Anda di halaman 1dari 14

P-phase precipitation and its eect on martensitic transformation

in (Ni,Pt)Ti shape memory alloys


Y. Gao
a
, N. Zhou
a
, F. Yang
a
, Y. Cui
a,1
, L. Kovarik
a,2
, N. Hatcher
b
, R. Noebe
c
,
M.J. Mills
a
, Y. Wang
a,
a
Ohio State University, Department of Materials Science and Engineering, 2041 College Road, 477 Watts Hall, Columbus, OH 43210, USA
b
Interdisciplinary Centre for Advanced Materials Simulations, Ruhr-Universita t Bochum, Germany
c
NASA Glen Research Center, 21000 Brookpark Road, Cleveland, OH 44135, USA
Received 26 August 2011; received in revised form 18 November 2011; accepted 19 November 2011
Abstract
A new precipitate phase named P-phase has recently been identied in (Ni,Pt)Ti high temperature shape memory alloys. In order to
understand the roles played by the ne coherent P-phase precipitates in determining the martensitic transformation temperature (M
s
),
strength of the B2 matrix phase, dimensional stability and shape memory eect of the alloys, a phase eld model of P-phase precipitation
is developed. Model inputs, including lattice parameters, precipitatematrix orientation relationship, elastic constants and free energy
data, are obtained from experimental characterization, ab initio calculations and thermodynamic databases. Through computer simu-
lations, the shape and spatial distribution of the P-phase precipitates, as well as the compositional and stress elds around them, are
quantitatively determined. On this basis, the elastic interaction energy between the P-phase precipitates and a martenstic nucleus is cal-
culated. It is found that both the chemical non-uniformity and stress eld associated with the P-phase precipitates are in favor of the
martensitic transformation. Their relative contributions to the increase in M
s
temperature are quantied as a function of aging time
and the result seems to agree with the experimental measurements. The shape and spatial distribution of the P-phase precipitates pre-
dicted by the simulations also agree well with experimental observations.
2011 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
Keywords: Coherent precipitate; Elastic interaction; Nucleation; Aging eect; Phase eld simulation
1. Introduction
Recent work has shown that the new class of high tem-
perature shape memory alloys (HTSMAs) (i.e. NiTi with
ternary additions such as Pt, Pd, replacing Ni or Hf and
Zr replacing Ti) have several important advantages relative
to the conventional NiTi binary shape memory alloys
(SMAs) relatively high transformation temperatures
and good dimensional stability [1]. A unique microstruc-
tural feature of these HTSMAs is the presence of ne
coherent precipitates. For example, recent experimental
characterization of (Ni,Pt)Ti alloys [2] has shown ne pre-
cipitates of an intermetallic phase named P-phase after
aging at $600 C for 100 h (Fig. 1). These precipitates have
a monoclinic structure and are fully coherent with the
matrix phase. Their shapes are more or less cuboidal, with
edge length 200400 nm and faces nearly parallel to {100}.
Spatial alignment of the precipitates along h100i directions
is also obvious from Fig. 1. Previous studies [36] have
shown that the precipitates not only strengthen the matrix
but also increase martensitic transformation temperature
M
s
. In addition, SMAs stabilized by ne precipitates exhi-
bit minimal accumulation of non-recoverable strain during
thermomechanical cycling [1,5,6]. However, the mecha-
1359-6454/$36.00 2011 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.actamat.2011.11.043

Corresponding author. Tel.: +1 614 292 0682; fax: +1 614 292 1537.
E-mail address: wang.363@osu.edu (Y. Wang).
1
Present address: IMDEA Materials, Science Park of the University
Carlos III of Madrid Mediterranean Avenue, No. 22, 28918 Leganes,
Spain.
2
Present address: EMSL, Pacic Northwest National Laboratory,
Richland, WA 99301, USA.
www.elsevier.com/locate/actamat
Available online at www.sciencedirect.com
Acta Materialia 60 (2012) 15141527
nisms by which the precipitates inuence the martensitic
transformation and dislocation plasticity in the B2 matrix
phase in these HTSMAs have not been investigated.
Phase eld modeling and simulation have been used
widely to treat both diusional and displacive phase transfor-
mations, as well as dislocationprecipitate interactions
[716]. In this study, we formulate a phase eld model to
study P-phase precipitation and its eect on martensite nucle-
ation. In particular, we document through systematic com-
puter simulations (a) the shape and spatial distribution of
coherent P-phase precipitates and (b) the stress and concen-
tration elds around a P-phase precipitate and their eects on
the B2 ! B19 martensitic transformation. Interaction of the
precipitates with dislocations and its eects on transforma-
tion induced plasticity in the B2 matrix and on dimensional
stability of the alloy will be reported in a separate paper.
The concentration and stress elds surrounding a coher-
ent precipitate are determined by the interplay between
chemical free energy and coherency elastic strain energy.
In this study, the chemical free energy models for the
matrix and precipitate phases in NiPtTi are formulated
based on existing thermodynamic databases [1719], while
the coherency elastic strain energy is formulated using
Khachaturyan and Shatalovs microelasticity theory [20].
In the following sections, thermodynamic modeling and
transformation strain calculation are rst presented (Sec-
tions 2 and 3), followed by descriptions of the phase eld
model (Section 4). Simulation results and their implications
are presented and discussed respectively in Sections 5 and
6. The major ndings are summarized in Section 7.
2. Thermodynamic modeling of B2 phase in TiNiPt
The accuracy of phase eld simulations and predictions
relies on the accuracy of available thermodynamic dat-
abases, especially the free energy of the B2 matrix phase
in the case of NiTi based SMAs [9]. Because of the limited
amount of thermodynamic data for the TiNiPt ternary
system, the assessment of the B2 phase chemical free energy
is rst carried out via the CALPHAD method utilizing
available experimental data from the literature [1719].
The B2 phase in TiNiPt can be described by a two-sub-
lattice model (Ni,Pt,Ti)
1
(Ni,Pt,Ti)
2
:
G
B2
y
1
Ni
; y
1
Pt
; y
1
Ti
; y
2
Ni
; y
2
Pt
; y
2
Ti
_ _
y
1
Ni
y
2
Ni
G
0
Ni;Ni
y
2
Pt
G
0
Ni;Pt
y
2
Ti
G
0
Ni;Ti
_ _
y
1
Pt
y
2
Ni
G
0
Pt;Ni
y
2
Pt
G
0
Pt;Pt
y
2
Ti
G
0
Pt;Ti
_ _
y
1
Ti
y
2
Ni
G
0
Ti;Ni
y
2
Pt
G
0
Ti;Pt
y
2
Ti
G
0
Ti;Ti
_ _
RT y
1
Ni
lny
1
Ni
y
1
Pt
lny
1
Pt
y
1
Ti
lny
1
Ti
_ _
RT y
2
Ni
lny
2
Ni
y
2
Pt
lny
2
Pt
y
2
Ti
lny
2
Ti
_ _
y
1
Ni
y
1
Ti
y
2
Ni

