You are on page 1of 13

Calphad 60 (2018) 16–28

Contents lists available at ScienceDirect

Calphad
journal homepage: www.elsevier.com/locate/calphad

Revised thermodynamic description of the Fe-Cr system based on an T


improved sublattice model of the σ phase

Aurélie Jacob , Erwin Povoden-Karadeniz, Ernst Kozeschnik
TU Wien, Institute of Materials Science and Technology, Getreidemarkt 9, 1060 Vienna, Austria

A R T I C L E I N F O A B S T R A C T

Keywords: The Fe-Cr system is re-assessed, focusing on an improved modeling of σ phase. The three sublattice model
Calphad (Cr,Fe)10(Cr,Fe)4(Cr,Fe)16 is parameterized to model the σ phase, solving discrepancies between computed and
σ phase experimental site fractions of previous descriptions. Taking into account relative metastability trends of end-
Crystal structure member compounds from first-principles analysis, only two additional interaction parameters of the σ-FeCr
DFT
description were required for the reproduction of the chemical structure, thermodynamic properties and phase
boundaries.

1. Introduction first-principles data have become an invaluable source of crystal


structural chemical data [7], allowing for a revision of the thermo-
A considerable amount of studies has been dedicated to the Fe-Cr dynamic description of the σ phase in order to achieve a better agree-
system due to its technological importance, often related to the occur- ment of phase equilibria, compositions and structural chemistry. For the
rence of the intermetallic topologically close-packed Frank-Kasper σ model revision of σ phase in the present study, all to the authors´
phase. Due to its brittleness and wide solubility potential for various knowledge available phase stability and crystal chemistry data are
elements, formation of σ phase is known to be crucial for materials collected and critically reviewed. A modified sublattice description is
properties and, thus, technological usability of many high-alloyed steel based on structural-chemical data [1,2,8] and new first-principles
grades. Therefore, understanding and prediction of its temperature- and analyses of σ phase compound enthalpies. Parameterization of the Fe-Cr
composition-dependent stability is desirable. Robust phase prediction alloy phases as well as liquid is re-adjusted, correspondently. The re-
within the Calphad spirit requires physically correct thermodynamic sulting improvements of the calculated phase diagram and thermo-
modeling from low-order binary and ternary systems to multi-compo- dynamic data in comparison with previous Calphad assessments
nents. A proper thermodynamic phase description for this aim starts [3–6,9–12] are presented.
with correct reproduction of known crystal structural chemistry, and
more precisely, fractions of different sorts of atoms on different crys- 2. Summary and assessment of previous studies of the Fe-Cr
tallographic sites, which often reveal a distinct coordination number. In system
Calphad modeling of intermetallic solid solutions, this kind of data is
“translated” to sublattice descriptions. Their experimental identifica- Xiong et al. [13] gave an extensive review about the thermo-
tion, however, is exceptionally challenging even by the use of high- dynamics of this system, which has been available until 2010, including
resolution analysis [1,2]. more than ten Calphad assessments. The most accepted version of the
As a consequence, previous sublattice suggestions for the σ-FeCr calculated phase diagram is given by Andersson and Sundman [3] to-
phase [3–6] have often represented only a vague approximation of its gether with a later modification of the liquid description from Lee [12].
real crystal chemistry. Whereas rough model approximations may re- Their thermodynamic modeling [3,12] is commonly used in multi-
produce phase boundaries in binary diagrams well, they become par- component thermodynamic database.
ticularly crucial at extensions to multi-component systems and applied These widely-used assessments [3,12] put up with the following
calculations in technological alloys. For instance, due to insufficiently problems, however. The sublattice description for the σ phase [3] does
precise calculated chemical potentials associated with inaccurately not reflect the site occupancies of elements correctly [8]. Moreover,
modeled site fractions, application for thermo-kinetic precipitation si- their calculated Cr solubility in the low temperature α-Fe phase is too
mulations often tends to inadequate particle evolution. Only recently, small with respect to the miscibility gap (see Fig. 7 of [13]), which


Corresponding author.
E-mail address: aurelie.jacob@tuwien.ac.at (A. Jacob).

http://dx.doi.org/10.1016/j.calphad.2017.10.002
Received 20 July 2017; Received in revised form 15 September 2017; Accepted 1 October 2017
Available online 21 December 2017
0364-5916/ © 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/).
A. Jacob et al. Calphad 60 (2018) 16–28

additionally affects the spinodal decomposition. Further, the proposed Paxton [22] and Dubiel and Inden [21], the eutectoid temperature can
magnetic parameters of Fe-Cr BCC do not reproduce the experimental be accepted to be between 773 and 802 K.
Curie temperature towards higher Cr-alloying (see Fig. 9 from Xiong The enthalpy of formation of the σ phase at T=923 K has been
et al. [13]). determined from adiabatic calorimeter by Dench [24]. Assessed en-
Xiong et al. [4] provide a better description of the miscibility gap as thalpy data by Müller et al. [25] took into account also other collected
well as of the Curie temperature and the magnetic moment. Their as- data [26] and suggested slightly higher values than the results by Dench
sessment of the Fe-Cr system is based on the revised lattice stability of [24].
pure iron from Chen and Sundman [14], providing a physically im-
proved description of magnetism and no thermal discontinuities of the 2.1.2. Crystal structure of σ phase
Gibbs energy function. So far, the revised lattice stabilities have only The σ phase crystallizes as a tetragonal structure in the P42/mnm
been used in the thermodynamic modeling of the Fe-Cr [4] and Fe-C space group. 30 atoms per unit cell are distributed on five different
[15] systems. For proper extension into ternary and multi-component Wyckoff positions [8]. Thus, the formula A212B415C814D812E814 can be
systems, refinement of all involved lattice stabilities would be required. given, where each letter represents a Wyckoff position, the subscript the
However, associated research activities are still ongoing [16], and number of atoms per position and the higher indices the coordination
combination to full descriptions of higher-order systems and im- number.
plementations into more comprehensive thermodynamic databases, The crystal structure of the σ phase in the Fe-Cr system was in-
such as MatCalc-compatible [17] databases being developed by the vestigated by Yakel [1] by XRD measurements of single crystals. He
authors of this study, is not performed at present. Consequently, we are confirmed the space group P42/mnm and determined that each Wyckoff
still relying on the first-generation lattice stabilities from Dinsdale [18]. position is mixed occupied by Fe and Cr. Details of the crystal structure
Xiong et al. [4] used the simplified two-sublattice model suggestion and information on the atomic positions are given in Table 1.
[8] for the σ phase, which, they claimed, would however be unlikely From structural-chemical investigation of samples annealed at dif-
appropriate in multi-component extensions. ferent temperatures up to 1400 h for the lower temperature, Yakel [1]
The re-calculated phase diagram of the Fe-Cr system based on the concluded that temperature does not influence the site preference sig-
present study is shown in Fig. 1. As expected, the FCC γ-loop is observed nificantly. Cieślak et al. [27] studied the site preferences for different
on the Fe-rich side. The BCC phase has a large miscibility gap between compositions in samples at 973 K annealed for up to 6650 h. The results
Fe-rich (BCC) and Cr-rich (BCC’) alloy compounds. σ phase is stable at of these two studies are shown in Fig. 2.
elevated temperatures. The studies [1,27] of site occupancies (Fig. 2) in σ-FeCr reveal that
Fe prefers the A and D positions, whereas Cr sits preferentially on B, C
2.1. σ phase in the Fe-Cr system and E. The two studies were done at different temperatures, T=923 K
[1] and T=973 K [27], respectively, but provide similar results, con-
2.1.1. Phase stability of σ phase firming the suggestion made by Yakel [1].
The σ phase was discovered for the first time in 1907 in the Fe-Cr
system. It is nowadays known to be present in more than forty binary 2.1.3. Thermodynamic modeling
systems [8]. In the Fe-Cr system, the σ phase is stable from ~44 to Several thermodynamic models [3–5,11], distinguished by the
~50 at% Cr [8]. It is formed by a congruent reaction BCC ↔ σ at number of sublattices and atoms per sublattice, have been proposed for
temperatures between 1093 and 1098 K [19,20] and decomposes by a the σ phase in the Fe-Cr system since the establishment of the com-
eutectoid reaction at T=773 K [21] into BCC+BCC’. pound energy formalism (CEF) [28] for non-stoichiometric phases. The
Reported experimental temperatures of the eutectoid reaction are CEF allows for the energetic description of interactions among different
controversial [21–23]. Williams and Paxton [22] determined this value sorts of atoms, sharing the same sublattice, as well as the combination
to be between 783 and 823 K. According to Dubiel and Inden [21], this of different sublattices. This made the CEF an essential pre-requisite for
temperature lies between 773 and 805 K, whereas Koyano et al. [23] the physically proper modeling of intermetallic phases with relatively
determined this value to be below 783 K, but traces of carbides may complex crystallographic order, such as it is the case for the group of
have blurred the results. Dubiel and Inden [21] annealed their C-free topologically close-packed phases including σ phase. In the CEF, the
Fe-Cr samples for four years, and, thus, likely approached the equili- mathematical model divides the phase into different sublattices.
brium state. Considering the temperature range given by Williams and The molar Gibbs energy of a phase can be formulated as:
Gm = G ref + Gid + Gex . (1)
The first term, G ref , represents the surface of reference. In the σ
phase, with full five-sublattice description, each of the five sublattices
represent distinct Wyckoff positions, and G ref is written

