Chapter 9

Solid state phase transformation

9.1 Pure metals

Solid
(metals)

Allotropy

crystal structure with lowest Gibbs free energy

Eo (binding energy) depends little on atomic arrangement
eg. Na bcc changes to hexagonal at 36 K , heat of fusion = Eo/1000
bonding depends on electronic structure.
- change in crystal structure.

If the same crystal structure exist in different temperature regimes some physical properties are continuous
between two regimes. The thermal expansion coefficient of iron fig. 9.1.
9.2 Alloys
9.2.1 Diffusion controlled phase transformation
9.2.1.1 General classification
From fig. 9.2 there are several reactions concerning solid
state phase transformation.
1. Dissolution or precipitation of a second phase on
crossing of one phase boundary,
+ (a-b, a-b)
2. Transformation of a crystal structure into another crystal
structure of the same composition on crossing two phase
boundaries:
(c-d, c-d)
3. Decomposition of a phase into several new phases on
crossing 3 phase boundaries:
+ (e-f, e-f)
A special case is eutectiod decomposition (e-f)
1

At a given temperature and pressure:

G = H - TS
H the enthalpy binding energies between next
neighbor atoms.
HAA, HBB , HAB
Hm the total binding enthalpy or the enthalpy of mixing
Hm = NAAHAA + NBBHBB + NAB HAB
Nij total number of bonds between atom i and j
(i = A, B : j = A, B)
- Hij or Hm is positive
-The enthalpy in this model is not the gain of enthalpy
( H < 0 when gain enthalpy) of an atom during its
transition from the free to the bonded state.
-It is the difference between the bonded state and a
reference state with absolute minimum enthalpy.
-A small value of H refers to a more strongly bonded state
than a large enthalpy.
9.2.1.2 Thermodynamics of decomposition
Consider: + or is formed from homogeneous
phase
This reaction can be
1. has the same crystal structure as but with different
composition (decomposition)
2. has a crystal structure and composition different from
. This is the case of precipitation.

If N the total number of atoms

z - the number of next neighbor atoms (coordination no)
c - the concentration of B-atoms
N.c = no. of B atoms
z.c = each B-atom has B-atom as next neighbors
Atomic concentration of B-atom
NBB = Nzc2
2
means all B-atom in B-B bonds are counted twice.

The same as NAA and NBB

Hm = N.z[(1-c)HAA+cHBB+2c(1-c)H0]
H0 = HAB (HAA + HBB)
H0 - exchange energy: H < 0 gained,
H > 0 lost
If AB bonds are generated from AA and BB-bond
H0 - 0 ideal solution binding enthalpy does not
depend on the arrangement of atoms.
S
Sv
+
Sm
(the vibration entropy)
(the entropy of mixing)
Sv - nearly independent of the atomic arrangement
S
Sm
m

Sm
9.4
= k ln m
9.5a
- the number of possible distinctly different
arrangement
Sm = -Nk [c ln c + (1-c) ln (1-c)]
9.6
Sm - always positive
c = 0.5 symmetric
c = 0 , c =1 infinite slope means
To obtain pure elements from alloys is very difficult
c 0, dG infinitely (large)
From eq. 9.3 and 9.6; the free energy of a regular
solution
Gm (Gibbs free energy of mixing) =.
9.7

Fig. 9.3 shows enthalpy, entropy and the Gibbs free energy
There are 2 different cases
1. H0 < 0, Gm (c) has a single minimum
2. H0 > 0, Gm (c) has 2 minima (at sufficiently low temp)
at c1 and c2
- The entropy of mixing approaches infinity when A and B
become pure elements.
- The change of enthalpy remains finite for all compositions.
- The free energy will always decrease for small
concentrations.
for large concentration: H > TS
for small concentration: H < TS

- In the range c1 < c < c2 the system can reduce its free
energy if it decomposes into the mixture of 2 phases
with composition c1 and c2
- The free energy of the mixture the common tangent
that intersects the G (c) curve at c1 and c2
- The points of tangency are usually not the minima of
the Gm (c) curve.
- If the tangent is a horizontal line Gm (c1) = Gm (c2) ,
the points of tangency are minima.
-If H0 < 0 , - the atoms of each element favor unlike
partners as next neighbors
- the long-range ordering is preferred over
decomposition.
- For c < c1 and c > c2;
- the solid solution has always a lower free
energy than a phase mixture.
c1 and c2 - solubility limits of the solid solution
c1(T) and c2 (T) - solvus line in binary phase system
A-B
- For the most simple case; HAA = HBB
Gm is symmetry c = 0.5
c1 and c2 (c2 = 1-c1) minima