i
L
i
NiTi;Ni
y
1
Ni
y
1
Ti
_ _
i
y
2
Pt

i
L
i
NiTi;Pt
y
1
Ni
y
1
Ti
_ _
i
_
y
2
Ti

i
L
i
NiTi;Ti
y
1
Ni
y
1
Ti
_ _
i
_
y
1
Ni
y
1
Pt
y
2
Ni

i
L
i
NiPt;Ni
y
1
Ni
y
1
Pt
_ _
i
y
2
Pt

i
L
i
NiPt;Pt
y
1
Ni
y
1
Pt
_ _
i
_
y
2
Ti

i
L
i
NiPt;Ti
y
1
Ni
y
1
Pt
_ _
i
_
y
1
Ti
y
1
Pt
y
2
Ni

i
L
i
TiPt;Ni
y
1
Ti
y
1
Pt
_ _
i
y
2
Pt

i
L
i
TiPt;Pt
y
1
Ti
y
1
Pt
_ _
i
_
y
2
Ti

i
L
i
TiPt;Ti
y
1
Ti
y
1
Pt
_ _
i
_
y
2
Ni
y
2
Ti
y
1
Ni

i
L
i
Ni;NiTi
y
2
Ni
y
2
Ti
_ _
i
y
1
Pt

i
L
i
Pt;NiTi
y
2
Ni
y
2
Ti
_ _
i
_
y
1
Ti

i
L
i
Ti;NiTi
y
2
Ni
y
2
Ti
_ _
i
_
y
2
Ni
y
2
Pt
y
1
Ni

i
L
i
Ni;NiPt
y
2
Ni
y
2
Pt
_ _
i
y
1
Pt

i
L
i
Pt;NiPt
y
2
Ni
y
2
Pt
_ _
i
_
y
1
Ti

i
L
i
Ti;NiPt
y
2
Ni
y
2
Pt
_ _
i
_
y
2
Pt
y
2
Ti
y
1
Ni

i
L
i
Ni;PtTi
y
2
Pt
y
2
Ti
_ _
i
y
1
Pt

i
L
i
Pt;PtTi
y
2
Pt
y
2
Ti
_ _
i
_
y
1
Ti

i
L
i
Ti;PtTi
y
2
Pt
y
2
Ti
_ _
i
_
1
Fig. 1. TEM images of the P-phase precipitates aged at 600 C for 100 h.
Y. Gao et al. / Acta Materialia 60 (2012) 15141527 1515
where y
1
Ni
; y
1
Pt
; y
1
Ti
; y
2
Ni
; y
2
Pt
and y
2
Ti
are the site fractions,
representing mole fractions of elements Ni, Pt and Ti on
the two sublattices (the superscript) with
0:5y
1
Ni
0:5y
2
Ni
x
Ni
2a
0:5y
1
Pt
0:5y
2
Pt
x
Pt
2b
0:5y
1
Ti
0:5y
2
Ti
x
Ti
2c
y
1
Ni
y
1
Pt
y
1
Ti
1 2d
y
2
Ni
y
2
Pt
y
2
Ti
1 2e
where x is the mole fraction of each species in the crystal. The
rst three terms in the expression describe the contributions
from the Gibbs free energies of the so-called end member
phases in which each sublattice is occupied by only one kind
of species. The fourth and fth term describe the contribu-
tions from mixing on each sublattice. The other terms are
the excess Gibbs free energy that describes the interactions be-
tween atoms on the same sublattice, as in a regular solution
model. At any given temperature and initial composition,
the phases must also be at internal equilibrium determined
by
@G
B2
@y
0 for all site fractions (y = y
1
Ni
; y
1
Pt
; y
1
Ti
; y
2
Ni
;
y
2
Pt
or y
2
Ti
). The free energy model and the assessed param-
eters used in Eq. (1) are given in Appendix A.
3. Crystallography of B2 to P-phase transformation
3.1. Crystal structure of the P-phase and its orientation
relationship with the B2 matrix
The P-phase has a monoclinic crystal structure with
a
p
= 0.745 nm, b
p
= 1.292 nm, c
p
= 1.422 nm and b =
100.45

[2]. The lattice parameter of the B2 matrix is


a
B2
= 0.3087 nm [2,21]. Each monoclinic unit cell of the P-
phase (Fig. 2a) includes 96 atoms in16 layers that are parallel
to the (111) plane of the B2 matrix phase. The P-phase has a
stoichiometry of Ti
44
Ni
36
Pt
16
, which indicates an excess of
Ni and a deciency of Ti in the structure as compared to
the B2 structure. The 16 Pt atoms forman ordered structure
on the Ni sublattice and Ni substitutes for four sites (Wyck-
o notation 4e) on the Ti sublattice [2].
According to the orientation relationship between the
B2 phase and the P-phase observed in the experiment [2]
(100
p
==11

2
B2
; 010
p
==1

10
B2
; 001
p
==

2
B2
), the
following lattice correspondence has been proposed by
Kovarik et al. [2]: 100
p
! 11

2
B2
; 010
p
! 3

30
B2
and
001
p
!

2
B2
. This lattice correspondence is schemati-
cally shown in Fig. 3. The space group of the B2 structure
and the monoclinic P-phase are Pm

3m and C2/c, respec-


tively. Rotations in the point group of B2 structure include
three fourfold (h100i), four threefold (h111i), and six
twofold (h110i) rotation axes, totaling 24 elements com-
bining with identity. Meanwhile, there are only two rota-
tion elements for the P-phase, which are one twofold
rotation and its identity. Their intersection group contains
two common symmetry elements and thus there are 12
crystallographic equivalent variants for the P-phase, which
can be divided into four groups (see Table 1). It has been
observed that within each group, three variants stack into
a layered structure along h111i (Fig. 4).
3.2. Coherency of P-phase/B2 matrix interface
It is dicult to directly observe the structure of P-phase/
B2 matrix interfaces since at room temperature, where the
high resolution STEM investigations have been carried
Fig. 2. (a) Ni sublattice (001)
P
planes in the P-phase (Ti sublattice planes
are not included). (b) The arrangement of Pt atoms in the (001)
P
Pt rich
planes. Each of these planes contain four Pt atoms and two Ni atoms [2].
Fig. 3. (a) Non-periodical arrangement of the three variants of the
monoclinic P-phase within a single P-phase particle. (b) A monoclinic unit
cell of the P-phase and its OR with the B2 parent phase [2].
Table 1
Symmetry related orientation variants of the P-phase.
Variant no. a b c a b
1

211 03

3 111
2 211 03

3 2

111
3 21

1 033 2

33 1

11
4 2

11 033 23

3 11

1
5 1

21

303

3 111
6 121

303

32

3 1

11
7 12

1 303

323

111
8

121 303 32

3 11

1
9 11

2 3

30

2 111
10 112 3

30

32 11

1
11 1

12 330

332

111
12

112 330 3

32 1

11
1516 Y. Gao et al. / Acta Materialia 60 (2012) 15141527
out, the matrix has already transformed to B19 martensitie.
The martensite adjacent to the P-phase particles is often not
in an orientation suitable for high resolution imaging. How-
ever, on occasion this is not the case, as shown in the high
angle annular dark eld (HAADF) image in Fig. 5. This
HAADF image was obtained with a FEI Titan 80300 with
C
s
-correction on the electron probe, and operated at
300 kV. In the image the precipitate is being viewed along
the 0

