Thermodynamic study of sodium-iron oxides Part II. Ternary phase diagram of the Na-Fe-O system

Article abstract of DOI:10.1016/S0040-6031(03)00137-0

Studies on ternary phase diagrams of the Na-Fe-O system have been carried out from the thermodynamic point of view. Thermodynamic data of main ternary Na-Fe oxides Na₄FeO₃(s), Na₃FeO₃(s), Na₅FeO₄(s) and Na₈Fe₂O ₇(s) have been assessed. A user database has been created by reviewing literature data together with recent DSC and vapor pressure measurements by the present authors. New ternary phase diagrams of the Na-Fe-O system have been constructed from room temperature to 1000 K. Stable conditions of the ternary oxides at 800 K were presented in predominance diagram as functions of oxygen pressure and sodium pressure.

Full text of DOI:10.1016/S0040-6031(03)00137-0

Thermochimica Acta 405 (2003) 67–72
Thermodynamic study of sodium–iron oxides
Part II. Ternary phase diagram of the Na–Fe–O system
Jintao Huang∗, Tomohiro Furukawa, Kazumi Aoto
Oarai Engineering Center, Japan Nuclear Cycle Development Institute, Narita-cho 4002,
Oarai-machi, Ibaraki-ken 311-1393, Japan
Abstract
Studies on ternary phase diagrams of the Na–Fe–O system have been carried out from the thermodynamic point of view.
Thermodynamic data of main ternary Na–Fe oxides Na4FeO3(s), Na3FeO3(s), Na5FeO4(s) and Na8Fe2O7(s) have been
assessed. A user database has been created by reviewing literature data together with recent DSC and vapor pressure mea-
surements by the present authors. New ternary phase diagrams of the Na–Fe–O system have been constructed from room
temperature to 1000 K. Stable conditions of the ternary oxides at 800 K were presented in predominance diagram as functions
of oxygen pressure and sodium pressure.
© 2003 Elsevier Science B.V. All rights reserved.
Keywords: Na–Fe–O system; Phase diagram; Thermodynamics
1. Introduction
diagram at low oxygen potentials and oxygen–sodium
potential diagram at 853 K were presented by
Seetharaman and Du [4]. Possibility of formation
of Na2FeO2(s) was ruled out by them by thermody-
namic analysis. Sridharan et al. [5,6] did extensive
experiments by thermal analysis, solid-state reactions,
etc. . . . so that partial phase diagram over 773 K
was deduced. However, large discrepancy can be
found among the available phase diagrams mentioned
above. For example, Sridharan et al. suggested that
there exist two-phase lines Na4FeO3(s)–Na3FeO3(s)
and Na3FeO3(s)–NaFeO2(s) in 500–650 ◦C, while
Seetharaman et al. reported Na4FeO3(s)–NaFeO2(s)
and Na2O(s)–NaFeO2(s) two-phase lines within the
same temperature zone. Therefore, vapor pressure
measurements on some Na–Fe oxides and DSC ther-
mal analysis were done in the present laboratory. Ther-
modynamic functions of Na3FeO3 and Na4FeO3 were
evaluated once again. A user database was created by
reviewing thermodynamic data of other Na–Fe oxides
Many pioneers have studied ternary phase dia-
grams of the Na–Fe–O system but there still exist
discrepancies. Early phase diagram study on the
Na–Fe–O system had been done by Dai et al. [1,2].
Some isothermal phase diagrams of FeO(s)–Na2O(s),
FeO(s)–Na2Fe2O4(s) and Fe3O4(s)–Na2Fe2O4(s)
were given. Lindemer et al. [3] constructed ternary
Na–Fe–O phase diagram and Ellingham diagram in
which Na2FeO2(s), Na8Fe2O7(s), Na3Fe5O9(s) as
well as Na4Fe6O11(s), were included by using es-
timated thermodynamic data. Their work correctly
predicted the formation of Na(l)–Fe(s)–Na4FeO3(s)
at high temperatures though stability of some ternary
oxides involved had not been confirmed. Partial phase
∗
Corresponding author. Tel.: +81-29-267-4141x5928;
fax: +81-29-267-7148.
E-mail address: huang@oec.jnc.go.jp (J. Huang).
0040-6031/$ – see front matter © 2003 Elsevier Science B.V. All rights reserved.
doi:10.1016/S0040-6031(03)00137-0

