See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/308878452

Thermodynamics of the Fe-N and Fe-N-C Systems: The Fe-N and Fe-N-C
Phase Diagrams Revisited

Article in Metallurgical and Materials Transactions A · October 2016

DOI: 10.1007/s11661-016-3731-0

CITATIONS READS

47 5,237

4 authors, including:

O. Fabrichnaya Andreas Leineweber

Technische Universität Bergakademie Freiberg Technische Universität Bergakademie Freiberg
280 PUBLICATIONS 4,888 CITATIONS 256 PUBLICATIONS 4,272 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Andreas Leineweber on 15 November 2016.

The user has requested enhancement of the downloaded file.

Thermodynamics of the Fe-N and Fe-N-C Systems:
The Fe-N and Fe-N-C Phase Diagrams Revisited
HOLGER GÖHRING, OLGA FABRICHNAYA, ANDREAS LEINEWEBER,
and ERIC JAN MITTEMEIJER

Several thermodynamic descriptions of the Fe-N and Fe-N-C systems were proposed before
now. The results of these descriptions significantly deviate from more recently obtained
experimental data. The present work provides a revised thermodynamic description of these
systems. The new description for the Fe-N system agrees distinctly better with the experimental
data especially for the equilibrium of c0 -Fe4N1x and e-Fe3N1+z. The new thermodynamic
description for the Fe-N-C system considering the Fe-rich part of the system with less than 33
at. pct N and less than 25 at. pct C excellently agrees with the new experimental data for both
the temperatures of the invariant reactions and the phase boundaries. This in particular
concerns the temperature range of typical technical nitriding and nitrocarburizing treatments
[723 K to 923 K, (450 C to 650 C)], within which three invariant reactions occur in the ternary
system.

DOI: 10.1007/s11661-016-3731-0
 The Author(s) 2016. This article is published with open access at Springerlink.com

I. INTRODUCTION The constitution of compound layers developing upon

nitriding of Fe can be predicted by the Fe-N phase
THE Fe-C, Fe-N and Fe-N-C systems are highly diagram,[12] assuming local equilibrium in the solid
relevant for Fe-based components, in particular if these state, featuring the interstital solid solution phases a
are subjected to technically applied nitriding and nitro- (ferrite) and c (austenite) and the iron nitride phases
carburizing treatments.[1] The Fe-N-C equilibrium c0 -Fe4N1x and e-Fe3N1+z. In order to identify such
phases considered in the present work have been listed local equilibria, the Fe-N system to be considered, as
in Table I, as assembled from References 2 and 3. discussed above for the Fe-C system, represents
The binary Fe-C system is the basis of all technically metastable equilibrium states, corresponding to sup-
applied steels.[4] The stable phases in the binary Fe-C pression of the formation of N2 gas. In genuine
system[5] are the terminal interstitial solid solution equilibria, iron-nitride phases such as c0 and e do not
phases a (ferrite) and c (austenite), the liquid solution occur. Metastable equilibria in the Fe-N system can be
phase and graphite. However, due to kinetically investigated by gas-nitriding experiments using NH3/H2
obstructed precipitation of graphite, at carbon contents atmospheres, defining the chemical potential of N in the
of up to 25 at. pct, the iron carbide cementite, gas phase.[1] For data obtained from such gas-nitrided
h-Fe3C1d , occurs in metastable equilibria. Recently, specimens, furthermore the establishment of a steady
the non-stoichiometry of h in equilibrium with a and c state instead of a local equilibrium at the surface of the
with positive values of d, has been quantified.[6] On this specimens, i.e., equality of the rate of N dissolution and
basis, a new thermodynamic description for the cemen- recombination instead of equality of the chemical
tite phase has been presented,[7] which, in contrast to the potential of N in the gas phase and in the solid, has to
previous descriptions,[8–11] recognizes and well describes be considered: the N concentration will then be lower
its non-stoichiometric character. than that corresponding to local metastable equilibrium
with the gas atmosphere.[1,13] The effect becomes signif-
icant above approximately 853 K (580 C) for nitrided
HOLGER GÖHRING, Ph.D. Student, is with the Max Planck
Institute for Intelligent Systems (formerly Max Planck Institute for
ferrite (a)[13] and is more pronounced for increasing
Metals Research), Heisenbergstraße 3, 70569 Stuttgart, Germany. nitrogen content in the solid matrix and thus becomes
Contact e-mail: h.goehring@is.mpg.de OLGA FABRICHNAYA, significant at the surface of e-iron nitride (containing
Research Scientist, is with the Institute of Materials Science, TU >30 at. pct N) already at 723 K (450 C).[14]
Bergakademie Freiberg, Gustav-Zeuner-Str. 5, 09599 Freiberg, Ger- Until today, the experimental data set for the Fe-N
many. ANDREAS LEINEWEBER, formerly Research Scientist with
the Max Planck Institute for Intelligent Systems (formerly Max Planck system as compiled in Reference 12 is the most complete
Institute for Metals Research), is now Professor with the Institute of experimental description of the system and largely
Materials Science, TU Bergakademie Freiberg. ERIC JAN provides the basis for the optimisation of thermody-
MITTEMEIJER, Director, is with the Max Planck Institute for namic parameters in the present work. Recently, some
Intelligent Systems (formerly Max Planck Institute for Metals
Research), and also Professor with the Institute for Materials Science,
additional data on the constitution of the system have
University of Stuttgart, Stuttgart, Germany. been published[13,15] that are also considered in
Manuscript submitted March 16, 2016. the present work. CALPHAD-type thermodynamic
Article published online October 3, 2016

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 47A, DECEMBER 2016—6173

 Table I. Phases Considered in the Present Work, Their Crystal Structure, and Formula Units of Their Sublattice Models

Phase Space Group Structure Formula Unit

Ferrite, a-Fe[N,C] Im3m bcc Fe lattice with N and C on octahedral sites (3 per Fe atom) Fe(C,N,Va)3
Austenite, c-Fe[N,C] Fm3m fcc Fe with N and C on octahedral sites Fe(C,N,Va)
c0 -Fe4N1z Pm3m fcc-type Fe lattice, N and C ordered on one octahedral site per unit Fe4(C,N,Va)
cell
e-Fe3(N,C)1þx P63 22, P312 hcp-type Fe lattice, N and C on every second octahedral site with Fe(C,N,Va)1=2
different types of ordera
h-Fe3C1d Pnma distorted hcp-type Fe lattice, C in trigonal prisms Fe3(C,Va)
a
The chosen formula unit of the sublattice model does not assume a specific state of order. It is, however, noted that octahedral sites adjacent in
the c direction cannot be occupied simultaneously.[14,30]

descriptions for the Fe-N system have been published in factors derived from N and C diffusivities in ternary e at
References 8 and 16 through 24. In general, for the 823 K (550 C)[58] and 853 K (580 C)[51] showed
intermediate phases c0 and e, these descriptions only agreement with the thermodynamic descriptions of e
consider random mixing of nitrogen on an interstitial given in References 20 and 24 and disagreement with
sublattice, and thus excess Gibbs energy parameters the one from Reference 22.
have to be introduced to describe the deviation of the In the present work, new thermodynamic assessments
real system from such ideal behavior. Theoretical of the Fe-N and Fe-N-C systems are presented, using the
approaches to describe the thermodynamics of nitrogen newly obtained data in the optimization process in order
ordering and disordering have been presented for to eliminate the discrepancies associated with the previous
c0 [25–29] and e.[14,27,30–33] thermodynamic descriptions. As a result, for the first time
Upon nitrocarburizing of Fe, a simultaneous uptake a description of the Fe-N-C system was obtained that is
of N and C into the substrate occurs.[1] Upon only compatible with all experimental data and thus is suit-
nitriding of technical steel, interactions of N and C have able for nitriding and nitrocarburizing applications.
to be considered as well, due to (initial) C present in the Furthermore, significant improvements in the description
substrate. A first systematic study of the ternary Fe-N-C of the binary Fe-N system were achieved: (i) a simpler
system has been provided in Reference 34. Subsequent model for the c0 phase, accounting for its homogeneity
work has been presented in References 35 and 36. All range but using fewer parameters than the thermody-
these early works have in common that they do not namic description from Reference 22 and (ii) an improved
include the possibility of an a + e equilibrium. Further- description of the c0 + e equilibrium at high N contents.
more, the appearance of c is concluded to occur at a
temperature as low as 838 K (565 C) in Reference 36,
in flagrant contrast with later experimental data.[15] II. THERMODYNAMIC MODEL OF THE FE-N
In contrast to these early experimental works,[34–36] AND FE-N-C SOLID SOLUTION PHASES
the observation of microstructures forming upon nitrid-
The Fe-N and Fe-N-C solid solution phases can be
ing of Fe-C alloys and C-containing steels and upon
described by the compound-energy formalism,[59,60] also
nitrocarburizing pure Fe, Fe-C alloys, and C-containing
called Hillert-Staffansson approach.[61] In the following,
steels, which do contain interfaces between a and e[37–47]
only an Fe-N-C solid solution phase is considered; a
makes it very likely that equilibrium between a and e
similar treatment is used for an Fe-N (and an Fe-C)
does occur, albeit in a narrow temperature range. Later
solid solution phase.
works[15,48–53] confirmed the occurrence of such
The interstitial solution of N and C in a phase u is
microstructures. Except for an early work,[54] the a + e
considered as a mixture of the hypothetical compounds
equilibrium is taken into account in all thermodynamic
Fea Cc , Fea Nc and Fea Vac (with Va standing for
descriptions of the Fe-N-C system in the litera-
vacancies) with a and c being stoichiometric indices
ture.[20,22,24,55–57] However, systematic experimental
determined by the crystal structure of the phase u. The
work to investigate the occurrence of the a + e equilib-
total Gibbs energy of the phase u with the formula
rium[48,53] showed that the experimentally determined
Fea (C,N,Va)c per formula unit reads
temperatures of the invariant reactions leading to the
appearance of this equilibrium deviate from the tem- Gum ¼ yuC  GuFe:C þ yuN  GuFe:N þ yuVa  GuFe:Va
peratures as predicted using each of the thermodynamic
þ cRTðyuC ln yuC þ yuN ln yuN þ yuVa ln yuVa Þ þ Gu;ex
descriptions from the literature.[20,22,24,55–57] Addition-
ally, a recent experimental study of the constitution in þ Gu;mag ;
the system Fe-N-C for the temperature range above ½1
853 K (580 C),[15] investigating both the phase bound-
aries and the temperatures of the invariant reactions with yuC , yuN
and yuVa
representing the fractions of
involving c, showed bad agreement with again each of sublattice occupancies of C, N, and Va, respectively,
the thermodynamic descriptions from the litera- recognizing that the first sublattice is always completely
ture.[20,22,24,55–57] Finally, analysis of the thermodynamic occupied by Fe, i.e., yuFe ¼ 1, the Gibbs energies  GuFe:C ,

