Metals

V. Transformation in the solid state

J. McCord

(mostly from Gottstein)

Pure metals allotropic modifications

Crystal structure of a metal is not necessarily stable at all temperatures
Crystal structure determined by the lowest Gibbs free energy G
Binding energy E0 in metals relatively constant
Differences in electronic structure responsible for instability in crystal
structure
Phase transitions

Allotropic modifications
Several phase transformations in the solid state possible
Example thermal expansion coefficient of iron

5.2

Allotropic modifications

Fig. 5.1 Allotropic modifications of selected elements

5.3

bcc

Polymorphism of Iron
bcc

fcc
bcc
bcc

fcc

bcc

Fig. 5.3 Temperature dependence of the lattice

constant of iron (note the different scale on the
left and right side).
Fig. 5.2 Cooling curve for pure iron
(note F).

Similar temperature dependence for same structures

Also phase transitions of different origin magnetic Fe transition
5.4

Diffusion controlled phase transformations

in alloys

Dissolution or precipitation of a second

phase on crossing of one phase boundary
+ (e.g. a' - b')

Transformation of a crystal structure

into another crystal structure of the same
composition on crossing two phase
boundaries
(e.g. c - d, c - d')
Decomposition of a phase into several
new phases on crossing (up to three)
phase boundaries
+ (e.g. e - f, e' - f')
Eutectoid decomposition
(e" - f")
Fig. 5.4 Phase diagram of a binary alloy AB with phase
5.5
transitions in the solid state.

Thermodynamics of decomposition
Formation of new phase is from a homogeneous phase

Fig. 5.5 Phase diagram of a binary alloy AB with phase

transitions in the solid state.

Two possibilities depending on

Decomposition
has the same crystal structure as , but has a different composition

Precipitation
has a crystal structure and a composition different from
5.6

Quasi chemical model (mostly enthalpy H)

Gibbs free energy

G = H - TS

Enthalpy H given by the binding enthalpies

HAA, HBB, HAB

between next neighbor

AA-, BB-, or AB-atoms

Total binding enthalpy

Hm = NAAHAA + NBBHBB + NABHAB

Nij is the total number of bonds between atoms i and j (i = A,B and j = A,B)

Nz = total number of atoms coordination number

1
2
Hm = N z (1 c ) H AA + 2c (1 c ) H AB + c 2HBB

2
1
= N z (1 c ) H AA + cHBB + 2c (1 c ) H0
1
2
H0 = H AB [H AA + HBB ]
2
Exchange energy H0 < 0 decrease in energy with AB
Exchange energy H0 > 0 increase in energy with AB
Ideal solution with H0 = 0

5.7

Entropy S = Sv + Sm Sm
Vibrational entropy Sv (vibrational modes =
independent of atomic arrangement)

Entropy of mixing Sm

Mixing entropy Sm

Possible arrangements m

N!
N A ! NB !

Fig 5.6 Entropy of mixing for ideal solution. Note

that S approaches for c 0 and c 1 with +/infinite slope. As a result, it is impossible to
produce perfectly pure crystals.

Mixing entropy Sm = k ln m

m =

Sm / c

...

Sm = Nk c ln c + (1 c ) ln(1 c )
Sm > 0
Symmetric with regard to c = 0.5
Approaches c = 0 and c = 1 with infinite slope
5.8

Free energy of solid solution

Gm =

1
Nz (1 c ) H AA + cHBB + 2c (1 c ) H0 + NkT c ln c + (1 c ) ln (1 c )
2

dG dG
=
=0
dc1 dc2
dG dG
=
0
dc1 dc2

Fig 5.7 Concentration dependence of free energy of a solid

solution for different heat of fusion H0.

Rule of common tangent

Gg = Gm ( c1 ) +

c c1
Gm ( c2 ) + Gm ( c1 ) )
(
c2 c1
5.9

Phase diagram with a miscibility gap

zH
c1 (T ) exp 0
kT

Eutectic phase diagram

solubility limit

Tc

c1(T)

Fig 5.8 Theoretical phase diagram of a solid solution with a

miscibility gap.

Depending on the behavior of the solidus temperature

Eutectic phase diagram (shown)
Peritectic phase diagram

5.10

The Au-Ni phase

diagram

Phase diagram of a
system with limited
solubility
Miscibility gap

temperature (C)

High solidus temperature

liquid

solid solution

Au-rich solid solution +

Ni-rich solid solution

Characteristics
Common G(c) curve for
both components
weight% Ni
Fig 5.9 Phase diagram of AuNi.

