0% found this document useful (0 votes)
101 views29 pages

Slides For A Course On Phase-Field Models Given in 2016, Session 1

This was the first session of a one-week course on phase-field modeling given in 2016 at the Indian Institute of Science, Bangalore, India

Uploaded by

Mathis Plapp
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
101 views29 pages

Slides For A Course On Phase-Field Models Given in 2016, Session 1

This was the first session of a one-week course on phase-field modeling given in 2016 at the Indian Institute of Science, Bangalore, India

Uploaded by

Mathis Plapp
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 29

Basics and application of

phase-field modeling in
materials science
1 : phase transitions, order
parameters
Mathis Plapp

Laboratoire de Physique de la Matière Condensée


CNRS/Ecole Polytechnique, 91128 Palaiseau, France
Branched growth structures
A snowflake

Ken Libbrecht, www.its.caltech.edu/~atomic/snowcrystals


Solidification microstructures

Hexagonal
cells (Sn-Pb)

Dendrites (Co-Cr)

Eutectic colonies Peritectic composite (Fe-Ni)


Interfaces: scales and descriptions
Macroscopic view:
Domains are separated
by sharp boundaries
Free-boundary problems

Mesoscopic view:
Fields are continuous but exhibit
interfaces with internal structure
Phase-field models

Microscopic view:
Matter consists of atoms
Each atom is considered individually
Atomistic methods
Course outline

Monday: Statistical Mechanics, Phase transitions, Ising model

Tuesday: Continuum description of the Ising ferromagnet

Wednesday: Phase separation in binary mixtures

Thursday: Phase-field models for solidification

Friday: Outlook on various aspects of phase-field modeling


Bottom-up approach:
from the Ising model
to « model A » (Allen-Cahn)
• J. S. Langer, « An introduction to the kinetics of first-order
phase transitions », in Solids far from equilibrium, Cambridge
university press (1991)
• N. Provatas and K. Elder, « Phase-field methods in materials
science and engineering », Wiley-VCH (2010)
• M. Plapp, « Phase-field models », in Multiphase microfluidics:
The diffuse interface method, CISM lectures 538, Springer (2012)
• M. Plapp, « Phase-field models », in Handbook of Crystal
Growth, Elsevier (2015)
Some statistical physics
• We will consider a system of fixed volume in contact with a
thermostat (constant temperature). The relevant statistical
ensemble is the canonical ensemble, the relevant thermodynamic
potential is the (Helmholtz) free energy F
• At equilibrium, the probabilities for all states obey the
Boltzmann distribution:

Pi  exp  Ei 
1 1

Z k BT

Z   exp  Ei  F  k B T ln Z
i
Magnetic domains

• In magnetic
materials, magnetic
moments can show a
cooperative behavior
• Can change
macroscopic behavior
(ferromagnetism)
The Ising model
• Toy model for
ferromagnetism
• N Atoms carry a single spin
that can have values ±1
• Spins interact only with
their nearest neighbors, and
with an external magnetic
field h

N
H   J  S i S j  h S i
i,j i 1
Simple : single spin
A single spin can have only two states: + and -

Z  exph   exp h   2 coshh 


Probabilities:
exp h  exp  h 
P  P 
Z Z
Magnetization:
1 
m  S   1P   1P  tanh h   log Z
 h
Paramagnetic behavior

 h 
m  tanh  
 k BT 

• Strong magnetic fields align the


spins
• High temperature creates
disorder (high entropy)
• Competition between energy
and entropy
Complete partition function:
you can try, but …
A system of N Ising spins has in total 2N microstates.
In the presence of interactions, the energies are not easy to
enumerate
→ the calculation of Z is difficult and complicated !!!!

One possible solution :

Numerical simulations with


the Monte Carlo method
The Metropolis algorithm
We want to evaluate a partition function (or an energy, or a
magnetization) : how to find the « important » configurations ?

• Choose an arbitrary initial state. Calculate its energy.


