Physical Metallurgy 2 ( 2016/2017)
Diffusionless (Non-diffusive)
Transformation
Dr. Deni Ferdia n , M. S c
Disclaimer: This lecture note is edited from different sources for the solely of teaching and learning
purposes. It may contain copyrighted materials from their respective owners; therefore, apart form
teaching and learning purposes, this lecture note may not be reproduced, stored, or transmitted in any
form or by any means.
1
Possible Transformations of Steel
Martensite
T Martensite
Strength
Ductility
bainite
fine pearlite
coarse pearlite
spheroidite
General Trends
Outline
4
Introduction
Martensitic transformations
Crystallographic theory of martensitic
transformations
Martensitic transformation in nonferrous metals
(SME)
2
5
INTRODUCTION
Introduction
6
Non-diffusive or diffusionless transformations are a class of phase
changes that do not occur by the long-range diffusion of atoms,
where, say, their diffusion flow down a concentration gradient
according to Fick‘s Law;
In this type of transformation, the process is rather by some form
of cooperative, homogeneous movement of many atoms that
results in a change in crystal structure.
Displacive transformations do not change the composition of the
parent phase, but rather only the crystal structure.
Consequently, displacive transformations do not require long
range atomic movement
3
7
Because no atomic migration is necessary, displacive transformations
usually progress in a time-independent fashion, with the speed of the
interface between the two phases able to move at nearly the speed of
sound.
The prototype of this behavior is a martensitic transformation, which
is equivalent to the deformation of the parent crystal lattice into
martensite.
Because of their nature, martensitic transformations are frequently
referred to as “displacive”, “shearlike” or “diffusionless”
transformation.
The ordered movement of large numbers of atoms lead some to refer
to these as military transformations in contrast to civilian diffusion-
based phase changes that occurs randomly.
The name martensite is after the German scientist Adolf Martens
(1850–1914); it was used originally to describe the hard micro
constituent found in quenched steels.
Many materials known to exhibit the same type of solid-state
phase transformation.
Martensite occurs in, for example, nonferrous alloys, pure
metals, ceramics, minerals, inorganic compounds, solidified
gases and polymers.
4
9
It has long been recognized that martensitic transformations are
diffusionless; this is why martensite can grow at temperatures below
100 K where the mobility of atoms is negligible.
It is not implied that all martensitic transformations occur at low
temperatures; indeed, many of them occur at comparatively high
temperatures, but these too are diffusionless.
Martensite can also grow at speeds in excess of 1000 m/s, speeds
which are inconsistent with the diffusion of atoms.
This great speed of formation makes martensite nucleation and growth
a difficult process to study experimentally.
The chemical composition of martensite can be measured and shown
to be identical to that of the parent austenite
10
MARTENSITE
TRANSFORMATION
5
Martensite Transformation – (Fe-0.18C-0.2Si-0.9Mn-2.9Ni-1.5Cr-
0.4Mo)
11
Why study martensitic transformations?
12
They occur in many different metal, ceramic and polymer systems, and
generally important to understand.
Steels represent the classical example of a mechanically hard
martensite.
Also, there are remarkable devices that exploit the consequence of
martensitic transformation (i.e. shape memory effect) such as stents
that open up once at body temperature.
Mechanically speaking, the martensites in this case are generally soft.
6
Massive vs. Martensitic transformations
13
There are two basic types of diffusionless transformations:
One is the massive transformation in which a diffusionless
transformation takes place without a definite orientation relationship.
The interphase boundary (between parent and product phases)
migrates so as to allow the new phase to grow; it is, however, a civilian
transformation because the atoms move individually.
The other is the martensitic transformation in which the change in
phase involves a definite orientation relationship because the atoms
have to move in a coordinated manner. There is always a change in
shape which means that there is a strain associated with the
transformation. The strain is a general one, meaning that all six
(independent) coefficients can be different.
Driving forces
14
Non-diffusive transformations require larger driving forces than for
diffusional transformations. Why?
In order for a transformation to occur without long range diffusion, it
must take place without a change in composition.
This leads to the so-called T0 concept, which is the temperature at
which the new phase can appear with a net decrease in free energy at
the same composition as the parent (matrix) phase.
The temperature, T0, at which segregation-less transformation
becomes possible (i.e. a decrease in free energy would occur), is always
less than the solvus (liquidus) temperature.