5.11

System with two different phases and

atomic% Zr

temperature (C)

(b)

(a)

c1

c2
weight% Zr

Fig 5.10 (a) Composition dependence of the free energy in a system with two different
phases and and (b) NbZr phase diagram

Two phases and with different crystal structures

Separate G(c) curves for each phase G(c) and G(c)
Homogeneous or solid solution for c < c1 or c > c2
Phase mixture of and for c1 < c < c2

5.12

Nucleation and spinodal decomposition

Process of decomposition

Decomposition would
lead to an increase of
free energy

Decomposition leads
to a gain of energy
(spontaneous
decomposition)

Fig 5.11 Connection of

decomposition of a solution
with its free energy. The
curve's points of inflection
are at cw and cw.

by a nucleation process
Formation of nucleus is with equilibrium concentration of new phase
by spontaneous or spinodal decomposition
Equilibrium composition attained in the course of time
Up-hill diffusion

5.13

 in more detail
d 2G
0 Spinodal decomposition
2
dc

d 2G
0
dc 2

d 2G
0
dc 2

d 2G
0 Nucleation and growth
2
dc

Fig 5.14 Schematic G(x) curve for a fixed

temperature for the discussion.
5.14

Nucleation vs. spinodal decomposition

Nucleus + diffusion

Equal terminal state

Nucleation
(a)
and
growth
Distance

(b)

Spinodal
decomposition

Continuous process

Distance
Fig 5.12 Schematic diagram of the change in concentration and dimensions
during decomposition of a solution by (a) nucleation and growth and (b)
spinodal decomposition.

5.15

Concentration
profiles f(t)
(a)

8 min

Developing of periodic wave-like

pattern
Wave length of spinodally
decomposed structures is small!
Fine lamellae composites
(b)

Fig 5.13 Numerically calculated concentration

profiles for spinodally decomposed Al-37%Zn
aged at 100C for (a) 8 min., (b) 15 min., and
(c) 23 min. The dashed lines indicate the
equilibrium concentration of decomposition
(c)

15 min

23 min

5.16

Range of spinodal
decomposition
d 2G
0 Spinodal decomposition
2
dc
spinodal line

Ta

d G
0 Nucleation and growth
2
dc

H AA = HBB
Spinodal curve c(T)
cw (1 cw ) =

G(Ta)

kT
2zH0

Coherent spinodal line in reality deeper

due to elastic energy contributions.
Fig 5.14 Schematic phase diagram and free
energy curve of a solid solution when a
miscibility gap occurs. G(Ta) = free energy at
aging temperature Ta.

cw

5.17

Range of
nucleation and
growth

Region between the spinodal

curve and the solubility limit on
both sides

T
Ta

Undercooling T
Occurrence of precipitation by
1. Nucleation
2. Growth

Range of
nucleation
and growth

G(Ta)

Fig 5.15 Schematic phase diagram and free

energy curve of a solid solution when a
miscibility gap occurs. G(Ta) = free energy at
aging temperature T (same as 5.14). The
typical
decomposition
procedure
is
illustrated. The sample is quenched from the
solid solution range () to a certain
temperature T below the solubility limit into
the two-phase region ( + ).
5.18

Decomposition
procedure
T

~ r2

Change of free energy G due to the formation

of a spherical nucleus with radius r

G( r ) = ( g t el )

4 3
r + 4 r 2
3

Gc
rc

specific interfacial energy of the phase

boundary
gt free energy gain per unit volume during
phase transformation
el distortion energy per unit volume or elastic
energy density
3

G
=0
r

Gc =

rc , Gc

Critical radius rc =

16

~ r3

3( g t el )2

2
g t el

Fig 5.16 Dependence of the

free energy of a spherical
nucleus on its radius.
5.19

Elastic distortion energy of a precipitate

Hard precipitate in a soft matrix
high el

shape factor

Fig 5.17 Elastic shape

factor for a rotation
ellipsoid with aspect
ratio c/b.

sphere

E 2
c
el =
(c c )
1
b
needle

disc
low el

aspect ratio c/b

Concentrations c and c - composition of matrix and precipitate

Young's modulus E and Poisson ratio of the phase
Atomic size factor (distortion depending on difference of lattice parameter with c)
Form factor
5.20

Structure and shape of phase boundaries

(a)

(b)

(c)

Fig 5.18 Structure of (a) coherent, (b) semi-coherent, and (c) incoherent phase boundaries.