G ref = ∑ yA2a yB4f yC8i yD8i ′ yE8j GABCDE


σ
,
ABCDE (2)
x
where yi expresses the site fraction of each constituent on each sub-
lattice.
The second term Gid represent the ideal entropy of mixing and is
defined as

Gid = RT ∑ yi ln (yi ).
i (3)
The last term represents the excess Gibbs energy, i.e. the term
summing up all non-ideal mixing contributions of the phase. This part
represents the focus during the thermodynamic optimization process.
The thermodynamic modeling of σ-FeCr given by Hertzman [11]
Fig. 1. Equilibrium phase stability diagram of Fe-Cr as calculated from the present and Anderson and Sundman [3] was proposed at a time where the
parameters.
quantitative analysis of site distributions in complex intermetallic

17
A. Jacob et al. Calphad 60 (2018) 16–28

Table 1
Crystal structure of σ-CrFe as given by Yakel [1].

Space group P42/mnm (no. 136)


Pearson symbol tP30
Lattice parameters a= 8.7961Å [1]
c=4.5605 Å [1]
Model: FeCr

Wyckoff position x y z CN
2a (blue) – A 0 0 0 12
4f (red) – B 0.39864 x 0 15
8i (orange) – C 0.46349 0.13122 0 14
8i’ (green) – D 0.73933 0.0661 0 12
8j (grey) – E 0.18267 x 0.25202 14

describe the σ phase from 13.3% to 66.7% of A atoms only. Herein, this
model cannot fully reflect the full homogeneity range of the σ phase in
other systems where this phase is also present. Anyway, the model was
applied in Fe-Cr by Chuang et al. [5] and by Westman [6]. The model
has also been used to model σ phase in Ni-V [29] and Mo-Re [7].
Chuang et al.'s [5] data were not used in ternary extensions due to
unclear formulations of model parameters and Westman did not publish
his results. Nevertheless, his parameterization has already been used in
ternary extensions, such as, Fe-Cr-C [31] and Fe-Cr-Nb [32].
Joubert [8] published a two sublattice alternative, which solved the
modeling problem of σ phase composition ranges. Xiong et al. [4] used
this model, i.e. (Cr,Fe)10(Cr,Fe)20, for σ-FeCr. However, Joubert [8] also
noted the simplifications of the model in terms of site distribution.
Xiong et al. [4] also stated some doubt on the model extension to multi-
component systems.
Calculated phase boundaries using Andersson and Sundman [3],
Westman [6] and Xiong et al. [4] descriptions of σ-FeCr are shown in
Fig. 3. Associated site occupancies of Fe and Cr on the different sub-
lattices are also displayed in the Fig. 4.
Fig. 2. Site occupancies of Fe on the different Wyckoff positions in the σ phase according Clearly, results using (Fe)8(Cr)4(Cr,Fe)18 and (Fe)10(Cr)4(Cr,Fe)16
to Yakel [1] and Cieslak [27].

phases was far less accurate than today. Thus, Andersson and Sundman
[3] used, “for convenience”, the three sublattices model
(Fe)8(Cr)4(Cr,Fe)18. This sublattice model allows Fe on the D site, Cr on
the B site and a mixing on A, C and E sites (See Table 1). While the
model properly reproduces the homogeneity range of σ-FeCr, An-
dersson [3] already speculated that it might be inappropriate for Ni-
containing σ phase. Furthermore, this model wrongly assigned the
moles of atoms in the different sublattices [8]. Joubert [8] revealed that
in several systems (e.g. Ni-V [29]), the homogeneity range of σ phase
cannot be represented by (D)8(B)4(A,C,E)18. Nevertheless, the set of
parameters from [3] have been widely used in extensions to ternary [8]
and multicomponent thermodynamic databases (e.g., TCFE, mc_fe).
In 1997, a scientific group [30] proposed a new sublattice model for
the σ phase, (Fe)10(Cr)4(Cr,Fe)16. This description indeed also re-
presented a simplification compared to the real crystallographic case.
Nevertheless, it obeys the rule to combine experimentally determined
Wyckoff positions of same coordination numbers, i.e. (2A+8D)(4B)
(8C+8E). While, in the binary Fe-Cr system, this description requires Fig. 3. Comparison of calculated σ-FeCr stability region according to Andersson and
only two end-members, none of them lies within the composition range Sundman [3], Westman [6] (i.e. only modification of the σ phase compared to Andersson
of the observed stability of σ phase. Moreover, this model allows to and Sundman [3]) and Xiong et al. [4] compared to Cook and Jones [19].

18
A. Jacob et al. Calphad 60 (2018) 16–28

Fig. 4. Calculated site occupancies of Fe and Cr on different


sublattices according to the modeling of a) Andersson and
Sundman [3], b) Westman [6] and c) Xiong et al. [4] compared to
experimental site occupancy [1,27].

[3,6] are close to each other, whereas the σ-FeCr phase boundaries 2.1.4. DFT calculations of σ phase
using (Cr,Fe)10(Cr,Fe)20 [4] deviate significantly. It was mentioned that Sluiter et al. [33] reported calculated enthalpies of formation for σ-
the slight “lining off” to the Fe-rich side is an intrinsic consequence of FeCr within the Ising model in the framework of density functional
the model and does not reproduce an experimental feature [19]. None theory (DFT). All of their calculations were done in a non-magnetic
of the up to now available model parameterizations can represent the state. Later, Havránková et al. [34], by using DFT, confirmed the results
crystal chemistry of the σ-Fe-Cr phase properly, as indicated in Fig. 4 by of Sluiter et al. [33]. Both studies reveal that the enthalpy of formation
the deviations of site occupancies from the experimental ones [1,27]. of the different configurations are positive and that the most stable
Previous modeling alternatives of σ phase can be summarized as configurations are - following the sequence of Wyckoff position
suitable approaches for specific questions in a distinct low-order A212B415C814D812E814 as given in Table 1 - FeCrCrFeCr and FeFeCrFeCr.
system, either the reproduction of experimental phase boundaries, or Site occupancies in σ-FeCr have been studies by DFT calculations
that of site occupancies. For the proper extension of σ phase modeling taking into account magnetism by Kabliman et al. [35] as a function of
to multicomponent systems, it is relevant to cope with both of these temperature for the composition Fe0.5Cr0.5. Their results are consistent
features. A proper approach will be presented in Section 3 of this paper. with the experimental data from Yakel [1] and Cieślak et al. [27].
This, in turn, will reflect a higher physical meaningfulness of the as-
sessed description and guarantee the usefulness of the concerning da- 2.2. Solution phases
tabases for kinetic simulations of σ phase evolution during complex
heat treatments of technological alloys [17]. The total Gibbs energy of the solution phase φ is given the same as
in Eq. (1). Its Gibbs energy of reference is
φ φ
G ref , φ = x Cr GCr + xFe GFe , (4)

19
A. Jacob et al. Calphad 60 (2018) 16–28

Table 2
Calculated enthalpy of formation at T=0 K as obtained from DFT.