6174—VOLUME 47A, DECEMBER 2016 METALLURGICAL AND MATERIALS TRANSACTIONS A


GuFe:N , and  GuFe:Va of the hypothetical non-magnetic two-phase equilibria. The available data are composi-
compounds Fea Cc , Fea Nc , and Fea Vac , the so-called tions and activities at the phase boundaries. If instead of
end-members, with Fe:C, Fe:N, and Fe:Va denoting activities, the nitriding potential, a (technical) process
that the second sublattice is fully occupied by C, N, and parameter,[1] was given, the activities were calculated
Va, respectively, the excess Gibbs energy Gu;ex and the using the Gibbs-energy equations for various gas species
magnetic contribution Gu;mag . given in Reference 86. Newer data for the c  a + c0
The Gibbs energy of the end members is usually invariant equilibrium and the a + c0 two-phase equilib-
described as a temperature series according to e.g., for rium[13,87,88] and the c + c0 and c + e equilibria[15,87]
Fea Nc : were also included. As an additional information, the
activity curves for c0 from References 25 and 82 and for

GuFe:N  a  Gref  ref
Fe  c GN ¼ a þ bT e in References 17 and 89 were used. However, these
½2 so-called absorption isotherms obtained from gaseous
þ cT ln T þ d1 T2 þ d2 T1 þ d3 T3 :
nitriding of Fe specimens are affected by the establish-
with the reference Gibbs energies for Fe and N,  Gref ment of steady states instead of true metastable equilib-
Fe
and  Gref rium at the surface of the specimen at higher
N and the model parameters a, b, c, di . The
reference state is usually the SER state, i.e., the enthalpy temperatures and N contents,[1,13,14] making it impossi-
of the elements in their most stable state at 298 K ble to use all data above 823 K (550 C). Already at
(25 C) and 1  105 Pa. lower temperatures, but high N contents (>30 at. pct), a
The excess Gibbs energy is described as steady state instead of an equilibrium prevails at the
surface. Therefore, the affected data have not been used
Gu;ex ¼yuC yuN LuFe:C;N þ yuC yuVa LuFe:C;Va during the optimization. Furthermore, during the opti-
½3 mization process, agreement of the model with the
þ yuN yuVa LuFe:N;Va ; activity data has been considered less important than
agreement with the information on solid-solid
only considering binary interaction parameters LuFe:C;N ,
equilibria.
LuFe:C;Va and LuFe:N;Va , with their composition depen- Based on this experimental information, the param-
dence described by a Redlich-Kister series[62]: 0 0

X u eters  GcFe:N and  GcFe:Va of c0 and the parameters

0 e
LuFe:C;N ¼ k
LFe:C;N ðyuC  yuN Þk ; LFe:N;Va and 1 LeFe:N;Va of e were optimized.
½4
k

and analogously for LuFe:C;Va and LuFe:N;Va . In the present B. Ternary Fe-N-C Data
work, interaction parameters of zeroth and first order For the optimization process of the model parameters
are used (i.e., k ¼ 0; 1). These treatments correspond to for the ternary Fe-N-C system, primarily recently
a regular and a sub-regular solution model, respectively, published data was used. During the optimization, care
whereas Gu;ex ¼ 0 corresponds to an ideal solution was taken that the resulting invariant temperatures
model.[63] For the magnetic contribution Gu;mag of the comply with the ranges as determined experimentally in
a, c and h phases, the Inden model[64,65] is used, taking References 15 and 53. The second source of data was the
the magnetic moment bu and the Curie (for a and h) or location of the phase boundaries at 853 K and 893 K
Néel (for c) temperature TuCurie; or TuNeel; as (poten- (580 C and 620 C) as determined experimentally in
tially concentration-dependent) model parameters, as Reference 15. The experimental information that in the
described in detail in Reference 60. For c0 and e no considered C and N content ranges the off-diagonal
separate magnetic contribution is modeled (see components of the thermodynamic factor of e are
Section IV). positive[51,58] was used as a constraint for the model of
the e phase. Additionally, the N-solubility data in
C-containing c from Reference 90 was used.
For the C content of c0 , no reliable equilibrium data is
III. EMPLOYED DATA FOR THE available. EPMA investigations on specimens produced
THERMODYNAMIC PARAMETER for the investigations in References 51 through 53
OPTIMIZATION and 58 showed C contents in c0 which were always
below 1 at. pct. Therefore, during the optimization care
A. Binary Fe-N Data
was taken that this level of C content was not exceeded
For the binary system Fe-N, the data as assembled in considerably. In h only trace amounts of N have been
Reference 12[25,35,36,66–85] was chosen according to the found at temperatures £1073 K (800 C).[91] At lower
recommendations given there. The binary thermody- temperatures, h layers can be produced on Fe substrates
namic descriptions of the a and c phase were taken from by heat treatment in an atmosphere containing CO,
Reference 22; for the choice of parameters to be NH3, and H2 .[92] During the treatment, the substrate is
optimized, see Section IV. Thus, only equilibrium data gradually saturated with N that has diffused through
including the phases c0 or e have been used, in particular h,[93] eventually leading to formation of an e layer
the data for the a + c0 , c + c0 , c + e, and c0 + e underneath the h layer.[50] h produced under these

