MatSci 193/203:

Atomic Arrangements in Solids

Beyond the hard sphere: Crystal structure

prediction and Lennard-Jones solids

Evan Reed
October 7, 2016

MatSci 193/203: Atomic Arrangements in Solids

Evan Reed 1
Stanford University
 For additional reference

Chapter 3 in Kittel (on Coursework)

Ch 17 in de Graef

MatSci 193/203: Atomic Arrangements in Solids

Evan Reed 2
Stanford University
 MatSci 193/203 course details

Course objectives:
What: Common crystalline structures for
research-relevant materials
Why: Why these structures are what they are;
The nature of chemical interactions in materials
Implications: Macroscopic manifestations of
microscopic (atomic) structure (mechanical
properties, etc.)

MatSci 193/203: Atomic Arrangements in Solids

Evan Reed
3
Stanford University
Total energy calculations reveal the
lowest energy crystal structure

MatSci 193/203: Atomic Arrangements in Solids

Evan Reed 5
Stanford University
 Types of bonding in materials and
molecules
There are a variety of types of interatomic interactions in materials, including:

Covalent bonds, antibonds (eg. Si, Ge, C)

Metallic bonds (alkali metals like Na)
Ionic interactions (e.g. NaCl, GaAs, CdSe)
van der Waals or dispersion (e.g. Ar, Xe) (electrostatic interactions
associated with correlated quantum fluctuations of electrons)

Materials can exhibit a mixture of these types, e.g. partially ionic and covalent

MatSciMatSci
331: Atom-based
193/203: Atomic
methods
Arrangements
for materials
in Solids
modeling
Evan
EvanReed
Reed
7
Stanford
StanfordUniversity
University
Noble gas solids any many molecular solids
are van der Waals bonded

MatSci 331: Atom-based methods for materials modeling

Evan Reed
8
Stanford University
Lennard-Jones: One of the most simple
interatomic empirical potentials

MatSci 331: Atom-based methods for materials modeling

Evan Reed
9
Stanford University
 Lennard-Jones: One of the most simple
interatomic empirical potentials
The power of 6 in the attractive term
comes from dipole fluctuation
interaction between separated atoms.
This is a quantum interaction that
occurs even at zero temperature.

The choice of power 12 in the repulsive

portion of the potential is rather
arbitrary it could be another power or
another functional form.

r r
1.12 at minimum
s s
10
 Interaction range hierarchy

Energy V Electric field E = -V

Charge: ~ -1/ r
2
Charge-charge ~ 1/ r
Charge-dipole ~ 1/ r 2 Dipole: ~ -1/ r 3
Dipole-dipole ~ 1/ r 3
Dipole-induced dipole
(van der Waals or ~ 1/ r 6
dispersion)

s / r6

s /r
Coulomb interactions are the
most long ranged

van der Waals dominates in the

MatSci 331: Atom-based methods for materials modeling
11
absence of other types of
Evan Reed
Stanford University
interactions
Lennard-Jones: One of the most simple
interatomic empirical potentials
LJ is most appropriate for closed shell atoms, possibly small molecules as well.

Interactions are weak well depth for Ar is around 100 degrees K.

MatSci 193/203: Atomic Arrangements in Solids

Evan Reed 12
Stanford University
Lennard-Jones: One of the most simple
interatomic empirical potentials
LJ might be most appropriate for closed shell atoms, possibly small molecules
as well.

Interactions are weak well depth for Ar is around 100 degrees K.

MatSci 193/203: Atomic Arrangements in Solids

MatSci 331: Atom-based methods for materials modeling
Evan Reed
Evan Reed
13
Stanford University
Stanford University
Lennard-Jones: One of the most simple
interatomic empirical potentials
LJ might be most appropriate for closed shell atoms, possibly small molecules
as well.

Interactions are weak well depth for Ar is around 100 degrees K.

MatSci 193/203: Atomic Arrangements in Solids

Evan Reed 14
Stanford University
Lennard-Jones: One of the most simple
interatomic empirical potentials
LJ might be most appropriate for closed shell atoms, possibly small molecules
as well.

Interactions are weak well depth for Ar is around 100 degrees K.

MatSci 193/203: Atomic Arrangements in Solids

Evan Reed 15
Stanford University
 Experimental observations of nearest
neighbor distances for van der Waals solids
(all FCC structures)

MatSci 193/203: Atomic Arrangements in Solids

Evan Reed 16
Stanford University
Lennard-Jones: One of the most simple
interatomic empirical potentials
LJ might be most appropriate for closed shell atoms, possibly small molecules
as well.

Interactions are weak well depth for Ar is around 100 degrees K.