Compound according to Wyckoff position Compound according to Ferromagnetic Non magnetic References
A212B415C814D812E814 Calphad model
ΔfH (kJ.mol−1.atom−1) Lattice ΔfH (kJ.mol−1.atom−1) Lattice
parameter (Å) parameter (Å)

Cr2Cr4Cr8Cr8Cr8 Cr10Cr4Cr16 (Cr30) 13.2 – – – [51]


Cr2Fe4Cr8Cr8Cr8 Cr10Fe4Cr16 (Cr26Fe4) 14.11 a=8.735 14.04 a=8.71 Present work
c=4.55 c=4.54
Fe2Cr4Cr8Fe8Cr8 Fe10Cr4Cr16 (Cr20Fe10) 5.62 a=8.69 5.69 a=9.14 Present work
c=4.43 c=6.04
Fe2Fe4Cr8Fe8Cr8 Fe10Fe4Cr16 (Cr16Fe14) 4.62 a=8.64 4.94 a=8.64 Present work
c=4.39 c=4.44
Cr2Cr4Fe8Cr8Fe8 Cr10Cr4Fe16 (Cr14Fe16) 12.35 a=8.6 13.52 a=8.43 Present work
c=4.58 c=4.62
Cr2Fe4Fe8Cr8Fe8 Cr10Fe4Fe16 (Cr10Fe20) 14.89 a=8.66 19.74 a=8.48 Present work
c=4.58 c=4.51
Fe2Cr4Fe8Fe8Fe8 Fe10Cr4Fe16 (Cr4Fe26) 15.27 a=8.44 18.40 a=8.42 Present work
c=4.54 c=4.44
Fe2Fe4Fe8Fe8Fe8 Fe10Fe4Fe16 (Fe30) 17.3 – – – [51]
Fe10Fe4Fe16 (Fe30) – – 25.8 – [54]

attempts to fit both low and high temperature ΔHmix [24] resulted in a


wrong solubility of Cr in α-Fe and a too high stability of the miscibility
gap α-α’. The thermodynamic modeling of Xiong et al. [4] is, in prin-
ciple, in agreement with experimental data [21,39,40] and computed
results from Eich et al. [10]. The results from [10] only deviate from the
experimental data of Kuwano [39] at high Cr content for the compo-
sition range between 0.65 < x(Cr) < 0.95.
Bonny et al. [9] could only reproduce the mixing enthalpy at 0 K
and the miscibility gap by using an excessively complex Redlich-Kister
(RK) model up to 5th order interactions and excess heat capacities of
mixing, which lack any experimental fundament and are unlikely re-
flecting a physical base. We found that producing proper phase stabi-
lities using the DFT enthalpy of mixing at 0 K directly by Calphad
modeling was not possible. We speculate that this discrepancy is asso-
ciated with approximations made for the magnetic contributions in
DFT. The same problem occurred in the work of Xiong et al. [4].

2.2.2. FCC
Fig. 5. Calculated enthalpies of formation of σ phase at 0 K by DFT compared to previous The austeno-ferritic transformation in Fe-Cr has been investigated
studies [33,50,52,53].
experimentally in refs. [41,42]. Based on these results, several ther-
modynamic assessment have been carried out [3,4,11,43,44]. In the
Where xCr and xFe represent the molar fraction of Cr and Fe, respec- present work, the experimental data of refs. [41,42] are shown to result
tively. The Gibbs energy of mixing describing the excess Gibbs energy in the best optimization of phase boundaries.
is:
n 2.2.3. Liquid
i, φ
Gex ,φ = x Cr xFe ∑ LCr , Fe Solidus and liquidus data have been measured by thermocouple and
i=1 (5) optical pyrometer techniques at high temperature up to 2100 K by
i, φ Adcock [41].
The interaction parameters LCr , Fe among Cr and Fe can be tem-
The enthalpy of mixing of liquid was measured several times
perature dependent.
[45–49]. Among the existing data, huge scatter exists. One study pre-
dicted negative enthalpy [46] of mixing, suggesting a strong interaction
2.2.1. BCC between Fe and Cr, whereas others [47,48] suggest a repulsion of the
Various atomistic studies [10,36–38] have determined the enthalpy atoms in liquid Fe-Cr. These differences are most likely the result of
of mixing of BCC Fe-Cr at 0 K. Consistently, a negative enthalpy of impurities in the samples influencing the measurements and the ex-
mixing below 10 at. % Cr and a maximum around 50 at. % Cr were perimental technique used by Batalin et al. [47]. The studies of Thie-
obtained. The DFT calculations of Levesque [37] and the results of demann et al. [45] and Pavars et al. [49] suggest an almost ideal be-
embedded-atom method from Eich et al. [10] were used in Monte Carlo havior of the enthalpy of mixing, which is consistent with the
simulations [10,37] to determine the miscibility gap α/α’ consistent experimental results for the liquidus [47] and the phase diagram. Thus,
with available experimental data [39]. The miscibility gap has also the results of Thiedemann et al. [45] and Pavars et al. [49] are accepted
been determined experimentally by Mössbauer spectroscopy by Ku- in the present work. The liquid phase was recently re-modeled using the
wano [39], Dubiel and Inden [19] and Dubiel and Zukrowski [21,40]. description of solid phases as described by Andersson and Sundman [3]
The miscibility gap of the Fe-Cr BCC phase [21,40] has been re- by Cui et al. [50]. Their description of liquid was done with the mod-
produced by Xiong et al. [4] using newly suggested lattice stabilities of ified quasi-chemical model, which is not compatible with the software
pure Fe from Chen and Sundman [14]. Xiong et al. [4] reproduced the program MatCalc and cannot be implemented in our databases for
experimental mixing enthalpy at 1529 K [22] well, whereas further thermokinetic computations of multi-component systems.

20
A. Jacob et al. Calphad 60 (2018) 16–28

Table 3
Re-optimized model parameters of the Fe-Cr system.

