6a '''A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations. ]]
|[[File:Mgotemplat.PNG|thumb|250px|'''Fig. 6b''' A plot showing the variance of lattice parameter with temperature ]]
|}

The varience of lattice parameter with temperature can be seen in Fig. 6b. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

{|align="center"
|[[File:Volumeefr114kj.PNG|thumb|250px|'''Fig. 7 '''A plot showing the variance of primitive cell volume with Temperature for MgO ]]
|}
Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range. The following calculation can then be performed to find the thermal expansion coefficient. <div style="text-align: center;">
<math>
V(T)=0.000000534T^{2} + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

{|align="center"
|[[File:Alphagraph.PNG|thumb|250px|'''Fig. 8 '''A plot of the thermal expansion coefficient against temperature ]]
|}

Figure 8 shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

a<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5} K^{-1}
</math>
</div>

This calculated value is in agreeable magnitude with literature, with a 3.5498% percentage difference to the literature value, which states a thermal expansion coefficient of <math>3.11*10^{-5}</math>.<ref> O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref>

=== Molecular Dynamics ===
GULP Molecular Dynamics calculations were used to calculate the lattice parameters of MgO at temperatures ranging from 0-1000 K. A supercell composed of 32 primitive cells and a SF of 1 was used for the calculation. The supercell was used to find an optimal input that was a compromise between accuracy and CPU usage.

A supercell is required for MD calculations as the simulation in dependent on a large system of atoms with randomized velocities and randomized population of states.
{|align="center"
|[[File:HEATefr114.PNG|thumb|250px|'''Fig. ''' '''9''' A plot showing how the averaged and instantaneous primitive unit cell volumes vary with time ]]
|}
Fig. 9 show's how the volume of the supercell varies with time at 300K. Initially, the system is equilibrating, however after the first 200 fs the averaged volume starts tending towards an equilibrium.

{|align="center"
|[[File:Md volefr114.PNG|thumb|250px|'''Fig. 10 '''A plot of primitive unit cell volume against temperature for MgO - Molecular dynamics ]]
|[[File:Volcomp.PNG|thumb|250px|'''Fig. 10b '''A plot of primitive cell volume against temperature for both MD and QHA calculations ]]
|}
Fig. 10a shows there is a linear relationship between primitive unit cell volume and temperature. A thermal expansion coefficient of 3.0454*10<sup>-5</sup> was extracted from the data, this value is in the same magnitude with a percentage difference of 2.0772% from the literature value of 3.11*10<sup>-5</sup>. <ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref>
<div style="text-align: center;">
<math>
V(T)=0.0005689
</math>
<math>
\frac {dV}{dT}=0.0005689
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.0005689)
</math>

<math>
\alpha_V=3.0454*10^{-5}
</math>
</div>

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|'''Fig. 11''' A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig 12 a and b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. 12a''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. 12b ''' The DOS graph for CaO at shrinking factor 20 ]]
|}
QHA and MD methods were then applied to the calcium oxide model to determine plots of primitive unit cell volume against temperature to produce Fig. 13. The QHA calculation method produced a thermal expansion coefficient of 5.0271*10<sup>-5</sup> K<sup>-1</sup>and the MD calculation method produced a thermal expansion coefficient of 4.9243*10<sup>-5</sup> K<sup>-1</sup>from 300 to 1000K in comparison to a literature value of 4.0791*10<sup>-5 </sup>K<sup>-1</sup>.<ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref>[[File:Caofinahfdahfl.PNG|centre|thumb|'''Fig.13 '''A plot of primitive unit cell volume against temperature for MD and QHA methods. ]]

== Conclusion ==
The thermal expansion of MgO was investigated using two models, the quasi-harmonic model and the molecular dynamic models. In the quasi-harmonic model, phonon dispersion curves, phonon density of state curves and Helmholtz free energy were calculated at different grid sizes to determine that a grid of 18x18x18 was the optimal grid size for use in the subsequent calculations of thermal expansion coefficient using the quasi-harmonic model. Although the quasi-harmonic model is less accurate than the molecular dynamics model, it still has the significant advantage of using much less CPU and is less expensiveThe molecular dynamic model was found to be more accurate however the model does have issues with CPU usage and cost.

The experiment was also extended to CaO via modification of both DLVis and GULP files. The optimal grid size for which was found to be 20x20x20. The thermal expansion coefficients for both CaO and MgO were determined

== References ==

Rep:Efr114MgOwiki

2017-03-24T11:48:19Z

Efr114: /* Quasi-Harmonic Approximation Method */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

The purpose of this experiment was to simulate the thermal expansion of MgO and to determine the thermal expansion coefficient for MgO using both the Quasi-Harmonic Approximation method and the Molecular Dynamics method.

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature.

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Fig. 4 a-e. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon.
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. 5a, this grid corresponds to one k-point sampled for the calculation. Fig. 5a has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 2 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|'''Fig. 5a''' Phonon DOS with a 1x1 grid ]]
|[[File:5x5efr114.PNG|thumb|200px|'''Fig. 5b''' Phonon DOS with a 5x5 grid ]]
|[[File:10x10newefr114.PNG|thumb|200px|'''Fig. 5c''' Phonon DOS with a 10x10 grid]]
|[[File:20x20efr114.PNG|thumb|200px|'''Fig. 5d''' Phonon DOS with a 20x20 grid]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|'''Fig. 5e''' Phonon DOS with a 30x30 grid ]]
|[[File:35x35efr114.PNG|thumb|200px|'''Fig. 5f''' Phonon DOS with a 35x35 grid ]]
|[[File:50x50.PNG|thumb|200px|'''Fig. 5g''' Phonon DOS with a 50x50 grid]]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure 5g, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation at 300K. The shinking factor was varied between 1 and 30 in steps of 5 and the optimum grid size was found to be between shrinking factor 15 and 20. The shrinking factor was then varied between 15 and 20 in steps of 1 to determine the optimal grid size. Table 2 shows the values of HFE and the difference from the most accurate value of HFE (<math> \Delta </math> HFE) for a range of shrinking factors.
{| class="wikitable"
!Shrinking factor
!HFE / eV
!<math> \Delta </math> HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|-
| colspan="3" |Table 2: the values of HFE and the difference from the
most accurate value of HFE for a range of

shrinking factors
|}

The accuracy of the HFE increases with increasing shrinking factor.
To obtain an accuracy of 1 meV, a grid of 2x2x2 should be used. For an accuracy of 0.5 meV, a grid of 2x2x2 should be used and to obtain an accuracy of 0.1 meV, a grid of 3x3x3 should be used.
The data in this simulation suggests that an 18x18x18 grid is the optimum grid size as it is the lowest grid size that still gives the energy to an accuracy of 6 decimal places.
Considering the qualitative visual inspection method as well as the energy calculations, an 18x18x18 grid was chosen for further calculations.

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The variance of Helmholtz free energy with temperature can be seen in Fig. 6a. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. 6a can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>component becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity.

{|align="center"
|[[File:Free energyefr114.PNG|thumb|250px|'''Fig. 6a '''A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations. ]]
|[[File:Mgotemplat.PNG|thumb|250px|'''Fig. 6b''' A plot showing the variance of lattice parameter with temperature ]]
|}

The varience of lattice parameter with temperature can be seen in Fig. 6b. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

{|align="center"
|[[File:Volumeefr114kj.PNG|thumb|250px|'''Fig. 7 '''A plot showing the variance of primitive cell volume with Temperature for MgO ]]
|}
Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range. The following calculation can then be performed to find the thermal expansion coefficient. <div style="text-align: center;">
<math>
V(T)=0.000000534T^{2} + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

{|align="center"
|[[File:Alphagraph.PNG|thumb|250px|'''Fig. 8 '''A plot of the thermal expansion coefficient against temperature ]]
|}

Figure 8 shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

a<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5} K^{-1}
</math>
</div>

This calculated value is in agreeable magnitude with literature, with a 3.5498% percentage difference to the literature value, which states a thermal expansion coefficient of 3.11*10^-5.<nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki>

=== Molecular Dynamics ===
GULP Molecular Dynamics calculations were used to calculate the lattice parameters of MgO at temperatures ranging from 0-1000 K. A supercell composed of 32 primitive cells and a SF of 1 was used for the calculation. The supercell was used to find an optimal input that was a compromise between accuracy and CPU usage.

A supercell is required for MD calculations as the simulation in dependent on a large system of atoms with randomized velocities and randomized population of states.
{|align="center"
|[[File:HEATefr114.PNG|thumb|250px|'''Fig. ''' '''9''' A plot showing how the averaged and instantaneous primitive unit cell volumes vary with time ]]
|}
Fig. 9 show's how the volume of the supercell varies with time at 300K. Initially, the system is equilibrating, however after the first 200 fs the averaged volume starts tending towards an equilibrium.

{|align="center"
|[[File:Md volefr114.PNG|thumb|250px|'''Fig. 10 '''A plot of primitive unit cell volume against temperature for MgO - Molecular dynamics ]]
|[[File:Volcomp.PNG|thumb|250px|'''Fig. 10b '''A plot of primitive cell volume against temperature for both MD and QHA calculations ]]
|}
Fig. 10a shows there is a linear relationship between primitive unit cell volume and temperature. A thermal expansion coefficient of 3.0454*10<sup>-5</sup> was extracted from the data, this value is in the same magnitude with a percentage difference of 2.0772% from the literature value of 3.11*10<sup>-5</sup>. <ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref>
<div style="text-align: center;">
<math>
V(T)=0.0005689
</math>
<math>
\frac {dV}{dT}=0.0005689
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.0005689)
</math>

<math>
\alpha_V=3.0454*10^{-5}
</math>
</div>

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|'''Fig. 11''' A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig 12 a and b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. 12a''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. 12b ''' The DOS graph for CaO at shrinking factor 20 ]]
|}
QHA and MD methods were then applied to the calcium oxide model to determine plots of primitive unit cell volume against temperature to produce Fig. 13. The QHA calculation method produced a thermal expansion coefficient of 5.0271*10<sup>-5</sup> K<sup>-1</sup>and the MD calculation method produced a thermal expansion coefficient of 4.9243*10<sup>-5</sup> K<sup>-1</sup>from 300 to 1000K in comparison to a literature value of 4.0791*10<sup>-5 </sup>K<sup>-1</sup>.<ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref>[[File:Caofinahfdahfl.PNG|centre|thumb|'''Fig.13 '''A plot of primitive unit cell volume against temperature for MD and QHA methods. ]]

== Conclusion ==
The thermal expansion of MgO was investigated using two models, the quasi-harmonic model and the molecular dynamic models. In the quasi-harmonic model, phonon dispersion curves, phonon density of state curves and Helmholtz free energy were calculated at different grid sizes to determine that a grid of 18x18x18 was the optimal grid size for use in the subsequent calculations of thermal expansion coefficient using the quasi-harmonic model. Although the quasi-harmonic model is less accurate than the molecular dynamics model, it still has the significant advantage of using much less CPU and is less expensiveThe molecular dynamic model was found to be more accurate however the model does have issues with CPU usage and cost.

The experiment was also extended to CaO via modification of both DLVis and GULP files. The optimal grid size for which was found to be 20x20x20. The thermal expansion coefficients for both CaO and MgO were determined

== References ==

Rep:Efr114MgOwiki

2017-03-24T11:47:53Z

Efr114: /* Extension: CaO: calculations and comparision */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

The purpose of this experiment was to simulate the thermal expansion of MgO and to determine the thermal expansion coefficient for MgO using both the Quasi-Harmonic Approximation method and the Molecular Dynamics method.

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature.

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Fig. 4 a-e. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon.
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. 5a, this grid corresponds to one k-point sampled for the calculation. Fig. 5a has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 2 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|'''Fig. 5a''' Phonon DOS with a 1x1 grid ]]
|[[File:5x5efr114.PNG|thumb|200px|'''Fig. 5b''' Phonon DOS with a 5x5 grid ]]
|[[File:10x10newefr114.PNG|thumb|200px|'''Fig. 5c''' Phonon DOS with a 10x10 grid]]
|[[File:20x20efr114.PNG|thumb|200px|'''Fig. 5d''' Phonon DOS with a 20x20 grid]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|'''Fig. 5e''' Phonon DOS with a 30x30 grid ]]
|[[File:35x35efr114.PNG|thumb|200px|'''Fig. 5f''' Phonon DOS with a 35x35 grid ]]
|[[File:50x50.PNG|thumb|200px|'''Fig. 5g''' Phonon DOS with a 50x50 grid]]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure 5g, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation at 300K. The shinking factor was varied between 1 and 30 in steps of 5 and the optimum grid size was found to be between shrinking factor 15 and 20. The shrinking factor was then varied between 15 and 20 in steps of 1 to determine the optimal grid size. Table 2 shows the values of HFE and the difference from the most accurate value of HFE (<math> \Delta </math> HFE) for a range of shrinking factors.
{| class="wikitable"
!Shrinking factor
!HFE / eV
!<math> \Delta </math> HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|-
| colspan="3" |Table 2: the values of HFE and the difference from the
most accurate value of HFE for a range of

shrinking factors
|}

The accuracy of the HFE increases with increasing shrinking factor.
To obtain an accuracy of 1 meV, a grid of 2x2x2 should be used. For an accuracy of 0.5 meV, a grid of 2x2x2 should be used and to obtain an accuracy of 0.1 meV, a grid of 3x3x3 should be used.
The data in this simulation suggests that an 18x18x18 grid is the optimum grid size as it is the lowest grid size that still gives the energy to an accuracy of 6 decimal places.
Considering the qualitative visual inspection method as well as the energy calculations, an 18x18x18 grid was chosen for further calculations.

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The variance of Helmholtz free energy with temperature can be seen in Fig. 6a. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. 6a can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>component becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity.

{|align="center"
|[[File:Free energyefr114.PNG|thumb|250px|'''Fig. 6a '''A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations. ]]
|[[File:Mgotemplat.PNG|thumb|250px|'''Fig. 6b''' A plot showing the variance of lattice parameter with temperature ]]
|}

The varience of lattice parameter with temperature can be seen in Fig. 6b. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

{|align="center"
|[[File:Volumeefr114kj.PNG|thumb|250px|'''Fig. 7 '''A plot showing the variance of primitive cell volume with Temperature for MgO ]]
|}
Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range. The following calculation can then be performed to find the thermal expansion coefficient. <div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

{|align="center"
|[[File:Alphagraph.PNG|thumb|250px|'''Fig. 8 '''A plot of the thermal expansion coefficient against temperature ]]
|}

Figure 8 shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

a<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5} K^{-1}
</math>
</div>

This calculated value is in agreeable magnitude with literature, with a 3.5498% percentage difference to the literature value, which states a thermal expansion coefficient of 3.11*10^-5.<nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki>

=== Molecular Dynamics ===
GULP Molecular Dynamics calculations were used to calculate the lattice parameters of MgO at temperatures ranging from 0-1000 K. A supercell composed of 32 primitive cells and a SF of 1 was used for the calculation. The supercell was used to find an optimal input that was a compromise between accuracy and CPU usage.

A supercell is required for MD calculations as the simulation in dependent on a large system of atoms with randomized velocities and randomized population of states.
{|align="center"
|[[File:HEATefr114.PNG|thumb|250px|'''Fig. ''' '''9''' A plot showing how the averaged and instantaneous primitive unit cell volumes vary with time ]]
|}
Fig. 9 show's how the volume of the supercell varies with time at 300K. Initially, the system is equilibrating, however after the first 200 fs the averaged volume starts tending towards an equilibrium.

{|align="center"
|[[File:Md volefr114.PNG|thumb|250px|'''Fig. 10 '''A plot of primitive unit cell volume against temperature for MgO - Molecular dynamics ]]
|[[File:Volcomp.PNG|thumb|250px|'''Fig. 10b '''A plot of primitive cell volume against temperature for both MD and QHA calculations ]]
|}
Fig. 10a shows there is a linear relationship between primitive unit cell volume and temperature. A thermal expansion coefficient of 3.0454*10<sup>-5</sup> was extracted from the data, this value is in the same magnitude with a percentage difference of 2.0772% from the literature value of 3.11*10<sup>-5</sup>. <ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref>
<div style="text-align: center;">
<math>
V(T)=0.0005689
</math>
<math>
\frac {dV}{dT}=0.0005689
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.0005689)
</math>

<math>
\alpha_V=3.0454*10^{-5}
</math>
</div>

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|'''Fig. 11''' A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig 12 a and b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. 12a''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. 12b ''' The DOS graph for CaO at shrinking factor 20 ]]
|}
QHA and MD methods were then applied to the calcium oxide model to determine plots of primitive unit cell volume against temperature to produce Fig. 13. The QHA calculation method produced a thermal expansion coefficient of 5.0271*10<sup>-5</sup> K<sup>-1</sup>and the MD calculation method produced a thermal expansion coefficient of 4.9243*10<sup>-5</sup> K<sup>-1</sup>from 300 to 1000K in comparison to a literature value of 4.0791*10<sup>-5 </sup>K<sup>-1</sup>.<ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref>[[File:Caofinahfdahfl.PNG|centre|thumb|'''Fig.13 '''A plot of primitive unit cell volume against temperature for MD and QHA methods. ]]

== Conclusion ==
The thermal expansion of MgO was investigated using two models, the quasi-harmonic model and the molecular dynamic models. In the quasi-harmonic model, phonon dispersion curves, phonon density of state curves and Helmholtz free energy were calculated at different grid sizes to determine that a grid of 18x18x18 was the optimal grid size for use in the subsequent calculations of thermal expansion coefficient using the quasi-harmonic model. Although the quasi-harmonic model is less accurate than the molecular dynamics model, it still has the significant advantage of using much less CPU and is less expensiveThe molecular dynamic model was found to be more accurate however the model does have issues with CPU usage and cost.

The experiment was also extended to CaO via modification of both DLVis and GULP files. The optimal grid size for which was found to be 20x20x20. The thermal expansion coefficients for both CaO and MgO were determined

== References ==

Rep:Efr114MgOwiki

2017-03-24T11:47:14Z

Efr114: /* Molecular Dynamics */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

The purpose of this experiment was to simulate the thermal expansion of MgO and to determine the thermal expansion coefficient for MgO using both the Quasi-Harmonic Approximation method and the Molecular Dynamics method.

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature.

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Fig. 4 a-e. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon.
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. 5a, this grid corresponds to one k-point sampled for the calculation. Fig. 5a has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 2 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|'''Fig. 5a''' Phonon DOS with a 1x1 grid ]]
|[[File:5x5efr114.PNG|thumb|200px|'''Fig. 5b''' Phonon DOS with a 5x5 grid ]]
|[[File:10x10newefr114.PNG|thumb|200px|'''Fig. 5c''' Phonon DOS with a 10x10 grid]]
|[[File:20x20efr114.PNG|thumb|200px|'''Fig. 5d''' Phonon DOS with a 20x20 grid]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|'''Fig. 5e''' Phonon DOS with a 30x30 grid ]]
|[[File:35x35efr114.PNG|thumb|200px|'''Fig. 5f''' Phonon DOS with a 35x35 grid ]]
|[[File:50x50.PNG|thumb|200px|'''Fig. 5g''' Phonon DOS with a 50x50 grid]]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure 5g, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation at 300K. The shinking factor was varied between 1 and 30 in steps of 5 and the optimum grid size was found to be between shrinking factor 15 and 20. The shrinking factor was then varied between 15 and 20 in steps of 1 to determine the optimal grid size. Table 2 shows the values of HFE and the difference from the most accurate value of HFE (<math> \Delta </math> HFE) for a range of shrinking factors.
{| class="wikitable"
!Shrinking factor
!HFE / eV
!<math> \Delta </math> HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|-
| colspan="3" |Table 2: the values of HFE and the difference from the
most accurate value of HFE for a range of

shrinking factors
|}

The accuracy of the HFE increases with increasing shrinking factor.
To obtain an accuracy of 1 meV, a grid of 2x2x2 should be used. For an accuracy of 0.5 meV, a grid of 2x2x2 should be used and to obtain an accuracy of 0.1 meV, a grid of 3x3x3 should be used.
The data in this simulation suggests that an 18x18x18 grid is the optimum grid size as it is the lowest grid size that still gives the energy to an accuracy of 6 decimal places.
Considering the qualitative visual inspection method as well as the energy calculations, an 18x18x18 grid was chosen for further calculations.