MatSci 193/203: Atomic Arrangements in Solids

Evan Reed 17
Stanford University
 Interatomic interactions described by
short-ranged pair potentials tend to form
close-packed structures

Van der Waals interactions and metallic bonds tend

to be reasonably well described by short-ranged pair
potentials between atoms, which provide the lowest
cohesive energies for close packed structures, FCC,
HCP. BCC structures are higher in energy.

MatSci 193/203: Atomic Arrangements in Solids

Evan Reed 18
Stanford University
Beyond the pair potential: A three body-potential
by Stillinger and Weber for tetrahedral structures

Z=4 for the diamond structure much lower than
Z=12 structures favored for LJ potential.
Hybridization of s and p valence oribtals in Si
results in density maximized between atoms in
covalently bonded materials. Such hybridization
induces angular dependence to bonding.
(Si diamond structure density shown)

MatSci 331: Atom-based methods for materials modeling

Evan Reed
19
Stanford University
A three body-potential by Stillinger and
Weber for tetrahedral structures
The usual many-body expansion

Two body term

Three body term gives explicitly

angular dependence to bond
energies

MatSci 193/203: Atomic Arrangements in Solids

Evan Reed 20
Stanford University
A three body-potential by Stillinger and
Weber for tetrahedral structures

Two body term

Three body term

MatSci 331: Atom-based methods for materials modeling

Evan Reed
21
Stanford University
 A three body-potential by Stillinger and
Weber for tetrahedral structures

Three body term

This term is minimized for angles of 109 degrees,

corresponding to tetrahedral bonding (sp3)
hybridization.

There are 7 empirical

parameters, determined from
lattice constant, cohesive
energy, liquid structure,
melting point. MatSci 331: Atom-based methods for materials modeling
22
Evan Reed
Stanford University
The Stillinger and Weber 3-body potential
comparison of crystal total energies
Stillinger-Weber Density functional theory
(more accurate)

Needs and Mujica,

PRB 1995.

Diamond is the lowest energy structure for both schemes,

but there is discrepancy for other crystal structures.
MatSci 193/203: Atomic Arrangements in Solids
Evan Reed 23
Stanford University
 Helmholtz and Gibbs energy governs
phase transitions at constant T
The vibrational spectra of crystals play a role in the
Helmholtz and Gibbs energies.

Vibrational entropy andMatSci

other entropic contributions lead to crystal
193/203: Atomic Arrangements in Solids
24
phase changes with
Evan Reed
Stanford temperature.
University
 Helmholtz and Gibbs energy governs
phase transitions at constant T

The quasiharmonic approximation treats each

vibration mode as a quantum harmonic oscillator to
provide the vibrational free energy.

Alloy configurations can also contribute to the free

energy of crystals.

States of matter in the lab are governed by a competition between

energy (e.g. crystals),MatSci
entropy(e.g. liquids, gasses), and kinetic
193/203: Atomic Arrangements in Solids
25
effects (e.g.Stanford
glasses).
Evan Reed
University F=U-TS.
 Crystals versus Glasses
Bad glass formers (e.g. pure
metals, pair potentials) crystallize
rapidly upon cooling from liquid.

Good glass formers, e.g. SiO2 are slow to crystallize,

kinetically quenched. Glassy phase is usually metastable.

MatSci 193/203: Atomic Arrangements in Solids

Evan Reed 26
Stanford University
 Crystallization rates for materials

Y. Shen

Silica crystallizes slowly at ambient pressure, but quickly at high

pressure (e.g. meteorite impacts)
Pure metals (e.g. Ag) crystallize approaching the speed of sound
GST exhibits non-Arrhenius behavior, resulting in fast crystallization at
high T and slow crystallization at low T
This is useful for phase change applications 27
Determining how to predict materials that form
glasses is a frontier materials challenge

Y. Shen

28
Determining how to predict materials that form
glasses is a frontier materials challenge

29
 Density Functional Theory provides total
energies and materials structure prediction

Modern techniques for

computation of materials
properties using
quantum methods have
been developing over
many decades.

MatSci 331: Atom-based methods for materials modeling

Evan Reed
30
http://nobelprize.org/nobel_prizes/chemistry/laureates/1998/index.html

Stanford University
Formal DFT is based on two key developments
Citations of the first paper proposing
a practical version of DFT

MatSci 331: Atom-based methods for materials modeling

Evan Reed
31
Stanford University
Simulations can be used to study phenomena that cannot be done in
experiments

Nitromethane detonation?

Computational materials science is a growth

field, even without algorithmic and theoretical
improvements.

However, theoretical and algorithmic

improvements often enable more than
Materials predictions will
advancing computer power can.

become more pervasive with

http://en.wikipedia.org/wiki/Moore%27s_law
time.

MatSci 331: Atom-based methods for materials modeling

Evan Reed
32
Stanford University