From eq. 9.10 composition of phase depends on

temperature
An example of the phase diagrams are shown in fig. 9.4
and 9.5. which corresponds to the terminal solubility of
solid solution in phase diagram
Fig. 9.4 - the maximum of the curve is not attained
because the alloy becomes liquid at a lower temperature.
depending on the behavior solidus temperature
(monotonic or with a minimum)
(or a peritectic or eutectic)

The common tangent intersect Gm (c) at minima and

the solubility limit c1 (T) differential eq. 9.7.
d Gm/dc |c=c1 = 0
9.9

4
c1 (T)

exp (- zH0/kT)

9.10

If the two phases and have different crystal

structures, there is no common curve Gm (c).
There are separate curves for each phase G (c) and
G (c) that intersect at a certain concentration.
- Since only the phase or phase mixture with the
smallest free energy is thermodynamically stable,
decomposition occurs.
a homogeneous c c1, c c2
a phase mixture of and for c1 < c < c2
c1 and c2 points of intersection of the G(c)-curves
with a common tangent fig. 9.6a.
9.2.1.3 Nucleation and spinodal decomposition
The process of decomposition can occur either by
-A nucleation process: a nucleus is formed with
equilibrium concentration c2
- Spontaneous decomposition (spinodal decomposition)
Fig. 9.7
-Alloys composition c1 decomposes into c1 and c1
The free energy of phase mixture GE tie line between
G (c1) and G (c1)
GE (c1) G1E > G (c1) G1
Such a decomposition is not stable
The system would return to the stable of a
homogeneous solid solution.

-Alloys composition c3 near the maximum of the

G (c) curve.
decomposition
can proceed spontaneously (free energy gain)
spinodal decomposition
Spinodal decomposition
- diffusion flux runs from low to high concentrations in
contrast to normal diffusion.
- negative diffusion coefficient, owing to a negative
thermodynamic factor.
Fig. 9.8 shows the difference between spinodal
decomposition and regular nucleation.
Nucleation
- a nucleus of is formed by thermal fluctuation with
correct concentration c
- grows by diffusion while the composition remains
unchanged (still c )
- the size of nucleus has to exceed the critical size for
stable growth.
- nucleus is preceded by an incubation time
Spinodal decomposition
- proceeds spontaneously without incubation time
- decomposition is amplified until equilibrium is attained

Nucleation and decomposition

-The terminal states are the same but intermediate
morphologies of the phase mixture are totally different.
Differences
During spinodal decomposition a periodic wave-like
pattern develop during the early stage.
the wave length of spinodal decomposed structures
is usually small ( ~ 50 A in Al-37Zn at 100o C, fig. 9.9)
Whereas during nucleation the shape of the
precipitated phase is determined by interfacial energy
and elastic energy.
Decomposition is possible if the curve has a concave
curvature d2G/dc2 0
d2G/dc2 = 0
The point of inflection in fig. 9.7, c and c and can
be calculated from second derivative of eq. 9.7.
The dependency c (T) defines the spinodal curve.
For the simple case HAA = HBB
c . ( 1 - c ) = kT/ 2zH0
9.11
Fig. 9.10
In the 2 phase regime: there are 2 regions.
1. The center spinodal decomposition can occur.
2. The region between the spinodal curve and the
solubility limits on both sides precipitation can occur by
nucleation and nucleus growth.