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The variance of Helmholtz free energy with temperature can be seen in Fig. 6a. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. 6a can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>component becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity.

{|align="center"
|[[File:Free energyefr114.PNG|thumb|250px|'''Fig. 6a '''A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations. ]]
|[[File:Mgotemplat.PNG|thumb|250px|'''Fig. 6b''' A plot showing the variance of lattice parameter with temperature ]]
|}

The varience of lattice parameter with temperature can be seen in Fig. 6b. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

{|align="center"
|[[File:Volumeefr114kj.PNG|thumb|250px|'''Fig. 7 '''A plot showing the variance of primitive cell volume with Temperature for MgO ]]
|}
Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range. The following calculation can then be performed to find the thermal expansion coefficient. <div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

{|align="center"
|[[File:Alphagraph.PNG|thumb|250px|'''Fig. 8 '''A plot of the thermal expansion coefficient against temperature ]]
|}

Figure 8 shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

a<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5} K^{-1}
</math>
</div>

This calculated value is in agreeable magnitude with literature, with a 3.5498% percentage difference to the literature value, which states a thermal expansion coefficient of 3.11*10^-5.<nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki>

=== Molecular Dynamics ===
GULP Molecular Dynamics calculations were used to calculate the lattice parameters of MgO at temperatures ranging from 0-1000 K. A supercell composed of 32 primitive cells and a SF of 1 was used for the calculation. The supercell was used to find an optimal input that was a compromise between accuracy and CPU usage.

A supercell is required for MD calculations as the simulation in dependent on a large system of atoms with randomized velocities and randomized population of states.
{|align="center"
|[[File:HEATefr114.PNG|thumb|250px|'''Fig. ''' '''9''' A plot showing how the averaged and instantaneous primitive unit cell volumes vary with time ]]
|}
Fig. 9 show's how the volume of the supercell varies with time at 300K. Initially, the system is equilibrating, however after the first 200 fs the averaged volume starts tending towards an equilibrium.

{|align="center"
|[[File:Md volefr114.PNG|thumb|250px|'''Fig. 10 '''A plot of primitive unit cell volume against temperature for MgO - Molecular dynamics ]]
|[[File:Volcomp.PNG|thumb|250px|'''Fig. 10b '''A plot of primitive cell volume against temperature for both MD and QHA calculations ]]
|}
Fig. 10a shows there is a linear relationship between primitive unit cell volume and temperature. A thermal expansion coefficient of 3.0454*10<sup>-5</sup> was extracted from the data, this value is in the same magnitude with a percentage difference of 2.0772% from the literature value of 3.11*10<sup>-5</sup>. <ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref>
<div style="text-align: center;">
<math>
V(T)=0.0005689
</math>
<math>
\frac {dV}{dT}=0.0005689
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.0005689)
</math>

<math>
\alpha_V=3.0454*10^{-5}
</math>
</div>

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|'''Fig. 11''' A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig 12 a and b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. 12a''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. 12b ''' The DOS graph for CaO at shrinking factor 20 ]]
|}
QHA and MD methods were then applied to the calcium oxide model to determine plots of primitive unit cell volume against temperature to produce Fig. 13. The QHA calculation method produced a thermal expansion coefficient of 5.0271*10<sup>-5</sup> K<sup>-1</sup>and the MD calculation method produced a thermal expansion coefficient of 4.9243*10<sup>-5</sup> K<sup>-1</sup>from 300 to 1000K in comparison to a literature value of 4.0791*10<sup>-5 </sup>K<sup>-1</sup>.<nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki> [[File:Caofinahfdahfl.PNG|centre|thumb|'''Fig.13 '''A plot of primitive unit cell volume against temperature for MD and QHA methods. ]]

== Conclusion ==
The thermal expansion of MgO was investigated using two models, the quasi-harmonic model and the molecular dynamic models. In the quasi-harmonic model, phonon dispersion curves, phonon density of state curves and Helmholtz free energy were calculated at different grid sizes to determine that a grid of 18x18x18 was the optimal grid size for use in the subsequent calculations of thermal expansion coefficient using the quasi-harmonic model. Although the quasi-harmonic model is less accurate than the molecular dynamics model, it still has the significant advantage of using much less CPU and is less expensiveThe molecular dynamic model was found to be more accurate however the model does have issues with CPU usage and cost.

The experiment was also extended to CaO via modification of both DLVis and GULP files. The optimal grid size for which was found to be 20x20x20. The thermal expansion coefficients for both CaO and MgO were determined

== References ==

Rep:Efr114MgOwiki

2017-03-24T11:46:35Z

Efr114: /* Grid Sizes for alternative solids. */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

The purpose of this experiment was to simulate the thermal expansion of MgO and to determine the thermal expansion coefficient for MgO using both the Quasi-Harmonic Approximation method and the Molecular Dynamics method.

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature.

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Fig. 4 a-e. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon.
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. 5a, this grid corresponds to one k-point sampled for the calculation. Fig. 5a has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 2 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|'''Fig. 5a''' Phonon DOS with a 1x1 grid ]]
|[[File:5x5efr114.PNG|thumb|200px|'''Fig. 5b''' Phonon DOS with a 5x5 grid ]]
|[[File:10x10newefr114.PNG|thumb|200px|'''Fig. 5c''' Phonon DOS with a 10x10 grid]]
|[[File:20x20efr114.PNG|thumb|200px|'''Fig. 5d''' Phonon DOS with a 20x20 grid]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|'''Fig. 5e''' Phonon DOS with a 30x30 grid ]]
|[[File:35x35efr114.PNG|thumb|200px|'''Fig. 5f''' Phonon DOS with a 35x35 grid ]]
|[[File:50x50.PNG|thumb|200px|'''Fig. 5g''' Phonon DOS with a 50x50 grid]]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure 5g, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation at 300K. The shinking factor was varied between 1 and 30 in steps of 5 and the optimum grid size was found to be between shrinking factor 15 and 20. The shrinking factor was then varied between 15 and 20 in steps of 1 to determine the optimal grid size. Table 2 shows the values of HFE and the difference from the most accurate value of HFE (<math> \Delta </math> HFE) for a range of shrinking factors.
{| class="wikitable"
!Shrinking factor
!HFE / eV
!<math> \Delta </math> HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|-
| colspan="3" |Table 2: the values of HFE and the difference from the
most accurate value of HFE for a range of

shrinking factors
|}

The accuracy of the HFE increases with increasing shrinking factor.
To obtain an accuracy of 1 meV, a grid of 2x2x2 should be used. For an accuracy of 0.5 meV, a grid of 2x2x2 should be used and to obtain an accuracy of 0.1 meV, a grid of 3x3x3 should be used.
The data in this simulation suggests that an 18x18x18 grid is the optimum grid size as it is the lowest grid size that still gives the energy to an accuracy of 6 decimal places.
Considering the qualitative visual inspection method as well as the energy calculations, an 18x18x18 grid was chosen for further calculations.

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The variance of Helmholtz free energy with temperature can be seen in Fig. 6a. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. 6a can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>component becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity.

{|align="center"
|[[File:Free energyefr114.PNG|thumb|250px|'''Fig. 6a '''A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations. ]]
|[[File:Mgotemplat.PNG|thumb|250px|'''Fig. 6b''' A plot showing the variance of lattice parameter with temperature ]]
|}

The varience of lattice parameter with temperature can be seen in Fig. 6b. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

{|align="center"
|[[File:Volumeefr114kj.PNG|thumb|250px|'''Fig. 7 '''A plot showing the variance of primitive cell volume with Temperature for MgO ]]
|}
Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range. The following calculation can then be performed to find the thermal expansion coefficient. <div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

{|align="center"
|[[File:Alphagraph.PNG|thumb|250px|'''Fig. 8 '''A plot of the thermal expansion coefficient against temperature ]]
|}

Figure 8 shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

a<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5} K^{-1}
</math>
</div>

This calculated value is in agreeable magnitude with literature, with a 3.5498% percentage difference to the literature value, which states a thermal expansion coefficient of 3.11*10^-5.<nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki>

=== Molecular Dynamics ===
GULP Molecular Dynamics calculations were used to calculate the lattice parameters of MgO at temperatures ranging from 0-1000 K. A supercell composed of 32 primitive cells and a SF of 1 was used for the calculation. The supercell was used to find an optimal input that was a compromise between accuracy and CPU usage.

A supercell is required for MD calculations as the simulation in dependent on a large system of atoms with randomized velocities and randomized population of states.
{|align="center"
|[[File:HEATefr114.PNG|thumb|250px|'''Fig. ''' '''9''' A plot showing how the averaged and instantaneous primitive unit cell volumes vary with time ]]
|}
Fig. 9 show's how the volume of the supercell varies with time at 300K. Initially, the system is equilibrating, however after the first 200 fs the averaged volume starts tending towards an equilibrium.

{|align="center"
|[[File:Md volefr114.PNG|thumb|250px|'''Fig. 10 '''A plot of primitive unit cell volume against temperature for MgO - Molecular dynamics ]]
|[[File:Volcomp.PNG|thumb|250px|'''Fig. 10b '''A plot of primitive cell volume against temperature for both MD and QHA calculations ]]
|}
Fig. 10a shows there is a linear relationship between primitive unit cell volume and temperature. A thermal expansion coefficient of 3.0454*10<sup>-5</sup> was extracted from the data, this value is in the same magnitude with a percentage difference of 2.0772% from the literature value of 3.11*10<sup>-5</sup>. <nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki><div style="text-align: center;">
<math>
V(T)=0.0005689
</math>
<math>
\frac {dV}{dT}=0.0005689
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.0005689)
</math>

<math>
\alpha_V=3.0454*10^{-5}
</math>
</div>

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|'''Fig. 11''' A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig 12 a and b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. 12a''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. 12b ''' The DOS graph for CaO at shrinking factor 20 ]]
|}
QHA and MD methods were then applied to the calcium oxide model to determine plots of primitive unit cell volume against temperature to produce Fig. 13. The QHA calculation method produced a thermal expansion coefficient of 5.0271*10<sup>-5</sup> K<sup>-1</sup>and the MD calculation method produced a thermal expansion coefficient of 4.9243*10<sup>-5</sup> K<sup>-1</sup>from 300 to 1000K in comparison to a literature value of 4.0791*10<sup>-5 </sup>K<sup>-1</sup>.<nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki> [[File:Caofinahfdahfl.PNG|centre|thumb|'''Fig.13 '''A plot of primitive unit cell volume against temperature for MD and QHA methods. ]]

== Conclusion ==
The thermal expansion of MgO was investigated using two models, the quasi-harmonic model and the molecular dynamic models. In the quasi-harmonic model, phonon dispersion curves, phonon density of state curves and Helmholtz free energy were calculated at different grid sizes to determine that a grid of 18x18x18 was the optimal grid size for use in the subsequent calculations of thermal expansion coefficient using the quasi-harmonic model. Although the quasi-harmonic model is less accurate than the molecular dynamics model, it still has the significant advantage of using much less CPU and is less expensiveThe molecular dynamic model was found to be more accurate however the model does have issues with CPU usage and cost.

The experiment was also extended to CaO via modification of both DLVis and GULP files. The optimal grid size for which was found to be 20x20x20. The thermal expansion coefficients for both CaO and MgO were determined

== References ==

Rep:Efr114MgOwiki

2017-03-24T11:45:36Z

Efr114: /* Conclusion */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

The purpose of this experiment was to simulate the thermal expansion of MgO and to determine the thermal expansion coefficient for MgO using both the Quasi-Harmonic Approximation method and the Molecular Dynamics method.

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature.

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Fig. 4 a-e. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon.
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. 5a, this grid corresponds to one k-point sampled for the calculation. Fig. 5a has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 2 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|'''Fig. 5a''' Phonon DOS with a 1x1 grid ]]
|[[File:5x5efr114.PNG|thumb|200px|'''Fig. 5b''' Phonon DOS with a 5x5 grid ]]
|[[File:10x10newefr114.PNG|thumb|200px|'''Fig. 5c''' Phonon DOS with a 10x10 grid]]
|[[File:20x20efr114.PNG|thumb|200px|'''Fig. 5d''' Phonon DOS with a 20x20 grid]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|'''Fig. 5e''' Phonon DOS with a 30x30 grid ]]
|[[File:35x35efr114.PNG|thumb|200px|'''Fig. 5f''' Phonon DOS with a 35x35 grid ]]
|[[File:50x50.PNG|thumb|200px|'''Fig. 5g''' Phonon DOS with a 50x50 grid]]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure 5g, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation at 300K. The shinking factor was varied between 1 and 30 in steps of 5 and the optimum grid size was found to be between shrinking factor 15 and 20. The shrinking factor was then varied between 15 and 20 in steps of 1 to determine the optimal grid size. Table 2 shows the values of HFE and the difference from the most accurate value of HFE (<math> \Delta </math> HFE) for a range of shrinking factors.
{| class="wikitable"
!Shrinking factor
!HFE / eV
!<math> \Delta </math> HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|-
| colspan="3" |Table 2: the values of HFE and the difference from the
most accurate value of HFE for a range of

shrinking factors
|}

The accuracy of the HFE increases with increasing shrinking factor.
To obtain an accuracy of 1 meV, a grid of 2x2x2 should be used. For an accuracy of 0.5 meV, a grid of 2x2x2 should be used and to obtain an accuracy of 0.1 meV, a grid of 3x3x3 should be used.
The data in this simulation suggests that an 18x18x18 grid is the optimum grid size as it is the lowest grid size that still gives the energy to an accuracy of 6 decimal places.
Considering the qualitative visual inspection method as well as the energy calculations, an 18x18x18 grid was chosen for further calculations.

=== Grid Sizes for alternative solids. ===
Crystals vary in both structure (e.g. FCC, BCC) and lattice parameters '''(a,b,c and math \alpha \beta \gamma)'''

'''Calcium oxide, CaO has the same structure as MgO'''
*# '''a similar oxide (eg: CaO) ?'''
*# '''a Zeolite (eg: Faujasite) ?'''
*# '''a metal (eg: lithium) ?'''

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The variance of Helmholtz free energy with temperature can be seen in Fig. 6a. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. 6a can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>component becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity.

{|align="center"
|[[File:Free energyefr114.PNG|thumb|250px|'''Fig. 6a '''A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations. ]]
|[[File:Mgotemplat.PNG|thumb|250px|'''Fig. 6b''' A plot showing the variance of lattice parameter with temperature ]]
|}

The varience of lattice parameter with temperature can be seen in Fig. 6b. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

{|align="center"
|[[File:Volumeefr114kj.PNG|thumb|250px|'''Fig. 7 '''A plot showing the variance of primitive cell volume with Temperature for MgO ]]
|}
Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range. The following calculation can then be performed to find the thermal expansion coefficient. <div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

{|align="center"
|[[File:Alphagraph.PNG|thumb|250px|'''Fig. 8 '''A plot of the thermal expansion coefficient against temperature ]]
|}

Figure 8 shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

a<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5} K^{-1}
</math>
</div>

This calculated value is in agreeable magnitude with literature, with a 3.5498% percentage difference to the literature value, which states a thermal expansion coefficient of 3.11*10^-5.<nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki>

=== Molecular Dynamics ===
GULP Molecular Dynamics calculations were used to calculate the lattice parameters of MgO at temperatures ranging from 0-1000 K. A supercell composed of 32 primitive cells and a SF of 1 was used for the calculation. The supercell was used to find an optimal input that was a compromise between accuracy and CPU usage.

A supercell is required for MD calculations as the simulation in dependent on a large system of atoms with randomized velocities and randomized population of states.
{|align="center"
|[[File:HEATefr114.PNG|thumb|250px|'''Fig. ''' '''9''' A plot showing how the averaged and instantaneous primitive unit cell volumes vary with time ]]
|}
Fig. 9 show's how the volume of the supercell varies with time at 300K. Initially, the system is equilibrating, however after the first 200 fs the averaged volume starts tending towards an equilibrium.

{|align="center"
|[[File:Md volefr114.PNG|thumb|250px|'''Fig. 10 '''A plot of primitive unit cell volume against temperature for MgO - Molecular dynamics ]]
|[[File:Volcomp.PNG|thumb|250px|'''Fig. 10b '''A plot of primitive cell volume against temperature for both MD and QHA calculations ]]
|}
Fig. 10a shows there is a linear relationship between primitive unit cell volume and temperature. A thermal expansion coefficient of 3.0454*10<sup>-5</sup> was extracted from the data, this value is in the same magnitude with a percentage difference of 2.0772% from the literature value of 3.11*10<sup>-5</sup>. <nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki><div style="text-align: center;">
<math>
V(T)=0.0005689
</math>
<math>
\frac {dV}{dT}=0.0005689
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.0005689)
</math>

<math>
\alpha_V=3.0454*10^{-5}
</math>
</div>

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|'''Fig. 11''' A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig 12 a and b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. 12a''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. 12b ''' The DOS graph for CaO at shrinking factor 20 ]]
|}
QHA and MD methods were then applied to the calcium oxide model to determine plots of primitive unit cell volume against temperature to produce Fig. 13. The QHA calculation method produced a thermal expansion coefficient of 5.0271*10<sup>-5</sup> K<sup>-1</sup>and the MD calculation method produced a thermal expansion coefficient of 4.9243*10<sup>-5</sup> K<sup>-1</sup>from 300 to 1000K in comparison to a literature value of 4.0791*10<sup>-5 </sup>K<sup>-1</sup>.<nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki> [[File:Caofinahfdahfl.PNG|centre|thumb|'''Fig.13 '''A plot of primitive unit cell volume against temperature for MD and QHA methods. ]]

== Conclusion ==
The thermal expansion of MgO was investigated using two models, the quasi-harmonic model and the molecular dynamic models. In the quasi-harmonic model, phonon dispersion curves, phonon density of state curves and Helmholtz free energy were calculated at different grid sizes to determine that a grid of 18x18x18 was the optimal grid size for use in the subsequent calculations of thermal expansion coefficient using the quasi-harmonic model. Although the quasi-harmonic model is less accurate than the molecular dynamics model, it still has the significant advantage of using much less CPU and is less expensiveThe molecular dynamic model was found to be more accurate however the model does have issues with CPU usage and cost.

The experiment was also extended to CaO via modification of both DLVis and GULP files. The optimal grid size for which was found to be 20x20x20. The thermal expansion coefficients for both CaO and MgO were determined

== References ==

Rep:Efr114MgOwiki

2017-03-24T11:36:59Z

Efr114: /* Conclusion */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

The purpose of this experiment was to simulate the thermal expansion of MgO and to determine the thermal expansion coefficient for MgO using both the Quasi-Harmonic Approximation method and the Molecular Dynamics method.

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature.

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Fig. 4 a-e. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon.
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. 5a, this grid corresponds to one k-point sampled for the calculation. Fig. 5a has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 2 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|'''Fig. 5a''' Phonon DOS with a 1x1 grid ]]
|[[File:5x5efr114.PNG|thumb|200px|'''Fig. 5b''' Phonon DOS with a 5x5 grid ]]
|[[File:10x10newefr114.PNG|thumb|200px|'''Fig. 5c''' Phonon DOS with a 10x10 grid]]
|[[File:20x20efr114.PNG|thumb|200px|'''Fig. 5d''' Phonon DOS with a 20x20 grid]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|'''Fig. 5e''' Phonon DOS with a 30x30 grid ]]
|[[File:35x35efr114.PNG|thumb|200px|'''Fig. 5f''' Phonon DOS with a 35x35 grid ]]
|[[File:50x50.PNG|thumb|200px|'''Fig. 5g''' Phonon DOS with a 50x50 grid]]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure 5g, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation at 300K. The shinking factor was varied between 1 and 30 in steps of 5 and the optimum grid size was found to be between shrinking factor 15 and 20. The shrinking factor was then varied between 15 and 20 in steps of 1 to determine the optimal grid size. Table 2 shows the values of HFE and the difference from the most accurate value of HFE (<math> \Delta </math> HFE) for a range of shrinking factors.
{| class="wikitable"
!Shrinking factor
!HFE / eV
!<math> \Delta </math> HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|-
| colspan="3" |Table 2: the values of HFE and the difference from the
most accurate value of HFE for a range of

shrinking factors
|}

The accuracy of the HFE increases with increasing shrinking factor.
To obtain an accuracy of 1 meV, a grid of 2x2x2 should be used. For an accuracy of 0.5 meV, a grid of 2x2x2 should be used and to obtain an accuracy of 0.1 meV, a grid of 3x3x3 should be used.
The data in this simulation suggests that an 18x18x18 grid is the optimum grid size as it is the lowest grid size that still gives the energy to an accuracy of 6 decimal places.
Considering the qualitative visual inspection method as well as the energy calculations, an 18x18x18 grid was chosen for further calculations.

