Thermodynamic Analysis of the Ni-Si-Ti System

Using Thermochemical Properties Determined from Ab Initio Calculations

Tatsuya Tokunaga
1
, Koji Hashima
2
*
, Hiroshi Ohtani
2
and Mitsuhiro Hasebe
2
1
Fukuoka Industrial Technology Center, Kitakyushu 807-0831, Japan
2
Department of Materials Science and Engineering, Kyushu Institute of Technology, Kitakyushu 804-8550, Japan
A thermodynamic analysis has been carried out on the Ni-Si-Ti ternary system using the CALPHAD method. A regular solution
approximation based on the sublattice model was adopted to describe the Gibbs energy for the individual phases in the binary and ternary
systems. The thermodynamic parameters for each phase were evaluated using available experimental data on the phase boundaries and other
related thermochemical properties. In addition to the experimental data, the enthalpy of formation for some binary and ternary compound phases
as determined by ab initio calculations was incorporated in the present analysis. There was good agreement between the calculated and the
experimental phase equilibria in the binary and ternary systems.
(Received November 20, 2003; Accepted January 20, 2004)
Keywords: thermodynamic analysis, nickel-silicon-titanium, ab initio calculations, metastable phase, isothermal section, phase equilibria,
calculation of phase diagrams
1. Introduction
The Ni-Si-Ti system contains some attractive compound
phases that can have uses in a variety of applications. For
example, the Ti
5
Si
3
phase is a candidate alloy for a high-
temperature structural material due to its high melting point.
Ni
3
(Si,Ti) alloy is a medium-temperature acid-resistant alloy,
and its environmental embrittlement has been intensely
studied.
1,2)
NiTi alloys are well known as shape memory
alloys, and the eect of Si addition on its transformation
temperatures and precipitated secondary phases has been
studied.
3)
Recently, the brazing of silicon nitride using Ni-Si-
Ti alloys as a brazing ller metal has been reported.
4)
Therefore, to develop these ternary alloys for various
purposes, information on the phase equilibria in the Ni-Si-
Ti system is required. The CALPHAD (CALculation of
PHAse Diagrams) approach
5)
provides a powerful tool for
obtaining information on physical parameters, such as
amounts and compositions of constituent phases, and the
liquidus and solidus temperatures.
In the present study, a thermodynamic analysis of the Ni-
Si-Ti ternary system has been carried out based on available
experimental data on the phase boundaries and thermochem-
ical properties. In addition to the experimental data, the
formation energies of some binary and ternary compound
phases obtained from ab initio calculations were incorporated
in the analysis.
2. Thermodynamic Description
To maintain consistency with other thermodynamic as-
sessments, the pure elements were described using the data
compiled by Dinsdale,
6)
with the Gibbs energy of each
element dened with respect to its stable state at 298.15 K
and 0.1 MPa.
The Gibbs energies of the liquid phase and the terminal
solid solution phases were described using the conventional
regular solution model as follows:
G

m
= x
Ni

Ni
x
Si

Si
x
Ti

Ti
RT(x
Ni
ln x
Ni
x
Si
ln x
Si
x
Ti
ln x
Ti
)
x
Ni
x
Si
L

Ni,Si
x
Ni
x
Ti
L

Ni,Ti
x
Si
x
Ti
L

Si,Ti
x
Ni
x
Si
x
Ti
L

Ni,Si,Ti
. (1)
where

G

i
is the lattice stability, and denotes the Gibbs
energy of element i in the phase state. R is the gas constant,
and x
Ni
, x
Si
, and x
Ti
are the mole fractions of Ni, Si, and Ti,
respectively. The parameter L

i, j
denotes the interaction
energy between elements i and j in the phase. L

Ni,Si,Ti
is
the ternary interaction parameter between elements Ni, Si,
and Ti, respectively. The composition dependencies of the
interaction parameters L

i, j
and L

Ni,Si,Ti
are expressed by eqs.
(2) and (3), respectively.
L

i, j
=
X
n
(x
i
x
j
)
n

n
L

i, j
. (2)
L

Ni,Si,Ti
= x
Ni

0
L

Ni,Si,Ti
x
Si

1
L

Ni,Si,Ti
x
Ti

2
L

Ni,Si,Ti
. (3)
In the case of the Si-Ti binary system, the Gibbs energies of
liquid and the terminal solid solution phases are given by an
equation similar to eq. (1), with x
Ni
= 0.
The Gibbs energy contributions due to ferromagnetic
ordering of the fcc phase were described using the model of
Hillert and Jarl
7)
mag
G

m
= RT ln([

1)f (t). (4)

where
mag
G

m
is the Gibbs energy of magnetic ordering, [ is
the average magnetic moment per atom in the phase
(expressed in units of Bohr magneton), and f (t) is a
polynomial function of t. The term t is dened as T,T

C
,
where T

C
is the critical temperature of the magnetic transition
in the phase. Both [

and T

C
can be concentration
dependent, and represented by an equation similar to eq. (2).
The Gibbs energy of compound phases with some form of
*
Graduate Student, Kyushu Institute of Technology, Present address: Kobe
Steel, Ltd., Shimonoseki 752-0953, Japan.
Materials Transactions, Vol. 45, No. 5 (2004) pp. 1507 to 1514
Special Issue on Advances in Computational Materials Science and Engineering III
#2004 The Japan Institute of Metals
homogeneity range, such as the Ni
3
Ti phase, was described
using the sublattice model.
8)
According to this model, the
Ni
3
Ti phase, for example, is divided into two sublattices,
where the rst sublattice is occupied predominantly by Ni
atoms, and the second one predominantly by Ti atoms, i.e.,
(Ni,Ti)
0.75
(Ti,Ni)
0.25
. The Gibbs energy of the Ni
3
Ti phase is
given by
G = y
I
Ni
y
II
Ni