7
Free energy - composition: T0
Diffusionless
transformation a,
product ∆Gγα T1
impossible at T1,
g,
Diffusionless parent
transformation
Common
possible at T2; G tangent
“T0” is defined by no ∆Gγα T1>T2
difference in free T2
energy between the
phases, ∆G=0 T2 corresponds to
next figure b.
16
T near T0
(a) (b)
equilibrium diffusionless
(a) Free energy temperature diagram for austenite and
martensite of fixed carbon concentration (C0) in (b)
(b) Free energy-composition diagram for the austenite and
martensite phases at the Ms temperature
8
Driving force estimation
The driving force for a martensitic transformation can be
estimated in exactly the same way as for other
transformations such as solidification.
Provided that an enthalpy (latent heat of transformation)
is known for the transformation, the driving force can be
estimated as proportional to the latent heat and the
undercooling below T0.
( − )
∆ = ∆
0
18
(a) (b)
(a) Iron-carbon phase diagram with T0 defined previously.
(b) Ms and Mf in relation to the TTT diagram for alloy C0
9
Heterogeneous nucleation
19
Why does martensite not form until well below the T0 temperature?
The reason is that a finite driving force is required to supply the energy
needed for (a) the interfacial energy of the nucleus and (b) the elastic
energy associated with the transformation strain.
The former is a small quantity (estimated to be 0.02 J.m-2) but the
elastic strain is large (estimated at 0.2 J.m-2 in the Fe-C system).
Therefore the following standard equation applies.
∗
16
Δ = ( )
3 Δ − Δ
Why does martensite require heterogeneous nucleation? The reason is
the large critical free energy for nucleation outlined above.
Theories of Martensite Nucleation
• A lenticular martensite nucleus with radius a and thickness c:
e.g.) s=0.2, γ=20mJ/m2, ΔGv=174mJ/m3
then, c*/a*=1/40, ΔG*=20eV (unable to overcome by thermal
fluctuation)
• Heterogeneity of Martensite nucleation: Dislocation, Inclusions but
not GB and free surface
10
21
CRYSTALLOGRAPHIC THEORY
OF MARTENSITE
Microstructure of martensite
22
The microstructural characteristics of martensite are:
The product (martensite) phase has a well defined
crystallographic relationship with the parent (matrix).
Martensite forms as platelets within grains.
Each platelet is accompanied by a shape change
The shape change appears to be a simple shear parallel
to a habit plane (the common, coherent plane between
the phases) and a uniaxial expansion (dilatation)
normal to the habit plane.
11
23
Successive sets of platelets form,
each generation forming
between pairs of the previous
set.
The transformation rarely goes
to completion because
of the strain associated with the
product that leads to back
stresses in the parent phase.
24
Martensite plate
remains
macroscopically
coherent with the
surrounding
austenite and even
the surface it
intersects.
12
Why tetragonal Fe-C martensite?
25
At this point, it is worth stopping to ask why a tetragonal
martensite forms in iron.
The answer has to do with the preferred site for carbon as
an interstitial impurity in BCC Fe.
Remember: Fe-C martensites are unusual for being so
strong (and brittle).
Possible sites
26
Possible sites for
interstitial atoms in the
fcc or hcp lattices
13
27
Most martensites are not significantly stronger than their parent
phases.
Interstitial sites:
FCC: octahedral sites radius = 0.052 nm
tetrahedral sites radius = 0.028 nm
BCC: octahedral sites radius = 0.019 nm
tetrahedral sites radius = 0.036 nm
Carbon atom radius is 0.08 nm.
Surprisingly, it occupies the octahedral site in the bcc Fe
structure, despite the smaller size of this site (compared to the
tetrahedral sites)
It is presumably because of the low modulus in the <100>
directions; in this case lower number of near and next nearest
neighbors compared to the tetrahedra interstitial position.
Interstitial sites for C in Fe
28
(a) FCC and (b) BCC
carbon occupies the
octahedral (left) and
tetrahedral (right) sites
One consequence of the
occupation of the octahedral
site in ferrite is that the carbon
atom has only two nearest
neighbors.
Each carbon atom therefore
distorts the iron lattice in its
vicinity; the distortion is a
tetragonal distortion.
14
29
If all the carbon atoms
occupy the same type of
site then the entire lattice
becomes tetragonal, as in
the martensitic structure.
Switching of the carbon
atom between adjacent
sites leads to strong
internal friction peaks at
characteristic temperatures
and frequencies.
Mechanisms
30
The mechanisms of military transformations are not entirely
clear.