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 47A, DECEMBER 2016—6175

conditions has been investigated by atom probe tomog- magnetic moment, so no magnetic contribution
raphy,[94] revealing, nevertheless, a maximum total was modeled*. The value of the parameter  GeFe:N was
impurity content of only 0.01 at. pct, also including N.
This supports the above experimental results of Refer-
ence 91 and is in contrast with the prediction of N *It was shown in an earlier work providing a new thermodynamic
description for h[7] that the magnetic moments obtained from fitting,
contents of>1 at. pct resulting from the thermodynamic specifically of cp curves, are far from experimentally obtained values.
description of Reference 22. Therefore, it was avoided to use such values for the thermodynamic
On the basis of these ternary experimental data, the descriptions of c0 and e in the present work.
0
binary parameters  GcFe:C of c0 and 0 LeFe:C;Va and
1 e
LFe:C;Va of e, which are only relevant for the ternary taken from Reference 22 as during re-optimization
system, and the ternary parameters 0 LcFe:C;N of c and attempts on the basis of the available equilibrium
0 e
LFe:C;N and 1 LeFe:C;N of e were optimized. data, the e phase became unreasonably stable at high
temperature. It was also attempted to include the
parameter  GeFe:C in the optimization. This led to the
unreasonable appearance of the e phase in the binary
IV. APPLIED MODELS; OPTIMIZATION Fe-C system instead of h for a large temperature
PROCESS range. Thus, it was decided to keep the value from
Reference 96 as also used in Reference 22. The opti-
The descriptions for the Gibbs energy of the pure
mization process revealed that several binary interac-
elements were taken from Reference 95. The formula
tion parameters of zeroth and first order (sub-regular
units of the sublattice models applied in the present
solution model) were necessary in order to obtain an
work for each phase have been listed in Table I. A
acceptable description of the e phase (see also the
thermodynamic database file is supplied as supplemen-
discussion in Sections I and especially V–D). The
tary data. The values for the parameters of the thermo-
introduction of ternary interaction parameters was
dynamic description for the a phase were taken from
not necessary.
Reference 22. For the c phase, only the parameter
0 c The resulting model parameters as determined in this
LFe:C;N was included in the optimization (see end of work and as taken from the literature are presented in
Section III–B). The values of the binary parameters of c Table II. If desired, the model for the Fe-N-C liquid
were taken from References 9 and 22. phase from Reference 56 can be included.
The thermodynamic parameters of c0 were completely
reassessed (see end of Sections III–A and III–B). In the
0
model for c0 from Reference 22,  GcFe:Va had been set V. DISCUSSION
equal to 4 GcFe:Va . In order to correctly model the
thermodynamics of c0 , then the interaction parameters A. The Binary Fe-N Phase Diagram
0 0
0 c
LFe:N;Va and 1 LcFe:N;Va (both T-dependent) had to be The temperatures and the compositions of the phases
introduced in Reference 22. In the present work, at the invariant equilibria in the Fe-N system, as
0
 c
GFe:Va was used as an optimization variable, eliminat- predicted by the thermodynamic description resulting
ing the need for any interaction parameters for c0 and from the present work, can be compared with the
thus reducing the number of parameters as compared to experimental data from References 12, 13, and 15 and
the model of Reference 22. Because of the lack of the previous predictions from References 22 and 24 in
accurate data for the C solubility in c0 , the parameter Table III. The agreement of these features of the Fe-N
 c
0 phase diagram with the experimental data is comparably
GFe:C was fixed to a value giving a reasonable good for the new and old[22,24] thermodynamic
homogeneity range of c0 in the ternary Fe-N-C system. descriptions.
The c0 phase shows ferromagnetic ordering with a The Fe-N phase diagram as calculated using the
somewhat concentration-dependent Curie temperature model parameters from the present work is shown in
around 763 K (490 C).[12] As there is no heat-capacity Figure 1. Various enlarged sections of the phase
data available, which would allow introducing the diagram are shown in Figure 2 to allow a more
magnetic moment as a fitting parameter, no magnetic detailed comparison with both the experimental
model is used . data[13,15,25,35,36,66,67,70–76,78–82,84,87,88] and the previous
The thermodynamic description for the h phase was predictions from References 22 and 24.
taken directly from Reference 7. No N solubility had to The homogeneity range of a agrees well with the
be modeled as explained in Section III–B. experimental data and with the homogeneity range
For the e phase, the sublattice model of resulting from the thermodynamic description of Refer-
Fe(C,N,Va)1=2 as used in References 22 and 55 through ence 22 (see Figure 2(a)). The newer experimental data
57 was also applied in the present work. The e phase from Reference 13 are better described by the thermo-
shows magnetic ordering with a Curie temperature dynamic description from Reference 24. This descrip-
strongly varying with, at least, the N content between 10 K tion, however, shows an a + e equilibrium below
and 550 K (260 C and 280 C).[12] As for the c0 phase, approximately 580 K (310 C), a temperature at which
the lack of heat-capacity data prevents fitting for the the a + c0 equilibrium is observed experimentally (see

6176—VOLUME 47A, DECEMBER 2016 METALLURGICAL AND MATERIALS TRANSACTIONS A

 Table II. Thermodynamic Model Parameters for the Solid Solution Phases as Determined in the Present Work and as Taken
from the Cited Literature to be Used with the Unary Gibbs-Energy Functions from Ref. [95]
a, Model Fe(C,N,Va)3
 a;non-mag a;non-mag
GFe:Va ¼  GFe
 a;non-mag a;non-mag
GFe:C   GFe  3 Ggra
C ¼ 322,050 + 75.667T
[9]


Ga;non- mag
 G a;non- mag gas
 3  G ¼ 93,562 + 165.07T[21]
Fe:N Fe 2 N2
0
LaFe:C;Va ¼ 190T[9]
baFe:Va ¼ baFe:C ¼ baFe:N ¼ baFe ¼ 2.22[9,21]
TaCurie;Fe:Va ¼ TaCurie;Fe:C ¼ TaCurie;Fe:N ¼ TaCurie;Fe ¼ 1043[9,21]

c, model Fe(C,N,Va)
 c;non-mag c;non-mag
GFe:Va ¼  GFe
 c;non-mag  c;non-mag
GFe:C  GFe   Ggra
C ¼ 77,207  15.877T
[9]

 c;non-mag a;non-mag
GFe:N   GFe  12  Ggas
N2 ¼ 20,277 + 245.3931T  21.2984TlnT
[22]

0
LcFe:C;Va ¼ 34,671[9]
0
LcFe:N;Va ¼ 26,150[16]
0
LcFe:C;N ¼ 8218
bcFe:Va ¼ bcFe:C ¼ bcFe ¼ 0.7[9]
TcNeel;Fe:Va ¼ TcNeel;Fe:C ¼ TcNeel;Fe ¼ 67[9]

c0 , model Fe4(C,N,Va)
0
a;non-mag
 c
GFe:C  4 GFe   Ggra
C ¼ 20,000
0
a;non-mag
 c
GFe:N  4 GFe  12  Ggas
N2 ¼ 37,744 + 72.786T

 4 Ga;non-mag ¼ 12,066 + 3.691T

0
 c
G Fe:Va Fe

e, model Fe(C,N,Va)1=2

GeFe:Va ¼  GeFe
1  gra

GeFe:C   GcFe  2 GC ¼ 52,905  11.9075T[96]
1  gas

GeFe:N   GaFe  ¼ 13,863 + 40.2123T[22]
4 GN2
0 e
LFe:C;Va ¼ 53059
1 e
LFe:C;Va ¼ 38,756
0 e
LFe:N;Va ¼ 8186  18.127T
1 e
LFe:N;Va ¼ 24,378 + 24.959T
0 e
LFe:C;N ¼ 20,772  32.504T
1 e
LFe:C;N ¼ 28,839

h, model Fe3(C,Va)
 h;non-mag
GFe:C  3HSER
Fe  HSER
C ¼ 8983 þ 658:38T  113:578T ln T  3:059  103 T2 þ 6:105  105 T1 [7]
 h;non-mag
G Fe:Va  3 Ga;non-mag ¼ 44,782  11.59T[7]
Fe
bhFe:C ¼ bhFe:Va ¼ 1.51[7]
ThCurie;Fe:C ¼ ThCurie;Fe:Va ¼ 485[10]

T in K, values of Gibbs energy and interaction parameters in J mol1.

also below) and, therefore, disagrees with the experi- on nitrided specimens[15] showed N contents of up to 1
mental phase-boundary data at low temperatures (cf. at. pct less than given in older works.[67,70,74,83] For the
Figure 2(a)). phase boundary c/c + c0 , the predictions by the ther-
The calculated phase boundaries of the c single-phase modynamic description from the present work lie
field agree well with the few available experimental data, between these values. For the phase boundary c/c + e,
see Figure 2(b). Experimental data on the phase bound- the data from Reference 15 is described better than the
aries c/c + c0 and c/c + e is somewhat contradictory: data from References 67, 70, 74, and 83. Overall, a good
recent investigations obtained by EPMA measurements representation of all data employed in the optimisation

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 47A, DECEMBER 2016—6177

 Table III. Comparison of Temperatures and Compositions of Phases Participating in the Invariant Reactions
0
Reaction Reference T [K (C)] xaN (at. pct) xcN (at. pct) xcN (at. pct) xeN (at. pct)
e Ð c0 present work (pred) 964 (691) 19.6 19.6
[12] (exp) 953 (680) 19.5 19.5
[15] (exp) 938–948 (665–675) N/A N/A
[22] (pred) 971 (698) 19.4 19.4
[24] (pred) 955 (682) 19.6 19.6
e Ð c+c0 present work (pred) 923 (650) 9.7 19.1 16.3
[12] (exp) 923 (650) 10.3 19.3 15.9
[15] (exp) 923–925 (650–652) N/A N/A N/A
[22] (pred) 923 (650) 9.7 19.1 16.1
[24] (pred) 923 (650) 10.3 19.6 15.9
c Ð a + c0 present work (pred) 865 (592) 0.39 9.0 19.2
[12] (exp) 865 (592) 0.40 8.8 19.3
[13] (exp) 866 (593) 0.44 N/A N/A
[22] (pred) 863 (590) 0.39 9.0 19.3
[24] (pred) 867 (594) 0.40 8.9 19.6
exp, Experimentally determined; pred, predicted by a thermodynamic description.