=== Grid Sizes for alternative solids. ===
Crystals vary in both structure (e.g. FCC, BCC) and lattice parameters '''(a,b,c and math \alpha \beta \gamma)'''

'''Calcium oxide, CaO has the same structure as MgO'''
*# '''a similar oxide (eg: CaO) ?'''
*# '''a Zeolite (eg: Faujasite) ?'''
*# '''a metal (eg: lithium) ?'''

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The variance of Helmholtz free energy with temperature can be seen in Fig. 6a. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. 6a can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>component becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity.

{|align="center"
|[[File:Free energyefr114.PNG|thumb|250px|'''Fig. 6a '''A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations. ]]
|[[File:Mgotemplat.PNG|thumb|250px|'''Fig. 6b''' A plot showing the variance of lattice parameter with temperature ]]
|}

The varience of lattice parameter with temperature can be seen in Fig. 6b. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

{|align="center"
|[[File:Volumeefr114kj.PNG|thumb|250px|'''Fig. 7 '''A plot showing the variance of primitive cell volume with Temperature for MgO ]]
|}
Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range. The following calculation can then be performed to find the thermal expansion coefficient. <div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

{|align="center"
|[[File:Alphagraph.PNG|thumb|250px|'''Fig. 8 '''A plot of the thermal expansion coefficient against temperature ]]
|}

Figure 8 shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

a<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5} K^{-1}
</math>
</div>

This calculated value is in agreeable magnitude with literature, with a 3.5498% percentage difference to the literature value, which states a thermal expansion coefficient of 3.11*10^-5.<nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki>

=== Molecular Dynamics ===
GULP Molecular Dynamics calculations were used to calculate the lattice parameters of MgO at temperatures ranging from 0-1000 K. A supercell composed of 32 primitive cells and a SF of 1 was used for the calculation. The supercell was used to find an optimal input that was a compromise between accuracy and CPU usage.

A supercell is required for MD calculations as the simulation in dependent on a large system of atoms with randomized velocities and randomized population of states.
{|align="center"
|[[File:HEATefr114.PNG|thumb|250px|'''Fig. ''' '''9''' A plot showing how the averaged and instantaneous primitive unit cell volumes vary with time ]]
|}
Fig. 9 show's how the volume of the supercell varies with time at 300K. Initially, the system is equilibrating, however after the first 200 fs the averaged volume starts tending towards an equilibrium.

{|align="center"
|[[File:Md volefr114.PNG|thumb|250px|'''Fig. 10 '''A plot of primitive unit cell volume against temperature for MgO - Molecular dynamics ]]
|[[File:Volcomp.PNG|thumb|250px|'''Fig. 10b '''A plot of primitive cell volume against temperature for both MD and QHA calculations ]]
|}
Fig. 10a shows there is a linear relationship between primitive unit cell volume and temperature. A thermal expansion coefficient of 3.0454*10<sup>-5</sup> was extracted from the data, this value is in the same magnitude with a percentage difference of 2.0772% from the literature value of 3.11*10<sup>-5</sup>. <nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki><div style="text-align: center;">
<math>
V(T)=0.0005689
</math>
<math>
\frac {dV}{dT}=0.0005689
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.0005689)
</math>

<math>
\alpha_V=3.0454*10^{-5}
</math>
</div>

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|'''Fig. 11''' A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig 12 a and b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. 12a''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. 12b ''' The DOS graph for CaO at shrinking factor 20 ]]
|}
QHA and MD methods were then applied to the calcium oxide model to determine plots of primitive unit cell volume against temperature to produce Fig. 13. The QHA calculation method produced a thermal expansion coefficient of 5.0271*10<sup>-5</sup> K<sup>-1</sup>and the MD calculation method produced a thermal expansion coefficient of 4.9243*10<sup>-5</sup> K<sup>-1</sup>from 300 to 1000K in comparison to a literature value of 4.0791*10<sup>-5 </sup>K<sup>-1</sup>.<nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki> [[File:Caofinahfdahfl.PNG|centre|thumb|'''Fig.13 '''A plot of primitive unit cell volume against temperature for MD and QHA methods. ]]

== Conclusion ==
The thermal expansion of MgO was investigated using two models, the quasi-harmonic model and the molecular dynamic models. In the quasi-harmonic model, phonon dispersion curves, phonon density of state curves and Helmholtz free energy were calculated at different grid sizes to determine that a grid of 18x18x18 was the optimal grid size for use in the subsequent calculations of thermal expansion coefficient using the quasi-harmonic model.

== References ==

Rep:Efr114MgOwiki

2017-03-24T11:27:59Z

Efr114: /* Molecular Dynamics */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

The purpose of this experiment was to simulate the thermal expansion of MgO and to determine the thermal expansion coefficient for MgO using both the Quasi-Harmonic Approximation method and the Molecular Dynamics method.

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature.

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Fig. 4 a-e. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon.
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. 5a, this grid corresponds to one k-point sampled for the calculation. Fig. 5a has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 2 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|'''Fig. 5a''' Phonon DOS with a 1x1 grid ]]
|[[File:5x5efr114.PNG|thumb|200px|'''Fig. 5b''' Phonon DOS with a 5x5 grid ]]
|[[File:10x10newefr114.PNG|thumb|200px|'''Fig. 5c''' Phonon DOS with a 10x10 grid]]
|[[File:20x20efr114.PNG|thumb|200px|'''Fig. 5d''' Phonon DOS with a 20x20 grid]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|'''Fig. 5e''' Phonon DOS with a 30x30 grid ]]
|[[File:35x35efr114.PNG|thumb|200px|'''Fig. 5f''' Phonon DOS with a 35x35 grid ]]
|[[File:50x50.PNG|thumb|200px|'''Fig. 5g''' Phonon DOS with a 50x50 grid]]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure 5g, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation at 300K. The shinking factor was varied between 1 and 30 in steps of 5 and the optimum grid size was found to be between shrinking factor 15 and 20. The shrinking factor was then varied between 15 and 20 in steps of 1 to determine the optimal grid size. Table 2 shows the values of HFE and the difference from the most accurate value of HFE (<math> \Delta </math> HFE) for a range of shrinking factors.
{| class="wikitable"
!Shrinking factor
!HFE / eV
!<math> \Delta </math> HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|-
| colspan="3" |Table 2: the values of HFE and the difference from the
most accurate value of HFE for a range of

shrinking factors
|}

The accuracy of the HFE increases with increasing shrinking factor.
To obtain an accuracy of 1 meV, a grid of 2x2x2 should be used. For an accuracy of 0.5 meV, a grid of 2x2x2 should be used and to obtain an accuracy of 0.1 meV, a grid of 3x3x3 should be used.
The data in this simulation suggests that an 18x18x18 grid is the optimum grid size as it is the lowest grid size that still gives the energy to an accuracy of 6 decimal places.
Considering the qualitative visual inspection method as well as the energy calculations, an 18x18x18 grid was chosen for further calculations.

=== Grid Sizes for alternative solids. ===
Crystals vary in both structure (e.g. FCC, BCC) and lattice parameters '''(a,b,c and math \alpha \beta \gamma)'''

'''Calcium oxide, CaO has the same structure as MgO'''
*# '''a similar oxide (eg: CaO) ?'''
*# '''a Zeolite (eg: Faujasite) ?'''
*# '''a metal (eg: lithium) ?'''

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The variance of Helmholtz free energy with temperature can be seen in Fig. 6a. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. 6a can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>component becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity.

{|align="center"
|[[File:Free energyefr114.PNG|thumb|250px|'''Fig. 6a '''A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations. ]]
|[[File:Mgotemplat.PNG|thumb|250px|'''Fig. 6b''' A plot showing the variance of lattice parameter with temperature ]]
|}

The varience of lattice parameter with temperature can be seen in Fig. 6b. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

{|align="center"
|[[File:Volumeefr114kj.PNG|thumb|250px|'''Fig. 7 '''A plot showing the variance of primitive cell volume with Temperature for MgO ]]
|}
Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range. The following calculation can then be performed to find the thermal expansion coefficient. <div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

{|align="center"
|[[File:Alphagraph.PNG|thumb|250px|'''Fig. 8 '''A plot of the thermal expansion coefficient against temperature ]]
|}

Figure 8 shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

a<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5} K^{-1}
</math>
</div>

This calculated value is in agreeable magnitude with literature, with a 3.5498% percentage difference to the literature value, which states a thermal expansion coefficient of 3.11*10^-5.<nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki>

=== Molecular Dynamics ===
GULP Molecular Dynamics calculations were used to calculate the lattice parameters of MgO at temperatures ranging from 0-1000 K. A supercell composed of 32 primitive cells and a SF of 1 was used for the calculation. The supercell was used to find an optimal input that was a compromise between accuracy and CPU usage.

A supercell is required for MD calculations as the simulation in dependent on a large system of atoms with randomized velocities and randomized population of states.
{|align="center"
|[[File:HEATefr114.PNG|thumb|250px|'''Fig. ''' '''9''' A plot showing how the averaged and instantaneous primitive unit cell volumes vary with time ]]
|}
Fig. 9 show's how the volume of the supercell varies with time at 300K. Initially, the system is equilibrating, however after the first 200 fs the averaged volume starts tending towards an equilibrium.

{|align="center"
|[[File:Md volefr114.PNG|thumb|250px|'''Fig. 10 '''A plot of primitive unit cell volume against temperature for MgO - Molecular dynamics ]]
|[[File:Volcomp.PNG|thumb|250px|'''Fig. 10b '''A plot of primitive cell volume against temperature for both MD and QHA calculations ]]
|}
Fig. 10a shows there is a linear relationship between primitive unit cell volume and temperature. A thermal expansion coefficient of 3.0454*10<sup>-5</sup> was extracted from the data, this value is in the same magnitude with a percentage difference of 2.0772% from the literature value of 3.11*10<sup>-5</sup>. <nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki><div style="text-align: center;">
<math>
V(T)=0.0005689
</math>
<math>
\frac {dV}{dT}=0.0005689
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.0005689)
</math>

<math>
\alpha_V=3.0454*10^{-5}
</math>
</div>

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|'''Fig. 11''' A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig 12 a and b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. 12a''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. 12b ''' The DOS graph for CaO at shrinking factor 20 ]]
|}
QHA and MD methods were then applied to the calcium oxide model to determine plots of primitive unit cell volume against temperature to produce Fig. 13. The QHA calculation method produced a thermal expansion coefficient of 5.0271*10<sup>-5</sup> K<sup>-1</sup>and the MD calculation method produced a thermal expansion coefficient of 4.9243*10<sup>-5</sup> K<sup>-1</sup>from 300 to 1000K in comparison to a literature value of 4.0791*10<sup>-5 </sup>K<sup>-1</sup>.<nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki> [[File:Caofinahfdahfl.PNG|centre|thumb|'''Fig.13 '''A plot of primitive unit cell volume against temperature for MD and QHA methods. ]]

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T11:13:12Z

Efr114: /* Molecular Dynamics */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

The purpose of this experiment was to simulate the thermal expansion of MgO and to determine the thermal expansion coefficient for MgO using both the Quasi-Harmonic Approximation method and the Molecular Dynamics method.

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature.

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Fig. 4 a-e. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon.
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|'''Fig. 5a''' Phonon DOS with a 1x1 grid ]]
|[[File:5x5efr114.PNG|thumb|200px|'''Fig. 5b''' Phonon DOS with a 5x5 grid ]]
|[[File:10x10newefr114.PNG|thumb|200px|'''Fig. 5c''' Phonon DOS with a 10x10 grid]]
|[[File:20x20efr114.PNG|thumb|200px|'''Fig. 5d''' Phonon DOS with a 20x20 grid]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|'''Fig. 5e''' Phonon DOS with a 30x30 grid ]]
|[[File:35x35efr114.PNG|thumb|200px|'''Fig. 5f''' Phonon DOS with a 35x35 grid ]]
|[[File:50x50.PNG|thumb|200px|'''Fig. 5g''' Phonon DOS with a 50x50 grid]]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation at 300K. The shinking factor was varied between 1 and 30 in steps of 5 and the optimum grid size was found to be between shrinking factor 15 and 20. The shrinking factor was then varied between 15 and 20 in steps of 1 to determine the optimal grid size. Table 2 shows the values of HFE and the difference from the most accurate value of HFE (<math> \Delta </math> HFE) for a range of shrinking factors.
{| class="wikitable"
!Shrinking factor
!HFE / eV
!<math> \Delta </math> HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|-
| colspan="3" |Table 2: the values of HFE and the difference from the
most accurate value of HFE for a range of

shrinking factors
|}

The accuracy of the HFE increases with increasing shrinking factor.
To obtain an accuracy of 1 meV, a grid of 2x2x2 should be used. For an accuracy of 0.5 meV, a grid of 2x2x2 should be used and to obtain an accuracy of 0.1 meV, a grid of 3x3x3 should be used.
The data in this simulation suggests that an 18x18x18 grid is the optimum grid size as it is the lowest grid size that still gives the energy to an accuracy of 6 decimal places.
Considering the qualitative visual inspection method as well as the energy calculations, an 18x18x18 grid was chosen for further calculations.

=== Grid Sizes for alternative solids. ===
Crystals vary in both structure (e.g. FCC, BCC) and lattice parameters '''(a,b,c and math \alpha \beta \gamma)'''

'''Calcium oxide, CaO has the same structure as MgO'''
*# '''a similar oxide (eg: CaO) ?'''
*# '''a Zeolite (eg: Faujasite) ?'''
*# '''a metal (eg: lithium) ?'''

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The variance of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>component becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity.

{|align="center"
|[[File:Free energyefr114.PNG|thumb|250px|'''Fig. 6a '''A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations. ]]
|[[File:Mgotemplat.PNG|thumb|250px|'''Fig. 6b''' blah A plot showing the varience of lattice parameter with temperature ]]
|}

The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

{|align="center"
|[[File:Volumeefr114kj.PNG|thumb|250px|'''Fig. 7''' ]]
|}
Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.<div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

{|align="center"
|[[File:Alphagraph.PNG|thumb|250px|'''Fig. 8''' ]]
|}

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

a<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5} K^{-1}
</math>
</div>

This calculated value is in agreeable magnitude with literature, with a 3.5498% percentage difference to the literature value, which states a thermal expansion coefficient of 3.11*10^-5.<nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki>

'''There are several approximations that have been made that could result in the difference between theoretical and literature values.'''

=== '''An opportunity to speculate ...''' ===
* '''As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?'''
* '''In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?'''

=== Molecular Dynamics ===
GULP Molecular Dynamics calculations were used to calculate the lattice parameters of MgO at temperatures ranging from 0-1000 K. A supercell composed of 32 primitive cells and a SF of 1 was used for the calculation. The supercell was used to find an optimal input that was a compromise between accuracy and CPU usage.

A supercell is required for MD calculations as the simulation in dependent on a large system of atoms with randomized velocities and randomized population of states.
{|align="center"
|[[File:HEATefr114.PNG|thumb|250px|'''Fig. ''' The ]]
|}
Fig. blah show's how the volume of the supercell varies with time at 300K. Initially, the system is equilibrating, however after the first 200 fs the averaged volume starts tending towards an equilibrium.

{|align="center"
|[[File:Md volefr114.PNG|thumb|250px|'''Fig. ''' ]]
|[[File:Volcomp.PNG|thumb|250px|'''Fig. ''' ]]
|}
Figure blah shows there is a linear relationship between primitive unit cell volume and temperature. A thermal expansion coefficient of 3.0454*10<sup>-5</sup> was extracted from the data, this value is in the same magnitude with a percentage difference of 2.0772% from the literature value of 3.11*10<sup>-5</sup>. <nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki><div style="text-align: center;">
<math>
V(T)=0.0005689
</math>
<math>
\frac {dV}{dT}=0.0005689
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.0005689)
</math>

<math>
\alpha_V=3.0454*10^{-5}
</math>
</div>

=== Questions ===
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?

=== Comparison of QHA and MD ===

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}
QHA and MD methods were then applied to the calcium oxide model to determine plots of primitive unit cell volume against temperature to produce Fig. Blah. The QHA calculation method produced a thermal expansion coefficient of 5.0271*10<sup>-5</sup> K<sup>-1</sup>and the MD calculation method produced a thermal expansion coefficient of 4.9243*10<sup>-5</sup> K<sup>-1</sup>from 300 to 1000K in comparison to a literature value of 4.0791*10<sup>-5 </sup>K<sup>-1</sup>. [[File:Caofinahfdahfl.PNG|centre|thumb|Final ]]

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T11:05:01Z

Efr114: /* Density of States */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

The purpose of this experiment was to simulate the thermal expansion of MgO and to determine the thermal expansion coefficient for MgO using both the Quasi-Harmonic Approximation method and the Molecular Dynamics method.

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature.

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Fig. 4 a-e. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon.
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|'''Fig. 5a''' Phonon DOS with a 1x1 grid ]]
|[[File:5x5efr114.PNG|thumb|200px|'''Fig. 5b''' Phonon DOS with a 5x5 grid ]]
|[[File:10x10newefr114.PNG|thumb|200px|'''Fig. 5c''' Phonon DOS with a 10x10 grid]]
|[[File:20x20efr114.PNG|thumb|200px|'''Fig. 5d''' Phonon DOS with a 20x20 grid]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|'''Fig. 5e''' Phonon DOS with a 30x30 grid ]]
|[[File:35x35efr114.PNG|thumb|200px|'''Fig. 5f''' Phonon DOS with a 35x35 grid ]]
|[[File:50x50.PNG|thumb|200px|'''Fig. 5g''' Phonon DOS with a 50x50 grid]]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation at 300K. The shinking factor was varied between 1 and 30 in steps of 5 and the optimum grid size was found to be between shrinking factor 15 and 20. The shrinking factor was then varied between 15 and 20 in steps of 1 to determine the optimal grid size. Table 2 shows the values of HFE and the difference from the most accurate value of HFE (<math> \Delta </math> HFE) for a range of shrinking factors.
{| class="wikitable"
!Shrinking factor
!HFE / eV
!<math> \Delta </math> HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|-
| colspan="3" |Table 2: the values of HFE and the difference from the
most accurate value of HFE for a range of

shrinking factors
|}

The accuracy of the HFE increases with increasing shrinking factor.
To obtain an accuracy of 1 meV, a grid of 2x2x2 should be used. For an accuracy of 0.5 meV, a grid of 2x2x2 should be used and to obtain an accuracy of 0.1 meV, a grid of 3x3x3 should be used.
The data in this simulation suggests that an 18x18x18 grid is the optimum grid size as it is the lowest grid size that still gives the energy to an accuracy of 6 decimal places.
Considering the qualitative visual inspection method as well as the energy calculations, an 18x18x18 grid was chosen for further calculations.

=== Grid Sizes for alternative solids. ===
Crystals vary in both structure (e.g. FCC, BCC) and lattice parameters '''(a,b,c and math \alpha \beta \gamma)'''

'''Calcium oxide, CaO has the same structure as MgO'''
*# '''a similar oxide (eg: CaO) ?'''
*# '''a Zeolite (eg: Faujasite) ?'''
*# '''a metal (eg: lithium) ?'''

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The variance of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>component becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity.