10 zone (i.e. a h110i orientation relative to the B2


matrix [2]). The key observation is that there are no inter-
face or mist dislocations present at the interface, indicating
that the interface between the precipitate and the martensite
is fully coherent. This same result has been obtained for
other selected regions of particle/matrix interface that have
been suitable for imaging. Based on these observations we
assume in this study that the precipitate/B2 matrix interface
is also fully coherent because if the interface contains mist
dislocations, one would expect that the dislocation content
will be inherited as the B2 matrix transforms to martensite.
STEM observations during in situ heating experiments
above the M
s
temperature are required in order to directly
observe the precipitate/B2 matrix interfaces, which is
beyond the scope of this study.
3.3. Calculation of stress-free transformation strain
According to the lattice correspondence described
above, the transformations of the three non-parallel lattice
vectors during B2 to P-phase transition are calculated as:
e
B2
1
a
B2
0

1
1
2
_

_
_

_; e
B2
2
a
B2
0

3
3
0
_

_
_

_;
e
B2
3
a
B2
0

3
3
2
_

_
_

_ 3a
e
P
1

a
P

6
p
1
1
2
_

_
_

_; e
P
2

b
P

18
p
3
3
0
_

_
_

_;
e
P
2

b
P

22
p

11
p
cosb arcsin
2
3

11
p
cosb arcsin
2
3

22
p
sinb arcsin
2
3

_
_

_
3b
Thus, the deformation gradient matrix, which corresponds
to variant 9 in Table 1, can be calculated with constrains
T
9
e
B2
i
e
P
i
; i 13 as:
T
9

0:9860 0:0005 0:0001
0:0005 0:9860 0:0001
0:0047 0:0047 0:9806
_

_
_

_ 4
According to the experimental observations (Fig. 4) [2], all
P-phase particles consist of three orientation variants
belonging to the same group (Table 1), with each individual
variant forming {111}
B2
plates in a randomly layered
structure. The three variants are rotated 120 with respect
to each other and have almost equal volume fraction,
which minimizes the strain energy because there is a small
lattice mismatch between two adjacent variants [2]. In or-
der to describe the lattice mist between the precipitate
and the matrix phases at the particle level, the stress-free
transformation strain (SFTS) is calculated by averaging
the three variants in a single P-phase particle, i.e. coarse-
graining the multi-layers into a homogenous structure as
shown in Fig. 6.
The SFTS of the coarse-grained P-phase particles can be
calculated as:
T
111

1
3
T
1
T
5
T
9

0:9860 0:0016 0:0016
0:0016 0:9860 0:0017
0:0016 0:0016 0:9806
_

_
_

_
5a
Fig. 4. High-angle annular dark eld image of the precipitate along the
zone axis directions h112i [2].
Fig. 5. HAADF image of an interface between the P-phase precipitate
(along a 0

10 zone, which is equivalent to a h110i zone of the B2 matrix)


and the B19 martensite. The continuity of lattice fringes from the
martensite to the precipitate indicates that the precipitate is coherent with
respect to the matrix.
Y. Gao et al. / Acta Materialia 60 (2012) 15141527 1517
e
T
111

T
T
111
T
111
I
2
%
1
3
e
1
e
5
e
9

0:0157 0:0016 0:0016


0:0016 0:0157 0:0016
0:0016 0:0016 0:0157
_

_
_

_ 5b
e
T

111

1
3
e
2
e
7
e
11

0:0157 0:0016 0:0016
0:0016 0:0157 0:0016
0:0016 0:0016 0:0157
_

_
_

_
5c
e
T
1

11

1
3
e
3
e
6
e
12

0:0157 0:0016 0:0016
0:0016 0:0157 0:0016
0:0016 0:0016 0:0157
_

_
_

_
5d
e
T
11

1

1
3
e
4
e
8
e
10

0:0157 0:0016 0:0016
0:0016 0:0157 0:0016
0:0016 0:0016 0:0157
_

_
_

_
5e
where T
(111)
is the averaged deformation gradient matrix
for the three variants within the (111) group and e
(111)
is
the averaged SFTS. I is the identity matrix. Such a
coarse-graining process reduces the 12 variants down to
four, characterized by the SFTS given above. Correspond-
ingly, four order parameters instead of 12 are used in the
phase eld simulations, which will be described in the next
section.
4. Phase eld model
4.1. Energy formulation
According to Landaus theory of phase transformations,
a set of four order parameters, {g
1
, g
2
, g
3
, g
4
}, is introduced
to characterize the B2 to P-phase transformation, with
{g
1
= g
2
= g
3
= g
4
= 0} representing the B2 phase and
{g
i
= 1, g
ji
= 0} (where i = 14 and j = 14) representing
the four coarse-grained P-phase variants. In addition, an
interpolation function hg
i
g
3
i
10 15g
i
6g
2
i
is used
to connect the equilibrium free energy functions of the
two phases through the order parameters. The nal form
of the chemical free energy density in the phase eld model
(in a unit of J m
3
) can be expressed as:
f x
Ni
; x
Pt
; fg
p
; p 14g
V
1
m
G
B2
1

4
p1
hg
p
G
p

4
p1
hg
p

_ _
6
where x
Ni
and x
Pt
are mole fractions of Ni and Pt respec-
tively, V
m
is the molar volume and G
B2
and G
p
are the equi-
librium Gibbs free energies of the B2 matrix and P-phase.
G
B2
at 873 K is tted to available thermodynamic data
(see Section 2) (in a unit of 10
6
J mol
1
):
G
B2
0:4493 2:2897x
Ni
1:6992x
Pt
3:4637x
2
Ni
1:0x
Ni
x
Pt
3:1846x
2
Pt
7
and G
p
could be approximated as a second order polyno-
mial function of Ni and Pt concentrations [22]:
G
p
0:4781 2:3586x
Ni
1:0980x
Pt
3:0x
2
Ni
3:0x
2
Pt
8
where the constants are determined by two conditions: (a) a
common tangent plane going through the equilibrium B2
matrix concentration and the equilibrium P-phase concen-
tration determined from experimental characterization; (b)
the curvature of the polynomial at the equilibrium concen-
tration of the P-phase yielding the desired interfacial energy
between the two phases for the gradient energy coecient
chosen. The two equilibrium free energy functions (Eqs.
(7) and (8)) are plotted in Fig. 7. For an alloy of Ni
0.3
Pt
0.2-
Ti
0.5
, the equilibrium Ni and Pt concentrations determined
by the common-tangent construction are x
B2
Ni
0:285;
x
B2
Pt
0:207 and x
P
Ni
0:375; x
P
Pt
0:167, which are close to
the equilibrium concentrations determined by experiment
at 873 K [2].
Assuming that the lattice mist between the two coexis-
ting phases is associated with both chemical and structural
non-uniformities, the SFTS eld is formulated as a function
of both concentrations and structural order parameters:
e
T
ij
r v
B2
Ni
d
ij
g
1
x
Ni
v
B2
Pt
d
ij
g
2
x
Pt