In solid: the nucleation of precipitates depends on

- The differences of volumes of mother phase and the
precipitates.
- The difference has to be compensated by elastic
distortion with the elastic energy (Eel).
- The elastic energy increases with increasing nucleus
volume V
Eel = el .V
el distortion energy per unit volume = elastic energy
density

If a spherical nucleus with radius r is formed, the free

energy increase.
G (r) = (-g +el ) . 4/3 r3 + 4r2

9.12

r the specific energy of the - phase boundary

g free energy gain per unit volume during phase
transformation. (=Gliq-Gsolid /V)
The maximum of the curve G (r) the critical size of
nucleus (r0)
r0 = 2 / (gu - el )
9.13
The energy of the interface boundary and the elastic
distortion energy play a major role since the small
changes of r0 result in large changes of the nucleation
rate.
The elastic distortion energy (per unit volume) for a
hard precipitate in a soft matrix can be calculated.
el = {(E 2)/(1-( })c c )2 . (c/b)
9.14
c , c - concentrations
E - Youngs modulus of
- Poisson ratio of
- d (ln a)/dc = the atomic size factor
(a, c lattice parameter)
- distortion
- the form factor

If the precipitate has elliptic shape,

c ellipsoidal radius parallel to the rotation axis
b the respective perpendicular ellipsoidal radius.
The relation between shape factor and c/b shows in fig.9.11
The precipitates:
Sphere distortion energy is the largest
- has the smallest surface for a given volume
- the shape of a precipitate is a compromise
between a minimum of distortion energy and
total surface energy.
1) If the precipitate and matrix have the same lattice
parameter, ( ~ 0. eq. 9.14) as in the system Al-Ag:
- the elastic energy play a negligible role and the
precipitates have the spherical shape.
2) If a precipitate and matrix have a large different crystal
lattice parameter, >> 0.
- plate-shaped precipitates are preferred as in the system
Al-Cu.
3) If the interfacial energy is anisotropic: different interfacial
plane have different energy.
- the area of the interface and its specific energy plays a
role, the plate-shaped precipitates are preferred.
- in system Al-Ag
In the beginning, the spherical precipitates, semicoherent
interface with large energy anisotropy, are formed.
During annealing, the precipitates become plate-shaped.

9.2.1.4 Metastable phases

The nucleation rate ~ exp (-G/RT)
G0 = G(n) =

9.15

(16/3) 3 /(gu - el )2

- strongly depends on interface energy

- depends on structure
There are 3 types of interface, coherent, partially
coherent and incoherent, fig. 9.12.
Coherent precipitates
- all lattice planes of matrix continue through the
precipitate
- there will be slight elastic distortions because of
differences of the lattice parameter in precipitate and
matrix.

Partially coherent interface

- the difference in lattice parameter between precipitates
and matrix is too large
- the edge dislocations are formed at the interface to
compensate the elastic distortions
- most lattice planes of the matrix are continued through the
precipitate but a few terminate in the interface and create the
interfacial edge dislocations.
Incoherent interface
- crystal structure of precipitate and matrix are different
- large interfacial energy the work of nucleation is high,
strongly delays nucleation especially at low temperatures
where diffusion proceeds slowly.
In this case metastable phases are formed because the
gain free energy is not as large as for the equilibrium phase
( phase, fig. 9.13)
metastable phase The gain free energy is small
coherent and partial coherent the energy of interface
boundary is low
9
high nucleation rate

- A metastable phase is thermodynamically unstable, its

free energy is higher than that of the equilibrium phase
- Frequently, several metastable phases are
successively formed until the equilibrium phase is
attained. Consecutive metastable phases have a
continuously lower free energy, fig 9.13.
- The first phases formed are coherent phases (or
decomposition zones) with a size of a few atomic
layers called Guinier-Prestion zone
- Guinier-Prestion zone decreases with increasing
temperature until above a critical temperature, further
decomposition zones stop and incoherent phases
appear, fig. 9.14.
- At low temperatures, diffusion effectively stops and
decomposition cannot take place.
There is a certain temperature range in which these
metastable phases exist. This is the temperature
range used for the thermal treatment of age
hardened alloys

9.2.1.5 Age hardening

The principle of age hardening
precipitate the second phase during annealing of a
supersaturated solid solution.
Al-Cu, fig. 9.15
- max solubility of Cu in Al is 5.65% at 548oC
- T < 300oC the solubility decreases to less than 1%
- Composition used for age hardening in fig. 9.15 b
at high temp
+ at low temp - intermediate phase Al2Cu
- tetragonal structure
fcc
- single phase, quenched supersaturated
heat increase strength, fig. 9.16