G
Ni
3
Ti
Ni:Ni
y
I
Ti
y
II
Ti


G
Ni
3
Ti
Ti:Ti
y
I
Ni
y
II
Ti


G
Ni
3
Ti
Ni:Ti
y
I
Ti
y
II
Ni


G
Ni
3
Ti
Ti:Ni
0.75RT(y
I
Ni
ln y
I
Ni
y
I
Ti
ln y
I
Ti
)
0.25RT(y
II
Ni
ln y
II
Ni
y
II
Ti
ln y
II
Ti
) G
ex
. (5)
where

G
Ni
3
Ti
i:j
is the Gibbs energy of the hypothetical Ni
3
Ti
phase, in which all the sites in Sublattice I are occupied by
element i, and all the sites of Sublattice II are occupied by
element j, respectively. The colon separates the constituent
elements in the sublattices. The site fraction of the elements
in each sublattice is denoted by y
i
. The term G
ex
denotes the
excess energy term described by an equation similar to eq.
(2).
Ansara et al.
9)
have shown that the Wagner-Schottky
model and the sublattice formalism are equivalent, and for an
A
m
B
n
phase with the formalism (A,B)
m
(A,B)
n
, for example,
the following parameter constraint must hold.

G
A
m
B
n
A:B


G
A
m
B
n
B:A
=

G
A
m
B
n
A:A


G
A
m
B
n
B:B
. (6)
Therefore, one of the four parameters in eq. (6) is not
independent, and this constraint was used for the NiTi phase.
The Gibbs energies of the ternary compound phases,
Ni
a
Si
b
Ti
c
, were described as follows:
G
Ni
a
Si
b
Ti
c
m
=
a
a b c

G
fcc
Ni

b
a b c

G
diamond
Si

c
a b c

G
hcp
Ti
ZG
f
Ni
a
Si
b
Ti
c
.
(7)
where ZG
f
Ni
a
Si
b
Ti
c
is the Gibbs energy of formation per mole
of atoms of the compound that can be expressed using the
following equation.
ZG
f
Ni
a
Si
b
Ti
c
= A B T. (8)
3. Ab Initio Calculations
There is no experimental data on the enthalpy of formation
for the Ni-Si-Ti ternary compound phases, and a prediction of
the formation energy of these ternary compound phases is
very useful for assessing the phase equilibria of the present
study. Recently, Ohtani et al.,
10)
and Ohtani and Hasebe
11)
have calculated some ordered and compound phases in the
various binary systems using the Full Potential Linearized
Augmented Plane Wave (FLAPW) method.
12)
They have
shown that the calculated results agree satisfactorily with the
experimental values. In the present study, the formation
energy of the Ni-Si-Ti ternary compound phases was
calculated using the FLAPW method with a General Gradient
Approximation (GGA).
13)
Furthermore, the phase stability of
the metastable binary phases was also calculated. This
estimation was also useful in the present analysis, as the
phase stability of the metastable phases cannot be evaluated
experimentally. The variation in the total energy with volume
at constant shape of unit cell was calculated for various Ni-
Si-Ti ternary compound phases. The formation energy,
ZE

form
, for the phase was dened from the total energy,
E

tot
, by subtracting the concentration weighted averages of
the pure elements, as shown in eq. (9)
ZE