Phase Model Parameters (J.mol−1)

Liquid (Cr,Fe) 0 Liquid


LCr , Fe = −5.257 × T
1 Liquid
LCr , Fe = −5419.8
BCC (Cr,Fe:Va) 0 BCC_A2
LCr , Fe = 24600 − 14.98 × T
1 BCC_A2
LCr , Fe = 500 − 1.5 × T
2 BCC_A2
LCr , Fe = −14000 + 9.15 × T
0
TC Cr , Fe = +932.5
1
TC Cr , Fe = −548.2
Bmag Cr , Fe = +2.15
FCC (Cr,Fe:Va) 0 FCC_A1
LCr , Fe = 28767 − 21 × T
1 FCC_A1
LCr , Fe = 33057 − 20.7 × T
σ phase (Cr,Fe)10(Cr,Fe)4(Cr,Fe)16 0 σ
GCr:Cr:Cr = +132000 + 30× 0GCr BCC_A2

0 σ
GCr:Fe:Cr = +140486 + 26× 0GCr BCC_A2
+4× 0GFe
BCC_A2

0 σ
GFe:Cr:Cr = +56200 + 20× 0GCr BCC_A2
+10× 0GFe
BCC_A2

0 σ 0 BCC_A2 0 BCC_A2
GFe:Fe:Cr = +46200 + 16× GCr +14× GFe
0 σ
GCr:Cr:Fe = +123500 − 110 × T + 14× 0GCr BCC_A2
+16× 0GFe
BCC_A2

0 σ 0 BCC_A2 0 BCC_A2
GCr:Fe:Fe = 148800 + 10× GCr +20× GFe
0 σ
GFe:Cr:Fe = 152700 + 26× 0GCr BCC_A2
+4× 0GFe
BCC_A2

0 σ 0 BCC_A2
GFe:Fe:Fe = 173333+30× GFe
0 σ
LFe:Cr:Cr , Fe = −170400 − 70 × T
0 σ
LFe:Fe:Cr , Fe = −330839 − 10 × T

Fig. 7. Phase stability of σ phase of the present work compared with previous thermo-
Fig. 6. Gibbs energy of σ phase calculated at T=1000 K obtained with the DFT enthalpies
dynamic [3,4,6] modeling and experimental data [18].
at 0 K, modified DFT, and from previous thermodynamic modeling [3,4,6]. The symbols
show the chemical composition of the end-members according to the model used by each
authors. contributions of non-ideal mixing to the molar Gibbs energy of the
phase. This will particularly be hampering the model extension to
A complete literature review was provided by Xiong et al. [13]. The multicomponent systems. Here, we thus follow the convention of the
interested reader is invited to refer to this paper for further details. Ringberg meeting [26], i.e. combining Wyckoff positions of same co-
ordination number in the sublattice description
(Cr,Fe)10(Cr,Fe)4(Cr,Fe)16. This model is rather similar to the one from
3. Revisions of thermodynamic modeling
Westman [6]. Yet, whereas Westman described σ phase with a
(Fe)10(Cr)4(Fe,Cr)16 model, we allow Fe and Cr on each sublattice in
3.1. σ phase
accordance with its structural-chemical analysis [1]. Importantly, this
allows to cover the composition range of phase stability in closer
3.1.1. Sublattice description:
agreement to the real physico-chemical nature of the σ phase.
Our literature survey of Section 2 suggests that there is a need for an
The total Gibbs energy with the three sublattice model
improved thermodynamic description of the σ phase in terms of its
(Cr,Fe)10(Cr,Fe)4(Cr,Fe)16 is given by
proper computed phase stability and site occupancies. Indeed, a full
sublattice description, considering all of the five Wyckoff positions in
the form of independent sublattices, will result in a large number of
end-member compounds and even more potential interactions for

21
A. Jacob et al. Calphad 60 (2018) 16–28

Fig. 8. Site occupancy of Fe in σ phase as calculated from the


present thermodynamic modeling compared with experimental
data [1,26].

Several authors reported the enthalpy of formation at 0 K calculated


with DFT. Depending on the study, different DFT codes have been used.
The reported enthalpies of formation as function of the chemical
composition are shown in Fig. 5 together with the present results. In our
modeling, the atoms are constrained to the chosen sublattice models
and certain atomic positions, which are not necessarily the same as the
ones from previous DFT calculations [34,54,55]. Thus, the previous
calculations [34,54,55] cannot be directly used as input for the Calphad
parameter optimizations and are, consequently, not listed in Table 2.
The DFT calculations show that the Fe10Cr4Cr16 and Fe10Fe4Cr16 are the
two most favorable configurations of all the end-members in the CEF.
These results suggest that Fe on the first sublattice and Cr on the third
sublattice are preferred, which is in agreement with experimental data
[1,27].
All the data are within the same range of magnitude except for the
data given by Havránková et al. [34], which show slightly higher va-
lues.
Fig. 9. Calculated enthalpy of formation at T=923 K compared to previous thermo-
dynamic modeling [3–5] and experimental data from Dench [23] and assessed data [24]. 3.1.3. Parameter optimization:
All optimized model parameters of this work are summarized in
1 2 3
Gσ = yCr σ
yCr yCr GCr 1 2 3 σ 1 2 3 σ
: Cr : Cr + yCr yFe yCr GCr : Fe : Cr + yFe yFe yCr GFe : Fe : Cr
Table 3.
1 2 3 σ 1 2 3 σ 1 2 3 σ
Several difficulties are encountered during the optimization process
+ yCr yCr yFe GCr : Cr : Fe + yCr yFe yFe GCr : Fe : Fe + yFe yCr yFe GFe : Cr : Fe of σ phase. Clearly, a “reduced” sublattice description relative to the
1 2 3 σ 1 1 1 1
+ yFe yFe yFe GFe : Fe : Fe + RT [10(yCr lnyCr + yFe lnyFe ) complex nature of TCP phases does not fully take into account all non-
2 2 2 2 3 3 3 3 ideal mixing contributions. In Calphad modeling, the Gibbs energy
+4(yCr lnyCr + yFe lnyFe )+16(yCr lnyCr + yFe lnyFe )]
curve of a phase with a certain homogeneity range is obtained by the
1 2 3 3 1 2 3 3
+ yFe yCr yCr yFe LFe : Cr : Cr , Fe + yFe yFe yCr yFe LFe : Fe : Cr , Fe (6) summation of the Gibbs energies of the different end-members (See Eq.
(6)). The end-members represent line compounds at a certain compo-
Superscripts in Eq. (6) denote sublattices 1–3. L denotes the impact
sition. By using the CEF[28], the different elements are placed in the
of non-ideal mixing between atoms separated by comma. Only inter-
different sublattices. The model can then represent the whole compo-
actions, which have shown to be relevant for the optimized σ-FeCr
sition range of the system and, thus, of the phase. However, even if the
description, are given.
sublattice model chosen covers the full homogeneity range of the phase,
it nevertheless cannot represent the complexity of the crystal structure
3.1.2. DFT data of compound energies of TCP phases in full extend [7]. In the present work, a three sublattice
In our model parameterization, at first, newly obtained compound model combining sublattices with the same coordination number and
enthalpies of formation at 0 K (see Table 2), analyzed by DFT, and the both elements is chosen, with Fe and Cr allowed on each sublattice in
ones from Sluiter [52] for the pure elements, are used to parameterize order to represent the full homogeneity range of the phase. The dif-
the end-members of (Cr,Fe)10(Cr,Fe)4(Cr,Fe)16 σ-FeCr. DFT calculations ferent end-members obtained by this sublattice model represent a fixed
were done with VASP [53] using the GGA:PBE approximation and the composition. In no matter, these compounds can represent the true
projector augmented waves (PAW) [54]. The ground state energy for nature of the crystal structure of the σ phase, which is reflected by their
each compound was obtained as follows: 1) the volume and ions were positive enthalpies of formation. Interaction parameters modeled with
relaxed; 2) then the stresses and forces were relaxed; 3) in the last step, R-K polynomials, are used to provide reasonable consistency with the
the tetrahedron smearing method was used for accurate calculations. physical nature of the σ phase. Modeled excess Gibbs energies of mixing

22
A. Jacob et al. Calphad 60 (2018) 16–28

Fig. 10. Calculated miscibility gap of the Fe-Cr system.