The small length scales mean that the reactions propagate at
high rates - close to the speed of sound.
The high rates are possible because of the absence of long
range atomic movement (via diffusion).
Possible mechanisms for martensitic transformations include:
(a) dislocation based
(b) shear based
Martensitic transformations strongly constrained by
crystallography of the parent and product phases.
This is analogous to slip (dislocation glide) and twinning,
especially the latter.
15
Shape change
31
(a) Undistorted cube;
(b) slipped cube with offsets
nb;
(c) twinned cube revealing
reorientation within twin.
Displacements are
proportional to distance
from twin plane.
32
Twinning on the (120) plane in a simple cubic crystal. Gray
circles represent original atom positions. Black circles are final
atom positions.
16
Twinning formation
33
(Left) Showing the twinning of an FCC structure. Black and white circles
represent atoms on different levels.
(Right) Graphical representation of a twinning shear occurring on a plane K1
in a direction d (shear direction)
Atomic model - the Bain model
34
For the case of FCC Fe transforming to BCT ferrite
(Fe-C martensite), there is a basic model known as the
Bain model, proposed by Bain in 1924.
The essential point of the Bain model is that it
accounts for the structural transformation with a
minimum of atomic motion.
Start with two FCC unit cells: contract by 20% in the z
direction, and expand by 12% along the x and y
directions.
17
Bain model
35
Bain correspondence for the α → α′ transformation. Possible interstitial sites for
carbon are shown by crosses (x). To obtain α′ the γ unit cell is contracted about 20%
on the C axis and expanded about 12% on the a axes
36
Orientation relationships in the Bain model are:
(111) <=> (011)α’
[101] <=> [111] α’
[110] <=> [100] α’
[112] <=> [011] α’
The experimentally observed orientation
relationships, however, are inconsistent
18
Does the Bain model fit the observation of the
undistorted habit plane?
37
Refer to the sphere/ellipsoid
model. Two vectors are needed to
form a plane:
Vector OA or OA′ is invariant but
OY′ is deformed by 12% (x’ and y’
axis).
In this plane the only vectors that
are not shortened or elongated by
the Bain distortion are OA or O'A'.
However in order to find a plane
in the FCC structure that is not The Bain deformation is here simulated by
the pure deformation in compressing a
distorted by the transformation sphere elastically to the shape of an oblate
requires that the vector OY' ellipsoid. As in the Bain deformation, this
(perpendicular to the diagram) “transformation” involves two expansion
must also be undistorted. axes and one contraction axis.
Modification of the Bain Model
• Internally twinned martensite model: The deformation of the
OY′ axis can be made to zero by introducing twinning or slip.
This can form by having alternate regions in the austenite
undergo the Bain strain along different contraction axes such
that the net distortions are compensated. Then the habit plane
becomes a macroscopically invariant plane.
• Experimental data on the habit planes: {111} to {225} to {259}
transition with increasing C content or Ni content. Also, a
transition occurs from dislocated martensite to twinned
martensite with increasing C or Ni. Thickness of twins in high
carbon {259} martensite: approx. 3nm.
19
Others Model for Martensite Transformation
39
Orientational relationship between the parent (fcc) and martensite (bcc) phases for the Bain (a),
the Nishiyama–Wassermann (b) and the Kurdjumov–Sachs (c) The blue atoms indicate a bcc
unit cell. The red arrows indicate part of the motion initiating the transformation. The dashed
arrows indicate the invariant direction which is shared by the parent and martensite phases.
Sandoval, New J. Phys. 11 (2009)
Orientation relationship (OR)
40
Kurdjumov–Sachs (111)F//(101)B (Kurdjumov and
(K-S) [1-10]F//[11-1]B Sachs, 1930)
[1-21]F 5.26° from [10-1]B
Nishiyama– (111)F//(101)B (Nishiyama, 1934;
Wasserman (N-W) [1-10]F 5.26° from [11-1]B Wassermann, 1935)
[1-21]F//[10-1]B
Greninger–Troiano (111)F//(101)B (Greninger and
(G-T) [1-10]F 2.5° from [11-1]B Troiano,
[1-21]F 2.5° from [10-1]B 1949)
Expresses the relationship between planes and directions in one phase, such as the
austenite matrix (F) with corresponding planes and directions in the other phase, i.e.
the martensite (B).