form
= E

tot
x

Ni
E
fcc Ni
tot
x

Si
E
diamond Si
tot
x

Ti
E
hcp Ti
tot
. (9)
where x

Ni
, x

Si
, and x

Ti
denote the mole fractions of Ni, Si, and
Ti in the phase, respectively.
The calculation of the total energy for each ternary
compound phase was based on the structural data compiled
by Villars.
14)
In his compilation, information on the atomic
positions for only the E-NiSiTi and V-Ni
4
Si
7
Ti
4
phases was
available, and therefore, the remaining phases referred to the
atomic positions of their prototype structures. In addition to
these ternary phases, calculations on some metastable binary
phases were also carried out. For the Ni
3
Si, Ti
3
Ni, and Ti
3
Si
phases with the D0
24
structure, the c,a ratio was also
optimized. The crystallographic data involved in the present
calculations are shown in Table 1. The R
mt
(mun-tin radius)
used in the present calculations was 2.0 a.u. in all cases of the
elements Ni, Si, and Ti. The plane wave basis set with
R
mt
K
max
= 9.0, which almost corresponds to the cut-o
energy of 20 Ry, was utilized. The term K
max
denotes the
maximum radius of a sphere centered at the origin of
reciprocal space, in which all the reciprocal lattice vectors
taken into the basis set are present. Care was taken to
calculate all the structures using the same k-point grid, which
included 2000 points in the rst Brillouin zone.
Each value of the formation energies referred to as, fcc Ni,
diamond Si, and hcp Ti, and the equilibrium lattice
parameters corresponding to the minimum total energy are
listed in Table 2. For the NiTi and Ni
3
Ti phases, the
experimental values of the formation energies
15)
are also
listed in Table 2.
4. Thermodynamic Analysis and Calculation of Phase
Diagrams
4.1 Ni-Si binary system
A thermodynamic analysis on this system has been
previously carried out by the authors group,
16)
which was
based on available experimental data and the results obtained
from ab initio energetic calculations. In that analysis,
although the Gibbs energy for the fcc phase was modeled
using a single equation to describe the order-disorder
transformation, the ordered fcc phase, Ni
3
Si([
1
), was treated
as an independent phase in this study to extrapolate this
system more easily to a higher order system. The phase
stability for a hypothetical Si
3
Ni phase with an L1
2
structure
was evaluated using the results obtained from ab initio
calculations.
16)
Except for those of the Ni
3
Si([
1
) phase, all
the parameters previously assessed by Tokunaga et al.
16)
were used. The calculated phase diagram is shown in Fig. 1,
and the parameters for the Ni
3
Si([
1
) phase are listed in
Table 3.
4.2 Ni-Ti binary system
The phase diagram of the Ni-Ti system has been studied
1508 T. Tokunaga, K. Hashima, H. Ohtani and M. Hasebe
many times before, and has been evaluated by Murray,
17)
although Ni
3
Ti had no homogeneity range in her evaluation.
Jia et al.,
18)
however, have reported a homogeneity range for
this phase.
The Ni-Ti system has been thermodynamically evaluated
by Kaufman and Nesor,
19)
Saunders,
20)
Liang and Jin,
21)
Bellen et al.,
22)
and Tang et al.
23)
In the work of Kaufman and
Nesor, all the intermediate phases were described as being
the stoichiometric phase. Liang and Jin modeled the NiTi
phase using the Wagner-Schottky model to describe the
homogeneity range, but treated Ni
3
Ti phase as a line
compound. Saunders, Bellen et al., and Tang et al., used
sublattice formalism to account for the homogeneity range of
both compound phases. In their work, the formula
(Ni,Ti)
0.75
(Ni,Ti)
0.25
was used for the Ni
3
Ti phase. On the
other hand, for a NiTi phase with a CsCl-type B2 super-
structure, the formulae (Ni,Va)
0.5
(Ni,Ti)
0.5
, (Ni,Ti)
0.5
-
(Ni,Ti)
0.5
, and (Ni,Ti,Va)
0.5
(Ni,Ti,Va)
0.5
were applied by
Saunders, Bellen et al., and Tang et al., respectively, where
Va denotes a vacancy. In the latter two modeling works, a
single Gibbs energy function for the bcc phase was used to
describe the A2/B2 transformation. According to the
Table 2 Calculated results of the enthalpy of formation and equilibrium
lattice constant for various binary and ternary phases in the Ni-Si-Ti
system.
Enthalpy of formation, ZH Lattice parameters
Phase
(kJ/(mole of atoms)) (nm)
Ab initio Experiment a b c
Binary phases
NiSi(B2) 25.99 0.28541
NiTi(B2) 34.59 33.1 1.1
15)
0.30165
SiTi(B2) 26.67 0.31161
Si
3
Ni(L1
2
) 20.60
16)
0.37446
Ni
3
Ti(L1
2
) 45.10 0.36255
Ti
3
Ni(L1
2
) 8.04 0.39458
Si
3
Ti(L1
2
) 4.51 0.38409
Ti
3
Si(L1
2
) 41.57 3.9753
Ni
3
Si(D0
24
) 40.02 0.49641 0.81908
Ni
3
Ti(D0
24
) 45.99 42.2 1.2
15)
0.51327 0.83577
Ti
3
Ni(D0
24
) 5.69 0.55958 0.90987
Ti
3
Si(D0
24
) 42.33 0.56223 0.91503
Ternary phases
G-Ni
16
Si
7
Ti
6
54.67 1.13232
G
/
-Ni
3
SiTi
2
59.20 0.48235 0.76027
V-Ni
4
Si
7
Ti
4
68.45 1.26079 0.49678
E-NiSiTi 81.86 0.61619 0.36779 0.70327
700
800
900
1000
1100
1200
1300
1400
1500
T
e
m
p
e
r
a
t
u
r
e
,

T
/

C
Ni 20 40 60 80 Si
Silicon Content, c
Si
(mol%)
Liquid
(Ni)
Ni
3
Si(
1
)
N
i
S
i
2
N
i
S
i

3
Ni
5
Si
2
()
Ni
2
Si()
Ni
2
Si()
Ni
3
Si
2
()
Fig. 1 The calculated Ni-Si phase diagram.
Table 1 Crystallographic data of various binary and ternary phases in the Ni-Si-Ti system.
Phase Strukturberichte
Pearson
Prototype
Lattice parameters (nm)
symbol
a b c
Binary phases
NiSi B2 cP2 ClCs
NiTi B2 cP2 ClCs 0.3007
SiTi B2 cP2 ClCs
Si
3
Ni L1
2
cP4 AuCu
3

Ni
3
Ti L1
2
cP4 AuCu
3

Ti
3
Ni L1
2
cP4 AuCu
3

Si
3
Ti L1
2
cP4 AuCu
3

Ti
3
Si L1
2
cP4 AuCu
3

Ni
3
Si D0
24
hP16 Ni
3
Ti
Ni
3
Ti D0
24
hP16 Ni
3
Ti 0.51010 0.83067
Ti
3
Ni D0
24
hP16 Ni
3
Ti
Ti
3
Si D0
24
hP16 Ni
3
Ti
Ternary phases
G-Ni
16
Si
7
Ti
6
cF116 Mn
23
Th
6
1.1187
G
/
-Ni
3
SiTi
2
hP12 MgZn
2
0.479 0.755
V-Ni
4
Si
7
Ti
4
tI60 Co
4
Ge
7
Zr
4
1.25229 0.49343
E-NiSiTi oP12 Co
2
Si 0.61484 0.36698 0.70173
Thermodynamic Analysis of the Ni-Si-Ti System 1509
Table 3 Evaluated thermodynamic parameters in the Ni-Si-Ti system.
Thermodynamic parameters, J/mol Reference
Liquid: model (Ni, Si, Ti)
1
L
Liquid
Si,Ti
= 244329.45 29.01807T (x
Si
x
Ti
) (69588.28 32.30618T)
37)
(x
Si
x
Ti
)
2
(53629.71)
L
Liquid
Ni,Si,Ti
= 70000 x
Ni
60000 x
Si
70000 x
Ti
Present work
(Ni): model (Ni, Si, Ti)
1
L
(Ni)
Ni,Ti
= 130333.64 20.22423T (x
Ni
x
Ti
) (46714.31)
Present work
T
(Ni)
C
= 633x
Ni
4670x
Ni
x
Ti
22)
[
(Ni)
= 0.52x
Ni
(Ti): model (Ni, Si, Ti)
1
L
(oTi)
Si Ti
= 243707.25 70.07653T
37)
(Ti): model (Ni, Si, Ti)
1
L
([Ti)
Si,Ti
= 195012.17 22.49328T
37)
Ni
3
Si(
1
): model (Ni, Si, Ti)
0.75
(Ni, Si, Ti)
0.25