1200
N contents in the low-T range. In this low-T range, the
agreement of the experimental data with the phase
1100 γ boundary from Reference 98, based on a thermody-
γ+ε ε namic description considering ordering of N on its
1000
α+ γ sublattice, is better, which is also shown in Figure 2(d).
temperature T/K

900 The thermodynamic description of Reference 98, how-

800 γ′ ever, gives multiple expressions for the phase boundary
α γ +γ ′ in order to cover the whole temperature range (see the
700 overlap of the two curves in Figure 2(d)). Moreover, in
600 α+ γ′ the high-T range, the agreement with the experimental
data is poor and there is a maximum in the proposed
500
phase boundary. This is thermodynamically only possi-
400 ble if the congruent transition e Ð c0 occurs at N
α+ ε contents as high as 21.2 at. pct, which is impossible
300
0 0.05 0.10 0.15 0.20 0.25 0.30 according to the model from the present work (maxi-
molar fraction xN mum N content of c0 is 20 at. pct) and also incompatible
with the prediction according to the model for c0 from
Reference 98.
Fig. 1—The Fe-N phase diagram as calculated using the thermody-
namic description from the present work, suppressing formation of A ‘‘potential phase diagram’’ using the activity of N
the N2 gas phase. At a temperature below 443 K (170 C), an a + e (reference state N2 gas at 1 9 105 Pa and at the
equilibrium is predicted, see the dotted lines (see discussion in Sec- considered temperature) as a variable is shown in
tion V–B). Figure 3(a), allowing comparison of the phase bound-
aries as calculated using the thermodynamic description
is given, also accounting for the error margins of the from the present work with the respective phase
usually applied EPMA method to determine these boundaries as calculated using the datasets from Refer-
phase-boundary compositions. ences 22 and 24, the phase boundary c0 /e as calculated
The c0 /c0 + e phase boundary agrees well with the using the expressions given in Reference 98, and the
experimental data from Reference 87, whereas the N phase boundaries as indicated by the experimental
content at the phase boundary a + c0 /c0 is lower than data.[13,25,68,69,71,76,77,79,82,85] The same diagram using
indicated by most of the experimental data and by the as a variable the often applied nitriding potential
phase boundary as calculated using the thermodynamic
description from Reference 22, but still agrees within pNH
rN ¼ 3=23 ; ½5
less than 0.1 at. pct (see Figure 2(c)). Better agreement pH2
could be achieved by introducing a more advanced
model considering N disorder, see Section V–C. a (technical process) parameter used for gaseous nitrid-
The thermodynamic description from the present ing, which is a measure for the activity of N, but
work reproduces the phase boundary c0 + e/e signifi- multiplied with  p1=2 (where  p ¼ 1  105 Pa is the
cantly better than the previous descriptions[22,24] (see pressure of the reference state, to obtain a dimensionless
Figure 2(d)). The data point from Reference 66 and variable[1]), is given in Figure 3(b). For the phase
semi-quantitative investigations in Reference 97 suggest boundary a/c0 , the thermodynamic dataset of the
that the phase boundary c0 + e/e might extend to lower present work describes the experimental data equally

6178—VOLUME 47A, DECEMBER 2016 METALLURGICAL AND MATERIALS TRANSACTIONS A

 1200 1100
[67]
[70]
1100 [76]
[74]
[73] 1050 [35]
[83]
1000 [74] α +γ [13]

temperature T/K
[75]
ε
temperature T/K
[15]
[76]
1000
900 [78]
[80] α γ+ε
800
[35]
[81]
γ
[84] 950
[13]
700 α + γ′ α+ γ
600
900 γ + γ′
(α + ε )
500 850
0 1 2 3 4 5 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18
-3
molar fraction xN x 10 molar fraction xN
(a) (b)
950 1000
[72] [66]
[25] [67]
[82] [70]
900 [87]
[88]
900 ε [71]
[74]
[79]

temperature T/K
γ′
temperature T/K

850 [36]
γ+ γ′ 800 [87]

800 γ′+ ε
700 γ′+ ε
750
α+ γ′ [98]
600
700
(α+ε)
650 500
0.190 0.192 0.194 0.196 0.198 0.200 0.202 0.20 0.22 0.24 0.26 0.28 0.30
molar fraction xN molar fraction xN
(c) (d)
present work [22] [24]

Fig. 2—Magnified sections of the Fe-N phase diagram as calculated using the thermodynamic description from the present work (solid lines) in
comparison to (i) the phase diagrams as calculated using the descriptions from Refs. [22] (dashed lines) and [24] (dotted lines) and (ii) various
experimental data. An a + e equilibrium is predicted using the description of Ref. [24] in the shown temperature range. (a) Low N-content range
with experimental data from Refs. [13, 35, 67, 73–76, 78, 80, 81] and [84]. (b) Equilibria involving the c phase compared with experimental data
from Refs. [13, 15, 35, 67, 70, 74] and [83]. (c) The homogeneity range of c0 compared with experimental data from Refs. [25, 72, 82, 87]
and [88]. Note that the thermodynamic description of Ref. [24] describes c0 as a stoichiometric compound with the formula Fe4:1 N. (d) The c0 +
e/e phase boundary compared with experimental data from Refs. [36, 66, 67, 70, 71, 74, 79] and [87]. Also, the phase boundary redrawn from
Fig. 4 in Ref. [98] (variant for low N content) is shown.

as well as the dataset published in Reference 22. For the (1381 C). A similar prediction is obtained from the
c range, the thermodynamic descriptions from the thermodynamic descriptions of the Fe-N system of
present work and from References 22 and 24 reproduce References 21 and 22. The thermodynamic descriptions
the experimental data well. However, the phase bound- from References 20 and 24, however, predict a congru-
ary c0 /e as calculated using the thermodynamic dataset ent transition c Ð e. Since there is no corresponding
of the present work agrees significantly better with the experimental data available, no conclusion can be drawn
experimental data than the phase boundaries resulting which variant is correct. Nevertheless, the shape of the c
from the previous descriptions of References 22 and 24. phase field as predicted by References 20 and 24 seems
The phase boundary c0 /e is even better described with unrealistic. Thus, the thermodynamic description
the expressions given in Reference 98. However, in that involving peritectic melting is the preferred one, at least
work direct least-squares fitting of the phase boundary until more experimental data is available.
was performed, yielding several expressions for different At low temperatures, an a + e equilibrium is predicted
temperature ranges. The thermodynamics of both the by the presently obtained thermodynamic description,
binary and ternary c0 and e phases are discussed in a which disappears upon heating at 443 K (170 C, see the
comparative manner in Sections V–C and V–D, dotted lines in Figure 1). The same feature at low temper-
respectively. ature is predicted using the thermodynamic descriptions
Including the thermodynamic description of the liquid from References 20, 22 and 24. The appearance of an
Fe-N phase from Reference 21 peritectic melting of e is a + e equilibrium at low temperatures is not necessarily a
predicted in the binary Fe-N system at 1654 K modeling artifact: experimental investigations[99–101]

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 47A, DECEMBER 2016—6179

 1000 1000
γ [69]
γ
[68]
900 [71]
900
[76]
[69]

temperature T/K
temperature T/K

α [77]
[79]
[68]
γ′
800 [25]
800 [71]
γ′ [82]
[76]
[77]
[85]
[79]
[13]
700 700 [25]
[82]
[85]
[13] α
600 ε 600

500 500 -2 -1 0 1
2
10
3
10 10 10 10 10
activity a nitriding potential rN °p
1/2
N
(a) (b)
present work [22] [24] [98]

Fig. 3—Potential phase diagrams as calculated using the thermodynamic description from the present work (solid lines), and as calculated using
the descriptions from Refs. [22] (dashed lines) and [24] (dotted lines). For comparison, the phase boundary c0 /e as given by Eqs. [14b/c] in
Ref. [98] (dash-dot lines) and experimental data for the phase boundaries from Refs. [13, 25, 68, 69, 71, 76, 77, 79, 82] and [85] are also shown.
(a) Using the activity of N as a variable (reference state N2 gas at 1  105 Pa and the respective temperature). (b) Using the nitriding potential
3=2
rN ¼ pNH3 =pH2 (multiplied with  p1=2 to obtain a dimensionless quantity) as a variable, which is a measure for the activity of N.[1]

suggested that e was in equilibrium with a at low have been discussed in detail in Reference 53, offering two
temperatures [approximately £550 K (280 C)]. Another possibilities realized by the previous various thermody-
work[97] excluded the possibility of the a + e equilibrium in namic descriptions for the Fe-N-C system.[20,22,24,55–57] In
the binary Fe-N system at low temperatures and discussed the first case, upon cooling, the a + e equilibrium is
the possible formation of the cubic a¢¢-Fe16N2 nitride as an replaced by the c0 + h equilibrium via a single transition
equilibrium phase, which was not included in the present reaction U2 , a + e Ð c0 + h. In the second case, upon
assessment. At least at 623 K (350 C) (and at higher cooling, the c0 + h equilibrium first appears via the
temperatures), precipitation of c0 from e was still pseudo-binary eutectoid reaction e4, e Ð c0 + h, dividing
observed.[102] However, due to the very slow kinetics at the e single phase field into two separate e single phase
those low temperatures it is difficult to reach a genuine fields. Subsequently, the second e single phase field
equilibrium state and this prohibits to draw a final vanishes via the ternary eutectoid reaction E2 , e Ð
conclusion. Finally, as a fine point, recent ab initio calcu- a + c0 + h. The thermodynamic description of the pre-
lations pertaining to 0 K (273 C)[103,104] indicate that a sent work reproduces the first sequence of invariant
mechanical mixture of pure a-Fe and e-Fe3N with a gross reactions (see Figure 4).
N content of 20 at. pct has a lower enthalpy than pure The temperature of the transitional reaction U2 as
c0 -Fe4N, supporting the occurrence of an a + e equilib- calculated using the dataset from the present work [839 K
rium at low temperature. It is noted that this point was not (566 C)] is only slightly below the value of the temper-
addressed specifically in these works.[103,104] ature for this reaction as determined experimentally
[842 ± 2 K (569 ± 2 C)].[53] The here predicted temper-
atures of the U1 and E1 invariant reactions are within the
B. The Ternary Fe-N-C System
boundaries determined experimentally for these reactions
A Scheil reaction scheme[105–107] illustrating the in Reference 15. The previous thermodynamic descrip-
sequence of invariant reactions as resulting from the tions for the system Fe-N-C[20,22,24,54–57] only describe a
present thermodynamic description of the Fe-N-C system part of the invariant temperatures correctly and give
is shown in Figure 4. The temperatures of the invariant significantly deviating values for other ones (see the
reactions in the Fe-N-C system as calculated using the discussion in Reference 53 and the examples in Table IV).
thermodynamic description of the present work can be In contrast, the new thermodynamic description predicts
compared in Table IV with the corresponding experi- correctly the values of all invariant temperatures, as
mental data and the predictions as obtained using the recently determined experimentally.
thermodynamic descriptions from References 22 and 24, Isothermal sections of the Fe-N-C phase diagram as
which are the two thermodynamic descriptions giving the calculated using the thermodynamic dataset derived in
best agreement with experimentally determined invariant the present work at the (technologically relevant) tem-
temperatures[15,53] according to the detailed discussion in peratures of 853 K and 893 K (580 C and 620 C) are
Reference 53. Both Figure 4 and Table IV use the desig- shown in Figure 5 together with the phase boundaries as
nations for the invariant reactions introduced in Refer- proposed in Reference 15 on the basis of EPMA inves-
ence 53. The possibilities for the sequence of invariant tigations on nitrocarburizing Fe specimens. These exper-
reactions in the system Fe-N-C below 853 K (580 C) imental data have been used in the optimization process.