4
p1
e
ij
phg
p
9
where v
B2
Ni

@a
B2
@x
Ni
1
a
B2
and v
B2
Pt

@a
B2
@x
Pt
1
a
B2
are the lattice expan-
sion coecients of the B2 phase, d
ij
is the Kronecker delta
function, and g
1
and g
2
are two hyperbolic tangent functions,
which are used to describe the concentration dependence of
the B2 lattice parameters:
g
1
x
Ni

2
x
B2
Ni
x
0
Ni
tanh
x
B2
Ni
x
0
Ni
2
x
Ni
x
0
Ni
_ _
_ _
g
2
x
Pt

2
x
B2
Pt
x
0
Pt
tanh
x
B2
Pt
x
0
Pt
2
x
Pt
x
0
Pt
_ _
_ _
:
Based on theoretical calculations of the concentration
dependence of the B2 lattice parameter [23,24], the lattice
constants for Ni
0.5
Pt
0.5
, Ni
0.5
Ti
0.5
and Pt
0.5
Ti
0.5
are 0.2995
nm, 0.3015 nm and 0.3137 nm respectively. By linear inter-
polation, the lattice expansioncoecients can be determined
as: v
B2
Ni
0:092, v
B2
Pt
0:013. The initial composition
Fig. 6. Schematic drawing of the coarse graining assumption.
1518 Y. Gao et al. / Acta Materialia 60 (2012) 15141527
(x
0
Ni
0:3; x
0
Pt
0:2) is chosen as a reference state for the
strain calculations.
The structural dependence of the SFTS, e
ij
(p), in Eq. (9)
can be expressed as:
e
ij
1 e
T
111
v
B2
Ni
x
P
Ni
x
0
Ni
d
ij
v
B2
Pt
x
P
Pt
x
0
Pt
d
ij

0:0126 0:0016 0:0016


0:0016 0:0126 0:0016
0:0016 0:0016 0:0126
_

_
_

_ 10a
e
ij
2 e
T

111
v
B2
Ni
x
P
Ni
x
0
Ni
d
ij
v
B2
Pt
x
P
Pt
x
0
Pt
d
ij

0:0126 0:0016 0:0016


0:0016 0:0126 0:0016
0:0016 0:0016 0:0126
_

_
_

_ 10b
e
ij
3 e
T
1

11
v
B2
Ni
x
P
Ni
x
0
Ni
d
ij
v
B2
Pt
x
P
Pt
x
0
Pt
d
ij

0:0126 0:0016 0:0016


0:0016 0:0126 0:0016
0:0016 0:0016 0:0126
_

_
_

_ 10c
e
ij
4 e
T
11

1
v
B2
Ni
x
P
Ni
x
0
Ni
d
ij
v
B2
Pt
x
P
Pt
x
0
Pt
d
ij

0:0126 0:0016 0:0016


0:0016 0:0126 0:0016
0:0016 0:0016 0:0126
_

_
_

_ 10d
The total elastic energy can then be formulated following
the KhachaturyanShatalov microelasticity theory [20]:
E
elastic

1
2

4
p;q1
_
--
d
3
k
2p
3
B
pq
~nf~g
p
g
k
f~g
q
g

k
11
where the integral is taken in the reciprocal space,
~
k, and~n is
a unit vector in the reciprocal space. Note that
~
k 0 is to be
excluded from the integration, which denes the principle
value of the integral. f~g
p
g
k
is the Fourier transform of
g. The asterisk denotes a complex conjugate. Considering
a system having free external boundaries, B
pq
~n can be
expressed as:
B
pq
~n
0 ~n 0
C
ijkl
e
ij
pe
kl
q n
i
r
p
ij
X
jk
r
q
kl
n
l
~n0
_
12
where X
1
ij
C
iklj
n
k
n
l
, r
p
ij
C
ijkl
e
kl
p
After incorporating the gradient terms, the total free
energy of the system becomes:
F
_
d
3
r f x
Ni
; x
Pt
; fg
p
g
_ _