{|align="center"
|[[File:Free energyefr114.PNG|thumb|250px|'''Fig. 6a '''A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations. ]]
|[[File:Mgotemplat.PNG|thumb|250px|'''Fig. 6b''' blah A plot showing the varience of lattice parameter with temperature ]]
|}

The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

{|align="center"
|[[File:Volumeefr114kj.PNG|thumb|250px|'''Fig. 7''' ]]
|}
Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.<div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

{|align="center"
|[[File:Alphagraph.PNG|thumb|250px|'''Fig. 8''' ]]
|}

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

a<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5} K^{-1}
</math>
</div>

This calculated value is in agreeable magnitude with literature, with a 3.5498% percentage difference to the literature value, which states a thermal expansion coefficient of 3.11*10^-5.<nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki>

'''There are several approximations that have been made that could result in the difference between theoretical and literature values.'''

=== '''An opportunity to speculate ...''' ===
* '''As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?'''
* '''In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?'''

=== Molecular Dynamics ===
GULP Molecular Dynamics calculations were used to calculate the lattice parameters of MgO at temperatures ranging from 0-1000 K. A supercell composed of 32 primitive cells and a SF of 1 was used for the calculation. The supercell was used to find an optimal input that was a compromise between accuracy and CPU usage.

A supercell is required for MD calculations as the simulation in dependent on a large system of atoms with randomized velocities and randomized population of states.

Fig. blah show's how the volume of the supercell varies with time at 300K. Initially, the system is equilibrating, however after the first 200 fs the averaged volume starts tending towards an equilibrium.

{|align="center"
|[[File:Md volefr114.PNG|thumb|250px|'''Fig. ''' ]]
|[[File:Volcomp.PNG|thumb|250px|'''Fig. ''' ]]
|}

{|align="center"
|[[File:HEATefr114.PNG|thumb|250px|'''Fig. ''' The ]]
|}
Figure blah shows there is a linear relationship between primitive unit cell volume and temperature. A thermal expansion coefficient of 3.0454*10<sup>-5</sup> was extracted from the data, this value is in the same magnitude with a percentage difference of 2.0772% from the literature value of 3.11*10<sup>-5</sup>. <nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki><div style="text-align: center;">
<math>
V(T)=0.0005689
</math>

<math>
\frac {dV}{dT}=0.0005689
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.0005689)
</math>

<math>
\alpha_V=3.0454*10^{-5}
</math>
</div>

=== Questions ===
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?

=== Comparison of QHA and MD ===

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}
QHA and MD methods were then applied to the calcium oxide model to determine plots of primitive unit cell volume against temperature to produce Fig. Blah. The QHA calculation method produced a thermal expansion coefficient of 5.0271*10<sup>-5</sup> K<sup>-1</sup>and the MD calculation method produced a thermal expansion coefficient of 4.9243*10<sup>-5</sup> K<sup>-1</sup>from 300 to 1000K in comparison to a literature value of 4.0791*10<sup>-5 </sup>K<sup>-1</sup>. [[File:Caofinahfdahfl.PNG|centre|thumb|Final ]]

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T10:59:35Z

Efr114: /* Quasi-Harmonic Approximation Method */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

The purpose of this experiment was to simulate the thermal expansion of MgO and to determine the thermal expansion coefficient for MgO using both the Quasi-Harmonic Approximation method and the Molecular Dynamics method.

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature.

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Fig. 4 a-e. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon.
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation at 300K. The shinking factor was varied between 1 and 30 in steps of 5 and the optimum grid size was found to be between shrinking factor 15 and 20. The shrinking factor was then varied between 15 and 20 in steps of 1 to determine the optimal grid size. Table 2 shows the values of HFE and the difference from the most accurate value of HFE (<math> \Delta </math> HFE) for a range of shrinking factors.
{| class="wikitable"
!Shrinking factor
!HFE / eV
!<math> \Delta </math> HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|-
| colspan="3" |Table 2: the values of HFE and the difference from the
most accurate value of HFE for a range of

shrinking factors
|}

The accuracy of the HFE increases with increasing shrinking factor.
To obtain an accuracy of 1 meV, a grid of 2x2x2 should be used. For an accuracy of 0.5 meV, a grid of 2x2x2 should be used and to obtain an accuracy of 0.1 meV, a grid of 3x3x3 should be used.
The data in this simulation suggests that an 18x18x18 grid is the optimum grid size as it is the lowest grid size that still gives the energy to an accuracy of 6 decimal places.
Considering the qualitative visual inspection method as well as the energy calculations, an 18x18x18 grid was chosen for further calculations.

=== Grid Sizes for alternative solids. ===
Crystals vary in both structure (e.g. FCC, BCC) and lattice parameters '''(a,b,c and math \alpha \beta \gamma)'''

Calcium oxide, CaO has the same structure as MgO
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The variance of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>component becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity.

{|align="center"
|[[File:Free energyefr114.PNG|thumb|250px|'''Fig. '''A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations. ]]
|[[File:Mgotemplat.PNG|thumb|250px|Fig. blah A plot showing the varience of lattice parameter with temperature ]]
|}

The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

{|align="center"
|[[File:Volumeefr114kj.PNG|thumb|250px|'''Fig. ''' ]]
|}
Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.<div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

{|align="center"
|[[File:Alphagraph.PNG|thumb|250px|'''Fig. ''' ]]
|}

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

a<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5} K^{-1}
</math>
</div>

This calculated value is in agreeable magnitude with literature, with a 3.5498% percentage difference to the literature value, which states a thermal expansion coefficient of 3.11*10^-5.<nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki>

There are several approximations that have been made that could result in the difference between theoretical and literature values.

=== An opportunity to speculate ... ===
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===
GULP Molecular Dynamics calculations were used to calculate the lattice parameters of MgO at temperatures ranging from 0-1000 K. A supercell composed of 32 primitive cells and a SF of 1 was used for the calculation. The supercell was used to find an optimal input that was a compromise between accuracy and CPU usage.

A supercell is required for MD calculations as the simulation in dependent on a large system of atoms with randomized velocities and randomized population of states.

Fig. blah show's how the volume of the supercell varies with time at 300K. Initially, the system is equilibrating, however after the first 200 fs the averaged volume starts tending towards an equilibrium.

{|align="center"
|[[File:Md volefr114.PNG|thumb|250px|'''Fig. ''' ]]
|[[File:Volcomp.PNG|thumb|250px|'''Fig. ''' ]]
|}

{|align="center"
|[[File:HEATefr114.PNG|thumb|250px|'''Fig. ''' The ]]
|}
Figure blah shows there is a linear relationship between primitive unit cell volume and temperature. A thermal expansion coefficient of 3.0454*10<sup>-5</sup> was extracted from the data, this value is in the same magnitude with a percentage difference of 2.0772% from the literature value of 3.11*10<sup>-5</sup>. <nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki><div style="text-align: center;">
<math>
V(T)=0.0005689
</math>

<math>
\frac {dV}{dT}=0.0005689
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.0005689)
</math>

<math>
\alpha_V=3.0454*10^{-5}
</math>
</div>

=== Questions ===
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?

=== Comparison of QHA and MD ===

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}
QHA and MD methods were then applied to the calcium oxide model to determine plots of primitive unit cell volume against temperature to produce Fig. Blah. The QHA calculation method produced a thermal expansion coefficient of 5.0271*10<sup>-5</sup> K<sup>-1</sup>and the MD calculation method produced a thermal expansion coefficient of 4.9243*10<sup>-5</sup> K<sup>-1</sup>from 300 to 1000K in comparison to a literature value of 4.0791*10<sup>-5 </sup>K<sup>-1</sup>. [[File:Caofinahfdahfl.PNG|centre|thumb|Final ]]

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T10:58:47Z

Efr114: /* Quasi-Harmonic Approximation */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

The purpose of this experiment was to simulate the thermal expansion of MgO and to determine the thermal expansion coefficient for MgO using both the Quasi-Harmonic Approximation method and the Molecular Dynamics method.

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature.

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Fig. 4 a-e. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon.
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation at 300K. The shinking factor was varied between 1 and 30 in steps of 5 and the optimum grid size was found to be between shrinking factor 15 and 20. The shrinking factor was then varied between 15 and 20 in steps of 1 to determine the optimal grid size. Table 2 shows the values of HFE and the difference from the most accurate value of HFE (<math> \Delta </math> HFE) for a range of shrinking factors.
{| class="wikitable"
!Shrinking factor
!HFE / eV
!<math> \Delta </math> HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|-
| colspan="3" |Table 2: the values of HFE and the difference from the
most accurate value of HFE for a range of

shrinking factors
|}

The accuracy of the HFE increases with increasing shrinking factor.
To obtain an accuracy of 1 meV, a grid of 2x2x2 should be used. For an accuracy of 0.5 meV, a grid of 2x2x2 should be used and to obtain an accuracy of 0.1 meV, a grid of 3x3x3 should be used.
The data in this simulation suggests that an 18x18x18 grid is the optimum grid size as it is the lowest grid size that still gives the energy to an accuracy of 6 decimal places.
Considering the qualitative visual inspection method as well as the energy calculations, an 18x18x18 grid was chosen for further calculations.

=== Grid Sizes for alternative solids. ===
Crystals vary in both structure (e.g. FCC, BCC) and lattice parameters '''(a,b,c and math \alpha \beta \gamma)'''

Calcium oxide, CaO has the same structure as MgO
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The variance of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>component becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity.

{|align="center"
|[[File:Free energyefr114.PNG|thumb|250px|'''Fig. '''A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations. ]]
|[[File:Mgotemplat.PNG|thumb|250px|Fig. blah A plot showing the varience of lattice parameter with temperature ]]
|}

The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

{|align="center"
|[[File:Volumeefr114kj.PNG|thumb|250px|'''Fig. ''' ]]
|}
Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.<div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

{|align="center"
|[[File:Alphagraph.PNG|thumb|250px|'''Fig. ''' ]]
|}

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

a<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5}
</math>
</div>

This calculated value is in agreeable magnitude with literature, with a 3.5498% percentage difference to the literature value, which states a thermal expansion coefficient of 3.11*10^-5.<nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki>

There are several approximations that have been made that could result in the difference between theoretical and literature values.

=== An opportunity to speculate ... ===
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===
GULP Molecular Dynamics calculations were used to calculate the lattice parameters of MgO at temperatures ranging from 0-1000 K. A supercell composed of 32 primitive cells and a SF of 1 was used for the calculation. The supercell was used to find an optimal input that was a compromise between accuracy and CPU usage.

A supercell is required for MD calculations as the simulation in dependent on a large system of atoms with randomized velocities and randomized population of states.

Fig. blah show's how the volume of the supercell varies with time at 300K. Initially, the system is equilibrating, however after the first 200 fs the averaged volume starts tending towards an equilibrium.

{|align="center"
|[[File:Md volefr114.PNG|thumb|250px|'''Fig. ''' ]]
|[[File:Volcomp.PNG|thumb|250px|'''Fig. ''' ]]
|}

{|align="center"
|[[File:HEATefr114.PNG|thumb|250px|'''Fig. ''' The ]]
|}
Figure blah shows there is a linear relationship between primitive unit cell volume and temperature. A thermal expansion coefficient of 3.0454*10<sup>-5</sup> was extracted from the data, this value is in the same magnitude with a percentage difference of 2.0772% from the literature value of 3.11*10<sup>-5</sup>. <nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki><div style="text-align: center;">
<math>
V(T)=0.0005689
</math>

<math>
\frac {dV}{dT}=0.0005689
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.0005689)
</math>

<math>
\alpha_V=3.0454*10^{-5}
</math>
</div>

=== Questions ===
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?

=== Comparison of QHA and MD ===

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}
QHA and MD methods were then applied to the calcium oxide model to determine plots of primitive unit cell volume against temperature to produce Fig. Blah. The QHA calculation method produced a thermal expansion coefficient of 5.0271*10<sup>-5</sup> K<sup>-1</sup>and the MD calculation method produced a thermal expansion coefficient of 4.9243*10<sup>-5</sup> K<sup>-1</sup>from 300 to 1000K in comparison to a literature value of 4.0791*10<sup>-5 </sup>K<sup>-1</sup>. [[File:Caofinahfdahfl.PNG|centre|thumb|Final ]]

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T10:54:51Z

Efr114: /* Helmholtz free energy (HFE) */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

The purpose of this experiment was to simulate the thermal expansion of MgO and to determine the thermal expansion coefficient for MgO using both the Quasi-Harmonic Approximation method and the Molecular Dynamics method.

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Fig. 4 a-e. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon.
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation at 300K. The shinking factor was varied between 1 and 30 in steps of 5 and the optimum grid size was found to be between shrinking factor 15 and 20. The shrinking factor was then varied between 15 and 20 in steps of 1 to determine the optimal grid size. Table 2 shows the values of HFE and the difference from the most accurate value of HFE (<math> \Delta </math> HFE) for a range of shrinking factors.
{| class="wikitable"
!Shrinking factor
!HFE / eV
!<math> \Delta </math> HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|-
| colspan="3" |Table 2: the values of HFE and the difference from the
most accurate value of HFE (delta HFE) for a range of

shrinking factors
|}

The accuracy of the HFE increases with increasing shrinking factor.
To obtain an accuracy of 1 meV, a grid of 2x2x2 should be used. For an accuracy of 0.5 meV, a grid of 2x2x2 should be used and to obtain an accuracy of 0.1 meV, a grid of 3x3x3 should be used.
The data in this simulation suggests that an 18x18x18 grid is the optimum grid size as it is the lowest grid size that still gives the energy to an accuracy of 6 decimal places.
Considering the qualitative visual inspection method as well as the energy calculations, an 18x18x18 grid was chosen for further calculations.

=== Grid Sizes for alternative solids. ===
Crystals vary in both structure (e.g. FCC, BCC) and lattice parameters (a,b,c and math \alpha \beta \gamma)

Calcium oxide, CaO has the same structure as MgO
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

{|align="center"
|[[File:Free energyefr114.PNG|thumb|250px|'''Fig. '''A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations. ]]
|[[File:Mgotemplat.PNG|thumb|250px|Fig. blah A plot showing the varience of lattice parameter with temperature ]]
|}

The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

{|align="center"
|[[File:Volumeefr114kj.PNG|thumb|250px|'''Fig. ''' ]]
|}
Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.<div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

{|align="center"
|[[File:Alphagraph.PNG|thumb|250px|'''Fig. ''' ]]
|}

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

a<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5}
</math>
</div>

This calculated value is in agreeable magnitude with literature, with a 3.5498% percentage difference to the literature value, which states a thermal expansion coefficient of 3.11*10^-5.<nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki>

There are several approximations that have been made that could result in the difference between theoretical and literature values.

=== An opportunity to speculate ... ===
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===
GULP Molecular Dynamics calculations were used to calculate the lattice parameters of MgO at temperatures ranging from 0-1000 K. A supercell composed of 32 primitive cells and a SF of 1 was used for the calculation. The supercell was used to find an optimal input that was a compromise between accuracy and CPU usage.

A supercell is required for MD calculations as the simulation in dependent on a large system of atoms with randomized velocities and randomized population of states.

Fig. blah show's how the volume of the supercell varies with time at 300K. Initially, the system is equilibrating, however after the first 200 fs the averaged volume starts tending towards an equilibrium.

{|align="center"
|[[File:Md volefr114.PNG|thumb|250px|'''Fig. ''' ]]
|[[File:Volcomp.PNG|thumb|250px|'''Fig. ''' ]]
|}

{|align="center"
|[[File:HEATefr114.PNG|thumb|250px|'''Fig. ''' The ]]
|}
Figure blah shows there is a linear relationship between primitive unit cell volume and temperature. A thermal expansion coefficient of 3.0454*10<sup>-5</sup> was extracted from the data, this value is in the same magnitude with a percentage difference of 2.0772% from the literature value of 3.11*10<sup>-5</sup>. <nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki><div style="text-align: center;">
<math>
V(T)=0.0005689
</math>

<math>
\frac {dV}{dT}=0.0005689
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.0005689)
</math>

<math>
\alpha_V=3.0454*10^{-5}
</math>
</div>

=== Questions ===
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?

=== Comparison of QHA and MD ===

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}
QHA and MD methods were then applied to the calcium oxide model to determine plots of primitive unit cell volume against temperature to produce Fig. Blah. The QHA calculation method produced a thermal expansion coefficient of 5.0271*10<sup>-5</sup> K<sup>-1</sup>and the MD calculation method produced a thermal expansion coefficient of 4.9243*10<sup>-5</sup> K<sup>-1</sup>from 300 to 1000K in comparison to a literature value of 4.0791*10<sup>-5 </sup>K<sup>-1</sup>. [[File:Caofinahfdahfl.PNG|centre|thumb|Final ]]

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T10:54:03Z

Efr114: /* Introduction */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

The purpose of this experiment was to simulate the thermal expansion of MgO and to determine the thermal expansion coefficient for MgO using both the Quasi-Harmonic Approximation method and the Molecular Dynamics method.

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Fig. 4 a-e. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon.
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation at 300K. The shinking factor was varied between 1 and 30 in steps of 5 and the optimum grid size was found to be between shrinking factor 15 and 20. The shrinking factor was then varied between 15 and 20 in steps of 1 to determine the optimal grid size. Table 2 shows the values of HFE and the difference from the most accurate value of HFE (delta HFE) for a range of shrinking factors.
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|-
| colspan="3" |Table 2: the values of HFE and the difference from the
most accurate value of HFE (delta HFE) for a range of

shrinking factors
|}

The accuracy of the HFE increases with increasing shrinking factor.
To obtain an accuracy of 1 meV, a grid of 2x2x2 should be used. For an accuracy of 0.5 meV, a grid of 2x2x2 should be used and to obtain an accuracy of 0.1 meV, a grid of 3x3x3 should be used.
The data in this simulation suggests that an 18x18x18 grid is the optimum grid size as it is the lowest grid size that still gives the energy to an accuracy of 6 decimal places.
Considering the qualitative visual inspection method as well as the energy calculations, an 18x18x18 grid was chosen for further calculations.

=== Grid Sizes for alternative solids. ===
Crystals vary in both structure (e.g. FCC, BCC) and lattice parameters (a,b,c and math \alpha \beta \gamma)

Calcium oxide, CaO has the same structure as MgO
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

{|align="center"
|[[File:Free energyefr114.PNG|thumb|250px|'''Fig. '''A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations. ]]
|[[File:Mgotemplat.PNG|thumb|250px|Fig. blah A plot showing the varience of lattice parameter with temperature ]]
|}

The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

{|align="center"
|[[File:Volumeefr114kj.PNG|thumb|250px|'''Fig. ''' ]]
|}
Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.<div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

{|align="center"
|[[File:Alphagraph.PNG|thumb|250px|'''Fig. ''' ]]
|}

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

a<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5}
</math>
</div>

This calculated value is in agreeable magnitude with literature, with a 3.5498% percentage difference to the literature value, which states a thermal expansion coefficient of 3.11*10^-5.<nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki>

There are several approximations that have been made that could result in the difference between theoretical and literature values.

=== An opportunity to speculate ... ===
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===
GULP Molecular Dynamics calculations were used to calculate the lattice parameters of MgO at temperatures ranging from 0-1000 K. A supercell composed of 32 primitive cells and a SF of 1 was used for the calculation. The supercell was used to find an optimal input that was a compromise between accuracy and CPU usage.

A supercell is required for MD calculations as the simulation in dependent on a large system of atoms with randomized velocities and randomized population of states.

Fig. blah show's how the volume of the supercell varies with time at 300K. Initially, the system is equilibrating, however after the first 200 fs the averaged volume starts tending towards an equilibrium.

{|align="center"
|[[File:Md volefr114.PNG|thumb|250px|'''Fig. ''' ]]
|[[File:Volcomp.PNG|thumb|250px|'''Fig. ''' ]]
|}

{|align="center"
|[[File:HEATefr114.PNG|thumb|250px|'''Fig. ''' The ]]
|}
Figure blah shows there is a linear relationship between primitive unit cell volume and temperature. A thermal expansion coefficient of 3.0454*10<sup>-5</sup> was extracted from the data, this value is in the same magnitude with a percentage difference of 2.0772% from the literature value of 3.11*10<sup>-5</sup>. <nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki><div style="text-align: center;">
<math>
V(T)=0.0005689
</math>

<math>
\frac {dV}{dT}=0.0005689
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.0005689)
</math>

<math>
\alpha_V=3.0454*10^{-5}
</math>
</div>

=== Questions ===
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?