6180—VOLUME 47A, DECEMBER 2016 METALLURGICAL AND MATERIALS TRANSACTIONS A

Table IV. Comparison of the Temperatures of the Invariant Reactions in the Ternary Fe-N-C System, Using the Designations for
the Invariant Reactions Introduced in Ref. [53]

Reaction Experiment This Work Ref. [22] Ref. [24]

[15]
U1 , c þ h Ð a þ e 868 to 873 (595 to 600) 873 (600) 867 (594) 952 (679)
E1, c Ð a þ c0 þ e 853 to 863 (580 to 590)[15] 857 (584) 859 (586) 857 (584)
U2 ; a þ e Ð c0 þ h 842 ± 2 (569 ± 2)[53] 839 (566) 783 (510) —
E2 ; e Ð c0 þ h 840 to 844 (567 to 571)[53] — — 833 (560)
e4 ; e Ð a þ c0 þ h 840 to 844 (567 to 571)[53] — — 825 (552)
All values in K (C); as predicted by the present thermodynamic description and the descriptions from Refs. [22] and [24] and as experimentally
determined in the cited literature.

equilibrium[51] also agree well with the predictions from

binary Fe–N ternary Fe–N–C binary Fe–C the present work.
e1 γ α+θ
At 893 K (620 C), the agreement with the experi-
1000 K (727 °C) mental data is also good, especially for the c single phase
c ε γ′ field, with the deviations between the data from Refer-
964 K (491 °C) α+γ+θ
ence 15 and the calculated phase boundaries being close
γ+ε+θ to the accuracy of EPMA. There is a clear deviation at
e2 ε γ+γ′
923 K (450 °C) 893 K (620 C) for the phase boundary c0 + e/e.
γ+γ′+ε U1 γ+θ α+ε However, the phase boundary given there is an estima-
873 K (600 °C) tion which is compatible with EPMA data presented in
e3 γ α+γ′
865 K (592 °C) the same work but not based on a thermodynamic
α+γ+γ′ model. In principle, the phase boundary as predicted by
α+γ+ε α+ε+θ the thermodynamic description from the present work is
E1 γ α+γ′+ε compatible with the EPMA data from Reference 15, see
857 K (584 °C) their Figure 8(b). Thermodynamic calculations at the
α+γ′+ε same temperature using the description from Refer-
ence 22 showed significantly higher N contents an lower
U2 α+ε γ′+θ C contents in e for the equilibria with c and especially h
839 K (566 °C) than determined in Reference 15 and predicted by the
thermodynamic description from the present work.
T α+γ′+θ γ′+ε+θ Both at 853 K and 893 K (580 C and 620 C), the
phase boundaries of the c0 phase show excellent agree-
Fig. 4—Scheil reaction scheme representing the sequence of invariant ment with those shown in Reference 15. The ternary c0
reactions predicted by the thermodynamic description from the pre- single phase field given there is, however, not based on
sent work, using the designations for the invariant reactions intro- quantitative experimental data, but has only estimative
duced in Ref. [53]. character. The c0 single phase field resulting from the
model from the present work agrees well with the
It was not possible to obtain an even better fit of the phase EPMA data mentioned in Section III–B.
boundaries without allowing the formation of a large The priority in the present work was to obtain a
miscibility gap in the e phase. The agreement with the reasonable representation of the phase boundaries of c
phase boundaries from Reference 15 at 853 K (580 C) is at 893 K (620 C) and to describe correctly the invariant
significantly better than as obtained by the predictions temperatures. On this basis, the parameter 0 LcFe:C;N was
from References 22 and 55 and comparable to the phase optimized. In order to obtain a better description of c in
boundaries resulting from the prediction from Refer- the range of higher temperatures, a temperature depen-
ence 24 as follows from Figure 8d in Reference 15. At the dence of 0 LcFe:C;N could be introduced as soon as more
same temperature, according to the descriptions from
References 56 and 57, the a + e equilibrium is non-exis- experimental data is available. Even though systemati-
tent or just disappearing. Therefore, the phase boundaries cally too low, the here determined prediction for the N
resulting from these descriptions cannot be compared to solubility in carbon containing c is already good; see the
experimental data or to the phase boundaries resulting comparison of experimental data[90] and the values
from the description from the present work. Further predicted by the thermodynamic description from the
experimental data points have been given in Figure 5(a): present work in Table V.
(i) the single data point from Reference 41 for the phase
boundary a + e/e shows higher N and lower C contents C. The Appropriateness of a Model for the c0 Phase
than predicted here; (ii) the data points for the same phase
boundary from Reference 46 (for 120 and 240 minutes), In the present work, a different approach for model-
however, agree well with the calculations from the present ing the thermodynamics of c0 than in Reference 22 has
0
work; (iii) recent experimental data measured at 853 K been used. Instead of setting  GcFe:Va equal to the Gibbs
(580 C) for either the a + e/e or the a + e + h/e energy of c-Fe and introducing interaction parameters,

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 47A, DECEMBER 2016—6181

 0.10 0.09
[41]
[46] 0.08 893 K
0.08 ε+θ [51] α +ε/ε
0.07
(620 °C)
[51] α +ε+θ/ε
γ+ε+θ ε+θ

molar fraction xC
α+γ
molar fraction xC

0.06
0.06 α+ε+θ 853 K ε
0.05 +θ
(580 °C)
0.04
ε 0.04
γ+θ
0.03
α+ε γ+ε γ′+ε
0.02 γ′+ε 0.02 γ γ+γ′+ε
α+γ′+ε α+γ′ 0.01 γ+γ′ γ′
γ′ α+γ
0 0
0.10 0.12 0.14 0.16 0.18 0.20 0.22 0.24 0.26 0 0.05 0.10 0.15 0.20 0.25
molar fraction xN molar fraction xN
(a) (b)
present work [15]

Fig. 5—Isothermal sections of the Fe-N-C phase diagram calculated using the thermodynamic description from the present work (solid lines)
compared with phase boundaries proposed in Ref. [15] (dashed lines). (a) At 853 K (580 C), also including separate data from Refs. [41, 46]
and [51]. (b) At 893 K (620 C).

Table V. The Solubility of N in C-containing c in Fe-N system, were discussed in detail in Reference 26. In
Equilibrium with N2 Gas at 1  105 Pa as Experimentally that work modeling was performed using three different
Determined in Ref. [90] and as Predicted by the Model approaches and the results were compared with the
from the Present Work thermodynamic description from Reference 22. The
‘‘Langmuir-type approach’’ in Reference 26 is identical
wN9102 wN9102 to the model applied in the present work. The other two
T wC (wt pct) (wt pct) models (‘‘Wagner–Schottky (WS) approach’’ and
[K (C)] (wt pct)a (exp, Ref. [90]) (calc, this work) ‘‘Gorsky–Bragg–Williams (GBW) approach’’) allow for
1323 (1050) 0.57 2.28 2.04 (dis)order of N. In Figure 1 in Reference 26, a function
1373 (1100) 0.44 2.25 2.03 characterizing the deviation of the thermodynamic data
0.75 2.12 1.82 from the expected values according to a Langmuir-type
1423 (1150) 0.43 2.18 1.95 model (yielding a constant value for this function), is
0.50 2.18 1.90 plotted showing the good fit of the WS and GBW models
0.76 2.00 1.75 to the experimental data from Reference 25. Thus, it was
0.76 1.94 1.75 concluded[26] that a model allowing for disorder is needed
1473 (1200) 0.46 2.08 1.86
in order to give a meaningful description of the thermo-
0.50 2.07 1.84
0.78 1.88 1.68 dynamics of the c0 phase, with the ‘‘WS approach’’
and the ‘‘GBW approach’’ giving equally meaningful
Compositions expressed in mass fractions wN and wC . descriptions.
a
Theoretically, for the experiments considered in Ref. [90], the Using the expressions for the Gibbs energy of c0 in
activity of C is zero at the surface of the specimen. However, as there is
no decarburizing medium in the gas phase, C remains in the substrate. Reference 109 it can be shown that also the WS and GBW
models, allowing for disorder of N, can be expressed in the
compound energy formalism[59] using a sublattice model
0