10

Fig. 9.16.
At low annealing temperatures (100o)
- hardness increases slowly but continuously until a
plateau is reached.
At slightly higher temperatures
- another rise of hardness is observed after the plateau.
- the second hardness approach maximum
At even higher temperatures
- the plateau at the maximum will be reached earlier but
the value of the plateau hardness decreases.
At 300oC
- the plateau is not obtained anymore, only the second
hardness occurs.
During heat treatment: the following phases may occur.
- GP I zones are single layers of copper atoms on {100}
aluminium lattice planes.
- GP II zones ( phase) are comprised of several parallel
copper layers on {100} planes cause a
tetragonal distortion of the second of the matrix fig.
9.17 and 9.18.
- intermediate phase
(CaF2 structure, coherent with the matrix on {100}
planes
- the equilibrium phase

11

The first hardening step

The plateau
- the formation of GP I and GP II zones
- GP zones are decomposition products. The number of
GP zones decreases with rising temperature.
- the hardness of the plateau decreases at higher
temperature.
The second hardening stage is obtained only at higher
temperatures
- artificial aging
- caused by the formation of the phase. Due to
precipitate coarsening at longer annealing times. The
precipitate spacing increases and the strength decreases,
the precipitate circumvented by dislocation.
Overaging
- the equilibrium phase

12

9.2.1.6 Growth kinetics of precipitates

The growth kinetics of nucleus are diffusion controlled
In binary alloy AB
c0 - the initial concentration
B the average concentration of the matrix,
changes with time
r0
jB - flux of B atom
r0 - precipitate size
R
4/3 r3 . d B/dt = jB (r0) . 4 r02

9.16

The precipitated volume fraction is

c c
c conc _ of _ matrix
X (t ) 0 B' 0
c0 cB c0 conc _ of _ particle
For a short time
X(t) = (2t/3)3/2

9.20

9.21

After a long time

X(t) = 1 2exp( t/)

R the average matrix volume per precipitate, a sphere

For steady state growth of the precipitates: the
concentration profile in front of the particle remains
unchanged stationary diffusion eq.
cB (r) = c0 (c0 cB) . (r0 /r)
cB is the equilibrium concentration at r = r0
From Ficks law
jB (r0) = - DB (c0 - cB) / r0
For the conservation of the no. of B atoms
{Conc change in matrix = conc of precipitate}
4/3 r3 (c0 - B ) = 4/3 ck r03
ck is the concentration in the precipitate

9.17

9.18

9.19

- the time constant, determined by the diffusion

constant
1/ = 3 DB (c0 cB)1/2 / ck1/3. R2

9.23

For a long annealing times

r0 ~ DB . t
a ~ DB . t
X ~ r03
X ~ t3/2
Fig. 9.21 shows good agreement with experiments,
(elastic distortion was ignored). In some cases, less
agreement, the growth kinetics can be controlled by
mechanism other than the atomic transport. (atoms B
moving across phase boundary)
13

For spherical particles

Chemical potential is determined by the curvature of the
surface.
The chemical potential difference between 2 particles with
different r1 and r2
= 2 (1/r1 1/r2 ) = kT (cB/B)

During annealing
- small precipitates are dissolved.
- large precipitates continue to grow Ostwald ripening
The driving force decrease of the total interface
boundary energy.
The total energy smallest, small no. of big particles
instead of the large no. of small
particles.
- the precipitate is to dissolve or to grow depends on the
chemical potential of atoms in the vicinity of a precipitate.
- there is always a flux of atoms from the higher to lower
chemical potential.
- In equilibrium, the chemical potential is constant
everywhere.

9.24

B - equilibrium concentration of the matrix for a flat

interface boundary (r )
- interface boundary energy
- atomic volume
The chemical potential difference is related to a
concentration gradient.
c = c(r1) c(r2) fig. 9.22 which cause diffusion
flux from the small to the large particle. The large particle
grows while the small particle dissolves. The average
particle size,
3 - 03 ~ DB t

9.25

Fig. 9.22
- This model, elastic distortions were not accounted for.
- If the precipitates are plates or needles, the shape
factor needed to be considered.
- r ~ t1/3 is observed. For each edge of a cuboidal particle
although with different kinetic coefficients (DB) for
different spatial directions.
-During aging: max hardness and overaging are due
14 to
Ostwald ripening.