=== Comparison of QHA and MD ===

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}
QHA and MD methods were then applied to the calcium oxide model to determine plots of primitive unit cell volume against temperature to produce Fig. Blah. The QHA calculation method produced a thermal expansion coefficient of 5.0271*10<sup>-5</sup> K<sup>-1</sup>and the MD calculation method produced a thermal expansion coefficient of 4.9243*10<sup>-5</sup> K<sup>-1</sup>from 300 to 1000K in comparison to a literature value of 4.0791*10<sup>-5 </sup>K<sup>-1</sup>. [[File:Caofinahfdahfl.PNG|centre|thumb|Final ]]

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T10:46:27Z

Efr114: /* Extension: CaO: calculations and comparision */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Fig. 4 a-e. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon.
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation at 300K. The shinking factor was varied between 1 and 30 in steps of 5 and the optimum grid size was found to be between shrinking factor 15 and 20. The shrinking factor was then varied between 15 and 20 in steps of 1 to determine the optimal grid size. Table 2 shows the values of HFE and the difference from the most accurate value of HFE (delta HFE) for a range of shrinking factors.
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|-
| colspan="3" |Table 2: the values of HFE and the difference from the
most accurate value of HFE (delta HFE) for a range of

shrinking factors
|}

The accuracy of the HFE increases with increasing shrinking factor.
To obtain an accuracy of 1 meV, a grid of 2x2x2 should be used. For an accuracy of 0.5 meV, a grid of 2x2x2 should be used and to obtain an accuracy of 0.1 meV, a grid of 3x3x3 should be used.
The data in this simulation suggests that an 18x18x18 grid is the optimum grid size as it is the lowest grid size that still gives the energy to an accuracy of 6 decimal places.
Considering the qualitative visual inspection method as well as the energy calculations, an 18x18x18 grid was chosen for further calculations.

=== Grid Sizes for alternative solids. ===
Crystals vary in both structure (e.g. FCC, BCC) and lattice parameters (a,b,c and math \alpha \beta \gamma)

Calcium oxide, CaO has the same structure as MgO
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

{|align="center"
|[[File:Free energyefr114.PNG|thumb|250px|'''Fig. '''A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations. ]]
|[[File:Mgotemplat.PNG|thumb|250px|Fig. blah A plot showing the varience of lattice parameter with temperature ]]
|}

The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

{|align="center"
|[[File:Volumeefr114kj.PNG|thumb|250px|'''Fig. ''' ]]
|}
Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.<div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

{|align="center"
|[[File:Alphagraph.PNG|thumb|250px|'''Fig. ''' ]]
|}

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

a<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5}
</math>
</div>

This calculated value is in agreeable magnitude with literature, with a 3.5498% percentage difference to the literature value, which states a thermal expansion coefficient of 3.11*10^-5.<nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki>

There are several approximations that have been made that could result in the difference between theoretical and literature values.

=== An opportunity to speculate ... ===
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===
GULP Molecular Dynamics calculations were used to calculate the lattice parameters of MgO at temperatures ranging from 0-1000 K. A supercell composed of 32 primitive cells and a SF of 1 was used for the calculation. The supercell was used to find an optimal input that was a compromise between accuracy and CPU usage.

A supercell is required for MD calculations as the simulation in dependent on a large system of atoms with randomized velocities and randomized population of states.

Fig. blah show's how the volume of the supercell varies with time at 300K. Initially, the system is equilibrating, however after the first 200 fs the averaged volume starts tending towards an equilibrium.

{|align="center"
|[[File:Md volefr114.PNG|thumb|250px|'''Fig. ''' ]]
|[[File:Volcomp.PNG|thumb|250px|'''Fig. ''' ]]
|}

{|align="center"
|[[File:HEATefr114.PNG|thumb|250px|'''Fig. ''' The ]]
|}
Figure blah shows there is a linear relationship between primitive unit cell volume and temperature. A thermal expansion coefficient of 3.0454*10<sup>-5</sup> was extracted from the data, this value is in the same magnitude with a percentage difference of 2.0772% from the literature value of 3.11*10<sup>-5</sup>. <nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki><div style="text-align: center;">
<math>
V(T)=0.0005689
</math>

<math>
\frac {dV}{dT}=0.0005689
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.0005689)
</math>

<math>
\alpha_V=3.0454*10^{-5}
</math>
</div>

=== Questions ===
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?

=== Comparison of QHA and MD ===

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}
QHA and MD methods were then applied to the calcium oxide model to determine plots of primitive unit cell volume against temperature to produce Fig. Blah. The QHA calculation method produced a thermal expansion coefficient of 5.0271*10<sup>-5</sup> K<sup>-1</sup>and the MD calculation method produced a thermal expansion coefficient of 4.9243*10<sup>-5</sup> K<sup>-1</sup>from 300 to 1000K in comparison to a literature value of 4.0791*10<sup>-5 </sup>K<sup>-1</sup>. [[File:Caofinahfdahfl.PNG|centre|thumb|Final ]]

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T10:37:03Z

Efr114: /* An opportunity to speculate */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Fig. 4 a-e. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon.
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation at 300K. The shinking factor was varied between 1 and 30 in steps of 5 and the optimum grid size was found to be between shrinking factor 15 and 20. The shrinking factor was then varied between 15 and 20 in steps of 1 to determine the optimal grid size. Table 2 shows the values of HFE and the difference from the most accurate value of HFE (delta HFE) for a range of shrinking factors.
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|-
| colspan="3" |Table 2: the values of HFE and the difference from the
most accurate value of HFE (delta HFE) for a range of

shrinking factors
|}

The accuracy of the HFE increases with increasing shrinking factor.
To obtain an accuracy of 1 meV, a grid of 2x2x2 should be used. For an accuracy of 0.5 meV, a grid of 2x2x2 should be used and to obtain an accuracy of 0.1 meV, a grid of 3x3x3 should be used.
The data in this simulation suggests that an 18x18x18 grid is the optimum grid size as it is the lowest grid size that still gives the energy to an accuracy of 6 decimal places.
Considering the qualitative visual inspection method as well as the energy calculations, an 18x18x18 grid was chosen for further calculations.

=== Grid Sizes for alternative solids. ===
Crystals vary in both structure (e.g. FCC, BCC) and lattice parameters (a,b,c and math \alpha \beta \gamma)

Calcium oxide, CaO has the same structure as MgO
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

{|align="center"
|[[File:Free energyefr114.PNG|thumb|250px|'''Fig. '''A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations. ]]
|[[File:Mgotemplat.PNG|thumb|250px|Fig. blah A plot showing the varience of lattice parameter with temperature ]]
|}

The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

{|align="center"
|[[File:Volumeefr114kj.PNG|thumb|250px|'''Fig. ''' ]]
|}
Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.<div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

{|align="center"
|[[File:Alphagraph.PNG|thumb|250px|'''Fig. ''' ]]
|}

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

a<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5}
</math>
</div>

This calculated value is in agreeable magnitude with literature, with a 3.5498% percentage difference to the literature value, which states a thermal expansion coefficient of 3.11*10^-5.<nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki>

There are several approximations that have been made that could result in the difference between theoretical and literature values.

=== An opportunity to speculate ... ===
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===
GULP Molecular Dynamics calculations were used to calculate the lattice parameters of MgO at temperatures ranging from 0-1000 K. A supercell composed of 32 primitive cells and a SF of 1 was used for the calculation. The supercell was used to find an optimal input that was a compromise between accuracy and CPU usage.

A supercell is required for MD calculations as the simulation in dependent on a large system of atoms with randomized velocities and randomized population of states.

Fig. blah show's how the volume of the supercell varies with time at 300K. Initially, the system is equilibrating, however after the first 200 fs the averaged volume starts tending towards an equilibrium.

{|align="center"
|[[File:Md volefr114.PNG|thumb|250px|'''Fig. ''' ]]
|[[File:Volcomp.PNG|thumb|250px|'''Fig. ''' ]]
|}

{|align="center"
|[[File:HEATefr114.PNG|thumb|250px|'''Fig. ''' The ]]
|}
Figure blah shows there is a linear relationship between primitive unit cell volume and temperature. A thermal expansion coefficient of 3.0454*10<sup>-5</sup> was extracted from the data, this value is in the same magnitude with a percentage difference of 2.0772% from the literature value of 3.11*10<sup>-5</sup>. <nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki><div style="text-align: center;">
<math>
V(T)=0.0005689
</math>

<math>
\frac {dV}{dT}=0.0005689
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.0005689)
</math>

<math>
\alpha_V=3.0454*10^{-5}
</math>
</div>

=== Questions ===
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?

=== Comparison of QHA and MD ===

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}
QHA and MD methods were then applied to the calcium oxide model to determine plots of primitive unit cell volume against temperature to produce Fig. Blah. The QHA calculation method produced a thermal expansion coefficient of 4.9243*10<sup>-5</sup> K<sup>-1</sup>and the MD calculation method produced a thermal expansion coefficient of 5.0271*10<sup>-5</sup> K<sup>-1</sup>from 300 to 1000K in comparison to a literature value of 4.0791*10<sup>-5 </sup>K<sup>-1</sup>
[[File:Caofinahfdahfl.PNG|centre|thumb|Final ]]

== Conclusion ==

== References ==

File:Caofinahfdahfl.PNG

2017-03-24T10:25:01Z

Efr114:

Rep:Efr114MgOwiki

2017-03-24T10:11:03Z

Efr114: /* Helmholtz free energy (HFE) */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Fig. 4 a-e. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon.
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
An opportunity to speculate
*Would this optimal grid size for MgO be appropriate for a calculation on;
*#a similar oxide (eg: CaO) ?
*#a Zeolite (eg: Faujasite) ?
*#a metal (eg: lithium) ?
An opportunity to speculate

=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation at 300K. The shinking factor was varied between 1 and 30 in steps of 5 and the optimum grid size was found to be between shrinking factor 15 and 20. The shrinking factor was then varied between 15 and 20 in steps of 1 to determine the optimal grid size. Table 2 shows the values of HFE and the difference from the most accurate value of HFE (delta HFE) for a range of shrinking factors.
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|-
| colspan="3" |Table 2: the values of HFE and the difference from the
most accurate value of HFE (delta HFE) for a range of

shrinking factors
|}

The accuracy of the HFE increases with increasing shrinking factor.
To obtain an accuracy of 1 meV, a grid of 2x2x2 should be used. For an accuracy of 0.5 meV, a grid of 2x2x2 should be used and to obtain an accuracy of 0.1 meV, a grid of 3x3x3 should be used.
The data in this simulation suggests that an 18x18x18 grid is the optimum grid size as it is the lowest grid size that still gives the energy to an accuracy of 6 decimal places.
Considering the qualitative visual inspection method as well as the energy calculations, an 18x18x18 grid was chosen for further calculations.

=== An opportunity to speculate ===
* Would this optimal grid size for MgO be appropriate for a calculation on;
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

{|align="center"
|[[File:Free energyefr114.PNG|thumb|250px|'''Fig. '''A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations. ]]
|[[File:Mgotemplat.PNG|thumb|250px|Fig. blah A plot showing the varience of lattice parameter with temperature ]]
|}

The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

{|align="center"
|[[File:Volumeefr114kj.PNG|thumb|250px|'''Fig. ''' ]]
|}
Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.<div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

{|align="center"
|[[File:Alphagraph.PNG|thumb|250px|'''Fig. ''' ]]
|}

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

a<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5}
</math>
</div>

This calculated value is in agreeable magnitude with literature, with a 3.5498% percentage difference to the literature value, which states a thermal expansion coefficient of 3.11*10^-5.<nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki>

There are several approximations that have been made that could result in the difference between theoretical and literature values.

=== An opportunity to speculate ... ===
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===
GULP Molecular Dynamics calculations were used to calculate the lattice parameters of MgO at temperatures ranging from 0-1000 K. A supercell composed of 32 primitive cells and a SF of 1 was used for the calculation. The supercell was used to find an optimal input that was a compromise between accuracy and CPU usage.

A supercell is required for MD calculations as the simulation in dependent on a large system of atoms with randomized velocities and randomized population of states.

Fig. blah show's how the volume of the supercell varies with time at 300K. Initially, the system is equilibrating, however after the first 200 fs the averaged volume starts tending towards an equilibrium.

{|align="center"
|[[File:Md volefr114.PNG|thumb|250px|'''Fig. ''' ]]
|[[File:Volcomp.PNG|thumb|250px|'''Fig. ''' ]]
|}

{|align="center"
|[[File:HEATefr114.PNG|thumb|250px|'''Fig. ''' The ]]
|}
Figure blah shows there is a linear relationship between primitive unit cell volume and temperature. A thermal expansion coefficient of 3.0454*10<sup>-5</sup> was extracted from the data, this value is in the same magnitude with a percentage difference of 2.0772% from the literature value of 3.11*10<sup>-5</sup>. <nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki><div style="text-align: center;">
<math>
V(T)=0.0005689
</math>

<math>
\frac {dV}{dT}=0.0005689
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.0005689)
</math>

<math>
\alpha_V=3.0454*10^{-5}
</math>
</div>

=== Questions ===
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?

=== Comparison of QHA and MD ===

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T10:07:31Z

Efr114: /* Helmholtz free energy (HFE) */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Fig. 4 a-e. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon.
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
An opportunity to speculate
*Would this optimal grid size for MgO be appropriate for a calculation on;
*#a similar oxide (eg: CaO) ?
*#a Zeolite (eg: Faujasite) ?
*#a metal (eg: lithium) ?
An opportunity to speculate

=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation at 300K. The shinking factor was varied between 1 and 30 in steps of 5 and the optimum grid size was found to be between shrinking factor 15 and 20. The shrinking factor was then varied between 15 and 20 in steps of 1 to determine the optimal grid size.
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|}

The accuracy of the HFE increases with increasing shrinking factor.
To obtain an accuracy of 1 meV, a grid of 2x2x2 should be used. For an accuracy of 0.5 meV, a grid of 2x2x2 should be used and to obtain an accuracy of 0.1 meV, a grid of 3x3x3 should be used.
The data in this simulation suggests that an 18x18x18 grid is the optimum grid size as it is the lowest grid size that still gives the energy to an accuracy of 6 decimal places.
Considering the qualitative visual inspection method as well as the energy calculations, an 18x18x18 grid was chosen for further calculations.

=== An opportunity to speculate ===
* Would this optimal grid size for MgO be appropriate for a calculation on;
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

{|align="center"
|[[File:Free energyefr114.PNG|thumb|250px|'''Fig. '''A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations. ]]
|[[File:Mgotemplat.PNG|thumb|250px|Fig. blah A plot showing the varience of lattice parameter with temperature ]]
|}

The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

{|align="center"
|[[File:Volumeefr114kj.PNG|thumb|250px|'''Fig. ''' ]]
|}
Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.<div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

{|align="center"
|[[File:Alphagraph.PNG|thumb|250px|'''Fig. ''' ]]
|}

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

a<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5}
</math>
</div>

This calculated value is in agreeable magnitude with literature, with a 3.5498% percentage difference to the literature value, which states a thermal expansion coefficient of 3.11*10^-5.<nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki>

There are several approximations that have been made that could result in the difference between theoretical and literature values.

=== An opportunity to speculate ... ===
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===
GULP Molecular Dynamics calculations were used to calculate the lattice parameters of MgO at temperatures ranging from 0-1000 K. A supercell composed of 32 primitive cells and a SF of 1 was used for the calculation. The supercell was used to find an optimal input that was a compromise between accuracy and CPU usage.

A supercell is required for MD calculations as the simulation in dependent on a large system of atoms with randomized velocities and randomized population of states.

Fig. blah show's how the volume of the supercell varies with time at 300K. Initially, the system is equilibrating, however after the first 200 fs the averaged volume starts tending towards an equilibrium.

{|align="center"
|[[File:Md volefr114.PNG|thumb|250px|'''Fig. ''' ]]
|[[File:Volcomp.PNG|thumb|250px|'''Fig. ''' ]]
|}

{|align="center"
|[[File:HEATefr114.PNG|thumb|250px|'''Fig. ''' The ]]
|}
Figure blah shows there is a linear relationship between primitive unit cell volume and temperature. A thermal expansion coefficient of 3.0454*10<sup>-5</sup> was extracted from the data, this value is in the same magnitude with a percentage difference of 2.0772% from the literature value of 3.11*10<sup>-5</sup>. <nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki><div style="text-align: center;">
<math>
V(T)=0.0005689
</math>

<math>
\frac {dV}{dT}=0.0005689
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.0005689)
</math>

<math>
\alpha_V=3.0454*10^{-5}
</math>
</div>

=== Questions ===
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?

=== Comparison of QHA and MD ===

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T09:46:07Z

Efr114: /* Phonon dispersion */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Fig. 4 a-e. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon.
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
An opportunity to speculate
*Would this optimal grid size for MgO be appropriate for a calculation on;
*#a similar oxide (eg: CaO) ?
*#a Zeolite (eg: Faujasite) ?
*#a metal (eg: lithium) ?
An opportunity to speculate

=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation which d
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|}

The accuracy of the HFE increases with shrinking factor.

Questions
* How does the free energy vary with grid size ?
*
* Which grid size is appropriate for calculations accurate to 1 meV, 0.5 meV and 0.1 meV per cell ?

=== An opportunity to speculate ===
* Would this optimal grid size for MgO be appropriate for a calculation on;
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

{|align="center"
|[[File:Free energyefr114.PNG|thumb|250px|'''Fig. '''A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations. ]]
|[[File:Mgotemplat.PNG|thumb|250px|Fig. blah A plot showing the varience of lattice parameter with temperature ]]
|}

The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

{|align="center"
|[[File:Volumeefr114kj.PNG|thumb|250px|'''Fig. ''' ]]
|}
Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.<div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

{|align="center"
|[[File:Alphagraph.PNG|thumb|250px|'''Fig. ''' ]]
|}

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

a<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5}
</math>
</div>

This calculated value is in agreeable magnitude with literature, with a 3.5498% percentage difference to the literature value, which states a thermal expansion coefficient of 3.11*10^-5.<nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki>

There are several approximations that have been made that could result in the difference between theoretical and literature values.

=== An opportunity to speculate ... ===
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===
GULP Molecular Dynamics calculations were used to calculate the lattice parameters of MgO at temperatures ranging from 0-1000 K. A supercell composed of 32 primitive cells and a SF of 1 was used for the calculation. The supercell was used to find an optimal input that was a compromise between accuracy and CPU usage.

A supercell is required for MD calculations as the simulation in dependent on a large system of atoms with randomized velocities and randomized population of states.

Fig. blah show's how the volume of the supercell varies with time at 300K. Initially, the system is equilibrating, however after the first 200 fs the averaged volume starts tending towards an equilibrium.

{|align="center"
|[[File:Md volefr114.PNG|thumb|250px|'''Fig. ''' ]]
|[[File:Volcomp.PNG|thumb|250px|'''Fig. ''' ]]
|}

{|align="center"
|[[File:HEATefr114.PNG|thumb|250px|'''Fig. ''' The ]]
|}
Figure blah shows there is a linear relationship between primitive unit cell volume and temperature. A thermal expansion coefficient of 3.0454*10<sup>-5</sup> was extracted from the data, this value is in the same magnitude with a percentage difference of 2.0772% from the literature value of 3.11*10<sup>-5</sup>. <nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki><div style="text-align: center;">
<math>
V(T)=0.0005689
</math>

<math>
\frac {dV}{dT}=0.0005689
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.0005689)
</math>

<math>
\alpha_V=3.0454*10^{-5}
</math>
</div>

=== Questions ===
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?

=== Comparison of QHA and MD ===

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T09:35:04Z

Efr114: /* Molecular Dynamics */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Figure Blah a-f. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon. ref
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
An opportunity to speculate
*Would this optimal grid size for MgO be appropriate for a calculation on;
*#a similar oxide (eg: CaO) ?
*#a Zeolite (eg: Faujasite) ?
*#a metal (eg: lithium) ?
An opportunity to speculate

=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation which d
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|}

The accuracy of the HFE increases with shrinking factor.

Questions
* How does the free energy vary with grid size ?
*
* Which grid size is appropriate for calculations accurate to 1 meV, 0.5 meV and 0.1 meV per cell ?