GcFe:Va has been used as a model parameter. Its value indicated by the formula unit Fe4(N,Va)(N,Va)3, i.e., as
should, therefore, not be interpreted as the Gibbs energy compared to the sublattice model applied in the present
of a hypothetical compound. The physical meaning of work with the formula unit Fe4(N,Va), with a second
0 interstitial sublattice, and ideal (WS) or regular (GBW)
the value of  GcFe:Va 4 GaFe can be understood as the interactions. Following the conclusion from Reference 26,
Gibbs energy of N-vacancy formation according to the we then tried to use a model equivalent to the ‘‘WS
formal reaction ð1  xÞ Fe4N + 4x a-Fe Ð Fe4N1x, as approach’’ from Reference 26 in the binary Fe-N system.
similarly described in References 6 and 7. As mentioned The optimization of the model parameters, however, gave
0
in Section IV, the parameter  GcFe:C was adjusted in a unreasonable results with e.g., c0 replacing the c phase or
way that the solubility of C in c0 agrees well with the no disorder in c0 at all. This is caused by a lack of any direct
experimental equilibrium values. Therefore, a physical experimental data quantifying disorder in c0 . Other
interpretation is even more difficult. Ab initio calcula- approaches to the thermodynamics of the c0 phase adopt
tions performed in Reference 108 indicated a positive the cluster variation method,[27–29,110] but the experimental
enthalpy of formation of c0 -Fe4C as it is the case in the data do not allow to prefer one or the other model:
present work. If realistic errors for the N-content determination in
Experimentally obtained data for the relationship of Reference 25 are assumed (approximately 5 pct), the
the activity of N and the N content of c0 ,[25] for the binary resulting deviations from the experimental results of the

6182—VOLUME 47A, DECEMBER 2016 METALLURGICAL AND MATERIALS TRANSACTIONS A

 0.200 model used in the present work can be ascribed to this
experimental uncertainty. Therefore, and also because the
0.199
[25] 723 K (450 °C)
solubility of C in c0 is considered, it was decided to adopt
molar fraction xN [25] 773 K (500 °C) the two-sublattice model in the present work, giving a
[25] 823 K (550 °C)
0.198
[82] 773 K (500 °C) reasonable agreement of the predicted activity curves and
the data from References 25 and 82 (see Figure 6).
0.197

0.196
D. The Thermodynamics of the e Phase
0.195 The values of the Gibbs energy of the end-members of
e,  GeFe:C and  GeFe:N have been taken from previous
0.194 descriptions.[22,96] The enthalpies of formation following
0 200 400 600 800 1000
activity aN
from these Gibbs-energy functions are compatible with
recent ab initio data from Reference 111, predicting
negative values close to the one following from the
Fig. 6—Comparison of the predicted relationship of N activity and applied Gibbs-energy function for various ordering
N content for binary c0 (solid lines) and the experimental data from
Refs. [25] and [82].
states of nitrides with the formula Fe2 N. The values
for the corresponding carbides with the formula Fe2 C
from Reference 111 are positive as it is the case for the
Gibbs-energy function applied in the present work,
whereas they are considerably smaller.
The relationship of the activity of N and the N content
in e for the binary Fe-N system as obtained from the
molar fraction xN

0.3 present thermodynamic description is shown in Figure 7.

The prediction agrees well with the (rather inaccurate)
[17] 623 K (350 °C) experimental data (errors in the range of 5 to 10 at. pct)
[17] 673 K (400 °C) from Reference 89, as follows from Figure 7. Consider-
0.25 [17] 723 K (450 °C) ing the more accurate data from Reference 17, shown in
[17] 773 K (500 °C) Figure 7 as well, good agreement occurs in the region of
[17] 823 K (550 °C)
[89] 823 K (550 °C)
low N content and low N activity; at higher N activities
lower N contents are predicted than experimentally
0.2 observed. It was not possible to reproduce these data
3 4
10 10 better without losing the good agreement with the two-
activity aN and three-phase equilibrium data in which e participates.
In the present work, the thermodynamics of the e phase
Fig. 7—Comparison of the relationship of N activity and N content
have been described focusing on correct description of the
in binary e resulting from the thermodynamic description from the available solid-solid equilibrium data. The description of
present work (lines), with T increasing from 623 K to 823 K (350 C the activity of N in e on the basis of the gas-solid
to 550 C) in steps of 50 K from top to bottom, and experimental equilibrium data from References 17 and 89 could be
data from Refs. [17] and [89]. better described by using models more explicitly consid-
ering the state of order in e than that in the present case.
Several approaches have been presented in the liter-
0.3 ature to describe the thermodynamic behavior of e in the
xN aN range of high N content, i.e., close to the maximum N
0.25
1518 content of xN ¼ 1=3. Descriptions on the basis of a
long-range order, GBW model of e have been presented
molar fraction xN , xC

1012
0.2 a =506
N 759 in References 14, 30, and 31, similar to the approach
a =759
N
506 mentioned above for c0 ,[26] and descriptions on the basis
aN=1012
0.15 of the cluster variation method have been presented in
aN=1518
References 27 and 32. Finally, ordering in e and also the
0.1 equilibrium with orthorhombic f-Fe2 N (not considered
xC 506 in the present work), which can formally be described as
0.05 759 ordered e, have been investigated recently by first-prin-
1012
1518 ciples calculations.[33] Recognizing the large homogene-
0 ity ranges of N and C in e and the necessary extension of
0 2 4 6 8 10 12
activity aC the binary model into the ternary Fe-N-C system, it is
reasonable to apply a sub-regular solution model for e
indicated by the formula unit Fe(C,N,Va)1=2 , and
Fig. 8—The relationships of N and C activities and N content (solid
lines) and C content (dotted lines) of e at 843 K (570 C) as pre-
therefore, handle all non-ideal (e.g., ordering) effects
dicted by the thermodynamic description from the present work and by introducing interaction parameters instead of apply-
as determined experimentally in Ref. [112] (individual data points). ing a physically more meaningful, but disproportionally

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 47A, DECEMBER 2016—6183

complex long-range order model. This approach has range; the past models either use considerably more
been adopted in the present work, thereby enabling model parameters to yield a similar description of
successful description of the equilibria in both the binary the experimental data or unrealistically model c0 as
Fe-N system and especially the ternary Fe-N-C system. a stoichiometric phase.
In the present work, the function for  GeFe:N has been 3. In the binary Fe-N system, the new thermodynamic
taken from Reference 22. Comparison of the Gibbs description reproduces the experimental data better
energy of e-Fe2 N as predicted by this function and by than previously published thermodynamic descrip-
the thermodynamic description for e from Reference 24, tions, especially the c0 + e equilibrium. The agree-
i.e., setting yeN ¼ 1=2 for the 1:1 model applied there, ment with the experimental data for both the N
shows that the corresponding Gibbs-energy values are content of e and the activity of N at the phase
virtually identical over a large temperature range. boundary c0 /e has been improved significantly.
The relationships of the N and C activities and the N 4. The thermodynamic descriptions available in liter-
and C contents of e at 843 K (570 C) have been plotted ature cannot reproduce recently obtained experi-
in Figure 8 together with the experimental data from mental data. Therefore, a thermodynamic
Reference 112. The agreement is very good for N. description of the ternary Fe-N-C system correctly
Compared to the experimental data, the predicted C describing especially the recently experimentally
contents are too low for high N activities; for low N observed temperatures of the invariant reactions
activities, the predicted values are closer to the exper- has been developed. The new thermodynamic
imentally determined ones. Similar experimental inves- description for the ternary Fe-N-C system also well
tigations have been performed in Reference 113, but are reproduces the recently obtained experimental
not included here recognizing the application of techni- phase boundaries in the system as well as the
cal steels, i.e., a high impurity content, for the specimens (positive) off-diagonal components of the thermo-
used in Reference 113. dynamic factor (pertaining to diffusion in e; as
A general trend visible in Figure 8 is the obvious determined from experiments).
mutual influence of N and C: an increase in the C
activity leads to a decrease of the N content and vice
versa. This can be discussed as follows:
Simultaneous interstitial diffusion of N and C in e is
governed by the thermodynamic factor ACKNOWLEDGMENTS
yi @li Open access funding provided by Max Planck Insti-
#ij ¼ ½6 tute for Intelligent Systems (formerly Max Planck
RT @yj
Institute for Metals Research).
with the chemical potential of component i (=N,C), li ,
being the proportional constant between the intrinsic
diffusion coefficients and the corresponding self-diffu- OPEN ACCESS
sion coefficients.[51,58,114] It has been found that at both
823 K and 853 K (550 C and 580 C) the off-diagonal This article is distributed under the terms of the
components of #ij are positive.[51,58] This information Creative Commons Attribution 4.0 International
has been used as a constraint during the optimization: License (http://creativecommons.org/licenses/by/4.0/),
the thermodynamic description from the present work which permits unrestricted use, distribution, and re-
results in positive values of the off-diagonal components production in any medium, provided you give appro-
of the thermodynamic factor over a large composition priate credit to the original author(s) and the source,
range of e. Only at low N and C contents, not covered provide a link to the Creative Commons license, and
by the experimentally observable homogeneity ranges of indicate if changes were made.
e, negative values of those off-diagonal components of
#ij occur.
The present thermodynamic description predicts that a
small miscibility gap occurs in e below approximately ELECTRONIC SUPPLEMENTARY MATERIAL
855 K (582 C) close to the line connecting Fe2 N and
Fe2 C. No experimental data exist to (in)validate this result.
The online version of this article (doi:10.1007/
s11661-016-3731-0) contains supplementary material,
which is available to authorized users.
VI. CONCLUSIONS