Fig. 11. Calculated spinodal decomposition in Fe-Cr compared


with previous calculations from Calphad [4], molecular dynamics
[10] and experiments [4,56,57].

may be considered as an approximation to the partial disordering of the with a difference of around 3 kJ, which is, as far as our experience goes,
phase. a typical deviation, found likewise for assessed intermetallic phases in
At first, we model the σ phase by using the energy of the end- many systems. For the less stable compounds, the Calphad assessment
members as obtained by DFT (we chose the energy for FM state from leads to large deviation of −8 to −10 kJ. Thereby, the σ phase is in-
Table 2). This however leads to a too unstable σ phase. Any attempt to deed significantly stabilized and temperature dependence was only
stabilize it entropically, by adding some negative temperature depen- necessary for the compound Cr2Cr4Fe8Cr8Fe8 (Cr10Cr4Fe16 in the 3SL
dence to the end-members provoke inevitably the creation of miscibility model, at x(Cr)=0.467) to stabilize the σ-FeCr on the Cr rich side. The
gaps, which is physically not correct and was experimentally not ob- difference between the data obtained by DFT and Calphad is in the
served. The prevention of miscibility gaps and stabilization of σ phase order of 8–10 kJ mol−1. This difference, which is higher than an ex-
could only be obtained with a relatively low weight given to the energy pected error (~5 kJ mol−1), is probably due to the problem of handling
of all end-members as obtained by DFT during the optimization. As a magnetism by DFT (see ΔHmix calculated by DFT [10,36–38]) and due
consequence, the Calphad-assessed enthalpies deviate from the DFT to the simplification made in the Calphad modeling in the CEF (i.e. the
data. For the two most stable end-members, Fe2Cr4Cr8Fe8Cr8 and constraint to a certain lattice, mixing on the same sublattice), leading to
Fe2Fe4Cr8Fe8Cr8, best reproduction of phase boundaries was obtained an error of magnitude of about 10 kJ mol−1.

23
A. Jacob et al. Calphad 60 (2018) 16–28

Fig. 12. Calculated Curie temperature compared to experiments [4] and previous ther-
modynamic modeling [3,4].

Fig. 14. a) Calculated ΔfH (T=1529 K) compared to experimental data [23] (open tri-
angles), thermodynamic modeling [4] and computation from molecular dynamics and
Monte Carlo [10]. b) Calculated ΔHmix at 300 K from the present work and previous
thermodynamic modeling [3,4] compared to enthalpy of mixing obtained by DFT at 0
Fig. 13. Calculated heat capacity at x(Cr)=0.13 compared to experiments [4,59,60] and
K [10,35–37].
previous thermodynamic calculations [3,4].

To illustrate the instability of the σ phase as obtained with DFT used


for the end-members, the Gibbs energy of σ phase at T=1000 K is
shown in Fig. 6 and compared with previous thermodynamic modeling
and the present optimization. The data as obtained by DFT and used in
the thermodynamic modeling is approximately 10 kJ higher than in all
thermodynamic assessments.
In the model optimization, interaction parameters were chosen such
as to distribute the atoms over the sublattice for a close match to the
sublattice occupations [1]. According to the calculation of the site oc-
cupancies compared to experiments [1], Calphad data yield to small
fractions for Fe in the first sublattice, whereas the third sublattice is
completely mixed, which is correct. The second sublattice is correctly
represented with an ideal mixing of species. The third sublattice with
highest occupation (i.e. a fraction of 0.533) has a high influence on the
stability of the σ phase. For closest agreement of computed and ex-
perimental phase boundaries, interaction parameters L(Fe:Cr:Cr,Fe)
and L(Fe:Fe:Cr,Fe) were parameterized. This keeps the σ phase parti-
cularly stable around ~50%, and reproduces the phase boundaries [19]
and the invariant reactions [4,21]. Fig. 15. Activities of Fe and Cr calculated at T=1173 K from the present thermodynamic
modeling compared to experiment [55] and previous modeling [3,4].
The phase stability of the σ phase, as calculated in the present work,
is shown in Fig. 7 and compared with previous thermodynamic mod-
eling and experimental data [19]. Andersson and Sundman [3] as well

24
A. Jacob et al. Calphad 60 (2018) 16–28

to proper phase boundaries, calculated site occupancies are in good


agreement to the experimental evidence. This is shown in Fig. 8.
Any attempt to get even closer to the data of the enthalpy from
Dench [24] in the parameter optimization leads to worse computation
of phase boundaries and invariant reactions. Thus, the deviation by our
thermodynamic optimization of 1 kJ mol−1, see Fig. 9, is likely in-
dicating the experimental uncertainty range and was decided to be
acceptable.
Compared to previous thermodynamic modeling of the σ-FeCr
[3–6], our description can reproduce the phase stability and the crystal
chemistry of the phase. Correct computed site fractions are of funda-
mental importance when using thermodynamic modeling in thermo-
kinetic precipitation simulations as associated chemical potentials are
also influencing the kinetic evolution of a steel matrix-precipitate
system.

3.2. Alloy phases

Fig. 16. Calculated FCC-BCC phase boundary according to the present thermodynamic The BCC phase is the predominant alloy phase in the Fe-Cr system
modeling.
and occurs in direct equilibrium with all the other phases. Thus, we
optimized its model parameters first. Since the BCC phase has adjacent
Table 4 phase boundaries to σ-FeCr, modifications of the thermodynamic de-
Invariant reactions as obtained from the present thermodynamic modeling compared scription of the latter also require adjustments of the description of the
with experiments and their respective references.
former. This, in turn, requires re-adjustments of the FCC and liquid
Reaction Calculated Experiments
phase descriptions in the Fe-Cr system.

C: L ↔ BCC 1782 K 1783 K, x(Cr)=0.21 [41] 3.2.1. BCC phase


x(Cr)=0.21 1783 K, x(Cr)=0.21 [13]a The BCC phase was considered for re-modeling since the thermo-
C: FCC ↔ BCC 1126 K 1119 K, x(Cr)=0.07 [13]a
x(Cr) = 0.073
dynamic description of Andersson and Sundman [3] does not allow Cr
C: BCC ↔ σ 1096 K 1093 K, x(Cr)=0.47 [19] to dissolve into BCC α-Fe and leads to a high Curie temperature com-
x(Cr)=0.46 1103 K, x(Cr)=0.47 [13]a pared to experiment [4,13]. Even though Xiong et al. [4] presented an
Eutectoid: σ ↔ BCC + BCC’ 788 K 783 K [21] accurate description of the alloy phases, their parameters are not being
x(Cr,BCC)=0.15 x(Cr,BCC)=0.143
adopted in the present study due to our focus of inserting our present
x(Cr,σ)=0.49 x(Cr,σ)=0.49
x(Cr,BCC’)=0.85 x(Cr,BCC’)=0.88 system description into an existing multi-component database. Thus,
drifting away from conventional pure element lattice stabilities [14]
a
-according to literature assessment. would necessitate enormous efforts of further Fe-base system re-opti-
mizations. Anyway, we will show, in the following, that proper phase
stabilities can be obtained in the Fe-Cr system with lattice stabilities
from Dinsdale [18].
In order to reproduce the experimental solubilities of Cr and Fe in
the BCC-phase, or, in other words, its miscibility gap [21,39,40], the
excess parameters of mixing of Cr and Fe from Xiong et al. [4] were
used as a starting point for the necessary re-adjustments. Recent mo-
lecular dynamics calculations of BCC-FeCr from Eich et al. [10] were
also considered for the optimization, as well as thermodynamic [24,56]
and solidus-liquidus data [41].
For the final description of the BCC phase, three interactions para-
meters were required, all of them with temperature dependence.
The miscibility gap between BCC+BCC’ is associated with the oc-
currence of a spinodal decomposition reaction. The calculated mis-
cibility gap obtained from the present thermodynamic modeling is
given in Fig. 10 and the calculated curves of spinodal decomposition are
drawn in Fig. 11.
There is a good agreement between the experimental data of the
miscibility gap [21,39,40] and the present calculations. The present
thermodynamic modeling leads to a slight deviation in the higher
Fig. 17. Calculated ΔHmix of liquid Fe-Cr at T=1960 K compared with experimental data, temperature range of the miscibility gap. Any attempt to improve this
T=1960 K [44], T=1863 K [45], T=1873 K [46], T=1973 K [47], T=1873-1973 K [48]
part of the phase diagram, adjusting temperature-dependences of 1st
and previous thermodynamic modeling [3,4,12,49].
and 2nd order interactions in BCC, lead to erroneous destabilization of
the liquid phase. With the present thermodynamic parameters, a max-
as Westman [6] obtained a too small single-phase region compared to imum difference of about 50 K from the experimental data of Kuwano
the experimental data of Cook and Jones [19], whereas the present [39] is obtained. The recent computational determination of the mis-
thermodynamic modeling as well as Xiong et al. [4] are in closer cibility gap by combination of molecular dynamics and Monte Carlo
agreement with [19]. simulation [10] shows a good agreement with experiments [21,39,40]
The advantage of our revised model, with the discussed combina- in comparison to previous computation [37]. The work of Eich et al.
tion rules for sublattices, over previous assessments is that, in addition [10] considered the experimental determination of the enthalpy of