=== An opportunity to speculate ===
* Would this optimal grid size for MgO be appropriate for a calculation on;
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

{|align="center"
|[[File:Free energyefr114.PNG|thumb|250px|'''Fig. '''A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations. ]]
|[[File:Mgotemplat.PNG|thumb|250px|Fig. blah A plot showing the varience of lattice parameter with temperature ]]
|}

The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

{|align="center"
|[[File:Volumeefr114kj.PNG|thumb|250px|'''Fig. ''' ]]
|}
Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.<div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

{|align="center"
|[[File:Alphagraph.PNG|thumb|250px|'''Fig. ''' ]]
|}

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

a<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5}
</math>
</div>

This calculated value is in agreeable magnitude with literature, with a 3.5498% percentage difference to the literature value, which states a thermal expansion coefficient of 3.11*10^-5.<nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki>

There are several approximations that have been made that could result in the difference between theoretical and literature values.

=== An opportunity to speculate ... ===
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===
GULP Molecular Dynamics calculations were used to calculate the lattice parameters of MgO at temperatures ranging from 0-1000 K. A supercell composed of 32 primitive cells and a SF of 1 was used for the calculation. The supercell was used to find an optimal input that was a compromise between accuracy and CPU usage.

A supercell is required for MD calculations as the simulation in dependent on a large system of atoms with randomized velocities and randomized population of states.

Fig. blah show's how the volume of the supercell varies with time at 300K. Initially, the system is equilibrating, however after the first 200 fs the averaged volume starts tending towards an equilibrium.

{|align="center"
|[[File:Md volefr114.PNG|thumb|250px|'''Fig. ''' ]]
|[[File:Volcomp.PNG|thumb|250px|'''Fig. ''' ]]
|}

{|align="center"
|[[File:HEATefr114.PNG|thumb|250px|'''Fig. ''' The ]]
|}
Figure blah shows there is a linear relationship between primitive unit cell volume and temperature. A thermal expansion coefficient of 3.0454*10<sup>-5</sup> was extracted from the data, this value is in the same magnitude with a percentage difference of 2.0772% from the literature value of 3.11*10<sup>-5</sup>. <nowiki><ref>O. Anderson and K. Zou, J. Phys. Chem. Ref., 2016, 19, 69.</ref></nowiki><div style="text-align: center;">
<math>
V(T)=0.0005689
</math>

<math>
\frac {dV}{dT}=0.0005689
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.0005689)
</math>

<math>
\alpha_V=3.0454*10^{-5}
</math>
</div>

=== Questions ===
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?

=== Comparison of QHA and MD ===

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T09:14:09Z

Efr114: /* Molecular Dynamics */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Figure Blah a-f. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon. ref
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
An opportunity to speculate
*Would this optimal grid size for MgO be appropriate for a calculation on;
*#a similar oxide (eg: CaO) ?
*#a Zeolite (eg: Faujasite) ?
*#a metal (eg: lithium) ?
An opportunity to speculate

=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation which d
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|}

The accuracy of the HFE increases with shrinking factor.

Questions
* How does the free energy vary with grid size ?
*
* Which grid size is appropriate for calculations accurate to 1 meV, 0.5 meV and 0.1 meV per cell ?

=== An opportunity to speculate ===
* Would this optimal grid size for MgO be appropriate for a calculation on;
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

{|align="center"
|[[File:Free energyefr114.PNG|thumb|250px|'''Fig. '''A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations. ]]
|[[File:Mgotemplat.PNG|thumb|250px|Fig. blah A plot showing the varience of lattice parameter with temperature ]]
|}

The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

{|align="center"
|[[File:Volumeefr114kj.PNG|thumb|250px|'''Fig. ''' ]]
|}
Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.<div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

{|align="center"
|[[File:Alphagraph.PNG|thumb|250px|'''Fig. ''' ]]
|}

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

a<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5}
</math>
</div>

This calculated value is in agreeable magnitude with literature, which states a thermal expansion coefficient of 4*10^-5 <nowiki><ref>https://www.hindawi.com/journals/jther/2014/123478<ref></nowiki>

There are several approximations that have been made that could result in the difference between theoretical and literature values.

=== An opportunity to speculate ... ===
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===
GULP Molecular Dynamics calculations were used to calculate the lattice parameters of MgO at temperatures ranging from 0-1000 K. A supercell composed of 32 primitive cells and a SF of 1 was used for the calculation. The supercell was used to find an optimal input that was a compromise between accuracy and CPU usage.

A supercell is required for MD calculations as the simulation in dependent on a large system of atoms with randomized velocities and randomized population of states

{|align="center"
|[[File:Md volefr114.PNG|thumb|250px|'''Fig. ''' ]]
|[[File:Volcomp.PNG|thumb|250px|'''Fig. ''' ]]
|}

{|align="center"
|[[File:HEATefr114.PNG|thumb|250px|'''Fig. ''' The ]]
|}

<div style="text-align: center;">
<math>
V(T)=0.0005689
</math>

<math>
\frac {dV}{dT}=0.0005689
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.0005689)
</math>

<math>
\alpha_V=3.0454*10^{-5}
</math>
</div>

=== Questions ===
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T08:55:56Z

Efr114: /* Quasi-Harmonic Approximation Method */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Figure Blah a-f. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon. ref
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
An opportunity to speculate
*Would this optimal grid size for MgO be appropriate for a calculation on;
*#a similar oxide (eg: CaO) ?
*#a Zeolite (eg: Faujasite) ?
*#a metal (eg: lithium) ?
An opportunity to speculate

=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation which d
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|}

The accuracy of the HFE increases with shrinking factor.

Questions
* How does the free energy vary with grid size ?
*
* Which grid size is appropriate for calculations accurate to 1 meV, 0.5 meV and 0.1 meV per cell ?

=== An opportunity to speculate ===
* Would this optimal grid size for MgO be appropriate for a calculation on;
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

{|align="center"
|[[File:Free energyefr114.PNG|thumb|250px|'''Fig. '''A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations. ]]
|[[File:Mgotemplat.PNG|thumb|250px|Fig. blah A plot showing the varience of lattice parameter with temperature ]]
|}

The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

{|align="center"
|[[File:Volumeefr114kj.PNG|thumb|250px|'''Fig. ''' ]]
|}
Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.<div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

{|align="center"
|[[File:Alphagraph.PNG|thumb|250px|'''Fig. ''' ]]
|}

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5}
</math>
</div>

This calculated value is in agreeable magnitude with literature, which states a thermal expansion coefficient of 4*10^-5 <nowiki><ref>https://www.hindawi.com/journals/jther/2014/123478<ref></nowiki>

There are several approximations that have been made that could result in the difference between theoretical and literature values.

=== An opportunity to speculate ... ===
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===
{|align="center"
|[[File:Md volefr114.PNG|thumb|250px|'''Fig. ''' ]]
|[[File:Volcomp.PNG|thumb|250px|'''Fig. ''' ]]
|}

{|align="center"
|[[File:HEATefr114.PNG|thumb|250px|'''Fig. ''' The ]]
|}

<div style="text-align: center;">
<math>
V(T)=0.0005689
</math>

<math>
\frac {dV}{dT}=0.0005689
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.0005689)
</math>

<math>
\alpha_V=3.0454*10^{-5}
</math>
</div>

=== Questions ===
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T08:54:57Z

Efr114: /* Quasi-Harmonic Approximation Method */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Figure Blah a-f. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon. ref
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
An opportunity to speculate
*Would this optimal grid size for MgO be appropriate for a calculation on;
*#a similar oxide (eg: CaO) ?
*#a Zeolite (eg: Faujasite) ?
*#a metal (eg: lithium) ?
An opportunity to speculate

=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation which d
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|}

The accuracy of the HFE increases with shrinking factor.

Questions
* How does the free energy vary with grid size ?
*
* Which grid size is appropriate for calculations accurate to 1 meV, 0.5 meV and 0.1 meV per cell ?

=== An opportunity to speculate ===
* Would this optimal grid size for MgO be appropriate for a calculation on;
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

{|align="center"
|[[File:Free energyefr114.PNG|thumb|250px|'''Fig. '''A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations. ]]
|[[File:Mgotemplat.PNG|thumb|250px|Fig. blah A plot showing the varience of lattice parameter with temperature ]]
|}

The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

{|align="center"
|[[File:Volumeefr114kj.PNG|thumb|250px|'''Fig. ''' ]]
|}
Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.<div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5}
</math>
</div>
[[File:Alphagraph.PNG|thumb|fig
]]

This calculated value is in agreeable magnitude with literature, which states a thermal expansion coefficient of 4*10^-5 <nowiki><ref>https://www.hindawi.com/journals/jther/2014/123478<ref></nowiki>

There are several approximations that have been made that could result in the difference between theoretical and literature values.

=== An opportunity to speculate ... ===
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===
{|align="center"
|[[File:Md volefr114.PNG|thumb|250px|'''Fig. ''' ]]
|[[File:Volcomp.PNG|thumb|250px|'''Fig. ''' ]]
|}

{|align="center"
|[[File:HEATefr114.PNG|thumb|250px|'''Fig. ''' The ]]
|}

<div style="text-align: center;">
<math>
V(T)=0.0005689
</math>

<math>
\frac {dV}{dT}=0.0005689
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.0005689)
</math>

<math>
\alpha_V=3.0454*10^{-5}
</math>
</div>

=== Questions ===
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T08:54:02Z

Efr114: /* Quasi-Harmonic Approximation Method */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Figure Blah a-f. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon. ref
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
An opportunity to speculate
*Would this optimal grid size for MgO be appropriate for a calculation on;
*#a similar oxide (eg: CaO) ?
*#a Zeolite (eg: Faujasite) ?
*#a metal (eg: lithium) ?
An opportunity to speculate

=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation which d
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|}

The accuracy of the HFE increases with shrinking factor.

Questions
* How does the free energy vary with grid size ?
*
* Which grid size is appropriate for calculations accurate to 1 meV, 0.5 meV and 0.1 meV per cell ?

=== An opportunity to speculate ===
* Would this optimal grid size for MgO be appropriate for a calculation on;
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

{|align="center"
|[[File:Free energyefr114.PNG|thumb|250px|'''Fig. '''A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations. ]]
|[[File:Mgotemplat.PNG|thumb|250px|Fig. blah A plot showing the varience of lattice parameter with temperature ]]
|}

The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.
[[File:Volumeefr114kj.PNG|thumb|fig
]]

Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.<div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5}
</math>
</div>
[[File:Alphagraph.PNG|thumb|fig
]]

This calculated value is in agreeable magnitude with literature, which states a thermal expansion coefficient of 4*10^-5 <nowiki><ref>https://www.hindawi.com/journals/jther/2014/123478<ref></nowiki>

There are several approximations that have been made that could result in the difference between theoretical and literature values.

=== An opportunity to speculate ... ===
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===
{|align="center"
|[[File:Md volefr114.PNG|thumb|250px|'''Fig. ''' ]]
|[[File:Volcomp.PNG|thumb|250px|'''Fig. ''' ]]
|}

{|align="center"
|[[File:HEATefr114.PNG|thumb|250px|'''Fig. ''' The ]]
|}

<div style="text-align: center;">
<math>
V(T)=0.0005689
</math>

<math>
\frac {dV}{dT}=0.0005689
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.0005689)
</math>

<math>
\alpha_V=3.0454*10^{-5}
</math>
</div>

=== Questions ===
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T08:52:15Z

Efr114: /* Molecular Dynamics */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Figure Blah a-f. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon. ref
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
An opportunity to speculate
*Would this optimal grid size for MgO be appropriate for a calculation on;
*#a similar oxide (eg: CaO) ?
*#a Zeolite (eg: Faujasite) ?
*#a metal (eg: lithium) ?
An opportunity to speculate

=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation which d
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|}

The accuracy of the HFE increases with shrinking factor.

Questions
* How does the free energy vary with grid size ?
*
* Which grid size is appropriate for calculations accurate to 1 meV, 0.5 meV and 0.1 meV per cell ?

=== An opportunity to speculate ===
* Would this optimal grid size for MgO be appropriate for a calculation on;
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

fig fig

[[File:Free energyefr114.PNG|thumb|Fig. Blah: A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations.]]The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.
[[File:Volumeefr114kj.PNG|thumb|fig
]]

Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.<div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5}
</math>
</div>
[[File:Alphagraph.PNG|thumb|fig
]]

This calculated value is in agreeable magnitude with literature, which states a thermal expansion coefficient of 4*10^-5 <nowiki><ref>https://www.hindawi.com/journals/jther/2014/123478<ref></nowiki>

There are several approximations that have been made that could result in the difference between theoretical and literature values.[[File:Mgotemplat.PNG|thumb|Fig. blah A plot showing the varience of lattice parameter with temperature ]]

=== An opportunity to speculate ... ===
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===
{|align="center"
|[[File:Md volefr114.PNG|thumb|250px|'''Fig. ''' ]]
|[[File:Volcomp.PNG|thumb|250px|'''Fig. ''' ]]
|}

{|align="center"
|[[File:HEATefr114.PNG|thumb|250px|'''Fig. ''' The ]]
|}

<div style="text-align: center;">
<math>
V(T)=0.0005689
</math>

<math>
\frac {dV}{dT}=0.0005689
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.0005689)
</math>

<math>
\alpha_V=3.0454*10^{-5}
</math>
</div>

=== Questions ===
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T08:51:27Z

Efr114:

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Figure Blah a-f. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon. ref
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
An opportunity to speculate
*Would this optimal grid size for MgO be appropriate for a calculation on;
*#a similar oxide (eg: CaO) ?
*#a Zeolite (eg: Faujasite) ?
*#a metal (eg: lithium) ?
An opportunity to speculate

=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation which d
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|}

The accuracy of the HFE increases with shrinking factor.

Questions
* How does the free energy vary with grid size ?
*
* Which grid size is appropriate for calculations accurate to 1 meV, 0.5 meV and 0.1 meV per cell ?

=== An opportunity to speculate ===
* Would this optimal grid size for MgO be appropriate for a calculation on;
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

fig fig

[[File:Free energyefr114.PNG|thumb|Fig. Blah: A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations.]]The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.
[[File:Volumeefr114kj.PNG|thumb|fig
]]

Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.<div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5}
</math>
</div>
[[File:Alphagraph.PNG|thumb|fig
]]

This calculated value is in agreeable magnitude with literature, which states a thermal expansion coefficient of 4*10^-5 <nowiki><ref>https://www.hindawi.com/journals/jther/2014/123478<ref></nowiki>

There are several approximations that have been made that could result in the difference between theoretical and literature values.[[File:Mgotemplat.PNG|thumb|Fig. blah A plot showing the varience of lattice parameter with temperature ]]

=== An opportunity to speculate ... ===
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===
{|align="center"
|[[File:Md volefr114.PNG|thumb|Fig. ]]|thumb|250px|'''Fig. ''' The ]]
|[[File:Volcomp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}
{|align="center"
|[[File:HEATefr114.PNG|thumb|Fig. ]]|thumb|250px|'''Fig. ''' The ]]
|}
[[File:HEATefr114.PNG|thumb|fig]]

<div style="text-align: center;">
<math>
V(T)=0.0005689
</math>

<math>
\frac {dV}{dT}=0.0005689
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.0005689)
</math>

<math>
\alpha_V=3.0454*10^{-5}
</math>
</div>

=== Questions ===
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T08:48:44Z

Efr114: /* Quasi-Harmonic Approximation Method */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Figure Blah a-f. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon. ref
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
An opportunity to speculate
*Would this optimal grid size for MgO be appropriate for a calculation on;
*#a similar oxide (eg: CaO) ?
*#a Zeolite (eg: Faujasite) ?
*#a metal (eg: lithium) ?
An opportunity to speculate

=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation which d
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|}

The accuracy of the HFE increases with shrinking factor.

Questions
* How does the free energy vary with grid size ?
*
* Which grid size is appropriate for calculations accurate to 1 meV, 0.5 meV and 0.1 meV per cell ?

=== An opportunity to speculate ===
* Would this optimal grid size for MgO be appropriate for a calculation on;
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

fig fig

[[File:Free energyefr114.PNG|thumb|Fig. Blah: A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations.]]The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.
[[File:Volumeefr114kj.PNG|thumb|fig
]]

Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.<div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5}
</math>
</div>
[[File:Alphagraph.PNG|thumb|fig
]]

This calculated value is in agreeable magnitude with literature, which states a thermal expansion coefficient of 4*10^-5 <nowiki><ref>https://www.hindawi.com/journals/jther/2014/123478<ref></nowiki>

There are several approximations that have been made that could result in the difference between theoretical and literature values.[[File:Mgotemplat.PNG|thumb|Fig. blah A plot showing the varience of lattice parameter with temperature ]]

=== An opportunity to speculate ... ===
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===
[[File:Md volefr114.PNG|thumb|Fig. ]]
[[File:HEATefr114.PNG|thumb|fig
]]
[[File:Volcomp.PNG|thumb]]
<div style="text-align: center;">
<math>
V(T)=0.0005689
</math>

<math>
\frac {dV}{dT}=0.0005689
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.0005689)
</math>

<math>
\alpha_V=3.0454*10^{-5}
</math>
</div>

=== Questions ===
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}

== Conclusion ==

== References ==

File:HEATefr114.PNG

2017-03-24T08:47:15Z

Efr114:

File:Volcomp.PNG

2017-03-24T08:45:28Z

Efr114:

Rep:Efr114MgOwiki

2017-03-24T08:24:17Z

Efr114: /* Molecular Dynamics */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Figure Blah a-f. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon. ref
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
An opportunity to speculate
*Would this optimal grid size for MgO be appropriate for a calculation on;
*#a similar oxide (eg: CaO) ?
*#a Zeolite (eg: Faujasite) ?
*#a metal (eg: lithium) ?
An opportunity to speculate

=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation which d
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|}

The accuracy of the HFE increases with shrinking factor.

Questions
* How does the free energy vary with grid size ?
*
* Which grid size is appropriate for calculations accurate to 1 meV, 0.5 meV and 0.1 meV per cell ?

=== An opportunity to speculate ===
* Would this optimal grid size for MgO be appropriate for a calculation on;
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

fig fig

[[File:Free energyefr114.PNG|thumb|Fig. Blah: A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations.]]The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.
[[File:Volumeefr114kj.PNG|thumb|fig
]]

Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.<div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5}
</math>
</div>
[[File:Alphagraph.PNG|thumb|fig
]]

There are several approximations that have been made that could result in the difference between theoretical and literature values.[[File:Mgotemplat.PNG|thumb|Fig. blah A plot showing the varience of lattice parameter with temperature ]]

=== An opportunity to speculate ... ===
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===
[[File:Md volefr114.PNG|thumb|Fig. ]]
<div style="text-align: center;">
<math>
V(T)=0.0005689
</math>

<math>
\frac {dV}{dT}=0.0005689
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.0005689)
</math>

<math>
\alpha_V=3.0454*10^{-5}
</math>
</div>

=== Questions ===
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}

== Conclusion ==

== References ==

File:Md volefr114.PNG

2017-03-24T08:22:11Z

Efr114:

Rep:Efr114MgOwiki

2017-03-24T08:19:41Z

Efr114:

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Figure Blah a-f. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon. ref
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
An opportunity to speculate
*Would this optimal grid size for MgO be appropriate for a calculation on;
*#a similar oxide (eg: CaO) ?
*#a Zeolite (eg: Faujasite) ?
*#a metal (eg: lithium) ?
An opportunity to speculate

=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation which d
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|}

The accuracy of the HFE increases with shrinking factor.

Questions
* How does the free energy vary with grid size ?
*
* Which grid size is appropriate for calculations accurate to 1 meV, 0.5 meV and 0.1 meV per cell ?