20
Role of slip and twinning in martensitic
transformation
This figure illustrates schematically
how dislocation glide or twinning of
the martensite can compensate such
as a Bain deformation and thereby
reduce the strain of the surrounding
austenite. The transformation shear
(s) is defined. Note how s can be
reduced by slip or twinning
Role of dislocations
42
Dislocations play an important, albeit hard to define, role in
martensitic transformations.
Dislocations in the parent phase (austenite) clearly provide sites
for heterogeneous nucleation.
Dislocation mechanisms are thought to be important for
propagation/growth of martensite platelets or laths.
Unfortunately, the transformation strain (and invariant plane)
does not correspond to simple lattice dislocations in the FCC
phase.
Instead, more complex models of interfacial dislocations are
required.
21
Role of slip and twinning in martensitic transformation
Fig. 12.5 (a) Formation of a martensite platelet in a crystal of austenite, (b) the
inhomogeneous twinning shear within the platelet
Role of slip and twinning in martensitic transformation
Fig. 6.11 Martensite habit planes in various types of steel
22
Role of slip and twinning in martensitic
transformation
Fig. 6.12 Transmission electron micrographs of (a) lath martensite and (b) twinned
martensite. Note the mid rib in the twinned martensite, which is thought to be the first
part of the plate to grow.
Self-accommodation by variants
46
A typical feature of martensitic transformations is that
each colony of martensite laths/plates consists of a
stack in which different variants alternate.
This allows large shears to be accommodated with
minimal macroscopic shear.
23
47
MARTENSITE STRUCTURE
48
24
Martensite Microstructure
49
50
25
51
Microstructures
52
(a), (b) Growth of
martensite with increasing
cooling below Ms.
Different martensite
morphologies in iron
alloys:
(c) low C (lath).
(d) medium C (plate).
(e) Fe-Ni (plate).
26
53
SHAPE MEMORY ALLOYS
Background
First record of shape
memory transformation
was found in
1932(AuCd)
Shape memory effect
was discovered in
Nickel-Titanium at the
Naval Ordnance
Laboratory(NiTiNOL)
Super elasticity
27
SME Demonstration (Nitinol wire)
55
Shape memory effect (SME)
56
General phenomenon associated with martensitic
transformations.
Characteristic feature = strain induced martensite
(SIM), capable of thermal reversion.
Ferroelasticity and Superelasticity also possible.
Md, Af, As, Ad, Ms, Mf temperatures.
28
Temperatures
57
Md Af
Ms T0
Ad As
Mf
The Md and Ad temperatures bracket T0 because they define the on-
cooling and on-heating temperatures at which the transformation is
possible with allowance for the effect of strain energy.
SME Definitions
58
Md: SIM possible below Md.
Af: reversion of SIM complete above Af (heating).
As: reversion of SIM starts above As (heating).
Ad: formation of parent phase possible above Ad.
Ms: martensite start temperature (cooling).
Mf: martensite finish temperature (cooling).
29
59
Classic alloy = Nitinol = NiTi (National Ordnance Laboratory)
Alloying for control of Ms.
Stress for SIM must be less than yield stress for plastic
deformation.
SME depends on incomplete transformation and elastic back
stresses to provide memory (>MS).
SME more effective in single crystals.
Alloying permits variations in the equilibrium transformation
temperature, for example (critical for bio applications, for
example).
Also variations in the maximum strain that can be recovered are
possible.
Super-elasticity
60
Super-elasticity is
simply reversible
(therefore
“elastic”)
deformation over
very large strain
ranges (many %).
Example: Ti-
50.2%Ni.
30
Mechanism of SME
Mechanism of
SME:
1) transformation;
2) martensite, self-
accommodated;
3) deformation by
variant growth;
4) heating causes
re-growth of parent
phase in original
orientation
Changes in microstructure of a shape memory alloy
(bi-crystal of austenitic CuAlNi )
62
31
Role of ordering
A key feature of the Ni-Ti alloys for shape memory applications
is that their compositions are all in the vicinity of 50Ni-50Ti
and that the high temperature phase is an ordered structure.
The low temperature monoclinic structure is also ordered (as is
the other, intermediate R phase which is trigonal).
The ordered structure (recall the discussion of ordered particle
strengthening) means that there is an appreciable resistance to
dislocation motion.
This is critical for favoring strain accommodation via
transformation and twinning as opposed to dislocation glide.
Self-accommodation
64
Micrograph with
diagram shows how
different variants of a
given martensitic phase
form so as to minimize
macroscopic shear
strains in a given region.