j
Ni
2
rx
Ni

2

j
Pt
2
rx
Pt

2
_

4
p1
j
g
2
rg
p
_ _
2
_
E
elastic
13
where j
Ni
, j
Pt
and j
g
are the gradient energy coecients
respectively for the Ni and Pt concentrations and the order
parameters.
4.2. Kinetic equations
The CahnHilliard equation and the time-dependent
GinsburgLandau (or AllenCahn equation) are used to
describe the time evolution of the concentration and order
parameter elds [9,16,25]:
1
V
2
m
@x
Ni
@t
r M
Ni
r
dF
dx
Ni
_ _ _ _
1
V
2
m
@x
Pt
@t
r M
Pt
r
dF
dx
Pt
_ _ _ _
14
@g
p
@t
L
dF
dg
p
; p 14
where M
Ni
and M
Pt
are the chemical mobilities of Ni and
Pt, and L is the kinetic coecient characterizing time
evolution of the order parameters.
Fig. 7. Gibbs free energies for B2 phase and P-phase as functions of Ni, Pt compositions in a NiTiPt ternary system.
Y. Gao et al. / Acta Materialia 60 (2012) 15141527 1519
5. Simulation results
As discussed earlier, interfaces between the B2 matrix
and the P-phase precipitates are assumed to be fully coher-
ent in this study. Correspondingly, the interfacial energy is
assumed to be 200 mJ m
2
, which yields a grid size of
l
0
= 1.687 nmin the phase eld simulations. Thus, a compu-
tation cell of 128l
0
128l
0
128l
0
grids represents a phys-
ical system of edge length 216 nm. The elastic constants
used in the simulations for the B2 Ni
30
Pt
20
Ti
50
are:
c
11
= 183 GPa, c
12
= 146 GPa and c
44
= 46 GPa [21]. All
dimensionless parameters used in the simulations are listed
in Appendix B.
5.1. Equilibrium shape of a P-phase precipitate
The equilibrium shape of an isolated P-phase precipitate
obtained from the simulation is shown in Fig. 8. It can be
seen that the precipitate has a slightly distorted cube shape
with smoothly curved edges and corners. Because the elas-
tic constants of the B2 matrix phase have a Zener anisot-
ropy factor (c
11
c
12
)/(2c
44
) = 2.5, and the three normal
strains (the diagonal terms) in the transformation strain
tensor are equal and far greater than the shear components
(o diagonal terms), the predicted habit plane of the pre-
cipitate is {11,1,1}, nearly parallel to {100}, as the elastic
soft directions lie along h100i. Because of the presence of
the shear components in the SFTS, the habit plane normal
deviates from that of {100} by $7.
5.2. Multi-particle correlation
In phase eld simulations, nucleation can be simulated
by using either the Langevin noise terms [26,27] or the
explicit nucleation algorithm [28,29]. In this study the
Langevin noise terms are used to generate microstructures
consisting of multi-particles in the simulations [30]. An
example is shown in Fig. 9. The spatial alignment is caused
by elastic interactions among the precipitates during nucle-
ation and growth, i.e. preferred nucleation and growth
along the elastically soft h100i directions [30]. Both the
shape and spatial alignment of the precipitates predicted
by the simulations agree well with the TEM images of
the P-phase precipitates shown in Fig. 1.
5.3. Concentration eld around a P-phase particle
The equal-concentration contours around a P-phase parti-
cle (on a (001) center cross-section) obtained from the phase
eld simulation are shown in Fig. 10. Comparing to the initial
composition of Ni
0.3
Pt
0.2
Ti
0.5
, the precipitation of P-phase
particles that have a composition of Ni
0.375
Pt
0.167
Ti
0.458
will
leave the matrix with lower Ni and higher Pt concentrations.
As shown clearly in Fig. 10a and b, the B2 matrix is depleted
in Ni and enriched in Pt near a growing P-phase particle.
When the particle reaches its equilibrium volume fraction,
the Ni and Pt concentrations in the B2 matrix become more
or less uniform(Fig. 10c and d). However, noticeable Ni con-
centration non-uniformity exists as shown in Fig. 10c. This is
caused by the elastic interaction between Ni atoms and the
coherency stress associated with the precipitates. Since the
principal SFTSs of the P-phase are all negative and the lattice
expansion coecients of the B2 phase are v
B2
Ni
0:092 and
v
B2
Pt
0:013; higher Ni and Pt concentrations around a P-
phase particle, in particular along the elastically soft direc-
tions, are expected as compared to the equilibriumconcentra-
tions of the B2 phase predicted by the thermodynamic
database. Because of the relatively small lattice mismatch
between the precipitate and matrix phases, concentration
non-uniformity caused by the elastic interactions around an
equilibrium precipitate, $0.01% for Ni and 60.001% for Pt,
are negligible as compared to Ni depletion and Pt enrichment
in the matrix associated with precipitate nucleation and
growth (Fig. 10a and b).
The maximum Pt concentration in the B2 matrix near a
growing P-phase precipitate is monitored as a function of
aging time (Fig. 11). The initial microstructure of the simula-
tion consists of a supercritical particle of $10 nm diameter.
The location of the maximum Pt concentration is found to
be near the corner of the cube-like precipitate. As one can
see from Fig. 11, the maximum Pt concentration in the B2
matrix increases with aging time with a decreasing rate, and
slowly approaches the equilibrium concentration. Note that
the GibbsThompsoneect, under the assumptionof a spher-
ical particle with similar size, will change the equilibrium Pt
concentration from 0.207 (given by the common-tangent
construction of the free energy surfaces in Fig. 7) to 0.2067.
5.4. Stress eld around a P-phase particle
Since the three principal SFTSs of a P-phase particle are
all negative, there are both compressive and tensile stresses
around the particle and mostly tensile stresses inside the
particle. Fig. 12 shows the variation of dierent stress com-
ponents around a P-phase particle. Since the stress compo-
nents r
13
and r
23
are orders of magnitude smaller than the
others, they are not included in Fig. 12. Signicant elastic
eects of P-phase precipitate on the martensitic transfor-
mation in the B2 matrix phase will be discussed below.
6. Discussion
6.1. Eect of Pt concentration in B2 on martensitic
transformation
The change of concentration in the B2 matrix caused by
P-phase precipitation (which depletes Ni while enriching Pt
in the matrix) shown in Figs. 10 and 11 may have a pro-
found impact on the M
s
temperature. Based on experimen-
tal measurements, the M
s
temperatures of non-aged
Ni
0.3
Pt
0.2
Ti
0.5
and Ni
0.2
Pt
0.3
Ti
0.5
are 259 C and 530 C,
respectively [3,31,32]. Within this temperature range, the
M
s
temperature increases almost linearly with Pt concen-
tration [33]. Thus a 0.1 at.% dierence in Pt concentration
1520 Y. Gao et al. / Acta Materialia 60 (2012) 15141527
Fig. 8. Three-dimensional views of the equilibrium shape of a P-phase particle (variant 1) obtained by the phase eld simulation.
Fig. 9. Multi-particle simulation results generated by Langevin noise induced nucleation in an initially supersaturated matrix.
Fig. 10. Concentration elds (in atomic fractions) around a P-phase particle along its central (100) cross-section. (a) Ni contour around a growing
precipitate; (b) Pt contour around a growing precipitate; (c) Ni contour around an equilibrium precipitate; (d) Pt contour around an equilibrium
precipitate.
Y. Gao et al. / Acta Materialia 60 (2012) 15141527 1521
could lead to a 2.7 C change in M
s
temperature. As a
result, the M
s
temperature for a B2 matrix with 20.7 at.%
Pt is estimated to be $278 C.
6.2. Eect of stress elds on martensitic transformation
Like any other stress-carrying defects such as disloca-
tions, the stress eld associated with the coherent P-phase
precipitates may aect the martensitic transformation of
the B2 matrix phase as well [9,3440]. To investigate this
eect, the elastic interaction energy between the P-phase
precipitate and a nucleating martensitic particle is calcu-
lated according to the following equation:
E
int
r
ij
e
M
ij
p 15
where r
ij
is the stress eld generated by the coherent P-
phase precipitates and e
M
ij
p is the SFTS of martensitic var-
iant p. A negative value of the interaction energy means
that the formation of martensite is favored by the presence
of the P-phase precipitates.
Since the space groups of B2 and B19 are Pm

3m and
Pmma, respectively, there are three twofold rotation axes
and identity elements in the intersection group. As a result,
six orientation variants of the martenstic phase can be gener-
ated after the transition. The lattice parameters of the B19
martensitic phases from experimental measurement and ab
initio calculations [4,21] are a
B19
= 0.470 nm,
b
B19
= 0.444 nm and c
B19
= 0.275 nm. The lattice corre-
spondence between the B2 and B19 phases has been pro-
posed as the following [33,41]: 100
B19
! 0

1
B2
; 010
B19
! 01

1
B2
; 001
B19
! 100
B2
. As a result, the SFTS can
be calculated as:
e
M
1
0:0507 0 0:0313
0 0:1014 0
0:0313 0 0:0507
_

_
_

_; e
M
2
0:0507 0 0:0313
0 0:1014 0
0:0313 0 0:0507
_

_
_

_;
e
M
3
0:0507 0:0313 0
0:0313 0:0507 0
0 0 0:1014
_

_
_

_; e
M
4
0:0507 0:0313 0
0:0313 0:0507 0
0 0 0:1014
_

_
_

_;
e
M
5
0:1014 0 0
0 0:0507 0:0313
0 0:0313 0:0507
_

_
_

_; e
M
6
0:1014 0 0
0 0:0507 0:0313
0 0:0313 0:0507
_

_
_

_:
16
The calculation results of the interaction energy are plotted
in Fig. 13. The locations with the most negative interaction
energy ($0.138 J mm
3
or 1.22 kJ mol
1
) are near the
edges of a cube-like shaped precipitate with edge length
of $116 nm. Based on ab initio calculations [21] of the
formation energy of B2 and B19 in (Ni,Pt)Ti system, the
enthalpy change of B2 to B19 transformation in Ni
0.3
Pt
0.2-
Ti
0.5
at 0 K is DH = 6 kJ mol
1
.
By assuming that both enthalpy change DH and entropy
change DS do not depend on temperature and the increase
in M
s
is the same as that in T
0
, at which temperature the
driving force of the martensitic transformation is zero,
the change of M
s
temperature by the elastic interaction
of a martensite nucleus with an existing P-phase particle
can be estimated using the data of M
s
(532 K) and A
s
(511 K) from experimental measurements [3]:
T
0