=== An opportunity to speculate ===
* Would this optimal grid size for MgO be appropriate for a calculation on;
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

fig fig

[[File:Free energyefr114.PNG|thumb|Fig. Blah: A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations.]]The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.
[[File:Volumeefr114kj.PNG|thumb|fig
]]

Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.<div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5}
</math>
</div>
[[File:Alphagraph.PNG|thumb|fig
]]

There are several approximations that have been made that could result in the difference between theoretical and literature values.[[File:Mgotemplat.PNG|thumb|Fig. blah A plot showing the varience of lattice parameter with temperature ]]

=== An opportunity to speculate ... ===
* What is the physical origin of thermal expansion ?
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===
<div style="text-align: center;">
<math>
V(T)=0.0005689
</math>

<math>
\frac {dV}{dT}=0.0005689
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.0005689)
</math>

<math>
\alpha_V=3.0454*10^{-5}
</math>
</div>

=== Questions ===
* Replot your data from the quasi-harmonic approximation calculations as, cell volume per formula unit vs temperature, and add some points from the MD runs at a few suitable temperatures.
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T08:15:21Z

Efr114: /* Molecular Dynamics */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Figure Blah a-f. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon. ref
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
An opportunity to speculate
*Would this optimal grid size for MgO be appropriate for a calculation on;
*#a similar oxide (eg: CaO) ?
*#a Zeolite (eg: Faujasite) ?
*#a metal (eg: lithium) ?
An opportunity to speculate

=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation which d
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|}

The accuracy of the HFE increases with shrinking factor.

Questions
* How does the free energy vary with grid size ?
*
* Which grid size is appropriate for calculations accurate to 1 meV, 0.5 meV and 0.1 meV per cell ?

=== An opportunity to speculate ===
* Would this optimal grid size for MgO be appropriate for a calculation on;
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

fig fig

[[File:Free energyefr114.PNG|thumb|Fig. Blah: A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations.]]The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.
[[File:Volumeefr114kj.PNG|thumb|fig
]]

Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.<div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5}
</math>
</div>
[[File:Alphagraph.PNG|thumb|fig
]]

There are several approximations that have been made that could result in the difference between theoretical and literature values.[[File:Mgotemplat.PNG|thumb|Fig. blah A plot showing the varience of lattice parameter with temperature ]]

=== An opportunity to speculate ... ===
* What is the physical origin of thermal expansion ?
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===

=== Questions ===
* Replot your data from the quasi-harmonic approximation calculations as, cell volume per formula unit vs temperature, and add some points from the MD runs at a few suitable temperatures.
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T08:10:14Z

Efr114: /* Phonon dispersion */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref> http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Figure Blah a-f. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon. ref
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
An opportunity to speculate
*Would this optimal grid size for MgO be appropriate for a calculation on;
*#a similar oxide (eg: CaO) ?
*#a Zeolite (eg: Faujasite) ?
*#a metal (eg: lithium) ?
An opportunity to speculate

=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation which d
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|}

The accuracy of the HFE increases with shrinking factor.

Questions
* How does the free energy vary with grid size ?
*
* Which grid size is appropriate for calculations accurate to 1 meV, 0.5 meV and 0.1 meV per cell ?

=== An opportunity to speculate ===
* Would this optimal grid size for MgO be appropriate for a calculation on;
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

fig fig

[[File:Free energyefr114.PNG|thumb|Fig. Blah: A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations.]]The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.
[[File:Volumeefr114kj.PNG|thumb|fig
]]

Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.<div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5}
</math>
</div>
[[File:Alphagraph.PNG|thumb|fig
]]

There are several approximations that have been made that could result in the difference between theoretical and literature values.[[File:Mgotemplat.PNG|thumb|Fig. blah A plot showing the varience of lattice parameter with temperature ]]

=== An opportunity to speculate ... ===
* What is the physical origin of thermal expansion ?
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===

=== Questions ===
* Replot your data from the quasi-harmonic approximation calculations as, cell volume per formula unit vs temperature, and add some points from the MD runs at a few suitable temperatures.
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?
index

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T08:09:54Z

Efr114: /* Phonon dispersion */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled <ref. http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html
</ref> |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Figure Blah a-f. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon. ref
{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
An opportunity to speculate
*Would this optimal grid size for MgO be appropriate for a calculation on;
*#a similar oxide (eg: CaO) ?
*#a Zeolite (eg: Faujasite) ?
*#a metal (eg: lithium) ?
An opportunity to speculate

=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation which d
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|}

The accuracy of the HFE increases with shrinking factor.

Questions
* How does the free energy vary with grid size ?
*
* Which grid size is appropriate for calculations accurate to 1 meV, 0.5 meV and 0.1 meV per cell ?

=== An opportunity to speculate ===
* Would this optimal grid size for MgO be appropriate for a calculation on;
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

fig fig

[[File:Free energyefr114.PNG|thumb|Fig. Blah: A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations.]]The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.
[[File:Volumeefr114kj.PNG|thumb|fig
]]

Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.<div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5}
</math>
</div>
[[File:Alphagraph.PNG|thumb|fig
]]

There are several approximations that have been made that could result in the difference between theoretical and literature values.[[File:Mgotemplat.PNG|thumb|Fig. blah A plot showing the varience of lattice parameter with temperature ]]

=== An opportunity to speculate ... ===
* What is the physical origin of thermal expansion ?
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===

=== Questions ===
* Replot your data from the quasi-harmonic approximation calculations as, cell volume per formula unit vs temperature, and add some points from the MD runs at a few suitable temperatures.
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?
index

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T08:03:57Z

Efr114: /* Quasi-Harmonic Approximation Method */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Figure Blah a-f. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon. ref http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html

{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
An opportunity to speculate
*Would this optimal grid size for MgO be appropriate for a calculation on;
*#a similar oxide (eg: CaO) ?
*#a Zeolite (eg: Faujasite) ?
*#a metal (eg: lithium) ?
An opportunity to speculate

=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation which d
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|}

The accuracy of the HFE increases with shrinking factor.

Questions
* How does the free energy vary with grid size ?
*
* Which grid size is appropriate for calculations accurate to 1 meV, 0.5 meV and 0.1 meV per cell ?

=== An opportunity to speculate ===
* Would this optimal grid size for MgO be appropriate for a calculation on;
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

fig fig

[[File:Free energyefr114.PNG|thumb|Fig. Blah: A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations.]]The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.
[[File:Volumeefr114kj.PNG|thumb|fig
]]

Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K, seen in figure .At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.<div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5}
</math>
</div>
[[File:Alphagraph.PNG|thumb|fig
]]

There are several approximations that have been made that could result in the difference between theoretical and literature values.[[File:Mgotemplat.PNG|thumb|Fig. blah A plot showing the varience of lattice parameter with temperature ]]

=== An opportunity to speculate ... ===
* What is the physical origin of thermal expansion ?
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===

=== Questions ===
* Replot your data from the quasi-harmonic approximation calculations as, cell volume per formula unit vs temperature, and add some points from the MD runs at a few suitable temperatures.
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?
index

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}

== Conclusion ==

== References ==

File:Volumeefr114kj.PNG

2017-03-24T08:00:41Z

Efr114:

File:Alphagraph.PNG

2017-03-24T07:59:33Z

Efr114:

Rep:Efr114MgOwiki

2017-03-24T07:52:11Z

Efr114: /* Quasi-Harmonic Approximation Method */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Figure Blah a-f. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon. ref http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html

{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
An opportunity to speculate
*Would this optimal grid size for MgO be appropriate for a calculation on;
*#a similar oxide (eg: CaO) ?
*#a Zeolite (eg: Faujasite) ?
*#a metal (eg: lithium) ?
An opportunity to speculate

=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation which d
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|}

The accuracy of the HFE increases with shrinking factor.

Questions
* How does the free energy vary with grid size ?
*
* Which grid size is appropriate for calculations accurate to 1 meV, 0.5 meV and 0.1 meV per cell ?

=== An opportunity to speculate ===
* Would this optimal grid size for MgO be appropriate for a calculation on;
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).
<div style="text-align: center;">
<math>A=U-TdS (3) </math>
</div>
MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

fig fig

[[File:Free energyefr114.PNG|thumb|Fig. Blah: A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations.]]The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K .

At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>
</div>

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.
<div style="text-align: center;">
<math>
V(T)=0.000560337T + 18.695980524
</math>

<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5}
</math>
</div>
compare this to literature.

Approximations in this calculation.[[File:Mgotemplat.PNG|thumb|Fig. blah A plot showing the varience of lattice parameter with temperature ]]

=== An opportunity to speculate ... ===
* What is the physical origin of thermal expansion ?
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===

=== Questions ===
* Replot your data from the quasi-harmonic approximation calculations as, cell volume per formula unit vs temperature, and add some points from the MD runs at a few suitable temperatures.
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?
index

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T07:49:33Z

Efr114: /* Quasi-Harmonic Approximation Method */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Figure Blah a-f. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon. ref http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html

{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
An opportunity to speculate
*Would this optimal grid size for MgO be appropriate for a calculation on;
*#a similar oxide (eg: CaO) ?
*#a Zeolite (eg: Faujasite) ?
*#a metal (eg: lithium) ?
An opportunity to speculate

=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation which d
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|}

The accuracy of the HFE increases with shrinking factor.

Questions
* How does the free energy vary with grid size ?
*
* Which grid size is appropriate for calculations accurate to 1 meV, 0.5 meV and 0.1 meV per cell ?

=== An opportunity to speculate ===
* Would this optimal grid size for MgO be appropriate for a calculation on;
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).

<math>A=U-TdS (3) </math>

MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

fig fig

[[File:Free energyefr114.PNG|thumb|Fig. Blah: A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations.]]The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K .

At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.

<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.

<math>
V(T)=0.000560337T + 18.695980524
</math>

<math>
\frac {dV}{dT}=0.000560337
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000560337)
</math>

<math>
\alpha_V=2.9996*10^{-5}
</math>

compare this to literature.

Approximations in this calculation.[[File:Mgotemplat.PNG|thumb|Fig. blah A plot showing the varience of lattice parameter with temperature ]]

=== An opportunity to speculate ... ===
* What is the physical origin of thermal expansion ?
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===

=== Questions ===
* Replot your data from the quasi-harmonic approximation calculations as, cell volume per formula unit vs temperature, and add some points from the MD runs at a few suitable temperatures.
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?
index

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T07:34:03Z

Efr114: /* Quasi-Harmonic Approximation Method */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Figure Blah a-f. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon. ref http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html

{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
An opportunity to speculate
*Would this optimal grid size for MgO be appropriate for a calculation on;
*#a similar oxide (eg: CaO) ?
*#a Zeolite (eg: Faujasite) ?
*#a metal (eg: lithium) ?
An opportunity to speculate

=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation which d
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|}

The accuracy of the HFE increases with shrinking factor.

Questions
* How does the free energy vary with grid size ?
*
* Which grid size is appropriate for calculations accurate to 1 meV, 0.5 meV and 0.1 meV per cell ?

=== An opportunity to speculate ===
* Would this optimal grid size for MgO be appropriate for a calculation on;
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).

<math>A=U-TdS (3) </math>

MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

fig fig

[[File:Free energyefr114.PNG|thumb|Fig. Blah: A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations.]]The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K .

At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.

<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.

<math>
V(T)=0.000560337T + 18.695980524
</math>

<math>
\frac {dV}{dT}=0.000001068
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068)
</math>

<math>
\alpha_V=5.7172*10^{-8}
</math>

compare this to literature.

Approximations in this calculation.[[File:Mgotemplat.PNG|thumb|Fig. blah A plot showing the varience of lattice parameter with temperature ]]

=== An opportunity to speculate ... ===
* What is the physical origin of thermal expansion ?
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===

=== Questions ===
* Replot your data from the quasi-harmonic approximation calculations as, cell volume per formula unit vs temperature, and add some points from the MD runs at a few suitable temperatures.
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?
index

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T07:31:54Z

Efr114: /* Quasi-Harmonic Approximation Method */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Figure Blah a-f. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon. ref http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html

{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
An opportunity to speculate
*Would this optimal grid size for MgO be appropriate for a calculation on;
*#a similar oxide (eg: CaO) ?
*#a Zeolite (eg: Faujasite) ?
*#a metal (eg: lithium) ?
An opportunity to speculate

=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation which d
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|}

The accuracy of the HFE increases with shrinking factor.

Questions
* How does the free energy vary with grid size ?
*
* Which grid size is appropriate for calculations accurate to 1 meV, 0.5 meV and 0.1 meV per cell ?

=== An opportunity to speculate ===
* Would this optimal grid size for MgO be appropriate for a calculation on;
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).

<math>A=U-TdS (3) </math>

MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

fig fig

[[File:Free energyefr114.PNG|thumb|Fig. Blah: A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations.]]The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K .

At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.

<math>
V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math>
\frac {dV}{dT}=0.000001068T + 0.000047391
</math>

<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.

<math>
V(T)=0.000560337T + 18.695980524
</math>
<math>
\frac {dV}{dT}=0.000001068
</math>
<math>
\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068)
</math>

compare this to literature.

y =
0.000560337x + 18.695980524
R² = 0.999100029

Approximations in this calculation.[[File:Mgotemplat.PNG|thumb|Fig. blah A plot showing the varience of lattice parameter with temperature ]]

=== An opportunity to speculate ... ===
* What is the physical origin of thermal expansion ?
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===

=== Questions ===
* Replot your data from the quasi-harmonic approximation calculations as, cell volume per formula unit vs temperature, and add some points from the MD runs at a few suitable temperatures.
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?
index

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T07:30:37Z

Efr114: /* Quasi-Harmonic Approximation Method */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Figure Blah a-f. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon. ref http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html

{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
An opportunity to speculate
*Would this optimal grid size for MgO be appropriate for a calculation on;
*#a similar oxide (eg: CaO) ?
*#a Zeolite (eg: Faujasite) ?
*#a metal (eg: lithium) ?
An opportunity to speculate

=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation which d
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|}

The accuracy of the HFE increases with shrinking factor.

Questions
* How does the free energy vary with grid size ?
*
* Which grid size is appropriate for calculations accurate to 1 meV, 0.5 meV and 0.1 meV per cell ?

=== An opportunity to speculate ===
* Would this optimal grid size for MgO be appropriate for a calculation on;
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).

<math>A=U-TdS (3) </math>

MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

fig fig

[[File:Free energyefr114.PNG|thumb|Fig. Blah: A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations.]]The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K .

At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.

<math> V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)

\frac {dV}{dT}=0.000001068T + 0.000047391

\alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.

<math> V(T)=0.000560337T + 18.695980524
</math>
<math> \frac {dV}{dT}=0.000001068
</math>

<math> \alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068)
</math>

compare this to literature.

y =
0.000560337x + 18.695980524
R² = 0.999100029

Approximations in this calculation.[[File:Mgotemplat.PNG|thumb|Fig. blah A plot showing the varience of lattice parameter with temperature ]]

=== An opportunity to speculate ... ===
* What is the physical origin of thermal expansion ?
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===

=== Questions ===
* Replot your data from the quasi-harmonic approximation calculations as, cell volume per formula unit vs temperature, and add some points from the MD runs at a few suitable temperatures.
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?
index

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T07:29:58Z

Efr114: /* Quasi-Harmonic Approximation Method */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Figure Blah a-f. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon. ref http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html

{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
An opportunity to speculate
*Would this optimal grid size for MgO be appropriate for a calculation on;
*#a similar oxide (eg: CaO) ?
*#a Zeolite (eg: Faujasite) ?
*#a metal (eg: lithium) ?
An opportunity to speculate

=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation which d
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|}

The accuracy of the HFE increases with shrinking factor.

Questions
* How does the free energy vary with grid size ?
*
* Which grid size is appropriate for calculations accurate to 1 meV, 0.5 meV and 0.1 meV per cell ?

=== An opportunity to speculate ===
* Would this optimal grid size for MgO be appropriate for a calculation on;
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).

<math>A=U-TdS (3) </math>

MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

fig fig

[[File:Free energyefr114.PNG|thumb|Fig. Blah: A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations.]]The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K .

At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.

<math> V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4)
</math>

<math> \frac {dV}{dT}=0.000001068T + 0.000047391
</math>
<math> \alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391)
</math>

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.

<math> V(T)=0.000560337T + 18.695980524
</math>
<math> \frac {dV}{dT}=0.000001068
</math>

<math> \alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068)
</math>

compare this to literature.

y =
0.000560337x + 18.695980524
R² = 0.999100029

Approximations in this calculation.[[File:Mgotemplat.PNG|thumb|Fig. blah A plot showing the varience of lattice parameter with temperature ]]

=== An opportunity to speculate ... ===
* What is the physical origin of thermal expansion ?
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===

=== Questions ===
* Replot your data from the quasi-harmonic approximation calculations as, cell volume per formula unit vs temperature, and add some points from the MD runs at a few suitable temperatures.
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?
index

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T07:29:23Z

Efr114: /* Quasi-Harmonic Approximation Method */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Figure Blah a-f. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon. ref http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html

{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
An opportunity to speculate
*Would this optimal grid size for MgO be appropriate for a calculation on;
*#a similar oxide (eg: CaO) ?
*#a Zeolite (eg: Faujasite) ?
*#a metal (eg: lithium) ?
An opportunity to speculate

=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation which d
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|}

The accuracy of the HFE increases with shrinking factor.

Questions
* How does the free energy vary with grid size ?
*
* Which grid size is appropriate for calculations accurate to 1 meV, 0.5 meV and 0.1 meV per cell ?

=== An opportunity to speculate ===
* Would this optimal grid size for MgO be appropriate for a calculation on;
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).

<math>A=U-TdS (3) </math>

MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

fig fig

[[File:Free energyefr114.PNG|thumb|Fig. Blah: A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations.]]The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K .

At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.

<math> V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4) </math>

<math> \frac {dV}{dT}=0.000001068T + 0.000047391 </math>
<math> \alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391) </math>

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.

<math> V(T)=0.000560337T + 18.695980524 </math>
<math> \frac {dV}{dT}=0.000001068</math>

<math> \alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068) </math>

compare this to literature.

y =
0.000560337x + 18.695980524
R² = 0.999100029

Approximations in this calculation.[[File:Mgotemplat.PNG|thumb|Fig. blah A plot showing the varience of lattice parameter with temperature ]]

=== An opportunity to speculate ... ===
* What is the physical origin of thermal expansion ?
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===

=== Questions ===
* Replot your data from the quasi-harmonic approximation calculations as, cell volume per formula unit vs temperature, and add some points from the MD runs at a few suitable temperatures.
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?
index

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T07:28:20Z

Efr114: /* Quasi-Harmonic Approximation Method */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Figure Blah a-f. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon. ref http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html

{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
An opportunity to speculate
*Would this optimal grid size for MgO be appropriate for a calculation on;
*#a similar oxide (eg: CaO) ?
*#a Zeolite (eg: Faujasite) ?
*#a metal (eg: lithium) ?
An opportunity to speculate

=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation which d
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|}

The accuracy of the HFE increases with shrinking factor.

Questions
* How does the free energy vary with grid size ?
*
* Which grid size is appropriate for calculations accurate to 1 meV, 0.5 meV and 0.1 meV per cell ?

=== An opportunity to speculate ===
* Would this optimal grid size for MgO be appropriate for a calculation on;
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).

<math>A=U-TdS (3) </math>

MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

fig fig

[[File:Free energyefr114.PNG|thumb|Fig. Blah: A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations.]]The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K .

At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume within this temperature range.

<math> V(T)=0.000000534T2 + 0.000047391T + 18.827344363 (4) </math>

<math> \frac {dV}{dT}=0.000001068T + 0.000047391 </math>
<math> \alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068T + 0.000047391) </math>

Figure blah shows the plot of thermal expansion coefficient against temperature for 100-450 K, it shows that thermal expansion coefficient is proportional to temperature within this temperature range.

At higher temperatures (500 - 1000 K), the data fits a linear trendline and an expression can be found for volume within this temperature range.

<math> V(T)=0.000560337T + 18.695980524 </math>
<math> \frac {dV}{dT}=0.000001068</math>
<math> \alpha_V=\frac{1}{V(0)}\frac{dV}{dT}=\frac{1}{V(0)}*(0.000001068) </math>

compare this to literature.

y =
0.000560337x + 18.695980524
R² = 0.999100029

Approximations in this calculation.[[File:Mgotemplat.PNG|thumb|Fig. blah A plot showing the varience of lattice parameter with temperature ]]

=== An opportunity to speculate ... ===
* What is the physical origin of thermal expansion ?
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===

=== Questions ===
* Replot your data from the quasi-harmonic approximation calculations as, cell volume per formula unit vs temperature, and add some points from the MD runs at a few suitable temperatures.
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?
index

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T07:20:00Z

Efr114: /* Quasi-Harmonic Approximation Method */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighboring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequency and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Figure Blah a-f. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon. ref http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html

{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
An opportunity to speculate
*Would this optimal grid size for MgO be appropriate for a calculation on;
*#a similar oxide (eg: CaO) ?
*#a Zeolite (eg: Faujasite) ?
*#a metal (eg: lithium) ?
An opportunity to speculate

=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation which d
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|}

The accuracy of the HFE increases with shrinking factor.