G
Ni3Si([1)
Ni:Ni
=

G
fcc
Ni

G
Ni3Si([1)
Ni:Si
= 49886.12 5.53901T 0.75

G
fcc
Ni
0.25

G
diamond
Si

G
Ni3Si([1)
Si:Ni
= 20600 0.25

G
fcc
Ni
0.75

G
diamond
Si

G
Ni3Si([1)
Si:Si
=

G
fcc
Si

G
Ni3Si([1)
Ni:Ti
= 889.15

G
Ni3Ti
Ni:Ti

G
Ni3Si([1)
Ti:Ni
= 8040.32 0.25

G
fcc
Ni
0.75

G
hcp
Ti
Present work

G
Ni3Si([1)
Ti:Ti
=

G
fcc
Ti

G
Ni3Si([1)
Si:Ti
= 4507.36 0.75

G
diamond
Si
0.25

G
hcp
Ti

G
Ni3Si([1)
Ti:Si
= 41574.11 0.25

G
diamond
Si
0.75

G
hcp
Ti
0
L
Ni3Si([1)
Ni,Si:Ni
=
0
L
Ni3Si([1)
Ni,Si:Si
= 50000
0
L
Ni3Si([1)
Ni:Ni,Si
=
0
L
Ni3Si([1)
Si:Ni,Si
= 3016.11 6.08310T
0
L
Ni3Si([1)
Ni:Si,Ti
= 10500
Ni
3
Si(
2
): model (Ni)
0.75
(Si, Ti)
0.25

G
Ni3Si([2)
Ni:Ti
= 6247.8275

G
Ni3Si([1)
Ni:Ti
Present work
0
L
Ni3Si([2)
Ni:Si,Ti
= 1,4 (2322.50 67.5T)
Ni
3
Si(
3
): model (Ni)
0.75
(Si, Ti)
0.25

G
Ni3Si([3)
Ni:Ti
= 65 0.04675T

G
Ni3Si([2)
Ni:Ti
Present work

L
Ni3Si([3)
Ni:Si,Ti
= 1,4 (2322.50 67.5T)
NiTi: model (Ni, Si, Va)
0.5
(Ni, Si, Ti)
0.5

G
NiTi
Ni:Ni
=

G
bcc
Ni
Present work

G
NiTi
Va:Ni
= 81198.65 13.702875T 0.5

G
bcc
Ni
25)

G
NiTi
Ni:Ti
= 41847.43 5.40673T 0.5

G
bcc
Ni
0.5

G
bcc
Ti

G
NiTi
Va:Ti
= 39351.22 8.296145T 0.5

G
bcc
Ti

G
NiTi
Ni:Si
=

G
NiTi
Si:Ni
= 25991.99 0.5

G
fcc
Ni
0.5

G
diamond
Si

G
NiTi
Si:Si
=

G
bcc
Si
Present work

G
NiTi
Si:Ti
= 26674.62 0.5

G
diamond
Si
0.5

G
hcp
Ti

G
NiTi
Va:Si
= 111813.89 30.005695T 0.5

G
diamond
Si
0.5 (

G
bcc
Ti


G
hcp
Ti
)
0
L
NiTi
Ni:Ni,Ti
=
0
L
NiTi
Va:Ni,Ti
= 72295.24 23.47071T
1
L
NiTi
Ni:Ni,Ti
=
1
L
NiTi
Va:Ni,Ti
= 24442.75
0
L
NiTi
Ni:Va,Ni
=
0
L
NiTi
Ni,Va:Ti
= 32012.19 13.247095T 25)
0
L
NiTi
Ni:Si,Ti
= 71231.43 8.57143T
Present work
0
L
NiTi
Ni,Si:Ti
= 52154.29 5.71429T
Ni
3
Ti: model (Ni, Ti)
0.75
(Ni, Si, Ti)
0.25

G
Ni3Ti
Ni:Ni
=

G
hcp
Ni

G
Ni3Ti
Ni:Ti
= 41421.96 7.35868T 0.75

G
fcc
Ni
0.25

G
hcp
Ti

G
Ni3Ti
Ti:Ni
= 5688.27 0.25

G
fcc
Ni
0.75

G
hcp
Ti

G
Ni3Ti
Ti:Ti
=

G
hcp
Ti

G
Ni3Ti
Ni:Si
= 40016.20 0.75

G
fcc
Ni
0.25

G
diamond
Si
Present work

G
Ni3Ti
Ti:Si
= 42326.41 0.25

G
diamond
Si
0.75

G
hcp
Ti
0
L
Ni3Ti
Ni:Ni,Ti
=
0
L
Ni3Ti
Ti:Ni,Ti
= 18274.75 16.21288T
0
L
Ni3Ti
Ni,Ti:Ni
=
0
L
Ni3Ti
Ni,Ti:Ti
= 50000
continued on the next page
1510 T. Tokunaga, K. Hashima, H. Ohtani and M. Hasebe
experimental results,
24)
the formula (Ni,Ti,Va)
0.5
(Ni,Ti,
Va)
0.5
should be used to describe the NiTi phase, and the
formula (Ni,Va)
0.5
(Ni,Ti)
0.5
is a good approximation of
reality, because the two sublattice should be identical in the
B2 structure. In the present study, the NiTi phase was treated
as an independent phase, as well as the Ni
3
Si([
1
) phase in the
Ni-Si system, and the formula (Ni,Va)
0.5
(Ni,Ti)
0.5
was
adopted. Except for NiTi, Ni
3
Ti, and the fcc phases, the
thermodynamic parameters of Bellen et al., were used,
because the parameters of Saunders were not revealed, and
those of Tang et al. were based on the results of Saunders.
The parameters for NiTi, Ni
3
Ti, and the fcc phases were
evaluated using all the experimental data employed by Bellen
et al., and the present results obtained from ab initio
calculations. Furthermore, the constraint described by eq.
(6) was introduced for the NiTi phase, where the parameters