32
65
Stress-strain-
temperature data
exhibiting the
shape memory
effect for a
typical NiTi
SMA.
Some videos
33
Surface relief
67
Micrographs
show a sequence
of temperatures
with surface
relief from the
martensite plates.
Stress versus Temperature
68
The stress applied to the material must be less than the critical resolved
shear stress for dislocation motion, because the latter is not recoverable;
SME= Shape Memory Effect; SE = Superelasticity
d/dT = ∆S/ = ∆H/(Te Critical
Stress for
Mf As Martensite
Stress Formation
SME SE
Critical
Stress for
Slip
Ms Af Temperature
34
Ni-Ti Alloys
69
X Ms Mf As Af
V > 25 < -140 < -64 > 25
Cr -100 < -180 < -58 > 25
Mn -116 < -180 < -63 > 10
Fe No < -180 -30 > 25
information
Co No No 0 > 25
information information
Cu > 25 < -100 ? > 25
TiNi0.95 70 60 108 113
TiNi 60 52 71 77
Ti0.95Ni -50 < -180 ? 20 (?)
[Wasilewski, SME in Alloys, p. 245]
SME requirements
70
For achieving a strong or technologically
useful SME, the following characteristics are
required.
High resistance to dislocation slip (to
avoid irreversible deformation).
Easy twin motion in the martensitic state
so that variants can exchange volume at
low stresses.
Crystallographically reversible
transformation from product phase back
to parent phase; ordered structures have
this property (whereas for a disordered
parent phase, e.g. most Fe-alloys, multiple
routes back to the parent structure exist.)
35
Photo-stimulated - Shape Memory Polymer
71
Nitinol Wires
Nitinol is generally doped with other materials like Cr,
Cu, Al, or Fe.
Flexinol is a popular brand of SMA wire.
Flexinol is designed to take more repeated stress cycles
than pure NiTi mixes.
Specifically designed to manufacturer’s needs.
0.0010 inch diameter wire can lift 7 grams in 1 second with a 20
mA current.
0.010 inch diameter wire can lift 930 grams in 1 second with a 1 A
current.
Wires are also made to change states at different temperatures
generally between -30 C and 120 C within 5 C.
36
Robotic muscle application
73
• To create a single direction of
movement the bias spring shown in
the upper portion of the finger
would hold the finger straight,
stretching the SMA wire, then the
SMA wire on the bottom portion of
the finger can be heated which will
cause it to shorten bending the joint
downwards.
• The heating takes place by running
an electric current through the wire;
the timing and magnitude of this
current can be controlled through a
computer interface used to
manipulate the joint.
Pseudoelastic applications
74
Commercial pseudoelastic
applications of SMAs:
Pseudoelastic eye glass
frames.
Golf club with embedded
pseudoelastic SMA insert.
37
Orthodontic application
75
Orthodontic
application of SMAs:
Nitinol braces used
for alignment
purposes in dental
applications.
A schematic showing
a NiTi drill used for
root canal surgery.
Medical applications: osteosynthesis
76
The NiTi plates are first cooled to
well below their transformation
temperature, then they are placed
on the set break just like titanium
plates.
However, when the body heats the
plate up to body temperature the
NiTi attempts to contract applying
sustained pressure on the break or
fracture for far longer than
stainless steel or titanium.
This steady pressure assists the
healing process and reduces
recovery time.
38
77
The shape memory effect is
induced in the wires simply by
heating them with an electric
current, which is easily supplied
through electrical wiring,
eliminating the need for large
hydraulic lines.
(a) Total view of the SMART Wing
model.
(b) Cut away view of the SMA
torque tubes as installed in the
model during Phase 1 of the
SMART Wing project
Smart wings
78
The shape memory
wire is used to
manipulate a flexible
wing surface.
The wire on the bottom
of the wing is
shortened through the
shape memory effect,
while the top wire is
stretched bending the
edge downwards, the
opposite occurs when
the wing must be bent
upwards.
39
Jet noise reductions
79
Engine noise levels during
take off and landing have
become more highly
regulated worldwide.
To reduce this noise, some
designers are installing
chevrons onto engines to
mix the flow of exhaust
gases and reduce engine
noise. Boeing variable geometry
chevron, flight testing
References
80
D. A Porter, and K.E Easterling, Phase Transformation in
Metals and Alloys, 2nd ed. Chapman and Hall, London,
1992.
R.W. Chan, P. Haasen, Physical Metallurgy, North-
Holland, 1996.
40