9.2.1.7 Eutectoid decomposition and discontinuous

precipitate
Solid state phase transformation proceeds by motion of
reaction front
Eutectoid: +
and - phase transformation proceed by motion of a
reaction front.
- occur simultaneously
- different compositions, possibly different
crystal structure
- lamellar structure, atoms diffuse in a short
range
- i.e. + Fe3C in steel

Eutectiod, fig. 9.24

- initiate at grain boundary
- if phase forms first and enrich in B, therefore, its
immediate environment is enriched in A (but deplete in
B) nucleus is generated.
- lamellar structure formed.
- lamellar spacing decreases with increasing
15
transformation rate, fig. 9.23.

Discontinuous precipitation
+
-Nucleation begins at grain boundary
- precipitation is not homogeneous in the grain interior. It
occurs at a few preferred locations, mostly at grain
boundaries discontinuous precipitation.
- connect to grain boundary migration
- diffusion proceeds faster in a grain boundary. Behind
the moving grain boundary the new phase is
precipitated
- plastic deformation can further increase the grain
boundary migration in addition to the chemical driving
force for phase transformation.
9.2.2 Martensitic transformations
- the spontaneous phase transformations without
concentration change.
- in Fe-C system (ferrite is suppressed)
- If the spontaneous change of the crystal structure is
associated with a volume change or change of shape
(which may be further complicated by the suppressed
change of composition, transformation proceeds by
successive displacive transformation of) needle or lath
shape (regions into the new crystal structure) will be
formed.
- Consecutively transformed regions will be limited in
their spatial extent by prior transformed regions.
- The volume fraction of martensite phase usually does
not depend on time but only on the temperature to which
the material was quenched.

-Volume fraction of martensite increases with decreasing

temperature, fig. 9.27.
-Ms (1% martensite volume fraction)
-Mf (99% martentsite volume fraction)
-Ms Mf depends on composition, fig. 9.28
-Ms decreases with increasing concentraion.
-On heating, martensite does not transform at Ms into fcc
austenite but superheating to As (austenite start).
As > Ms and As depends on composition, fig. 9.28
- In special case of Fe-C, martensite usually
decomposes on annealing into ferrite and cementite
16
instead of retransforming to austenite.

-By plastic deformation the difference between Ms and

As can be reduced.
-Assume T0 the thermodynamic equilibrium temp for
both phase (same free energy).
T0 = (Ms + As)/2

Martensite transformation in Fe-C system was given by

Bain, fig 9.29.
- pure iron: fcc bcc
- but with carbon: fcc tetragonal
- a0 lattice parameter of fcc
- the center of 2 next neighbor fcc unit cell contains a
tetragonal body-centered unit cell (bct) with lattice
parameter
a = a0/2 and c = a0
- to obtain bcc, a bct has to be compressed in c direction
and stretched in the plane perpendicular to c direction, by
which distortion the volume will be change by about 3-5%.
-This martensite transformation model, the atomic
positions do not change much and the next neighbors
remain next neighbors after phase transformation.
- Fe-C system: martensite transformation fcc bct
- c/a ratio in martensite increases with increasing carbon
concentration whereas the magnitude of the a-axis
independent of carbon content (or decreases slightly), fig.
9.30
-In fcc- phase a0 increases with increasing carbon
content. C in fcc located the octahedral interstitial sites of
iron lattice (which are the center of the cube or on the
center of its edges).
-- After transformation, C are located only on c-axis.
C
atoms are larger than the octahedral interstitial sites. This
leads to a slight increase of the lattice parameter of fcc in
c direction.

17

-Because of all C atoms are in c direction of martensite

c direction increases. Tetragonal martensite is formed.
-During tempering of martensite, C atom is allowed to
diffuse precipitate ; bcc martensite or ferrite is obtained.
-Solubility of C in is much larger than in
bct martensite is a strongly supersaturated solid
solution.
Tetragonal distortion reduces the dislocation mobility
(increase Peierls stress)
-The martensite transformation causes a substantial
plastic deformation besides the change of lattice structure.
- the compression and stretching of unit cell shear
deformation a shape change of the transformed region
and form relief on the surface, fig 9.31.
- the shape changes cause large elastic strain to the
immediate environment of the martensite which can be
reduced by plastic deformation within the martensite via
glide and twinning, fig 9.31 and 9.32.

18