Questions
* How does the free energy vary with grid size ?
*
* Which grid size is appropriate for calculations accurate to 1 meV, 0.5 meV and 0.1 meV per cell ?

=== An opportunity to speculate ===
* Would this optimal grid size for MgO be appropriate for a calculation on;
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).

<math>A=U-TdS (3) </math>

MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes tends to linearity

fig fig

[[File:Free energyefr114.PNG|thumb|Fig. Blah: A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations.]]The varience of lattice parameter with temperature can be seen in Fig. blah. The lattice parameter increases with increasing temperature, indicating positive thermal expansion coefficients. At lower temperatures (100 - 450 K), the data has a quadratic fit but at higher temperatures (500 - 1000 K), the data has a linear fit. The general trend of thermal expansion with increase in temperature can be explained by considering the effect of an increase of energy in the lattice system. The more energy input into the lattice, the higher energy phonons are more populated. As seen in the phonon dispersion plot, the higher frequency phonons are optical phonons - which correspond to out-of-phase vibrations with neighboring atoms, this causes increased vibration and interatomic seperation of the ions which results in expansion of the lattice.

Volume of the unit cell of MgO can now be calculated within the temperature range 100 - 1000 K .

At lower temperatures (100 - 450 K), the data fits a quadratic trendline and an expression can be found for volume for this temperature range.

V(T)=
0.000000534T2 + 0.000047391T + 18.827344363

dV/dT=0.000001068T + 0.000047391

coefficient of thermal expansion for MgO and compare this to literature.

R² = 0.999238853

y =
0.000560337x + 18.695980524
R² = 0.999100029

Approximations in this calculation.[[File:Mgotemplat.PNG|thumb|Fig. blah A plot showing the varience of lattice parameter with temperature ]]

=== An opportunity to speculate ... ===
* What is the physical origin of thermal expansion ?
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===

=== Questions ===
* Replot your data from the quasi-harmonic approximation calculations as, cell volume per formula unit vs temperature, and add some points from the MD runs at a few suitable temperatures.
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?
index

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T06:34:18Z

Efr114: /* Quasi-Harmonic Approximation Method */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighbouring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequenct and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Figure Blah a-f. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon. ref http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html

{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
An opportunity to speculate
*Would this optimal grid size for MgO be appropriate for a calculation on;
*#a similar oxide (eg: CaO) ?
*#a Zeolite (eg: Faujasite) ?
*#a metal (eg: lithium) ?
An opportunity to speculate

=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation which d
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|}

The accuracy of the HFE increases with shrinking factor.

Questions
* How does the free energy vary with grid size ?
*
* Which grid size is appropriate for calculations accurate to 1 meV, 0.5 meV and 0.1 meV per cell ?

=== An opportunity to speculate ===
* Would this optimal grid size for MgO be appropriate for a calculation on;
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).

<math>A=U-TdS (3) </math>

MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the <math>TdS</math> component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the <math> TdS </math>componant becomes increasingly dominant in the contribution to Helmholtz free energy and the plot becomes more linear.

[[File:Free energyefr114.PNG|thumb|Fig. Blah: A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations.]]
[[File:Mgotemplat.PNG|thumb|Fig. blah A plot showing the varience of lattice parameter with temperature ]]

=== Questions ===
* Comment on the shape of these curves.
* Compute the coefficient of thermal expansion for MgO
* How does this compare to that measured ? Find a measurement in the literature or on the web - at what temperature was the measurement made ?
* What are the main approximations in your calculation ?

=== An opportunity to speculate ... ===
* What is the physical origin of thermal expansion ?
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===

=== Questions ===
* Replot your data from the quasi-harmonic approximation calculations as, cell volume per formula unit vs temperature, and add some points from the MD runs at a few suitable temperatures.
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?
index

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}

== Conclusion ==

== References ==

Rep:Efr114MgOwiki

2017-03-24T06:31:32Z

Efr114: /* Optimisation- harmonic approximation */

== Introduction ==
[[File:Mgo beginning.PNG|thumb|206x206px|'''Fig. 1''' A labelled visualization of the conventional cell format of the input MgO crystal structure]]
Thermal expansion is the tendancy of matter to change in volume in relation to temperature, the understanding and calculation of which is essential for a vast array of fields, particularly physical sciences engineering.

The extent that a material can thermally expand can be given by the general thermal expansion coefficient, <math> \alpha_V </math>, which is definined in Equation 1.<ref> N. C. Corsepius,T. C. DeVore, B. A. Reisner and D. L. Warnaar, J. Chem. Edu., 2007, 84, 818-822 </ref>
<math> \alpha_{V} = \frac {1}{V}\bigg(\frac {\partial V}{\partial T}\bigg)_P (1) </math>

This equation can be used for solid, liquid and gaseous systems at constant pressure. A thermal expansion coefficient is an intrinsic property of the material. The coefficient's for most materials are positive values, however some materials do have negative coefficients: graphene has a negative coefficient of <math>-8.0*10^{-6}\ K^{-1} </math><ref> D. Yoon, Y-W. Son and H. Cheong, Nano Lett., 2011, 11, 3227-3231</ref>

== Methodology ==

=== MgO: The Model ===
Magnesium oxide is a face-centred cubic lattice, composed of Mg<sup>2+</sup> and O<sup>2-</sup> ions. The model used in this experiment is assumed to be ideal and non-defective with simple Buckingham repulsive potential interactions between the ions.The lattice parameters for Mgo in the input file are <math> a=b=c=2.9782 </math> and <math> \alpha=\beta=\gamma=60.000^{\circ} </math>

=== Computational Methods ===
''DLVisualise (DLVis)''was used to visualise the lattice and the output of the lattice calculations. DLVis is used to construct and modify lattices and provides a graphical user interface for GULP <ref>http://www.cse.clrc.ac.uk/cmg/DLV/ </ref>

''General Utility Lattice Program (GULP) ''was used to compute the lattice calculations using analytical solutions, it is a general purpose code that models atoms as hard spheres instead of separating the atom into it's components.<ref> http://www.staff.amu.edu.pl/~magchem/GULP/gman.html </ref>

=== Quasi-Harmonic Approximation ===
The Quasi-Harmonic Approximation (QHA), as it's name suggests, is a modification of the well- known classical mechanic harmonic oscillator system . QHA is dependant on volume, allowing the model to be used for thermal expansion calculations, the harmonic oscillator model could not be used for such calculations as it imply's that the interatomic distance between ions is independant of temperature. The QHA

The free energy of the crystal can be modeled using the QHA using Equation 2.

<math>F(T,V) =U(V)+E_{ZP}(V)-TS(T,V) (2)\ </math>

<math>U(V)</math> corresponds to the internal energy term, <math>E_{ZP}</math> corresponds to the sum of zero point energies and <math>S(T,V)</math> corresponds to entropy.

=== Molecular Dynamics ===
Molecular Dynamics (MD) is a classical mechanical model that randomly assigns the velocities of the particles of the system, which are scaled using the Boltzmann distribution, force, acceleration and hence trajectory can then be calculated at timesteps throughout the simulation which runs to a time-averaged equilibrium if allowed to run for an appropriate time. If the input system obeys the Ergodic hypothesis, the macroscopic thermodynamic properties, such as the thermal expansion coefficient and Helmholtz free energy can be calculated.

== Results and Discussion ==

=== MgO: Internal Energy ===
A single point calculation using GULP was used to measure the total lattice energy of an MgO ionic lattic , which produced an output file containing measurement of specific properties of the input structure such as , however it does not optimise or change the structure. The total lattice energy, the energy required to seperate every ion in the lattice to infinate separation, was calculated to be of -41.0753 eV per primitive cell. This value corresponds to 3963.1638 kJ mol<sup>-1</sup>.

=== Phonon dispersion ===
[[File:Phonodispersionefr114.PNG|thumb|300x300px|'''Fig. 2''' The phonon sispersion curve for MgO calculated using GULP and visualized using DLVis]]

All possible vibrations of a crystal can be assigned a k-vector. To plot the vibrational frequencys over the whole three-dimensional k space would be extremely difficult and computational chemists use a specific set of k-points that depict a connected path of specific symmetry points, which are k-points that have particularly high symmetry, to plot a phonon dispersion curve. The relationship between <math> \omega </math> and k-point is called the phonon dispersion relation.

The phonon dispersion operation in GULP was used to calculate the phonon dispersion curve for MgO, which is a plot of phonon vibration frequency <math> \omega </math> against k-point. The frequency was calculated for 50 k-points including <math> W, L, \Gamma , X, W, K </math>, seen in Fig. 2, where <math> \Gamma </math> corresponds to <math> (0,0,0) </math>

The phonon diagram of MgO was plotted, seen in Fig. 2. The diagram has 6 branches, which is concordant with theory of lattice vibrations which states that, for a 3 dimensional model with an n number of atoms in it's primitive cell, <math> No.\ of\ branches = 3n </math>, hence for a primitive cell of two atoms, 6 branches should be visible in the photon dispersion graph. There are two types of phonon: an acoustic phonon and an optical phonon.
Acoustic phonons are lower frequency and tend to 0 at the k-space origin, they correspond to in-phase vibrations of neighbouring atoms. These phonons exhibit more bonding character due to the in-phase movement away from the k-space origin. Optical phonons are high frequenct and have non-zero values at the k-space origin, they correspond to out-of-phase vibrations of neighbouring atoms. For the three-dimensional system with a primitive cell of n atoms, <math>No.\ of\ optical\ branches =3n-3 </math> and <math> No.\ of\ optical\ branches =3 </math>which can be clearly identified in Fig. 2. As seen in Fig. 2, branches merge with other branches and some branches appear to separate, this is due to the symmetry involved in k-space 'path'.[[File:Symettry .jpg|thumb|'''Fig. 3''' A diagram representing the reciprocal space of an FCC lattice, with the symmetry points labelled |left|150x150px]]The k-space coordinates of the symmetry points in Table 1 were used with the animator function DLVis to produce the animations seen in Figure Blah a-f. These are visualizations of the lowest frequency phonon for the symmetry points <math> W, L, \Gamma , X, W, K </math>. The <math>\Gamma</math> phonon can be represented within a primitive cell, however, the phonons become more complex and need to expand to larger supercells, to produce a visualisation of the whole phonon. ref http://nms.kcl.ac.uk/lev.kantorovitch/codes/lev00/MgO_vibrations/2.html

{|align="center"
|[[File:250efr114sig.gif|thumb|250px|'''Fig. 4a''' A .gif file of the animation of the lowest frequency phonon for symmetry point <math>\Gamma</math>]]
|[[File:Xnewefr114.gif|thumb|250px|'''Fig. 4b''' A .gif file of the animation of the lowest frequency phonon for symmetry point X]]
|[[File:Lnewefr114.gif|thumb|250px|'''Fig. 4c''' A .gif file of the animation of the lowest frequency phonon for symmetry point L]]
|[[File:Wnewefr114.gif|thumb|250px|'''Fig. 4d''' A .gif file of the animation of the lowest frequency phonon for symmetry point W]]
|[[File:Finalnewefr114bl.gif|thumb|250px|'''Fig. 4e''' A .gif file of the animation of the lowest frequency phonon for symmetry point K]]
|}

{| class="wikitable"
!Symmetry point
!Real Space
<math>(x,y,z)</math>
!Reciprocal space
<math>(k_x,k_y,k_z)</math>
|-
|<math>\Gamma</math>
|<math>(0,0,0)</math>
|<math>(0,0,0)</math>
|-
|X
|<math>\bigg(0, \frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(0, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|W
|<math>\bigg(\frac {1}{4},\frac {1}{2},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {\pi}{2a},\frac {\pi}{a}, \frac {3\pi}{2a}\bigg)</math>
|-
|L
|<math>\bigg(\frac {1}{2},\frac {1}{2},\frac {1}{2}\bigg)</math>
|<math>\bigg(\frac {\pi}{a}, \frac {\pi}{a},\frac {\pi}{a}\bigg)</math>
|-
|K
|<math>\bigg(\frac {3}{8},\frac {3}{8},\frac {3}{4}\bigg)</math>
|<math>\bigg(\frac {3\pi}{4a}, \frac {3\pi}{4a},\frac {3\pi}{2a}\bigg)</math>
|-
| colspan="3" |Table 1: A table of values for the real space and
reciprocal space coordinates for symmetry points
|}

=== Density of States ===
The phonon DOS operation in GULP was used to calculate a phonon density of states (DOS) plot, which is a plot of intensity against frequency of the phonon. A phonon DOS plot displays the density of vibrational energy states at a given frequency which is the number of vibrational energy states per unit frequency at any given frequency.

For a 'complete' DOS plot, every k-point in the Brillouin zone should be used in the calculation, however this calculation would require an extremely large amount of CPU and would be expensive to run. A sample of the points must be used, this is achieved by the use of a grid of points is then used. The construction of the grid is based on a shrinking factor (SF), an increase in SF results in more k-space values sampled.

Initially, a 1x1x1 grid (shrinking factor 1) was used in calculations, producing Fig. Blah, this grid corresponds to one k-point sampled for the calculation. Fig. blah has 4 peaks at frequencies of ''ca.'' 290,350, 680 and 810 cm<sup>-1. </sup>

These values were directly compared with the values from the photon dispersion graph, seen in Fig. 3 to conclude that the sampled k-point was the symmetry point L, this is also confirmed by the .log output file. A sample of one-k point does not accurately represent the phonon DOS of the whole crystal and the grid size must be increased to improve the accuracy. If the grid size is too high, too many k-points are sampled, increasing the CPU usage making the calculations longer and more expensive. A compromise between accuracy and CPU usage must be found.

{|align="center"
|[[File:1x1efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:5x5efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:10x10newefr114.PNG|thumb|200px|''Figure '':]]
|[[File:20x20efr114.PNG|thumb|200px|''Figure '' :]]
|}

{|align="center"
|[[File:30x30efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:35x35efr114.PNG|thumb|200px|''Figure '': ]]
|[[File:50x50.PNG|thumb|200px|''Figure '']]
|
|}

The optimal grid size was found by calculation of the phonon DOS, increasing in increments of SF 5. The optimal SF was found to be in between SF 30 and 35, following which calculations of phonon DOS, increasing in increments of SF 1 and starting from SF 30 were performed to result in a conclusion that a SF of 33 was the optimal value, this value was obtained from a qualitative method which increases the possibility for error. Above the optimal SF, seen in Figure blah, the photon DOS curve does becomes more smooth and there is a small shift frequency values, however the small gain in accuracy is not worth the extra processing required.
*How is the density of states related to the dispersion curves ?
An opportunity to speculate
*Would this optimal grid size for MgO be appropriate for a calculation on;
*#a similar oxide (eg: CaO) ?
*#a Zeolite (eg: Faujasite) ?
*#a metal (eg: lithium) ?
An opportunity to speculate

=== Helmholtz free energy (HFE) ===
Helmholtz free energy (HFE) calculations were run under the harmonic approximation which d
{| class="wikitable"
!Shrinking factor
!HFE / eV
!Delta HFE / eV
|-
|1
|40.930301
|0.0161
|-
|2
|40.926609
|0.0198
|-
|3
|40.926432
|0.0201
|-
|5
|40.926463
|0.0200
|-
|10
|40.926480
|0.0200
|-
|15
|40.926482
|0.0200
|-
|16
|40.946482
|1.0000*10<sup>6</sup>
|-
|17
|40.946482
|1.0000*10<sup>6</sup>
|-
|18
|40.946483
|0.0000
|-
|19
|40.946483
|0.0000
|-
|20
|40.946483
|0.0000
|}

The accuracy of the HFE increases with shrinking factor.

Questions
* How does the free energy vary with grid size ?
*
* Which grid size is appropriate for calculations accurate to 1 meV, 0.5 meV and 0.1 meV per cell ?

=== An opportunity to speculate ===
* Would this optimal grid size for MgO be appropriate for a calculation on;
*# a similar oxide (eg: CaO) ?
*# a Zeolite (eg: Faujasite) ?
*# a metal (eg: lithium) ?

=== Quasi-Harmonic Approximation Method ===
GULP optimisation calculations were used to calculate the Helmholtz free energy and lattice parameters of MgO at temperatures ranging from 0-1000 K. These calculations are under the quasi-harmonic approximation method. A shrinking factor of 18 was used for these calculations, due to conclusions from earlier stages of the experiment.

The varience of Helmholtz free energy with temperature can be seen in Fig. Blah. There is an inverse quadratic relationship between the Helmholtz free energy and temperature, confirmed by a R<sup>2</sup> value of 0.9995 for a quadratic fit trendline. The shape of the plot in Fig. Blah can be explained using the equation for Helmholtz free energy (Equation 3).

A=U-TS (3)

MgO has a negative and hence exothermic lattice energy. At lower temperatures, both the internal energy and the TS component contribute to the Helmholtz free energy resulting in a non-linear region. As temperature increases, the TS componant dominates the contribution to Helmholtz free energy and the plot becomes more linear.

[[File:Free energyefr114.PNG|thumb|Fig. Blah: A plot showing the variance of Helmholtz free energy with Temperature using Quasi-Harmonic SF 18 calculations.]]
[[File:Mgotemplat.PNG|thumb|Fig. blah A plot showing the varience of lattice parameter with temperature ]]

=== Questions ===
* Comment on the shape of these curves.
* Compute the coefficient of thermal expansion for MgO
* How does this compare to that measured ? Find a measurement in the literature or on the web - at what temperature was the measurement made ?
* What are the main approximations in your calculation ?

=== An opportunity to speculate ... ===
* What is the physical origin of thermal expansion ?
* As the temperature approaches the melting point of MgO how well do the phonon modes represent the actual motions of the ions ?
* In a diatomic molecule with an exactly harmonic potential would you expect the bond length to increase with temperature ? Why does it increase in the solid when we are using an quasi-harmonic approximation ?

=== Molecular Dynamics ===

=== Questions ===
* Replot your data from the quasi-harmonic approximation calculations as, cell volume per formula unit vs temperature, and add some points from the MD runs at a few suitable temperatures.
* How does the thermal expansion predicted by MD compare to that predicted by the quasi-harmonic approximation ?
* Why do the two methods produce different answers ? - how does the difference depend on temperature ?

=== An opportunity to speculate ===
* From your experience of computing an accurate Free Energy how large a cell should be used to perform reliably MD for MgO ?
* What would happen to the cell volume at high temperature in the quasi-harmonic approximation and in the MD ?
index

=== Extension: CaO: calculations and comparision ===
[[File:Cao input.PNG|thumb|Fig. BLAH A screenshot of the altered GULP library file with adjusted values]]
Methodology
DLVis- The input DLVis structure was altered using <code> Edit > Model > Modify Atoms </code> to change the atom type from Mg to Ca. The <code> Edit > Lattice Parameter </code> function was then used to uniput the conventional cell lattice parameter of 4.8108<ref> S. Speziale and S. R. Shieh, J. Geophys. Res., 111, 2006, 2203-2215 </ref>
GULP- The GULP library file also had to be altered to adjust for the different Buckingham potentials involved with CaO compared to MgO.
The same methodolody of the MgO experiment was then used.

The phonon dispersion curve was calculated and the optimal grid size of 20x20 was determined, which can be seen in Fig Blah a and Fig Blah b respectively.

As expected, the calculated phonon dispersion curve for CaO shows great similarity to that of MgO and three acoustic branches and three optic branches can be seen, which is concordant with theory. The DOS graph is for shrinking factor 20, which was found to be the optimal grid size, complimenting the earlier hypothesis of the expected similarity of optimal grid size for CaO compared with Mgo

{|align="center"
|[[File:Caophon.PNG|thumb|250px|'''Fig. ''' The phonon dispersion curve for CaO ]]
|[[File:Phonocaoelp.PNG|thumb|250px|'''Fig. ''' The DOS graph for CaO at shrinking factor 20 ]]
|}

== Conclusion ==

== References ==