1. New thermodynamic descriptions for the Fe-N

system and the Fe-N-C system have been developed REFERENCES
by focusing on the equilibria involving the c0 and e 1. Thermochemical Surface Engineering of Steels, 3rd ed., E.J.
phases. Mittemeijer, and M.A.J. Somers, eds., Thermochemical Sur-
2. A simple ideal-solution model for the c0 phase has face Engineering of Steels, Woodhead Publishing, Cam-
been used to describe successfully its homogeneity bridge, 2015.

6184—VOLUME 47A, DECEMBER 2016 METALLURGICAL AND MATERIALS TRANSACTIONS A

 2. D.H. Jack and K.H. Jack: Mater. Sci. Eng., 1973, vol. 11, 40. E.J. Mittemeijer, W.T.M. Straver, P.F. Colijn, P.J. van der
pp. 1–27. Schaaf, and J.A. van der Hoeven: Scr. Metall., 1980, vol. 14,
3. E.J. Mittemeijer: ASM Handbook, Vol. 4A, Steel Heat Treating pp. 1189–92.
Fundamentals and Processes, Chapter Fundamentals of Nitrid- 41. A. Wells and T. Bell: Heat Treat. Met., 1983, vol. 10, pp. 39–44.
ing and Nitrocarburizing, ASM International, New York, 2013, 42. H.C.F. Rozendaal, F. Colijn, and E.J. Mittemeijer: Surf. Eng.,
pp. 619–46. 1985, vol. 1, pp. 30–42.
4. E.J. Mittemeijer: Fundamentals of Materials Science, Springer, 43. A. Wells: J. Mater. Sci., 1985, vol. 20, pp. 2439–45.
Heidelberg, 2011. 44. A. Wells: Thin Solid Films, 1985, vol. 128, pp. L33–36.
5. H. Okamoto: J. Phase Equilib., 1992, vol. 13, pp. 543–65. 45. M.A.J. Somers and E.J. Mittemeijer: Surf. Eng., 1987, vol. 3,
6. A. Leineweber, S.L. Shang, and Z.K. Liu: Acta Mater., 2015, pp. 123–37.
vol. 86, pp. 374–84. 46. M.A.J. Somers, P.F. Colijn, W.G. Sloof, and E.J. Mittemeijer: Z.
7. H. Göhring, A. Leineweber, and E.J. Mittemeijer: CALPHAD, Metallkd., 1990, vol. 81, pp. 33–43.
2016, vol. 52, pp. 38–46. 47. H. Du, M.A.J. Somers, and J. Ågren: Metall. Mater. Trans. A,
8. J. Ågren: Metall. Trans. A, 1979, vol. 10A, pp. 1847–52. 2000, vol. 31A, pp. 195–211.
9. P. Gustafson: Scand. J. Metall., 1985, vol. 14, pp. 259–67. 48. M. Nikolussi, A. Leineweber, E. Bischoff, and E.J. Mittemeijer:
10. B. Hallstedt, D. Djurovic, J. von Appen, R. Dronskowski, A. Int. J. Mater. Res. (formerly Z. Metallkd.), 2007, vol. 98, pp.
Dick, F. Körmann, T. Hickel, and J. Neugebauer: CALPHAD, 1086–1092.
2010, vol. 34, pp. 129–33. 49. T. Wöhrle, A. Leineweber, and E.J. Mittemeijer: J. Heat Treatm.
11. R. Naraghi, M. Selleby, and J. Ågren: CALPHAD, 2014, vol. 46, Mat., 2010, vol. 65, pp. 243–48.
pp. 148–58. 50. T. Woehrle, A. Leineweber, and E.J. Mittemeijer: Metall. Mater.
12. H.A. Wriedt, N.A. Gokcen, and R.H. Nafziger: J. Phase Equilib., Trans. A, 2012, vol. 43A, pp. 2401–13.
1987, vol. 8, pp. 355–77. 51. H. Göhring, A. Leineweber, and E.J. Mittemeijer: Metall. Mater.
13. J. Stein, R.E. Schacherl, M.S. Jung, S. Meka, B. Rheingans, and Trans. A, 2015, vol. 46A, pp. 3612–26.
E.J. Mittemeijer: Int. J. Mater. Res. (formerly Z. Metallkd.), 52. H. Göhring, S. Kante, A. Leineweber, and E.J. Mittemeijer: Int.
2013, vol. 104, pp. 1053–65. J. Mater. Res. (formerly Z. Metallkd.), 2016, vol. 107, pp.
14. M.A.J. Somers, B.J. Kooi, L. Maldzinski, E.J. Mittemeijer, A.A. 203–216.
van der Horst, A.M. van der Kraan, and N.M. van der Pers: Acta 53. H. Göhring, A. Leineweber, and E.J. Mittemeijer: Metall. Mater.
Mater., 1997, vol. 45, pp. 2013–25. Trans. A, 2016, vol. 47A, pp. 4411–24.
15. T. Woehrle, H. Cinaroglu, A. Leineweber, and E.J. Mittemeijer: 54. S. Hertzman: Metall. Trans. A, 1987, vol. 18A, pp. 1753–66.
Int. J. Mater. Res. (formerly Z. Metallkd.), 2016, vol. 107, pp. 55. J. Slycke, L. Sproge, and J. Ågren: Scand. J. Metall., 1988,
192–202. vol. 17, pp. 122–26.
16. M. Hillert and M. Jarl: Metall. Trans. A, 1975, vol. 6A, 56. H. Du and M. Hillert: Z. Metallkd., 1991, vol. 82, pp. 310–16.
pp. 553–59. 57. P. Franke and H.J. Seifert, eds.: Landolt–Börnstein—Group IV
17. L. Maldzinski, Z. Przylecki, and J. Kunze: Steel Res., 1986, Physical Chemistry, Volume 19C1—Thermodynamic Properties
vol. 57, pp. 645–49. of Inorganic Materials Compiled by SGTE. Ternary Steel Sys-
18. J. Kunze: Steel Res., 1986, vol. 57, pp. 361–67. tems: Phase Diagrams and Phase Transition Data, Springer,
19. K. Frisk: CALPHAD, 1987, vol. 11, pp. 127–34. Berlin, 2012.
20. J. Kunze: Nitrogen and Carbon in Iron and Steel. Thermody- 58. T. Woehrle, A. Leineweber, and E.J. Mittemeijer: Metall. Mater.
namics, Akademie-Verlag, Berlin, 1990. Trans. A, 2013, vol. 44A, pp. 2548–62.
21. K. Frisk: CALPHAD, 1991, vol. 15, pp. 79–106. 59. M. Hillert: J. Alloys Compd., 2001, vol. 320, pp. 161–76.
22. H. Du: J. Phase Equilib., 1993, vol. 14, pp. 682–93. 60. H. Lukas, S.G. Fries, and B. Sundman: Computational Ther-
23. A.F. Guillermet and H. Du: Z. Metallkd., 1994, vol. 85, modynamics—The Calphad Method, Cambridge University
pp. 154–63. Press, Cambridge, 2007.
24. J. Kunze: Härt.-Tech. Mitt., 1996, vol. 51, pp. 348–55. 61. M. Hillert and L.-I. Staffansson: Acta Chem. Scand., 1970,
25. H.J. Grabke: Ber. Bunsen-Ges. Phys. Chem, 1969, vol. 73, vol. 24, pp. 3618–26.
pp. 596–601. 62. O. Redlich and A.T. Kister: Ind. Eng. Chem., 1948, vol. 40,
26. B.J. Kooi, M.A.J. Somers, and E.J. Mittemeijer: Metall. Mater. pp. 345–48.
Trans. A, 1996, vol. 27A, pp. 1055–61. 63. M. Hillert: Phase Equilibria, Phase Diagrams and Phase Trans-
27. M.I. Pekelharing, A.J. Böttger, and E.J. Mittemeijer: Philos. formations, Cambridge University Press, Cambridge, 2007.
Mag., 2003, vol. 83, pp. 1775–96. 64. G. Inden: in Project Meeting CALPHAD V, W. Pitsch, ed., 1976
28. S. Shang and A.J. Böttger: Acta Mater., 2005, vol. 53, pp. III.4–1–13.
pp. 255–64. 65. M. Hillert and M. Jarl: CALPHAD, 1978, vol. 2, pp. 227–38.
29. A.J. Böttger, D.E. Nanu, and A. Marashdeh: Comput. Mater. 66. E. Lehrer: Z. Techn. Phys., 1929, vol. 10, pp. 177–85.
Sci., 2014, vol. 95, pp. 8–12. 67. O. Eisenhut and E. Kaupp: Z. Elektrochem., 1930, vol. 36,
30. B.J. Kooi, M.A.J. Somers, and E.J. Mittemeijer: Metall. Mater. pp. 392–404.
Trans. A, 1994, vol. 25A, pp. 2797–814. 68. P.H. Emmett, S.B. Hendricks, and S. Brunauer: J. Am. Chem.
31. M.I. Pekelharing, A. Böttger, M.A.J. Somers, M.P. Steenvoorden, Soc., 1930, vol. 52, pp. 1456–64.
A.M. van der Kraan, and E.J. Mittemeijer: Mater. Sci. Forum, 69. E. Lehrer: Z. Elektrochem., 1930, vol. 36, pp. 383–92.
1999, vols. 318–320, pp. 115–20. 70. E. Lehrer: Z. Elektrochem., 1930, vol. 36, pp. 460–73.
32. S. Shang and A.J. Böttger: Acta Mater., 2003, vol. 51, 71. S. Brunauer, M.E. Jefferson, P.H. Emmett, and S.B. Hendricks:
pp. 3597–3606. J. Am. Chem. Soc., 1931, vol. 53, pp. 1778–86.
33. M.B. Bakkedal: Ph.D. Thesis, Technical University of Denmark, 72. C. Guillaud and H. Creveaux: Compt. Rend. Acad. Sci. Paris,
2015. 1946, vol. 222, pp. 1170–72.
34. K.H. Jack: Proc. R. Soc. London, Ser. A, 1948, vol. 185, 73. L.J. Dijkstra: Trans. AIME, 1949, vol. 185, pp. 252–60.
pp. 41–55. 74. L.B. Paranjpe, M. Cohen, M.B. Bever, and C.F. Floe: Trans.
35. E. Scheil, W. Mayr, and J. Müller: Arch. Eisenhüttenwes., 1962, AIME, 1950, vol. 188, pp. 261–67.
vol. 33, pp. 385–92. 75. A. Burdese: Metall. Ital., 1955, vol. 47, pp. 357–61.
36. F.K. Naumann and G. Langenscheid: Arch. Eisenhüttenwes., 76. N.S. Corney and E.T. Turkdogan: J. Iron Steel Inst., 1955,
1965, vol. 36, pp. 677–82. vol. 180, pp. 344–48.
37. D. Gerardin, H. Michel, J.P. Morniroli, and M. Gantois: Mem. 77. J.D. Fast and M.B. Verrijp: J. Iron Steel Inst., 1955, vol. 180,
Sci. Rev. Metall., 1977, vol. 74, pp. 457–67. pp. 337–43.
38. D. Gerardin, H. Michel, and M. Gantois: Scr. Metall., 1977, 78. W. Pitsch and E. Houdremont: Arch. Eisenhüttenwes., 1956,
vol. 11, pp. 557–61. vol. 27, pp. 281–84.
39. J. Matauschek and H. Trenkler: Härt.-Tech. Mitt., 1977, vol. 32, 79. A. Burdese: Ann. Chim. (Rome, Italy), 1959, vol. 49, pp.
pp. 177–81. 1873–1884.