Coherent phase boundaries (low - high el)

Lattice planes of the matrix continue through the precipitate
Elastic distortions
Semi-coherent phase boundaries (low - lowering of el)
Significant difference in lattice parameter
Interfacial edge dislocations in the interface compensate the elastic
distortions
Most lattice planes of the matrix continued through the
5.21

Structure and shape of phase boundaries

(a)

(b)

(c)

Fig 5.19 Structure of phase boundaries (a) coherent, (b) semi-coherent, and (c) incoherent.

Incoherent phase boundary (large - no el)

Crystal structure of both phases is different
Orientations of matrix and precipitate are different
Different crystal structure
Incoherent interphase boundary
Large interfacial energy hinders nucleation - large rc
5.22

Nucleation and growth

through metastable
states

with

Gc if3

Gc
N exp

kT

Increasing nucleation rate by

lowering Gc

free energy

Nucleation rate

solid-solution (
)

123456
G1 > G2 > G3 > G4 > G5 > G6
Formation of thermodynamically
unstable metastable states
Guinier-Preston Zones GP-I, GP-II
-phases

concentration
Fig 5.20 Schematic diagram of the free energy change for
5.23
successive metastable phases.

Existence of GP zones
Limited temperature range for the
existence of coherent GP Zones

Density N of GP zones

Limited to lower temperatures

Elimination of zones through thermal

activation
Number of Guinier-Preston zones
decreases with rising temperatures

Above a critical temperature only

incoherent phases appear

Fig 5.21 Temperature dependence of the equilibrium

density N of Guinier-Preston zones (existence curve). Due
to insufficient diffusion, the thermodynamic equilibrium
cannot be reached at low temperatures.

5.24

Age hardening in Al-Cu

Age hardening process for strengthening of alloys
Age hardened aluminum alloys
Precipitation of a second phase during annealing of a supersaturated solid
solution
(a)

(b)

Fig 5.22 (a) Al-Cu phase diagram and (b) detail of the Al-Cu phase
diagram (Al-rich side) indicating how the annealing takes place for
age hardening.
5.25

(a)

Characteristic plateaus
Increase in hardness

Formation of coherent phases

GP-I - single Cu layers on {100} Al
lattice planes
GP-II or several Cu layers on
{100} Al lattice planes

Further increase in hardness

Formation of partially coherent
AlCu-phase ( )
Formation of incoherent Al2Cu
( ) not happening

Hardness Hb

Age hardening
curves
GP-I & II

time (days)

(b)

Fig 5.23 (a) Age hardening curves

of Al-4%Cu-1%Mg. (b) Section
through a GP-I zone in AlCu.

Decrease in hardness
(increase in precipitation size)

5.26

equilibrium phase

Phases in Al-Cu

FCC Al phase

(a)

(b)

(c)

GP-II,

(d)

Fig 5.24 Crystal lattice of the

solid
solution
(a),
the
equilibrium phase (b), as
well as the metastable phases
' (c) and (d) " of the Al-Cu
system (Cu , Al )
5.27

Annealing for age hardening of Al-Cu

a

c (see. Fig. 5.22)

b

Fig 5.25 Annealing sequence for age hardening with corresponding change in
microstructure for Al-Cu.
5.28

Time-temperature-transformation - TTT diagram

Supersaturated solid solution

TTT diagram for

AlCu

Fig 5.26 Sketch of a TTT

diagram. The time t
required to reach a
specified state of precipitation (e.g., a given
volume fraction) is determined in an isothermal
experiment. In order to
prevent precipitation the
cooling rate for quenching
has to be large enough to
avoid an intersection of the
T(t) curve and the knee
describing the onset of
precipitation.

5.29

Growth kinetics of precipitates

Short times (X < 0.2)

r DB t

3
X( t ) r

X( t ) t 3 2

r : precipitate radius (assumption of spherical particle)

X : precipitate volume fraction

Long times (X > 0.9)

X( t ) = 1 2e t /

Neighboring precipitates compete for remaining solute atoms

Fig 5.27 Schematic diagram of the change in concentration and dimensions

during decomposition of a solution by nucleation and growth.