G
NiTi
Va:Ni
evaluated by Ansara et al.
25)
was accepted. And the
parameters L
NiTi
Ni,Va:Ni
and L
NiTi
Ni,Va:Ti
as evaluated by Ansara et
al.
25)
were also used. The optimized parameters for NiTi,
Ni
3
Ti, and the fcc phases are listed in Table 3, and the
calculated phase diagram of the Ni-Ti system is shown in
Fig. 2, together with the experimental data.
17,18,2630)
There is
good agreement between the experimental and the calculated
phase boundaries, and the assessed values of enthalpy of
formation for the NiTi and Ni
3
Ti phases agree satisfactorily
with the experimental data
15)
and the results obtained from ab
initio calculations, where the assessed values were
33.33 kJ/mol and 40.04 kJ/mol, respectively.
600
800
1000
1200
1400
1600
1800
T
e
m
p
e
r
a
t
u
r
e
,

T
/

C
Ti 20 40 60 80 Ni
Nickel Content, c
Ni
(mol%)
Ref. 27)
Ref. 30)
Ref. 29)
Ref. 17)
Ref. 18)
Ref. 26)
Ref. 28)
Ni
3
Ti
NiTi
N
i
T
i
2
Liquid
(Ni)
(Ti)
(Ti)
Fig. 2 The calculated Ni-Ti phase diagram.
Thermodynamic parameters, J/mol Reference
0
L
Ni3Ti
Ti:Si,Ti
= 50000
0
L
Ni3Ti
Ni:Si,Ti
= 23768.57 8.57143T
Present work
Ti
3
Si: model (Ti)
0.75
(Si)
0.25

G
Ti3Si
Ti:Si
= 50208.88 3.45426T 0.25

G
diamond
Si
0.75

G
hcp
Ti
37)
Ti
5
Si
3
: model (Si, Ti)
2,8
(Si, Ti)
3,8
(Ti)
3,8

G
Ti5Si3
Si:Si:Ti
= 1,8 (255250.87 73.29542T 5

G
diamond
Si
3

G
hcp
Ti
)

G
Ti5Si3
Ti:Si:Ti
= 1,8 (568823.86 23.18233T 3

G
diamond
Si
5

G
hcp
Ti
)

G
Ti5Si3
Si:Ti:Ti
= 1,8 (353572.99 70.11309T 2

G
diamond
Si
6

G
hcp
Ti
)
37)

G
Ti5Si3
Ti:Ti:Ti
= 1,8 (40000 20T 8

G
hcp
Ti
)
0
L
Ti5Si3
Si,Ti:Si:Ti
=
0
L
Si3Ti5
Si,Ti:Ti:Ti
= 1,8 (400000)
0
L
Ti5Si3
Si:Si,Ti:Ti
=
0
L
Si3Ti5
Ti:Si,Ti:Ti
= 1,8 (41983.41)
Ti
5
Si
4
: model (Ti)
5,9
(Si)
4,9

G
Ti5Si4
Ti:Si
= 1,9 (676945.62 36.72729T 4

G
diamond
Si
5

G
hcp
Ti
)
37)
TiSi: model (Ti)
0.5
(Si)
0.5

G
TiSi
Ti:Si
= 76649.28 6.68905T 0.5

G
diamond
Si
0.5

G
hcp
Ti
37)
TiSi
2
: model (Ti)
1,3
(Si)
2,3

G
TiSi2
Ti:Si
= 1,3 (178743.24 13.18593T 2

G
diamond
Si


G
hcp
Ti
)
37)
G-Ni
16
Si
7
Ti
6
: model (Ni)
16,29
(Si)
7,29
(Ti)
6,29

G
Ni16Si7Ti6
Ni:Si:Ti
= 1,29 (1585300.40 130.88101T 16

G
fcc
Ni
7

G
diamond
Si
6

G
hcp
Ti
)
Present work
G
/
-Ni
3
SiTi
2
: model (Ni)
3,6
(Si)
1,6
(Ti)
2,6

G
Ni3SiTi2
Ni:Si:Ti
= 1,6 (355198.62 7.03677T 3

G
fcc
Ni


G
diamond
Si
2

G
hcp
Ti
)
Present work
V-Ni
4
Si
7
Ti
4
: model (Ni)
4,15
(Si)
7,15
(Ti)
4,15

G
Ni4Si7Ti4
Ni:Si:Ti
= 1,15 (1026811.10 3.13988T 4

G
fcc
Ni
7

G
diamond
Si
4

G
hcp
Ti
)
Present work
E-NiSiTi: model (Ni)
1,3
(Si)
1,3
(Ti)
1,3

G
NiSiTi
Ni:Si:Ti
= 1,3 (245578.62 20.08640T

G
fcc
Ni


G
diamond
Si


G
hcp
Ti
)
Present work
G
//
-Ni
49
Si
37
Ti
14
: model (Ni)
4.9,10
(Si)
3.7,10
(Ti)
1.4,10