• Choose a possible move (for example, return a spin). Calculate
the change in energy DE between initial and final state.
• If DE is negative, accept the move
• If DE is positive, accept the move with a probability
P  exp DE
• In practice, draw a random number between 0 and 1, and
complete the move if it is smaller than P
• Repeat. After enough steps, the correct equilibrium is reached
Example: Ising model

kBT/J = 4
Example: Ising model

kBT/J = 1
Ising model: phase diagram

There is a phase transition between paramagnetic


and ferromagnetic state !
Interacting spins:
mean-field approximation
• A single spin in the average field created by the neighbors
• Each neighbor has a magnetization m, so the effective field is

h eff  h  zJm
• Self-consistent treatment: m must also be the result of

m  tanhheff   tanhh  zJm 


• For h=0, 3 solutions for T<Tc, 1 solution for T>Tc with

k B Tc  zJ
Alternative approach
Let us calculate directly the free energy as a function of m
F  U  TS
The internal energy U is the average of the Hamiltonian:

U  H   J  S iS j  h S i
i,j i

If we neglect the correlations between the spins, we get

Nz 2
U  J m  hNm
2
Entropy
Boltzmann : the thermodynamic entropy is given by the log
of the number of microstates

S  k B log W
N N!  1 m 
W     N   N 
 N   N  ! N  N  !  2 
Stirling’s formula, log N! ~ N log N for large N, yields

 1  m   1  m   1  m   1  m 
log W   N   log    log  
 2   2   2   2 
Free energy
Putting all parts together, we obtain

F zJ 2  1  m   1  m   1  m   1  m 
  m  hm  k B T  log    log  
N 2  2   2   2   2 

A Taylor expansion of all the log terms up to 4th order yields

F zJ 2  1 2 1 4
 f   m  hm  k B T  ln 2  m  m 
N 2  2 12 

f
kB
T  Tc m 
2 k BT 4
m  hm
2 12
Landau theory
The free energy can be written as

f  T  Tc m  m  hm , a, b  0
a 2 b 4
2 4

• Similar expressions can be written for all phase transitions


if the right order parameters are identified.
• The form of the polynomial depends only on the dimensionality
of the order parameter and the symmetries of the Hamiltonian
• Can predict singular behavior at the critical point
Critical behavior
The equilibrium magnetization minimizes the free energy:

f ' m  0  aT  Tc m  bm  h3

For h=0, solutions are m*=0 for T>Tc and

m 0
* m*

aTc  T
m 
*

b Tc T

for T<Tc
Magnetic susceptibility
The equilibrium magnetization depends on h !
For small magnetic fields, the equation can be linearized

m h   m 0   m 0 
h h
 
* * *

2aTc  T f ' ' m 0


*

The magnetic susceptibility is given by

m *
T  
1 1
 
h h0

2aTc  T  f ' ' m 0 
*

Inhomogeneous systems

 • The magnetization
varies between different
domains
• Idea: coarse-graining
• Need to take into
account interactions
between neighboring
cells
1
mI   Si
   i incellI
Gradient square term
• We suppose that ℓ is big enough to do thermodynamics
(entropy …)
• We need to re-evaluate the internal energy. For cell I:

U  J  I J
m 2

inner pairs
m m I
boundary pairs
J

• We use:

2 2
 m I m J  m I  m J  
1 m m

2 I J
2 2 2
Total free energy
J
F    f m I  
2
 mI  m J 2

I 2 I ,J
In the continuum limit: Ginzburg-Landau free energy
functional

K 
F    m   f m  dV
J
K
2

V 
2 

!! Attention !! Supposes that there is a scale hierarchy:

Lattice spacing << ℓ << scales of interest


Dynamics
• There is no conservation law or constraint attached to spins,
they can flip freely → the magnetization is non-conserved
• From linear response theory (or the master equation) we can
obtain that the rate of change is proportional to the
thermodynamic force, and the system evolves toward a
minimum of F

m
t
 
F
m

  K m  f m 
2

Allen-Cahn equation or model A
Summary
• The Ising model is a simple top model for the ferromagnetic
to paramagnetic transition
• Can be studied by computer simulations with the Monte
Carlo method
• Mean-field theory gives a qualitative explanation for the
phase transition
• Similar theories can be formulated for any phase transition
if the right order parameters are identified (Landau theory)
• For inhomogeneous (polydomain) systems, coarse-graining
yields a Ginzburg-Landau free energy functional with a square
gradient term and a double-well potential

You might also like