5.30

X( t ) = 1 2e t /

Growth kinetics of
precipitates

1 X = 2e t /
1 3DB ( c0 cB )
=

cK1/ 3R 2

1/ 3

DB

(b)
X( t ) = 1 2e t /
X( t ) t 3 2
(a)
incubation
time &
nucleation

Ostwald
ripening
growth

Fig 5.28 (a) Sketch of volume fraction of precipitates as function of time at constant temperature. Note that
nucleation gives rise to an incubation time. (b) Growth kinetics of C in -Fe. Solid line - precise theory;
dashed line - dependence according to the shown equation fro X(t) (ck concentration in the precipitate).
5.31

Ostwald ripening
Competitive" precipitate coarsening
Driving force is a decrease of the total interface boundary energy
Flux of atoms from the higher to lower chemical potential (from small
to large particles)
Spherical particles: determined by the curvature of the surface
1 1
c
= 2 = kT B
c B
r1 r2
Radi r1 and r2
b equilibrium concentration for a flat
interface boundary (r infinity)
atomic volume

Average particle size

DB if cB 2
r r
t DB t
kT
3

diffusion

3
0

Fig 5.29 The contribution of the interfacial energy leads to a chemical potential difference between
particles of different size. This causes a solute diffusion current from small to large particles.
5.32

Eutectoid decomposition and discontinuous

precipitation

Precipitation processes by nucleation and growth of individual nuclei

Alternatively
Solid state phase transformation proceeds by motion of a reaction front
Eutectoid decomposition and discontinuous precipitation
1. One phase decomposes into two different phases with different composition
2. Morphology of the phase transformation is a lamellar structure
3. Concurrent precipitation occurs by exchange of atoms via short-range
diffusion
Example

Pearlite reaction in steel

Decomposition of C-rich fcc --Fe in low-carbon -Fe and cementite (Fe3C)

5.33

Lamellar microstructure
Decomposition of Fe-0.8%C-X
Microstructure is determined by transformation rate
Lamellar structure ferrite (-Fe) and cementite (Fe3C)
(a)

(b)

slow cooling rate

fast cooling rate

(c)

Fig. 5.30 Microstructure of eutectoidally decomposed Fe-0.78%C-0.97%Mn. (a) Cooled

slowly (solidification at 680C), (b) cooled rapidly (solidification at 639C) and (c) incipient
forming of pearlite with advancing carbide lamella in Fe-0.8%C (iron carbide lamella are
light colored, ferrite lamella dark colored).
5.34

Mechanism of lamella formation

Reaction initiated at a grain boundary

Nucleation favored

Assumption:
Decomposition of +
phase forms the first nucleus

new -crystal nucleus

Local change of concentration

nucleus is generated
Local change of concentration

Lamellar microstructure
Thickness of the lamellae is determined by
the range of diffusion

Fig. 5.31 Illustration of the principle of

lamella generation during eutectoid
decomposition.

Lamellar spacing l decreases with increasing

transformation rate R
1
l
R
5.35

Schematics of laminar growth (Fe-C)

(bcc)

transverse diffusion

(fcc)

Fig. 5.32 Lamella arrangement and carbon concentration distribution in growth

direction during the pearlite reaction. A depletion or enrichment of C occurs at the
moving phase boundary, leading to a transverse diffusion.
5.36

Discontinuous precipitation
Reaction at the transformation front +

Only one new phase is formed (precipitation of )

Nucleation starts at grain boundaries

Discontinuous precipitation

Generation of lamellar microstructure

Grain boundary migration (similar to recystallization)

Grow through concurrent grain boundary

displacement in a lamellar morphology

Chemical driving force

Increase by (pre) plastic deformation

Analogous reverse process - dissolution of

precipitates by moving grain boundaries during the
recrystallization of two-phases

Fig. 5.33 Reaction front during

discontinuous precipitation in Al2.8%Ag-1%Ca.
5.37

Martensitic transformation

Increasing cooling rates

Diffusivity slows down
For high cooling rates diffusion fails to accomplish the necessary
concentration changes
Suppression of phase transformation

Increase of supercooling of unstable phase

Driving forces for transformation increase drastically.
Phase transformation with change of crystal structure (Fe-C from fcc to bcc)
Spontaneous change of crystal structure without change of concentration
Spontaneous phase transformations without concentration change
Martensitic transformations