G
Ni49Si37Ti14
Ni:Si:Ti
= 1,10 (698457.14 57.14286T 4.9

G
fcc
Ni
3.7

G
diamond
Si
1.4

G
hcp
Ti
)
Present work
H-NiSi
4
Ti
4
: model (Ni)
1,9
(Si)
4,9
(Ti)
4,9

G
NiSi4Ti4
Ni:Si:Ti
= 1,9 (625000.00

G
fcc
Ni
4

G
diamond
Si
4

G
hcp
Ti
)
Present work
Thermodynamic Analysis of the Ni-Si-Ti System 1511
4.3 Si-Ti binary system
The Si-Ti phase diagram has been thermodynamically
assessed by Kaufman,
31)
Murray,
32)
Vahlas et al.,
33)
Gong et
al.,
34)
and Engqvist et al.
35)
The results of these calculations
deviated from experimental data concerning the phase
diagram and/or thermodynamic data, respectively. Recently,
thermodynamic assessments have been carried out based on
new thermodynamic data, stability ranges of the Ti-silicides
and the homogeneity range of Ti
5
Si
3
phase by Seifert et al.,
36)
Tokunaga et al.,
37)
and Seifert.
38,39)
Seifert et al. attempted to
model the liquid phase using both the partially ionic liquid
model
40)
and the sub-regular model, and satisfactorily
reproduced the experimental phase equilibria using only the
partially ionic liquid model. The latter two analyses repro-
duced the experimental results using the sub-regular model
for the liquid phase. In the present study, the parameters
assessed by Tokunaga et al.,
37)
were accepted. In their
analysis, the Ti
5
Si
3
phase was described using a three
sublattice model with the formula (Si,Ti)
2
(Si,Ti)
3
Ti
3
, and
other compound phases were treated as a stoichiometric
phase. The calculated phase diagram is shown in Fig. 3, and
the set of the parameters used is listed in Table 3.
4.4 Ni-Si-Ti ternary system
In the Ni-Si-Ti ternary system, the isothermal sections
have been reported at 750

C, 1000

C, and 1100

C over the
entire composition range by Markiv et al.,
41)
Westbrook et
al.,
42)
and Hu et al.,
43)
respectively. Williams
44)
provided
some modication for the Ni corner of the 1000

C isotherm.
Lutskaya and Alisova
45)
reported on the phase relationship of
NiTi-SiTi section at 800

C. In their results, seven ternary

compound phases were identied in this system: G-
Ni
16
Si
7
Ti
6
,
4144)
G
/
-Ni
3
SiTi
2
,
4143,45)
G
//
-Ni
49
Si
37
Ti
14
,
4143)
V-Ni
4
Si
7
Ti
4
,
4143,45)
E-NiSiTi,
4143,45)
F-Ni
5
SiTi
6
42)
(or
Ni
37
Si
10
Ti
53
41)
), and H-NiSi
4
Ti
4
.
43)
According to the results
of Hu et al., the existence of F-Ni
5
SiTi
6
is doubtful, and they
pointed out that the martensitic transformation of the NiTi
phase, whose phase boundary is not far from the F-Ni
5
SiTi
6
phase, during cooling caused the complexities observed for
the phase. They also reported that all ternary compound
phases, except for the H-NiSi
4
Ti
4
phase, showed some range
in homogeneity. In the present study, the six ternary
compound phases, G-Ni
16
Si
7
Ti
6
, G
/
-Ni
3
SiTi
2
, G
//
-
Ni
49
Si
37
Ti
14
, V-Ni
4
Si
7
Ti
4
, E-NiSiTi, and H-NiSi
4
Ti
4
, were
taken into account, and treated as a stoichiometric phase. As
for the binary compound phases, Ni
3
Ti, NiTi, Ni
3
Si([
1
),
Ni
3
Si([
2
), and Ni
3
Si([
3
), these were treated as a phase
having a solid solubility of the third element, and the
formulae (Ni,Ti)
0.75
(Ni,Si,Ti)
0.25
, (Ni,Si,
Va)
0.5
(Ni,Si,Ti)
0.5
, (Ni,Si,Ti)
0.75
(Ni,Si,Ti)
0.25
, (Ni)
0.75
(Si,
Ti)
0.25
, and (Ni)
0.75
(Si,Ti)
0.25
were adopted, according to
the experimental results, respectively. The remaining phases
were assumed to be a pure binary phase due to a negligible
homogeneity range.
The thermodynamic parameters in the Ni-Si-Ti system
were optimized based mainly on the isothermal section at
1100

C. The results of Li et al.

46)
were also employed to
obtain the solid solubility of Ti in the Ni
3
Si([
1
) phase. In
addition to these experimental data, the formation energies
for various binary metastable and ternary compound phases
obtained from ab initio calculations were incorporated into
the CALPHAD approach. The entropy contribution to the
Gibbs energy for the ternary compound phases and the
interaction parameters causing the dissolution of the third
element into the binary compound phases were adjusted to t
the experimental phase equilibria data. The formation
entropy was assumed to be zero for the metastable binary
phases. For example, the Gibbs energy for the metastable
NiSi phase with the B2 structure, which corresponds the
phase stability of the counter phase for the NiTi phase, was
described using the following equation

G
NiTi
Ni:Si
= A 0.5

G
fcc
Ni
0.5

G
diamond
Si
. (10)
where the term A corresponds to ZE

form
in eq. (9). However,
the formation energy for the metastable Ni
3
Ti phase with the
L1
2
structure obtained from ab initio calculations was more
negative than that with the D0
24
structure assessed thermo-
dynamically in the present analysis, where the assessed value
was 40 kJ/mol. Therefore, the phase stability of the Ni
3
Ti
phase with the L1
2
structure was determined in accordance
with the following equation.

G
Ni
3
Si([
1
)
Ni:Ti


G
Ni
3
Ti
Ni:Ti
= ZE
Ni
3
Ti(L1
2
)
form
ZE
Ni
3
Ti(D0
24
)
form
. (11)
The parameter

G
NiTi
Va:Si
for the NiTi phase in the ternary
system was calculated using the constraint of eq. (6) in the Si-
Ti binary system. The optimized parameters of the ternary
system are listed in Table 3.
The isothermal section diagram at 1100

C is shown in
Fig. 4, together with the experimental isothermal section.
43)
When calculating the isothermal section diagram, the Ti
3
Si
phase was not taken into account, because in practice, the
Ti
3
Si phase only forms after lengthy anneals at high
temperatures, and the Ti
5
Si
3
phase is formed instead. In
reality, the Ti
3
Si phase has not been conrmed in the
experimental ternary isothermal section.
4143)
In Fig. 5, the
calculated isothermal section at 750

C is compared with the

experimental result given by Markiv et al.
41)
As shown in
Figs. 4 and 5, the calculated phase equilibria reproduced the
600
800
1000
1200
1400
1600
1800
2000
2200
T
e
m
p
e
r
a
t
u
r
e
,