25
A. Jacob et al. Calphad 60 (2018) 16–28

Fig. 18. Calculated activity of the Fe-Cr liquid at T=1873 K re-


ferring to pure liquid Fe and pure BCC Cr compared to previous
thermodynamic modeling and experiments [48,58].

mixing at high temperature given by Dench [24] and not the one ob-
tained by their own DFT, which was also problematic for the Calphad
reproduction of the phase diagram [4].
With the present thermodynamic modeling, more Cr compared to
Andersson and Sundman [3], but slightly less than Xiong [4] and Eich
et al. [10] is dissolved in the α-Fe at low temperature. As for the time
being, there is no experimental solubility data available in this part of
the phase diagram, it is difficult to judge about the best optimization. In
any case, it is important that a reasonable amount of Cr is allowed to
dissolve into α-Fe as this will influence important phase stability fea-
tures such as BCC → FCC transformation and spinodal decomposition in
multi-component, i.e. technologically relevant extensions. For instance,
ferrite decomposition to a Fe- and Cr-rich solid solution BCC+BCC’ is in
part responsible for embrittlement of Cr-rich steels at low temperatures.
Therefore, in the context of model extensions and their use in phase
stability simulations of steels, it is important that the spinodal decom-
position is correctly described by the modeling as shown in Fig. 11.
The spinodal, as calculated in the present work (Fig. 11), is in good
agreement with experiments [4,57,58]. The model parametrization for
Fig. 19. Calculated phase boundary Liquid – BCC compared to experimental works [40]. the ferromagnetic to paramagnetic transition in Fe-Cr BCC was recon-
sidered to obtain an improved reproduction of the experimental Curie
temperatures as a function of Cr-alloying. Consistency with the ex-
periment was obtained, as shown in Fig. 12, with the parametrization
include in Table 2. For the magnetic moment we chose +2.15 leading
to close agreement with the experimental Cp peak shape (see Fig. 13).

3.2.1.1. Thermodynamic properties. The enthalpy of mixing at high


temperature is calculated in Fig. 14 and compared to Dench [24]. For
comparison, results of Xiong [4] and Eich et al. [10] are also given.
The results of the calculated enthalpies at high temperature are
consistent with experiments and the previous Calphad assessment [4].
The calculations of several studies [10,36–38] show a small negative
enthalpy of mixing at the Fe-rich side. By considering a negative value
in their assessment, Xiong et al. [4] were also not successful to re-
produce a proper phase diagram. The same problem occurred in the
present work, and the negative value at the Fe-rich side was discarded
in the optimization procedure. The calculated ΔHmix at 300 K of the
Calphad modeling (present work and [3,4]) thus does not reproduce a
negative ΔHmix at low Cr content as given by the different DFT studies
Fig. 20. Calculated phase diagram according to the present work.
[10,36–38]. Even though, experimental thermodynamic properties
could be well reproduced as shown by the calculated activities at
1173 K compared to [56] in Fig. 15.

26
A. Jacob et al. Calphad 60 (2018) 16–28

3.2.2. FCC technological alloys, we believe that this effort will be highly worth-
The thermodynamic modeling of Andersson and Sundman [3] pro- while.
vide the best agreement among various Calphad assessments. The ex- It should be mentioned that the revised σ phase modeling does not
perimental data on the austeno-ferrite transformation from Adcock [41] produce problems for the calculated phase stabilities of the alloy phases
and Kirchner [42] were reproduced (see Fig. 16) with a relative small in the Fe-Cr system. Only relatively slight re-adjustments of the model
re-adjustment of interaction parameters relative to the previous sug- parameterizations are necessary for the reproduction of all phase
gestion [4]. boundaries.