Despite the name it is not only occurring in the Fe-C system, when the formation
of ferrite is suppressed.
5.38

Martensitic transformations
Structural change

Metal or alloy

fcc

hcp

Co, Fe-Mn

fcc

bcc

Fe-Ni

fcc

tetragonal bcc

Fe-C, Fe-(Mn, Cr, Ni,...)-C

fcc

tetragonal fcc

In-Tl, Mn-Cu

bcc

hcp

Li, Ti, Zr, many Ti and Zr alloys

bcc

distorted hcp

Cu-Al

bcc

tetragonal fcc

Cu-Zn, Cu-Sn

bcc

orthorhombic

Au-Cd

tetragonal

orthorhombic

U-Cr

Table
5.1
Important
martensitic transformations
5.39

Mixed microstructure
Spontaneous change of the crystal structure
Change of volume and shape
Suppressed change of composition

Fig 5.34 Martensite and

retained austenite in an
FeNiAl-alloy

Transformation proceeds by successive displacive transformations of needle

or lath shaped regions into the new crystal structure

Consecutively transformed regions are limited in their spatial extent by prior

transformed regions
5.40

austenitization-temperature 850C

Volume fraction of the martensite

phase depends on the temperature
to which the material was quenched

Not dependent on time

Ms - martensite start
1% martensitic volume fraction in
the microstructure

Mf - martensite finish
99% martensitic volume fraction

Range of existence of martensite

depends on composition

temperature (C)

Existence range of
martensite

martensite content (%)

Fig. 5.35 Existence curve of martensite in Fe-0.45%C.
Ac3 is the temperature of beginning transformation to
austenite during heating. In practice Ms, and Mf are
determined by martensite contents of 1% and 99%,
respectively.
5.41

Composition range

Ms - martensite start
As austenite start

Ms decreases with increasing

concentration

Fe-C
Decomposition on annealing into
ferrite and cementite
Not retransforming to austenite

(calculated)

By plastic deformation the difference

between the transformation start
temperatures Ms and As can be
reduced (Md and Ad).
Martensitic transformation
accompanied by deformation
Fig. 5.36 Concentration dependence of the range of
existence of martensite in FeNi.
5.42

Bain model of martensitic formation

(a)

(b)

(c)

Fig. 5.37 The Bain-model of martensitic formation in the FeC system and the corresponding
orientation correlation of austenite and martensite after Kurdjumov-Sachs. The open circles indicate
possible positions of carbon.

Crystallographic relationships observed during martensitic transformation in the

system Fe-C - Bain model

Transformation from fcc to bcc

bcc unit cell tetragonally distorted in the presence of carbon

a) Center of two next neighbor fcc unit cells (lattice parameter a0) contains a
tetragonal body-centered unit cell (bct)
5.43

Bain model of martensitic formation

(a)

(b)

(c)

Fig. 5.38 The Bain-model of martensitic formation in the FeC system and the corresponding
orientation correlation of austenite and martensite after Kurdjumov-Sachs. The open circles indicate
possible positions of carbon.

b) Obtaining a bcc unit cell from a bct unit cell by compression along the c direction
and stretching in the plane perpendicular to the c direction (volume change of
about 3 to 5%)
During martensite transformation the atomic positions change only slightly and
next neighbors remain next neighbors after phase transformation.
c) Bain correspondence is substantiated by X-ray crystallography investigations that
confirm an orientation relationship Fe-C {111} || {110} and<110> || <111>
5.44

Kurdjumov-Sachs
orientation relationship

Complicated irrationally
indexed Habit planes
(not described by simple
Bain model)

{111} || {110}
<101> || <111>
Fig. 5.39 Stereographic representation of the KurdjumovSachs orientation relationship. The stereogram is centered
on (111)fcc || (011)bcc. The unmarked neighboring pairs of
poles would superpose exactly for the Bain orientation. They
do not do so for the Kurdjumov-Sachs orientation.
5.45

surface
interior

Martensitic transformation of Fe-C lattice transformation from fcc to bct

austenite

tetragonal
martensite
axis ratio c/a

a or c axis length (10-10 m)

C stabilizes phase
C austenite is located on the
octahedral interstitial sites of the iron
lattice

a axis length (10-10 m)

Lattice parameters

After transformation the carbon

atoms are located only on the c axis
lattice parameter in the c-direction
increases, i.e. tetragonal martensite
(bct) is formed
weight % C