T
/

C
Ti 20 40 60 80 Si
Silicon Content, c
Si
(mol%)
Ti
5
Si
3
T
i
5
S
i
4
T
i
S
i
T
i
S
i
2
T
i
3
S
i
Liquid
(Ti)
(Ti)
Fig. 3 The calculated Si-Ti phase diagram.
1512 T. Tokunaga, K. Hashima, H. Ohtani and M. Hasebe
experimental data satisfactorily, except for the phase equi-
libria between the Ni
5
Si
2
(,), Ni
2
Si(o), and G-Ni
16
Si
7
Ti
6
phases at 750

C.
Figure 6 shows a comparison of the calculated isothermal
section diagram in the Ni-rich region with the homogeneity
region of the Ni
3
Si([
1
) phase reported by Li et al.
46)
The
calculated solubility of Ti in the Ni
3
Si([
1
) phase is rather
lower than the experimental result, at 17.0 mol%Ti, and such
a large solubility could not be reproduced simultaneously
with the phase equilibria at 750

C and 1100

C.
Although there is no experimental information on the
liquidus temperatures for comparison, the calculated liquidus
surface projection of the Ni-Si-Ti system is shown in Fig. 7.
5. Conclusions
A thermodynamic analysis of the Ni-Si-Ti ternary system
was carried out by combining the available experimental data
5
10
15
20
25
30
35
40
5 10 15 20 25 30 35 40
Present work
Ref. 46)
(Ni)
+
1
+G
Ni
3
Ti+G+G'
(Ni)+
1
+Ni
3
Ti
Ni
3
Ti+G'
+NiTi

1
+Ni
3
Ti
+G
++G"

1
Ni
S
i
l
i
c
o
n

C
o
n
t
e
n
t
,

c
S
i
(
m
o
l
%
)
Titanium Content, c
Ti
(mol%)
Ni
3
Ti

Fig. 6 Comparison of the calculated and experimental phase equilibria in

the Ni-rich region of the Ni-Si-Ti system at 900

C.
20
40
60
80
20 40 60 80
V
E
G
G'
G"
NiTi
Ni
3
Ti

Ti
5
Si
3
TiSi TiSi
2
Si
(Ni)
(Ti)

H
Ti
5
Si
4
NiTi
2
1200
1200
1
2
0
0
1
1
0
0
1
2
0
0
2
1
0
0
o
C
1
9
0
0
2
0
0
0 1
8
0
0
1
7
0
0
1600
1
6
0
0
1
5
0
0
1500
1
4
0
0
1300
1
3
0
0
1
4
0
0
1300
1
4
0
0
1400
Ni
Si
Ti
S
i
l
i
c
o
n

C
o
n
t
e
n
t
,

c
S
i
(
m
o
l
%
)
Titanium Content, c
Ti
(mol%)
Fig. 7 Calculated liquidus surface projection in the Ni-Si-Ti system.
20
40
60
80
20 40 60 80
V
E
G
G'
G"
NiTi Ni
3
Ti

Ti
5
Si
3
TiSi
(Ni)
(Ti)
Ti
5
Si
4
NiSi
2
NiSi

NiTi
2
TiSi
2
Ref. 41)
(Ni
37
Si
10
Ti
53
)
Ni
Si
Ti
S
i
l
i
c
o
n

C
o
n
t
e
n
t
,

c
S
i
(
m
o
l
%
)
Titanium Content, c
Ti
(mol%)
Fig. 5 Comparison of the calculated and experimental phase equilibria in
the Ni-Si-Ti system at 750

C.
20
40
60
80
20 40 60 80
V
E
G
G'
G"
NiTi Ni
3
Ti
Ti
5
Si
3
TiSi
TiSi
2
(Si)
(Ni)
(Ti)
Liquid
Liquid H
Ti
5
Si
4

Ni
Si
Ti
S
i
l
i
c
o
n

C
o
n
t
e
n
t
,

c
S
i
(
m
o
l
%
)
Titanium Content, c
Ti
(mol%)
(b)
20
40
60
80
20 40 60 80 Ni
Si
Ti
V
E
G
G'
G"
NiTi
Ni
3
Ti
Ti
5
Si
3
TiSi
TiSi
2
(Ni) Liquid
Liquid H
Ti
5
Si
4
S
i
l
i
c
o
n

C
o
n
t
e
n
t
,

c
S
i
(
m
o
l
%
)
Titanium Content, c
Ti
(mol%)

