Rajkamal.

A
Research scholar,
Dept of physics,
Srm university.
19.11.2014.
 Phonon

 Phonon fundamentals
-Part 1
 Phonon calculation in quantum espresso
-Part 2
 Program files and results
-Part 3
 Concept introduced –IGOR TAMM –Russian
physicst (1932)
 Greek word –sound or voice,because long-
wavelength phonons give rise to sound.
 The study of phonons is an important part
of condensed matter physics.
 Major Role – physical properties such as
thermal conductivity and electrical
conductivity.
PHONON
 quantization of vibrational energy.
 quantized lattice waves (collective oscillations
of atoms)travels sound speed.

 If ώ is the frequency of a phonon.

 Energy is in integer multiples of ħώ
 a phonon is a collective excitation in a
periodic, elastic arrangement of atoms or
molecules in condensed matter, such as
solids and some liquids.

 A phonon is a quantum mechanical

description of an elementary vibrational
motion in which a lattice of atoms or
molecules uniformly oscillates at a single
frequency.
 The connection between frequency and
wave-vector, ω=ω’(k), is known as a
dispersion relation.

 frequency- ω
 Wave-vector -k,
 Velocity of sound- ω’
Dispersion relation
Answer for 2nd question.
 Due to this relation….

Some large number of k the wave length

approaches interatomic distance.

 So wavelength of the wave going through

the solid cant be smaller than
interatomic distance. So it has upper
cutoff.
This is how….
 Answer for 1st question

 Large number of the k and small value of the wavelength,

we expect the continuum of the picture breakdown.
 This is gonna help us to understand the deviation
relation of frequency and wave vector.
Newton force on atom

If we wanted to include more distant neighbour interactions (up to the

Nth), our equation at this point would have the form

Going back to nearest neighbours, try a solution of the form

 The continuum and discrete result are
identical at small value of the q.

 When q=π/a different. There is deviation

those value of q.

 Periodicity because of sin function, region

betn -π/a and π/a is called brillouine zone.
1.In the long wavelength limit (small k),

Linear dispersion = continuous description is applicable

speed of sound
Obervations- what happens ?
 This question gonna make us to add a
new feature to this dispersion
relationship.

 I,e 1d Mono atomic lattice to 1d

Diatomic lattice
What if two atoms have same masses.ie m1=m2

 What would be the dispersion relationship?

Case 2:
 m1=m2,but lattice parameter is 2a .it
means two atoms per basis we have 1d
lattice with 2a lattice parameter.

 What would be the dispersion relationship?

One dimensional Diatomic lattice
If the two types of atom have different masses, a gap
opens up between the two branches.
The equation of motion

With wave solution

One dimensional Diatomic lattice

The solution of the diatomic lattice is

When k=0,ω =0 in lower soln and upper soln is non zero.

When q=π/2a,we get different soln.

The solution of the diatomic lattice is

When k=0,ω =0 in lower soln and upper soln is non zero.

When q=π/2a,we get different soln.

 Why it is acoustic and optical mode?
How?
Acoustic branch :setting ω=0 in
this{mentioned} equation. This equation will
be satisfied only if A=B

Optical branch :setting ω= in ………………………….

this{mentioned} equation. ………………………….
We get ,
A1M1 + A2M2=0
Why its is acoustic and optical mode? How?

In the case of Acoustic branch

if A=B atoms are moving in the same direction. The Net Dipole Moment
does not change that much..

In the case of Optical branch

 A1M1 + B2M2=0 which represents atoms are moving in opposite direction.
So, the Net Dipole Moment changes drastically.
 If Net Dipole Moment oscillating quite violently ,this type of vibration pattern
exited by a electromagnetic wave.
 That why it is called Optical branch.
 Aim - find dispersion relation ie relationship betn
w and k.
- Density of states

 How to find frequency w?

DFT
 Phonon frozen method
 drawback large supercells are needed
 accurately calculate the force constant
matrix-
 cause problems
 large time of calculation.
 Frozen Phonons

• The forces between every atom construct the

force constant matrix in the crystal. This force
constant matrix then allows us to calculate the
normal modes of at any particular wavevector q.

• DFT to calculate the forces on every atom using

the Hellman-Feynman theorem.

• This method of calculating the force constant

matrix by explicitly displacing atoms is called the
Frozen-Phonon method.
 Frozen Phonons

Advantage
• Quicker and computationally cheaper than
the Linear Response method, which utilizes
DFPT to calculate forces.

Drawback
• Large super cells - needed to accurately
calculate the Force constant matrix.
• Boundary conditions used in DFT
calculations cause problems -frozen phonon
method.
Dynamical matrix
This is what DFPT…..