MsAs
2
522 K
DS
DH
T
0
0:0115 kJ=K mol
T
new
0

DH1:22
DS
627K
M
new
s
% M
s
T
new
0
T
0
637 K
17
where M
new
s
is the martensitic transformation start temper-
ature with pre-existing P-phase particles. Comparing with
the chemical driving force for B2 to B19 transformation,
the maximum elastic interaction energy is nearly one-fth
of DH. As a result, the coherency stress of P-phase precip-
itates could make a major contribution to the driving force
for the B2 to B19 martensitic transformation. The M
s
tem-
perature could increase as much as 100 K.
From Fig. 13 it is seen that the edges of the cube-like
shaped particle are the most preferred sites for martensite
nucleation and dierent martensitic variants (indicated by
dierent colors online) are favored at dierent edges. How-
ever, since internally twinned martensitic plates are usually
formed in shape memory alloys, it is expected that internally
twinned martensitic particles consisting of large fractions of
the most favored variant will develop via an autocatalytic
eect [42,43].
Considering the relative contributions from the chemical
non-uniformity and stress eld associated with P-phase
precipitation, the most favorable nucleation site for the
martensitic phase should be located somewhere near the
edge center of the cuboidal P-phase particle. Correspond-
ing NEB calculations [42] for determining the nucleation
site, critical nucleus conguration and activation energy
are underway.
6.3. Eect of aging on M
s
temperature
As compared to the NiTi binary system where the
eects of Ni
4
Ti
3
precipitates on martensitic transforma-
tions have been investigated extensively [9,33,44,45], the
ternary element addition oers the opportunity to alter
the type, shape, spatial distribution and stress state of pre-
cipitates and hence their eects on the strength of the B2
matrix and the behavior of the martensitic transformation.
0.2
0.201
0.202
0.203
0.204
0.205
0.206
0.207
0 500 1000 1500 2000 2500 3000 3500 4000
reduced time
m
a
x
i
m
u
m

P
t

c
o
n
c
e
n
t
r
a
t
i
o
n
Fig. 11. Simulation results of maximum Pt concentration around a P-
phase particle during growth.
1522 Y. Gao et al. / Acta Materialia 60 (2012) 15141527
From experimental observations, the M
s
temperature of
TiNiPt alloys increases signicantly with aging time.
As shown in Fig. 14, the M
s
temperature for a nominal
Ni
0.3
Pt
0.2
Ti
0.5
alloy has a strong and highly non-linear
dependence on the aging time, with a sharp increase at
the initial stage of aging. The above simulation results
and analysis could provide some insight into the physical
origin of such a dependence of M
s
on aging time.
The phase eld simulation results show that both the
chemical non-uniformity (enrichment in Pt) and elastic
stress eld in the B2 matrix generated by P-phase precipita-
tion alter the driving force for the martensitic transforma-
tion and hence alter the M
s
temperature. As shown in
Fig. 11, the Pt concentration in the B2 matrix increases
monotonically as the precipitates grow, reaching a maxi-
mum when the precipitate phase reaches its equilibrium
volume fraction. Meanwhile, as the formation of 0.2% (vol-
ume fraction) martensite is usually dened as the incubation
period [46,47], the 0.2% fraction of the matrix phase with
the lowest elastic interaction energy has been monitored
in our simulation and its cut-o energy is used in the corre-
sponding M
s
temperature calculations. As the P-phase par-
ticles grow, their elastic interaction energy (magnitude) with
martensite nuclei increases sharply at the beginning and
then continues to increase but at a much slower rate, as
shown in Figs. 15 and 16, where the change of the cut-o
elastic interaction energy in the B2 matrix phase is plotted
as a function of particle volume and aging time,
respectively.
Based on the calculated Pt concentration (Fig. 11) and
elastic interaction energy (Fig. 16) around a growing
P-phase particle as a function of aging time, the change
Fig. 12. Stress elds around a P-phase particle: (a) r
11
, (b) r
22
, (c) r
33
and (d) r
12
. Axis 1 is along [100]
B2
; axis 2 is along [010]
B2
and axis-3 is along
[001]
B2
.
Fig. 13. Three-dimensional elastic interaction energy iso-surface (0.1 J mm
3
or 0.886 kJ mol
1
) between a P-phase precipitate and martensite nucleus.
Y. Gao et al. / Acta Materialia 60 (2012) 15141527 1523
of M
s
temperature can be determined. The result is plotted
in Fig. 17. Since the diusivity data for this ternary system
are not available, the time scale of our phase eld simula-
tions is in a reduced unit. Therefore, only qualitative com-
parison with the experimental results can be made. By
comparing Fig. 17 with Fig. 14, it is readily seen that the
general trend of M
s
variation with aging time predicted
by the phase eld simulation agrees well with the experi-
mental measurement. This also indicates that the elastic
interaction between the P-phase precipitates and martens-
itic nuclei may dominate the aging eect on M
s
in this ter-
nary alloy system.
The eects of the P-phase precipitates on the martensitic
transformation found in this ternary alloy are similar to
that found in the NiTi binary system [9,44]. For example,
when coherent Ni
4
Ti
3
particles start to precipitate out, a
jump in M
s
was also observed and M
s
was found to
increase with aging temperature within a certain tempera-
ture range [44].
Comparing the magnitudes of the M
s
changes, the calcu-
lations slightly overestimate the aging eect primarily at
the earliest time increments. This could be due to the fact
that the data used for elastic constants and transformation
enthalpy are obtained from ab initio calculations at 0 K,
which is a rough approximation for the actual values at
the aging temperature. Comparing to the elastic constants
of NiTi binary alloy measured from experiments at room
temperature (C
11
= 162 GPa, C
12
= 129 GPa,
C
44
= 34 GPa) [48], the elastic constants calculated at 0 K
are already signicantly higher. Since the M
s
temperature
is much higher than room temperature, the elastic moduli
may further soften at the transformation temperature. This
will lead to an overestimation of the elastic interaction
energy.
6.4. Limitations of the model
As mentioned above, a homogeneous modulus approx-
imation is used in our simulations for simplicity, which
assumes both the parent and product phases have the same
elastic constants. However, based on the ab initio calcula-
tion results listed in Table 2 [21], the C
11
of the B2 and B19
phases in Ni
31
Pt
19
Ti
50
are signicantly dierent. The ab ini-
tio calculations were performed using the full-potential lin-
earized augmented plane wave method (FLAPW) [49] and
converged with respect to both k-point mesh and repeated-
cell size. A system size of 16 atoms was found to be
sucient in determining the elastic properties. However,
Fig. 14. Experimental result of the eect of aging time at 500 C on the M
s
temperature of a nominal Ni
30
Pt
20
Ti
50
alloy.
Fig. 15. Simulation results of the lowest interaction energies around a
particle with dierent volume.
Fig. 16. Simulation results of the lowest interaction energy around a P-
phase particle as a function of reduced aging time.
0
20
40
60
80
100
120
0 500 1000 1500 2000 2500 3000 3500
reduced time
c
h
a
n
g
e