(Ti)
(a)
Fig. 4 (a) Calculated isothermal section of the Ni-Si-Ti system at 1100

C.
(b) Experimental isothermal section of the Ni-Si-Ti system at 1100

C.
43)
Thermodynamic Analysis of the Ni-Si-Ti System 1513
on the phase equilibria, and the results obtained from ab initio
calculations. The results are summarized as follows:
(1) Some modication of the thermodynamic parameters
for the Ni-Si and Ni-Ti binary systems were made based on
available experimental data, and the results obtained from ab
initio calculations. The calculated phase diagrams of the Ni-
Si and Ni-Ti system agree well with the experimental results.
(2) The phase stability for some binary phases and ternary
compound phases was evaluated using ab initio calculations.
The calculated phase diagrams of the Ni-Si-Ti ternary system
agree well with the available experimental results. Therefore,
it is considered that the incorporation of the phase stability
based on ab initio calculations into the conventional
CALPHAD framework is promising or useful.
Acknowledgements
The authors acknowledge the use of the THERMO-CALC
software program for calculating the equilibrium phase
diagrams.
REFERENCES
1) Y. Kaneno, M. Wada, H. Inoue and T. Takasugi: Mater. Trans. 42
(2001) 418421.
2) T. Nakamura, Y. Kaneno, H. Inoue and T. Takasugi: Collected
Abstracts of the 2003 Autumn Meeting of the Japan Inst. Metals,
pp. 487.
3) S. F. Hsieh, S. K. Wu and H. C. Lin: J. Alloy. Compd. 339 (2002) 162
166.
4) M. Naka and H. Takase: Preprints of the National Meeting of JWS,
No. 73 (2003) pp. 314315.
5) N. Saunders and A. P. Miodownik: CALPHAD, (Pergamon, Elsevier
Science Ltd., Oxford, 1998).
6) A. T. Dinsdale: CALPHAD 15 (1991) 317425.
7) M. Hillert and M. Jarl: CALPHAD 2 (1978) 227238.
8) M. Hillert and L.-I. Staansson: Acta Chem. Scand. 24 (1970) 3618
3626.
9) I. Ansara, N. Dupin, H. L. Lukas and B. Sundman: In: P. Nash and B.
Sundman, editors. Applications of Thermodynamics in the Synthesis
and Processing of Materials, Proc. Materials Week 94, Rosemont,
USA, October, 1994, TMS, (1995) 273283.
10) H. Ohtani, Y. Chen and N. Tatara: Ceramics Japan 37 (2002) 529533.
11) H. Ohtani and M. Hasebe: CAMP-ISIJ 16 (2003) 1519.
12) P. Blaha, K. Schwarz, P. Sorantin and S. B. Trickey: Comput. Phys.
Commun. 59 (1990) 399415.
13) J. P. Perdew, J. A. Chevary, S. H. Vosko, K. A. Jackson, M. R.
Pederson, D. J. Singh and C. Fiolhais: Phys. Rev. B 46 (1992) 6671
6687.
14) P. Villars: Pearsons Handbook: Crystallographic Data for Interme-
tallic Phases, (ASM International, Materials Park, OH, 1997).
15) Q. Guo and O. J. Kleppa: J. Alloy. Compd. 269 (1998) 181186.
16) T. Tokunaga, K. Nishio, H. Ohtani and M. Hasebe: CALPHAD 27
(2003) 161168.
17) J. L. Murray: Phase Diagrams of Binary Titanium Alloys, (ASM
International, Materials Park, OH, 1987) 197211.
18) C. C. Jia, K. Ishida and T. Nishizawa: In: J. E. Morral, R. S. Schiman
and S. M. Merchant, editors. Experimental Methods of Phase Diagram
Determination, Pub. TMS, Pittsburgh, PA (1994) 3138.
19) L. Kaufman and H. Nesor: CALPHAD 2 (1978) 81108.
20) N. Saunders: Thermo Tech Ltd., Guilford: Unpublished work, cited in
Ref. 22 (1992).
21) H. Liang and Z. Jin: CALPHAD 17 (1993) 415426.
22) P. Bellen, K. C. H. Kumar and P. Wollants: Z. Metallkd. 87 (1996)
972978.
23) W. Tang, B. Sundman, R. Sandstromand C. Qiu: Acta Mater. 47 (1999)
34573468.
24) R. J. Wasilewski, S. R. Butler, J. E. Hanlon and D. Worden: Metall.
Trans. 2 (1971) 229238.
25) I. Ansara, N. Dupin, H. L. Lukas and B. Sundman: J. Alloy. Compd.
247 (1997) 2030.
26) A. Taylor and R. W. Floyd: J. Inst. Metals 80 (1951) 577587.
27) D. M. Poole and W. Hume-Rothery: J. Inst. Metals 83 (1954) 473480.
28) Yu. A. Bagariatskii and Yu. D. Tyapkin: Zh. Neorg. Khim. 3 (1958)
934935. (in Russian); translated in Russ. J. Inorg. Chem. 3 (1958)
151158.
29) G. R. Purdy and J. G. Parr: Trans. AIME 221 (1961) 636639.
30) G. F. Bastin and G. D. Rieck: Metall. Trans. 5 (1974) 18171826.
31) L. Kaufman: CALPHAD 3 (1979) 4576.
32) J. L. Murray: Phase Diagrams of Binary Titanium Alloys, (ASM
International, Materials Park, OH, 1987) 289294.
33) C. Vahlas, P. Y. Chevalier and E. Blanquet: CALPHAD 13 (1989)
273292.
34) S. F. Gong, A. Robertsson, H. T. G. Hentzell and X.-H. Li: J. Appl.
Phys. 68 (1990) 45354549.
35) J. Engqvist, C. Myers and J.-O. Carlsson: J. Electrochem. Soc. 139
(1992) 31973205.
36) H. J. Seifert, H. L. Lukas and G. Petzow: Z. Metallkd. 87 (1996) 213.
37) T. Tokunaga, K. Nishio and M. Hasebe: Collected Abstracts of the
2001 Autumn Meeting of the Japan Inst. Metals, pp. 250.
38) H. J. Seifert: Max-Planck-Institut fur Metallforschung, Institut fur
Werkstowissenschaft, Stuttgart, Germany: Unpublished work, cited
in Ref. 39.
39) Y. Du, J. C. Schuster, H. J. Seifert and F. Aldinger: J. Am. Ceram. Soc.
83 (2000) 197203.
40) M. Hillert, B. Jansson, B. Sundman and J.

Agren: Metall. Trans. A 16A
(1985) 261266.
41) V. Ya. Markiv, E. I. Gladyshevskii, P. I. Kripyakevich and T. I.
Fedoruk: Izv. Akad. Nauk SSSR, Neorg. Mater. 2 (1966) 13171319.
(in Russian); translated in Inorg. Mater. 2 (1966) 11261128.
42) J. H. Westbrook, R. K. Di Cerbo and A. J. Peat: Rep. 58-rc-2117,
General Electronic Research Laboratory, USA, 1958.
43) X. Hu, G. Chen C. Ion and K. Ni: J. Phase Equilib. 20 (1999) 508514.
44) K. J. Williams: J. Inst. Metals 99 (1971) 310315.
45) N. V. Lutskaya and S. P. Alisova: Izv. Acad. Nauk SSSR, Metally N3
(1992) 194196. (in Russian); translated in Russian Metallurgy N3
(1992) 180182.
46) Y. Li, T. Zhang, Z. Zheng and Y. Zhu: Acta Met. Sin. 26 (1990) A172
A176.
1514 T. Tokunaga, K. Hashima, H. Ohtani and M. Hasebe