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 47A, DECEMBER 2016—6185

80. M. Nacken and J. Rahmann: Arch. Eisenhüttenwes., 1962, 98. B.J. Kooi, M.A.J. Somers, and E.J. Mittemeijer: Metall. Mater.
vol. 33, pp. 131–40. Trans. A, 1996, vol. 27A, pp. 1063–71.
81. H.A. Wriedt: J. Phase Equilib., 1987, vol. 8, pp. 355–77. 99. D.K. Inia, M.H. Pröpper, W.M. Arnoldbik, A.M. Vredenberg,
82. H.A. Wriedt: Trans. TMS-AIME, 1969, vol. 245, pp. 43–46. and D.O. Boerma: Appl. Phys. Lett., 1997, vol. 70, pp. 1245–47.
83. D. Atkinson and C. Bodsworth: J. Iron Steel Inst., 1970, vol. 208, 100. E.H. Du Marchie van Voorthuysen, B. Feddes, N.G. Chechenin,
pp. 587–93. D.K. Inia, A.M. Vredenberg, and D.O. Boerma: Phys. Stat. Sol.
84. K. Abiko and Y. Imai: Trans. Jpn. Inst. Met, 1977, vol. 18, A, 2000, vol. 177, pp. 127–33.
pp. 113–24. 101. E.H. Du Marchie, D.O. van Voorthuysen, Boerma, and N.C.
85. H.H. Podgurski and F.N. Davis: Acta Metall., 1981, vol. 29, Chechenin: Metall. Mater. Trans. A, 2002, vol. 33A, 2593–98.
pp. 1–9. 102. A. Leineweber, J. Aufrecht, and E. J. Mittemeijer: Int. J. Mater.
86. M.W. Chase Jr.: NIST-JANAF Thermochemical Tables, Fourth Res. (formerly Z. Metallkd.), 2006, vol. 97, pp. 753–59.
Edition, J. Phys. Chem. Ref. Data, Monograph 9, 1998. 103. Y. Imai, M. Sohma, and T. Suemasu: J. Alloys Compd., 2014,
87. Data shown in J. Kunze: Nitrogen and Carbon in Iron and Steel. vol. 611, pp. 440–45.
Thermodynamics. Akademie-Verlag, Berlin, 1990, cited as Z. 104. J.-S. Chen, C. Yu, and H. Lu: J. Alloys Compd., 2015, vol. 625,
Przylecki, L. Maldzinski, private communication. pp. 224–30.
88. M.A.J. Somers and E.J. Mittemeijer: Metall. Trans. A, 1990, 105. E. Scheil: Arch. Eisenhüttenwes., 1936, vol. 9, pp. 571–73.
vol. 21A, pp. 189–204. 106. H.L. Lukas, E.-T. Henig, and G. Petzow: Z. Metallkd., 1986,
89. W.D. Jentzsch and S. Böhmer: Krist. Techn., 1977, vol. 12, vol. 77, pp. 360–67.
pp. 1275–83. 107. B. Predel, M. Hoch, and M. Pool: Phase Diagrams and Hetero-
90. H. Schenck, M.G. Frohberg, and F. Reinders: Stahl Eisen, 1963, geneous Equilibria, Springer, Berlin, Heidelberg, 2004.
vol. 83, pp. 93–99. 108. C.M. Fang, M.A. van Huis, B.J. Thijsse, and H.W. Zandbergen:
91. A. Kagawa and T. Okamoto: Trans. Jpn. Inst. Met., 1981, Phys. Rev. B, 2012, vol. 85, pp. 054116-1–054116-7.
vol. 22, pp. 137–43. 109. B.J. Kooi: Iron-Nitrogen Phases: Ph.D. thesis, TU Delft, 1995.
92. T. Gressmann, M. Nikolussi, A. Leineweber, and E.J. 110. M.I. Pekelharing, A.J. Böttger, M.A.J. Somers, and E.J. Mittemeijer:
Mittemeijer: Scr. Mater., 2006, vol. 55, pp. 723–26. Metall. Mater. Trans. A, 1999, vol. 30A, pp. 1945–53.
93. M. Nikolussi, A. Leineweber, and E.J. Mittemeijer: Philos. Mag., 111. C.M. Fang, M.A. van Huis, J. Jansen, and H.W. Zandbergen:
2010, vol. 90, pp. 1105–22. Phys. Rev. B, 2011, vol. 84, pp. 094102-1–094102-10.
94. H.S. Kitaguchi, S. Lozano-Perez, and M.P. Moody: Ultrami- 112. S. Pietzsch, S. Böhmer, and H.-J. Spies: in Proceedings of the
croscopy, 2014, vol. 147, pp. 51–60. Second International Conference on Carburizing and Nitriding with
95. A.T. Dinsdale: CALPHAD, 1991, vol. 15, pp. 317–425. Atmospheres, 1995, pp. 295–300.
96. J.-O. Andersson: CALPHAD, 1988, vol. 12, pp. 9–23. 113. S. Hoja, H. Klümper-Westkamp, F. Hoffmann, and H.-W. Zoch:
97. S. Malinov, A.J. Böttger, E.J. Mittemeijer, M.I. Pekelharing, and J. Heat Treatm. Mat., 2010, vol. 65, pp. 22–29.
M.A.J. Somers: Metall. Mater. Trans. A, 2001, vol. 32A, 114. J.S. Kirkaldy and D.J. Young: Diffusion in the Condensed State,
pp. 59–73. The Institute of Metals, London, 1987.

6186—VOLUME 47A, DECEMBER 2016 METALLURGICAL AND MATERIALS TRANSACTIONS A

View publication stats