o
f

M
s

t
e
m
p
e
r
a
t
u
r
e

(
K
)
Fig. 17. Change of M
s
temperature as a function of reduced time
estimated from phase eld simulation results.
1524 Y. Gao et al. / Acta Materialia 60 (2012) 15141527
the ground state C
0
elastic constant of the B2 structure was
found to be unstable while the B19 structure exhibited
comparable elastic constants to that of the binary alloy.
In future work, a more accurate model [5052], includ-
ing elastic modulus inhomogeneity, will be applied to cal-
culate the stress eld provided that the elastic constants
of both B2 phase and P-phase at high temperature are
known.
In the present calculation of the SFTS for the P-phase, a
coarse-graining assumption has been made in which the
SFTSs of three monoclinic variants are averaged. To
document the eect of such coarse-graining in SFTS calcu-
lations on the coherency stress eld around a precipitate,
parallel simulations have been carried out using respectively
the averaged SFTS of three variants and the individual
SFTS of the three variants in a layered structure (Fig. 18).
The results show similar stress eld distributions around
the P-phase precipitate in the two cases. Slight dierences
can be seen near the P-phase interface which is due to the
laminate structure. Considering the fact that the layer thick-
ness in the P-phase particle is only $1.4 nm and the typical
dimension of the P-phase particles ($200 nm) far exceeds
the layer thickness, the eect of such coarse-graining may
be neglected. However, if the particle size is comparable
to the layer thickness, then the coarse-graining may result
in larger errors.
7. Summary
A phase eld model that describes P-phase precipitation
in Ti(Ni,Pt) ternary high temperature shape memory alloys
is developed. The major model inputs such as lattice param-
eters, precipitatematrix orientation relationship, elastic
constants and free energy data are determined based on
experimental characterization, ab initio calculations and
thermodynamic databases. The dilatational components
of the stress-free transformation strain (SFTS) of the P-
phase precipitates are found to be 1.57% and the shear
components to be 0.16%. Under the assumption of isotropic
interfacial energy, the simulation results show that the P-
phase particles have slightly distorted cuboidal shapes and
are aligned along the elastically soft h100i directions. The
predicted particle shape and spatial alignment agree well
with experimental observations.
The concentration non-uniformity around both a grow-
ing precipitate and an equilibrium one has been determined
quantitatively. Around a growing P-phase particle, the
enrichment in Pt and depletion in Ni in the B2 matrix could
lead to an increase in the M
s
temperature up to $19 C.
The coherency stress elds caused by the P-phase precip-
itates are calculated and their eects on the B2 to B19
martensitic transformation are investigated by calculating
the elastic interaction energy between a B19 martensite
nucleus and the existing P-phase precipitates. It is shown
that the elastic interaction could contribute as much as
1.22 kJ mol
1
to the driving force for the B2 to B19 transfor-
mation, by which the M
s
temperature could increase as
much as 100 C.
Considering both the chemical and mechanical eects
analyzed above, the variation of M
s
temperature during
aging is obtained. It is found that the magnitude of the
elastic interaction energy between the ne P-phase precipi-
tates and martensite nucleus increases sharply at early
stages of precipitation, which may be responsible for the
experimentally observed sharp increase in M
s
temperature
at the initial stage of aging.
In conclusion, ne coherent precipitates could be
utilized in practice as an ecient way of increasing the
M
s
temperature in shape memory alloys. This is indeed
the case in the recently developed high temperature shape
memory alloy NiTiPt [3,4], and may also have relevance
for NiTi with ternary additions of Pd and Hf [5,6,5355].
Table 2
Elastic constants of NiTiPt (GPa) [21].
NiTi-B2 Ni
31
Pt
19
Ti
50
-B2 Ni
31
Pt
19
Ti
50
-B19
C
11
183 167 242
C
12
146 168 163
C
44
46 53 51
Fig. 18. Stress elds (r
11
) around a P-phase particle calculated using (a) an averaged stress-free transformation strain (SFTS) and (b) SFTSs of individual
variants having a laminate structure.
Y. Gao et al. / Acta Materialia 60 (2012) 15141527 1525
Acknowledgements
This work was supported by the NASA Fundamental
Aeronautics Program, Supersonics Project (Dale Hopkins,
API), NSF under grant DMR1008349 (YW) and US
Department of Energy, Oce of Basic Energy Sciences un-
der grant DE-SC0001258 (YG, FY and MJM).
Appendix A. Free energy model and parameters of the B2
phase in TiNiPt
The assessed parameters are as listed (unit: J mol
1
,
298 K < T < 1728 K) below.
G
0
Ni;Ni
16645:85228:596T 44:192T lnT 0:0096814T
2
G
0
Pt;Pt
26256:138243:976552T 49:1052T ln
T 0:00496494T
2
4:027610
8
T
3
15948T
1
G
0
Ti;Ti
23195:7122:197944T 3:167T lnT 0:00822826T
2
G
0
Ni;Pt
9961:694236:286276T 46:6486T ln
T 0:00732367T
2
2:013810
8
T
3
7974T
1
G
0
Ni;Ti
71442:08257:929208T 46:0893T ln
T 0:009618675T
2
1:0671610
7
T
3
72636T
1
G
0
Pt;Ti
63428:513:001T
G
0
Pt;Ni
9961:694236:286276T 46:6486T ln
T 0:00732367T
2
2:013810
8
T
3
7974T
1
G
0
Ti;Ni
71442:08257:929208T 46:0893T ln
T 0:009618675T
2
1:0671610
7
T
3
72636T
1
G
0
Ti;Pt
89500:675129:232248T 26:1361T
lnT 0:00163116T
2
2:013810
8
T
3
7974T
1
L
0
NiTi;Ni
103000 32T
L
1
NiTi;Ni
23000
L
0
Ni;NiTi
103000 32T
L
1
Ni;NiTi
23000
L
0
Ni;NiPt
45000
L
0
Pt;NiTi
45000
L
0
Ti;NiTi
100000
L
0
PtTi;Pt