3.2.3. Liquid 4. Conclusion


In the present work, the enthalpy of mixing, as given by
Thiedemann et al. [45] and Pavars et al. [49], and experimental solidus Thermodynamic model descriptions of the Fe-Cr system have been
and liquidus data and the congruent melting temperature [41] were revised with the focus of physico-chemically correct site occupancies of
considered for the re-optimization of the liquid phase. The enthalpy of solute atoms in the model description of σ phase. Wyckoff positions and
mixing of liquid on the Fe-rich side is close to zero J.mol−1, thus, the first-principles compound energies are considered for the optimization
liquid phase follows almost Raoult's law, i.e. ideal behavior of mixture. of model parameters.
Nevertheless, as a consequence of the considerable non-ideality of the The presented and discussed three sublattice model
BCC-description, excess Gibbs energy of mixing is required in the Cal- (Cr,Fe)10(Cr,Fe)4(Cr,Fe)16 shows to be more respectful to the structural-
phad description in order to reproduce the experimental solidus and chemical nature of the σ-phase compared to previous assessments. This
liquidus data. A small temperature-dependence for the 0th order R-K is reflected by proper trends of calculated site occupancies as function
polynomial was used, as well as a 1st order interaction parameter of composition and temperature. The consistency of the present model
without temperature dependence. This, in turn, required optimization is further reflected in the close agreement between experimental and
of temperature the dependence of the 1st and 2nd order R-K poly- calculated phase diagram data. Moreover, the model could be kept
nomials of the BCC phase for the correct reproduction of the high- relatively simple; only two interactions parameters for the non-ideal
temperature section of the system (see Table 4). mixing of Fe and Cr in Fe-Cr σ have been adapted.
The results for the enthalpy of mixing of liquid at T=1960 K are
given in Fig. 17 and compared with experiments [45–49] and previous Acknowledgements
thermodynamic modeling [3,4,12,50].
Our calculated enthalpies of mixing are in good agreement with the The computational results presented in this work have been
experimental work of Thiedeman [45] and Pavars et al. [49] as well as achieved using the Vienna Scientific Cluster (VSC).
the activity calculated at T=1873 K and compared with experiments Funding by the Austrian Science Fund (FWF): P13 (SFB ViCoM) is
[49,59] (the calculation refers to pure liquid of Fe and Cr as given gratefully acknowledged.
experimentally), see Fig. 18.
The respective solidus and liquidus obtained from the present Appendix A. Supplementary material
thermodynamic modeling is shown in Fig. 19 and compared with the
experimental results of Adcock [41]. In principle, good agreement is Supplementary data associated with this article can be found in the
found. Only at higher Cr content, the calculations deviate from the online version at doi:10.1016/j.calphad.2017.10.002.
measurements, similar as for previous modeling. Indeed, at high tem-
perature, Cr volatizes, which affects the chemical composition during References
the measurement. Thus, the experimental data in this part of the phase
diagram likely contain considerable uncertainty and are not given high [1] H.L. Yakel, Atom distributions in sigma phases. I. Fe and Cr atom distributions in a
weight in the parameter optimization. binary sigma phase equilibrated at 1063, 1013 and 923 K, Acta Crystallogr. Sect. B.
39 (1983) 20–28, http://dx.doi.org/10.1107/S0108768183001974.
[2] H. Yakel, Atom distributions in sigma phases. II. Estimations of average site-occu-
3.2.4. Equilibrium phase transformations pation parameters in a sigma phase containing Fe, Cr, Ni, Mo and Mn, Acta
The full phase diagram of Fe-Cr was calculated from the present Crystallogr. Sect. B Struct. Sci. 39 (1983) 28–33, http://dx.doi.org/10.1107/
S0108768183001986.
thermodynamic parameters (Table 3) and is presented in Fig. 1 and [3] J.-O. Andersson, B. Sundman, Thermodynamic properties of the Cr-Fe system,
compared to previous assessments and experimental phase diagram Calphad 11 (1987) 83–92, http://dx.doi.org/10.1016/0364-5916(87)90021-6.
data in Fig. 20. [4] W. Xiong, P. Hedström, M. Selleby, J. Odqvist, M. Thuvander, Q. Chen, An im-
proved thermodynamic modeling of the Fe–Cr system down to zero kelvin coupled
The calculated invariant reactions obtained from the present ther- with key experiments, Calphad 35 (2011) 355–366, http://dx.doi.org/10.1016/j.
modynamic modeling are given in Table 4 in comparison to experiment calphad.2011.05.002.
data. [5] Y. Chuang, J.-C. Lin, Y.A. Chang, A thermodynamic description and phase re-
lationships of the Fe-Cr system: Part I The bcc phase and the sigma phase, Calphad
To summarize, the agreement of the present calculated phase dia-
11 (1987) 57–72.
gram (Fig. 20), special points and transformations compared to ex- [6] S. Westman, Unpublished work, 2000.
periments [19,21,41,42] are good. The present thermodynamic mod- [7] R. Mathieu, N. Dupin, J.C. Crivello, K. Yaqoob, A. Breidi, J.M. Fiorani, et al.,
eling allows for a higher amount of Cr being dissolved in α-Fe at low CALPHAD description of the Mo-Re system focused on the sigma phase modeling,
Calphad Comput. Coupling Phase Diagr. Thermochem. 43 (2013) 18–31, http://dx.
temperature compared to Andersson et al. [3]. The calculated decom- doi.org/10.1016/j.calphad.2013.08.002.
position of BCC agrees with the experimental observation [21]. The re- [8] J.-M. Joubert, Crystal chemistry and Calphad modeling of the σ phase, Prog. Mater.
calculated phase diagram is also in agreement with the previous one Sci. 53 (2008) 528–583, http://dx.doi.org/10.1016/j.pmatsci.2007.04.001.
[9] G. Bonny, D. Terentyev, L. Malerba, New contribution to the thermodynamics of Fe-
given by Xiong et al. [4]. Cr alloys as base for ferritic steels, J. Phase Equilibria Diffus. 31 (2010) 439–444,
http://dx.doi.org/10.1007/s11669-010-9782-9.
3.2.5. Consequences of thermodynamic modeling [10] S.M. Eich, D. Beinke, G. Schmitz, Embedded-atom potential for an accurate ther-
modynamic description of the iron-chromium system, Comput. Mater. Sci. 104
The present improvement of the sublattice model for σ-FeCr is the (2015) 185–192, http://dx.doi.org/10.1016/j.commatsci.2015.03.047.
first step towards more physically-based steel databases. Indeed, to [11] S. Hertzman, B. Sundman, A thermodynamic analysis of the Fe-Cr system, Calphad
change the thermodynamic modeling of σ phase in various other sys- 6 (1982) 67–80.
[12] B.-J. Lee, Revision of thermodynamic descriptions of the Fe-Cr & Fe-Ni liquid
tems in accordance will be a challenging task. Nevertheless, in view of phases, Calphad 17 (1993) 251–268.
the associated improvements of the usability of multicomponent data- [13] W. Xiong, M. Selleby, Q. Chen, J. Odqvist, Y. Du, Phase equilibria and thermo-
bases for predictive simulations of phase transformations in dynamic properties in the Fe-Cr system, Crit. Rev. Solid State Mater. Sci. 35 (2010)