5.46

Fig. 5.40 Dependence of the lattice parameters of austenite and martensite on carbon content

Mechanism of martensite transformation

Shear deformation (diffusionless transformation)
Simplest case Co: fcc hcp

{111}fcc = {0001}hcp

Invariant plane (Habit plane)

Unconstrained
transformation

(a)

(b)

(c)

Constrained
transformation

(d)

Fig. 5.41 (a) Illustration of a martensitic transformation fcc hcp by shear deformation, i.e., stacking
sequence changes from ABAB... to ABCABC... . (b) Shearing within a matrix. (c) Internal plastic
deformation of the transformed phase to keep the original shape. (d) Martensite crystal aM between
two grain boundaries.
5.47

Shape change

Shape changes lead to elastic compatibility strains in the immediate

environment of the martensite plates
Reduced by plastic deformation within the martensite via glide and twinning
Overall deformation corresponds to a shear parallel to an undistorted plane

Undistorted plane {111}fcc = (0001)hex during the cobalt transformation

Habit plane irrational for Fe-C

(a)

(b)

(c)

(d)

Fig. 5.42 The formation of martensite changes the shape of a the original crystal (a) by shear
(b) slip (c) or twinning (d) in martensite can largely compensate for this shape change.
5.48

Plastic deformation during martensitic

transformation

Martensitic transformation causes a substantial plastic deformation

Compression and stretching of the unit cell by shear deformation

Shape change of the transformed region

Surface relief
surface

(a)
(b)
Fig. 5.43 (a) Formation of a characteristic surface texture by martensitic transformation.
(b) Plate martensite in Fe-33.2%Ni with internal twinning.
5.49

TTT diagram and

martensitic
transformation

Not isothermal!
Cooling rate must be high enough
to avoid the pearlite or bainite
reaction.

perlite-start

end of trans.
temperature (C)

Special material properties are

achieved by martensitic
transformation

austenization temperature 970C

bainite

martensitic start

1. Martensite
2. Pearlite
3. Bainite

time (sec)

Fig. 5.41 TTT-diagram of an alloy steel (Fe-C-Cr) with cooling curves. Different cooling rates yield
martensite (1), pearlite (2) or bainite (3). Only for very low cooling rates occurs the transformation at
equilibrium temperature (F - ferrite; P - pearlite; M, - martensite start)

5.50

Shape memory effect

Characteristics
Regaining original after deformation by a respective heat treatment
Requirements
1. Martensitic transformation with Ms As (very little hysteresis)
2. Deformation only by martensitic transformation and twinning, i.e.,
suppression of dislocation slip
Alloys: AuCd, Fe3Pt, NiAl, NiTi (Nitiniol)

Cooling through the transformation temperature Mf

Crystallographically equivalent variants are adjusted in such a way that the

shape of the specimen remains unchanged

After deformation - material heated to temperature Af

Martensite disappears and original shape is restored

5.51

Shape memory effect

martensite
and suitable
variants

stress induced
growth of
variants

single
crystal

-phase
heat
treatment

austenite
formation and
shape
recovery

martensite
single
variant
state

quenched in
water
(T < Mf)

reaction to
applied
stress

reversible
strain limit

reheating
into the
-zone

Fig. 5.44 Schematic illustration of the shape-memory-effect works in a single crystal.

5.52

Pseudo- or superelasticity

Similar effect to shape memory effect (same materials)

Introduction of martensitic phase by applied mechanical stress

Large strain

stress
(arb. units)
austenite to martensite
transformation

strain
Fig. 5.45 Typical stress strain curve of a superelastic material (in comparison to steel).

Good damping properties - energy of mechanical impact consumed by the

martensitic transformation
5.53

(a)

Comparison
shape-memory effect
and superelasticity

Fig. 5.46 (a)(c) Schematic illustration of the

mechanism of the shape-memory effect and
superelasticity, in which solid lines represent the
shape-memory path and dotted lines represent
the superelasticity path. Martensite forms upon
cooling below Mf. However, the macroscopic
shape of the sample does not change because
different crystallographic equivalent forms occur
with the same frequency. Application of an
external stress induces the growth of the
crystallographic variant that causes the largest
shear deformation in the direction of the applied
stress. At high stress only this variant survives.
Heating above Af gives rise to a transformation
into the original phase and hence to shape
recovery.

(b)

(c)

5.54