125052:7 39:631T
L
0
Pt;PtTi
34135:1 24:302T
L
0
Ti;PtTi
34:135:1 24:302T
L
0
NiTi;Ti
100000
L
0
PtTi;Ti
125052:7 39:631T
Appendix B. Dimensionless parameters used in simulations
G
B2
0:89864:5794x
Ni
3:3984x
Pt
6:9273x
2
Ni
2:0x
Ni
x
Pt
6:3692x
2
Pt
G
p
0:95614:7172x
Ni
2:1960x
Pt
6:0x
2
Ni
6:0x
2
Pt
j
Ni
20
j
Pt
20
j
g
0:2
c
11
91:084
c
12
72:668
c
44
22:895
V
2
m
M
Ni
1:0
V
2
m
M
Pt
1:0
L 5:0
References
[1] Ronald Noebe, Glen Bigelow, Darrell Gaydosh et al. In: Grummon
DS, Mitchell MR, Mertmann M, editors. SMST 2010: proceedings of
the international conference on shape memory and superelastic
technology, Pacic Grove, California. New York: Springer; 2010.
[2] Kovarik L, Yang F, Garg A, Diercks D, Kaufman M, Noebe RD,
et al. Acta Mater 2010;58:4660.
[3] Noebe R, Draper S, Gaydosh D, Garg A, Lerch B, Penney N,
et al. In: Berg B, Mitchell MR. Proft J, editors. SMST 2006:
proceedings of the international conference on shape memory and
superelastic technologies. Materials Park (OH): ASM International;
2008. p. 409.
[4] Noebe R, Gaydosh D, Padula II S, Garg A, Biles T, Nathal M, et al.
In: Armstrong WD, editor. Smart structures and materials 2005:
active materials: behavior and mechanics, SPIE proceedings, vol.
5761. Bellingham (WA): SPIE; 2005.
[5] Sasaki TT, Hornbuckle BC, Thompson GB, Weaver ML, Noebe RD,
et al. IFES 2010 - 52nd International Field Emission Symposium,
Sydney, Australia: Imaging & Microscopy; 2010.
[6] Bigelow GS, Garg A, Padula SA, et al. Scripta Mater 2011;64:725.
[7] Wang YZ, Chen LQ. Simulation of microstructural evolution using
the eld method. New York: Wiley; 2000.
[8] Wang YZ, Li J. Acta Mater 2010;58:1212.
[9] Zhou N, Shen C, Wagner M, Eggeler G, Mills MJ, Wang Y, et al.
Acta Mater 2010;58:6685.
[10] Boettinger WJ et al. Annu Rev Mater Res 2002;32:163.
[11] Chen LQ. Annu Rev Mater Res 2002;32:113.
[12] Moelans N, Blanpain B, Wollants P. Calphad- Comput Coupl Phase
Diagrams Thermochem 2008;32:268.
[13] Steinbach I. Modell Simul Mater Sci Eng 2009;17:073001.
[14] Zhou N, Shen C, Mills MJ, Li J, Wang Y, et al. Acta Mater
2011;59:3484.
[15] Zhou N, Shen C, Mills MJ, Wang Y, et al. Acta Mater 2007;55:5369.
[16] Cahn JW, Hillard J. J Chem Phys 1958;28:258.
[17] Bellen P, Kumar KCH, Wollants P, et al. Z Metall 1996;87:972.
[18] Mei Li, Hana Wei, Changrong L, et al. J Alloys Compd 2008;461:189.
[19] Lu XG, Sundman B, Agren J, et al. Calphad Comput Coupl Phase
Diagrams Thermochem 2009;33:450.
[20] Khachaturyan AG. Theory of structural transformations in sol-
ids. New York: Wiley-Interscience; 1983.
[21] Hatcher N, Kontsevoi OY, Freeman AJ, et al. In: ESOMAT 2009:
proceedings of the 8th European symposium on martensitic trans-
formations. Prague: EDP Science; 2009.
[22] Hu SY, Murray J, Weiland H, et al. Comput Coupl Phase Diagrams
Thermochem 2007;31:303.
[23] Che XL, Li JH, Dai Y, et al. Sci China Ser E Technol Sci
2009;52:2681.
[24] Bozzolo G, Mosca H, del Grosso M, et al. Intermetallics 2008;16:668.
[25] Wang Y, Banerjee D, Su CC, et al. Acta Mater 1998;46:2983.
[26] Wang Y, Wang HY, Chen LQ, Khachaturyan AG, et al. J Am Ceram
Soc 1995;78:657.
1526 Y. Gao et al. / Acta Materialia 60 (2012) 15141527
[27] Shen C, Wang Y, in: Furrer DU, Semiatin SL, editors. ASM
handbook, vol. 22A. Fundamentals of modeling for materials
processing. Materials Park (OH): ASM International; 2010.
[28] Simmons JP, Shen C, Wang Y, et al. Scripta Mater 2000;43:935.
[29] Simmons JP, Wen YH, Shen C, Wang Y, et al. Mater Sci Eng A
2004;365:136.
[30] Wang Y, Khachaturyan AG. Philos Mag A 1995;72:1161.
[31] Rios O, Noebe R, Biles T, Garg A, Palczer A, Scheiman D, et al. In:
Armstrong WD, editor. Smart structures and materials 2005: active
materials: behavior and mechanics, SPIE proceedings, vol. 5761.
Bellingham (WA): SPIE; 2005.
[32] Zarinejad M, Liu Y, Tong Y, et al. Intermetallics 2009;17:914.
[33] Otsuka K, Ren X. Prog Mater Sci 2005;50:511.
[34] Khalil-Alla J, Dlouhy A, Eggeler G. Acta Mater 2002;50:4255.
[35] Gall K, Sehitoglu H, Chumlyakov YI, Kireeva IV, Maier HJ. J Eng
Mater Technol Trans ASME 1999;121:19.
[36] Gall K, Sehitoglu H, Chumlyakov YI, Kireeva IV, Maier HJ. J Eng
Mater Technol Trans ASME 1999;121:28.
[37] Tirry W, Schryvers D. Nat Mater 2009;8:752.
[38] Tirry W, Schryvers D. Acta Mater 2005;53:1041.
[39] Ke C, Ma X, Zhang XP, et al. Acta Mater Sin 2010;46:921.
[40] Wang QY, Zheng YF, Liu Y. Mater Lett 2011;65:74.
[41] Huang XY, Ackland GJ, Rabe KM, et al. Nat Mater 2003;2:307.
[42] Shen C, Li J, Wang Y, et al. Metall Mater Trans A 2008;39A:976.
[43] Wang Y, Khachaturyan AG. Mater Sci Eng A Struct Mater Prop
Microstruct Process 2006;438:55.
[44] Zheng Y, Jiang F, Li L, et al. Acta Mater 2008:56;736.
[45] Guo W et al. Acta Mater 2011;59:3287.
[46] Entwisle AR. Metall Mater Trans B 1971;2:2395.
[47] Shih CH, Averbach BL, Cohen M, et al. J Metals Trans AIME
1955;203:709.
[48] Mercier O, Melton KN, Gremaud G, Hagi J, et al. J Appl Phys
1980;51:1833.
[49] Wimmer E, Krakauer H, Weinert M, et al. Phys Rev B 1981;24:864.
[50] Wang YU, Jin YM, Khachaturyan AG, et al. J Appl Phys
2002;92:1351.
[51] Shen Y, Li Y, Li Z, et al. Scripta Mater 2009;60:901.
[52] Zhou N, Shen C, Mills MJ, et al. Acta Mater 2008;56:6156.
[53] Bigelow GS, Padula SA, Garg A, et al. Metall Mater Trans A
2010;41A:3065.
[54] Shimizu S, Xu Y, Okunishi E, Tanaka S, Otsuka K, Mitose K, et al.
Mater Lett 1998;34:23.
[55] Meng XL, Cai W, Zheng YF, Zhao LC, et al. Mater Sci Eng Struct
Mater Prop Microstruct Process 2006;438:666.
Y. Gao et al. / Acta Materialia 60 (2012) 15141527 1527

Anda mungkin juga menyukai