27
A. Jacob et al. Calphad 60 (2018) 16–28

125–152, http://dx.doi.org/10.1080/10408431003788472. microstructure evolution in binary alloys and its application to the FeCr case, Acta
[14] Q. Chen, B. Sundman, Modeling of thermodynamic properties for BCC, FCC, liquid, Mater. 95 (2015) 324–334, http://dx.doi.org/10.1016/j.actamat.2015.05.040.
and amorphous iron, J. Phase Equilibria. 22 (2001) 631–644, http://dx.doi.org/10. [39] H. Kuwano, Mössbauer effect study on the miscibility gap of the iron-chromium
1007/s11669-001-0027-9. binary system, Trans. Jpn. Inst. Met. 26 (1985) 473–481 〈http://biz3.bioweb.ne.jp/
[15] R. Naraghi, M. Selleby, J. Ågren, Thermodynamics of stable and metastable struc- jim/journal/e/26/07/473.html〉.
tures in Fe-C system, Calphad Comput. Coupling Phase Diagr. Thermochem. 46 [40] S.M. Dubiel, J. Zukrowski, Fe-rich border and activation energy of phase decom-
(2014) 148–158, http://dx.doi.org/10.1016/j.calphad.2014.03.004. position in a Fe-Cr alloy, Mater. Chem. Phys. 141 (2013) 18–21, http://dx.doi.org/
[16] I. Roslyakova, B. Sundman, H. Dette, L. Zhang, I. Steinbach, Modeling of Gibbs 10.1016/j.matchemphys.2013.05.023.
energies of pure elements down to 0K using segmented regression, Calphad 55 [41] F. Adcock, Alloys of iron research Part X - The chromium-iron constitutional dia-
(2016) 165–180, http://dx.doi.org/10.1016/j.calphad.2016.09.001. gram, J. Iron Steel Inst. 129 (1931) 99.
[17] 〈www.matcalc-engineering.com〉. [42] G. Kirchner, T. Nishizawa, B. Uhrenius, The distribution of chromium between
[18] A.T. Dinsdale, SGTE data for pure elements, Calphad 15 (1991) 317–425. ferrite and austenite and the thermodynamics of the a/g equilibrium in the Fe-Cr
[19] A.J. Cook, F. Jones, The brittle constituent of the iron-chromium system (sigma and Fe-Mn system, Mettalurgical Trans. 4 (1973) 167–174.
phase), Iron Steel Inst. - Pap. 217 (1943) 217–226. [43] L. Kaufman, Discussion of thermodynamics of a-y transformation in Fe-Cr alloys,
[20] R. Vilar, G. Cizeron, Evolutions structurales developpées au sein de la phase sigma Mettalurgical Trans. 5 (1974) 1688–1689.
Fe-Cr, Acta Metall. 35 (1987) 1229–1239. [44] J. Chipman, Thermodynamic of the a-y transformation in the Fe-Cr alloys, Metall.
[21] S.M. Dubiel, G. Inden, On the miscibility gap in the Fe-Cr system: a moessbauer Trans. 5 (1974) 521–523.
study on long term annealed alloys, Z. Fuer Met. 78 (1987) 544–549. [45] U. Thiedemann, M. Rösner-Kuhn, D.M. Matson, G. Kuppermann, K. Drewes,
[22] R. Williams, H. Paxton, The nature of ageing of binary iron-chromium alloys around M.C. Flemings, et al., Mixing enthalpy measurements in the liquid ternary system
500 °C, J. Iron Steel Inst. 185 (1957) 358–374. iron-nickel-chromium and its binaries, Steel Res. 1 (1998) 3–7.
[23] T. Koyano, U. Mizutani, H. Okamoto, Evaluation of the controversial s–>(Cr)+(a- [46] Y. Iguchi, S. Nobori, K. Saito, T. Fuwa, A calorimetric study of heats of mixing of
Fe) eutectoid temperature in the Fe-Cr system by heat treatment of mechanically liquid iron alloys Fe-Cr, Fe-Mo, Fe-W, Fe-W, Fe-V, Fe-Nb, Fe-Ta, J. Iron Steel Inst.
alloyed powder, J. Mater. Sci. Lett. 14 (1995) 1237–1240. Jpn. 68 (1982) 89–96.
[24] W.A. Dench, Adiabatic high-temperature calorimeter for the measurement of heats [47] G. Batalin, V. Kurach, V. Sudavstova, Enthalpies of mixing of Fe-Cr and Fe-Ti melts,
of alloying, Trans. Faraday Soc. 59 (1963). Zhurnal Fiz. Khimii. 58 (1984) 481–483.
[25] F. Müller, O. Kubaschewski, The thermodynamic properties and the equilibrium [48] V. Shumikhin, A.K. Biletsky, G. Batalin, V. Anishin, Study of thermodynamic and
diagram of the system chromium-iron, High. Temp. - High. Press. 1 (1969) kinetic parameters of dissolution of solid materials in ferrocarbonic melts, Arch. Für
543–551. Das. Eisenhüttenwes. 52 (1981) 143–146.
[26] H. Martens, P. Duwez, Heat evolved and volume change in the alpha-sigma-trans- [49] I.A. Pavars, B. Baum, P.V. Gel’d, Thermophysical and thermodyanmic properties of
formation in Cr-Fe alloys, Trans. Am. Inst. Min. Metall. Eng. 206 (1956) (614–614). liquid alloys of iron and chromium, Teplofiz. Vysok. Temp. 8 (1970) 72–76.
[27] J. Cieślak, M. Reissner, S.M. Dubiel, J. Wernisch, W. Steiner, Influence of compo- [50] S. Cui, I.-H. Jung, Thermodynamic modeling of the Cu-Fe-Cr and Cu-Fe-Mn systems,
sition and annealing conditions on the site-occupation in the σ-phase of Fe–Cr and Calphad 56 (2017) 241–259, http://dx.doi.org/10.1016/j.calphad.2017.01.004.
Fe–V systems, J. Alloy. Compd. 460 (2008) 20–25, http://dx.doi.org/10.1016/j. [51] M.H.F. Sluiter, Ab initio lattice stabilities of some elemental complex structures,
jallcom.2007.05.098. Calphad 30 (2006) 357–366, http://dx.doi.org/10.1016/j.calphad.2006.09.002.
[28] M. Hillert, The compound energy formalism, J. Alloy. Compd. 320 (2001) 161–176, [52] G. Kresse, J. Furthmüller, Efficiency of ab-initio total energy calculations for metals
http://dx.doi.org/10.1016/S0925-8388(00)01481-X. and semiconductors using a plane-wave basis set, Comput. Mater. Sci. 6 (1996)
[29] A. Watson, F.H. Hayes, Some experiences modelling the sigma phase in the Ni-V 15–50.
system, J. Alloy. Compd. 320 (2001) 199–206, http://dx.doi.org/10.1016/S0925- [53] G. Kresse, D. Joubert, From ultrasoft pseudopotentials to the projector augmented-
8388(00)01472-9. wave method, Phys. Rev. B. 59 11–19. 〈http://prb.aps.org/abstract/PRB/v59/i3/
[30] I. Ansara, T.G. Chart, A.F. Guillermet, F.H. Hayes, U.R. Kattner, D.G. Pettifor, et al., p1758_1〉 (accessed 13 September 2013).
Themodynamic modelling of solutions and alloys, Calphad 21 (1997) 171–218. [54] J. Pavlů, J. Vřešťál, M. Šob, Ab initio study of formation energy and magnetism of
[31] A.V. Khvan, B. Hallstedt, C. Broeckmann, A thermodynamic evaluation of the Fe-Cr- sigma phase in Cr-Fe and Cr-Co systems, Intermetallics 18 (2010) 212–220, http://
C system, Calphad Comput. Coupling Phase Diagr. Thermochem. 46 (2014) 24–33, dx.doi.org/10.1016/j.intermet.2009.07.018.
http://dx.doi.org/10.1016/j.calphad.2014.01.002. [55] E.A. Kabliman, A.A. Mirzoev, A.L. Udovskii, First-principles simulation of an or-
[32] A. Jacob, C. Schmetterer, A. Khvan, A. Kondatriev, D. Ivanov, B. Hallstedt, Liquidus dered sigma phase of the Fe-Cr system in the ferromagnetic state, Phys. Met.
projection and thermodynamic modeling of Cr-Fe-Nb ternary system, Calphad 54 Metallogr. 108 (2009) 435–440, http://dx.doi.org/10.1134/
(2016) 1–15, http://dx.doi.org/10.1016/j.jallcom.2008.10.156. S0031918×09110027.
[33] M.H.F. Sluiter, K. Esfarjani, Y. Kawazoe, Site occupation reversal in the Fe-Cr sigma [56] J. Vrestal, J. Tousek, A. Rek, Thermodynamic activity of components in the Fe-Cr a-
phase, Phys. Rev. Lett. 75 (1995) 3142–3146. solid solution, Kov. Mater. 4 (1978) 393–399.
[34] J. Havránková, J. Vřešťál, L. Wang, M. Šob, Ab initio analysis of energetics of σ- [57] D. Chandra, L.H. Schwartz, Mössbauer effect study of the 475 °C decomposition of
phase formation in Cr-based systems, Phys. Rev. B. 63 (2001) 1–5, http://dx.doi. Fe-Cr, Metall. Trans. 2 (1971) 511–519, http://dx.doi.org/10.1007/BF02663342.
org/10.1103/PhysRevB.63.174104. [58] Y. Imai, M. Izumiyama, T. Masumoto, Phase transformation of Fe-Cr binary system
[35] E. Kabliman, P. Blaha, K. Schwarz, O.E. Peil, A.V. Ruban, B. Johansson, at about 500 °C, Sci. Rep. Res. Inst. Tohoku Univ. Ser. A - Phys. Chem. Metall. S
Configurational thermodynamics of the Fe-Cr σ phase, Phys. Rev. B. 84 (2011) (1966) 56.
184206, http://dx.doi.org/10.1103/PhysRevB.84.184206. [59] N. Maruyama, S. Ban-Ya, Measurement of activities in liquid Fe-Ni, Fe-Co AND Ni-
[36] P. Olsson, I. a. Abrikosov, L. Vitos, J. Wallenius, Ab initio formation energies of Co alloys by a transportation method, Nippon Kinzoku Gakkaishi/J. Jpn. Inst. Met.
Fe–Cr alloys, J. Nucl. Mater. 321 (2003) 84–90, http://dx.doi.org/10.1016/S0022- 42 (1978) 992–999.
3115(03)00207-1. [60] R.N. Hajra, R. Subramanian, H. Tripathy, A.K. Rai, S. Saibaba, Phase stability of
[37] M. Levesque, E. Martínez, C.-C. Fu, M. Nastar, F. Soisson, Simple concentration- Fe–15wt.% Cr alloy: a calorimetry and modeling based approach to elucidating the
dependent pair interaction model for large-scale simulations of Fe-Cr alloys, Phys. role of magnetic interactions, Thermochim. Acta. 620 (2015) 40–50, http://dx.doi.
Rev. B. 84 (2011) 184205, http://dx.doi.org/10.1103/PhysRevB.84.184205. org/10.1016/j.tca.2015.10.002.
[38] I. Dopico, P. Castrillo, I. Martin-Bragado, Quasi-atomistic modeling of the

28

You might also like