 Dynamical matrices are known - can generate easily IFC

using Fourier transformation and phonon frequencies
interpolated for any wave vectors.

Phonon density of states

Its same as the density of states in DFT.

 Einstein and Debye model

 Debye model almost same as

Discrete picture.

 Debye model gives very good

agreement in treating of
specific heat of solid.
Program files and results
 Simulation
The Multiple Phonon simulation in Quantum Espresso
consists of four steps:

 Step 1. PWSelf consistent calculation of electron density. Outputs are

wave functions.

 Step 2. PHPhonon calculation from linear response, with output on a

rough grid.

 Step 3. Q2RFourier transform to real space and obtain force

constants by interpolation.

 Step 4. MATDYNCalculation of all phonons from dynamical matrix,

given the force constants.
 Self consistent calculation
 &control
calculation='scf‘
 It calculates the total
………. energy of a periodic
………. crystal from its
outdir='/home/k equilibrium position.
/
&system
……….
……….
/
&electrons
………
/
ATOMIC_SPECIES
……..  This total energy help us
ATOMIC_POSITIONS to find IFC
……..

K_POINTS AUTOMATIC

444111
 Calculation of Dynamical matrices on a grid of q-
vectors (will require several minutes).

 Phonons of Si at X
&inputph
tr2_ph=1.0d-14,
amass(1)=28.0855,
prefix='SIPH',
ldisp=.true
outdir='/home/kathirvel/es
presso/tmp/'
fildyn='si.dynX',
/
4.0 4.0 4.0

Prompt> mpirun –np 8 ph.x < si.ph.in > si.ph.out

 Calculation of IFC's in real space.

 &input file containing dynamical

fildyn='si.dynX', matrix

zasr='simple', To impose acoustic sum rules

flfrc='si444.fc'
/
File containing the IFC's, and
a list of q-points for which the
frequencies are to be
calculated

prompt> mpirun –np 8 matdyn.x <input file> output file

 The calculation of Phonon dispersions
along selected high-symmetry lines
 &input Indicate type of Acoustic
asr='simple', Sum Rules used for the Born
amass(1)=58.6934, effective charges
flfrc='ni444.fc',
flfrq='ni.freq'
/ File where matdyn writes the
36 interpolated frequencies
0.0 0.0 0.0 0.0
0.1 0.0 0.0 0.0
………………..
High symmetry k points

prompt> matdyn.x <input file> output file

 The calculation of Phonon DOS

 &input Dos calculation, default set ‘true’

asr='simple',
dos=.true.
amass(1)=58.6934, DOS written file
flfrc='ni444.fc',
fldos='ni.phdos',
deltaE=1.d0, Energy level
nk1=6,nk2=6,nk3=6
/
The grid of q-vectors used to calculate the
DOS ( "nk1", "nk2", "nk3",using the
Monkhorst-Pack logic).

Prompt> matdyn.x < phdos.in > phdos.out

 Things you need to know………………………..

Self-consistent field Dynamical matrix

 Scf step as usual.  In the calculation of Dynamical matrix If

grid of q-vectors increases, program
 Prefix of the Scf and all ph,q2r,freq,phdos running time will also increases.(not to
should be same. specify more than 4).

 Note : relaxation prefix would be visible to

phonon calculation so, prefix of relax and Scf
must be different if you do sequence calculation  Form the output of Dynamical matrix
of Phonon. As per my experience it happened calculation (flfrc='grap.fc‘-contains force
when I did dynamical matrix calculation, It constant)we can calculate Acoustic Sum
took Energy values with ‘K’ points from relax Rule.
calculation not from Scf.
 In matdyn calculation we can proceed to two different types
of calculations which are mentioned below.

1.Phonon dispersions 2.Phonon DOS

 In this matdyn.in calculation we must to specify the list of q-
points for which the frequencies are to be calculated. These q
points can be generate from X-crysden which are in crystal co-
ordinates ,actual need is Cartesian co-ordinates in the units of
2pi/a.  If you choose option "dos=.true." you
must specify the grid of q-vectors
 To get this Cartesian co-ordinates we calculated band structure used to calculate the DOS.
for given atom using high symmetry points in crystal co-
ordinates which is generated from X-crysden. From the output
file of band structure calculation,we could extract Cartesian co-
ordinates which is in the units of 2pi/a.

 we can use these co-ordinates as a high symmetry q points for

which the frequencies are to be calculated( matdyn.in).

 The file selected in flfrc, "si.freq", contains a list of frequencies

in a format that can be further processed by another auxiliary
code, "plotband.x", the same used for band structure plotting.
 Phonon calculation for Graphene

Phonon dispersion Phonon density of states

 Phonon calculation for Silicon

Phonon dispersion Phonon density of states