68
J. Huang et al. / Thermochimica Acta 405 (2003) 67–72
in literatures. Thus, new ternary phase diagrams and
chemical potential diagrams in the Na–Fe–O system
were constructed from room temperature to 1000 K by
using the Thermo-Calc code. Stability of the ternary
oxides was quantitatively discussed.
2.1.2. Fe–O system
This system is extensively investigated so that orig-
inal data given in the SSUB database were utilized for
the calculation, in which the three well-known solid
solutions, Hematite, Magnetite and Wustite were in-
cluded. Further discussion in this part is no longer
necessary.
2. Thermodynamic aspects
2.1.3. Na–Fe system
No further treatment was done for this binary sys-
tem because no compounds were found and only very
limited solutions were reported.
Thermodynamic basis of equilibrium calcula-
tion and the construction of phase diagram by
the Thermo-Calc can be found in literature [7] as
well as the User’s guide. Present study employed
TERN-module to create the ternary phase diagrams
of the Na–Fe–O system. POLY-module was used to
generate the potential diagram. The SSUB database
provided by SGTE is employed as the main database
for the calculation. A user database was built up for
the calculation. Thermodynamic data source and as-
sessment of literatures and experimental results are
given as the following.
2.2. Ternary Na–Fe oxides
2.2.1. NaFeO2(s)
Data assessed by SGTE was used. Thermodynamic
data of NaFeO2(s) used for the present calculation
are given in detail in Table 1 together with other
ternary Na–Fe oxides. The Gibbs energy of formation
of NaFeO2(s) can be expressed as,
ꢀfG◦(Na4FeO3) = −701849.4 + 204.19 × T
(1)
2.1. Binary system in Na–Fe–O
2.2.2. Na4FeO3(s)
2.1.1. Na–O system
This compound has been studied so many times that
thermodynamic evaluation are possible. Among the
available data of Na4FeO3(s), Bhat and Borgstedt’s re-
sult [9] seems more reliable because their result agrees
quite well with experimental data given by Gross and
Wilson [10] and Shaiu et al. [11]. Therefore, the fol-
A schematic binary Na–O phase diagram can be
found in literature [8]. The three binary oxides in the
system, Na2O(s), Na2O2(s) and NaO2(s), were taken
into account in the present calculation but no solutions
were included for lack of necessary data.
Table 1
Thermodynamic data used for the Na–Fe–O phase diagram calculation
NaFeO2
Na4FeO3
Na3FeO3
Na5FeO4
Na8Fe2O7
ꢀf G◦(298) kJ mol−1
ꢀf H◦(298) kJ mol−1
S◦(298) J mol−1 K−1
−639.98
−698.18
88.3
−1107.52
−1206.13
208.9
−1068.3
−1162.6
172.0
−1462.7
−1596.0
246.3
−2524.3
−2746.0
438.1
Cp(T) = a + b × T – c/T2 (J mol−1 K−1
)
a
b
c
100.77
0.0186
212.49
0.0383
3280000
181.7
0.03339
2967000
262.61
0.04818
4455000
444.3
0.08156
7422000
1479000
G(T) = −g1 + g2 × T – g3×T × ln(T) − g4×T2 + g5/T (J mol−1
)
g1
g2
g3
g4
g5
723394
455.159
80.55
1282201
1244.166
212.49
1228246
1071.530
181.69
0.016695
1483500
1691381
1551.917
262.60
0.024000
2228000
2906986
2603.706
444.3
0.04078
3711000
0.006653
0.0
0.019155
1642000

J. Huang et al. / Thermochimica Acta 405 (2003) 67–72
69
lowing expression is employed in the present study.
2.2.5. Na8Fe2O7
The standard enthalpy of formation was given as
ꢀfH◦(Na8Fe2O7) = −2746.0 kJ mol−1 by Stuve
et al. [14]. Lindemer et al. estimated the entropy of
ꢀfG◦(Na4FeO3) = −1212202 + 351.10 × T
(2)
Since experimental data, such as heat capacities,
enthalpy increments and Gibbs energy functions of
Na4FeO3(s) are not available, estimated data have
to be used. So, Lindemer’s estimation of entropy
S◦(298) = 208.9 J mol−1 K−1 was employed. Heat
capacity Cp(T) given by MALT2 database [12] was
employed that was estimated from those of its com-
ponent oxides. ꢀfH◦(298) was calculated according
to the following formula:
Na8Fe2O7 as about S◦(298) = 438.1 J mol−1 K−1
.
So, ꢀfH◦(Na8Fe2O7) was able to be calculated by
Eq. (2), i.e. ꢀfH◦(298) = −2524.3 kJ mol−1. Its
heat capacity was roughly calculated from those of
Na3FeO3(s) and Na5FeO4(s).
ꢀfG◦(Na8Fe2O7) = −2754068 + 771.86 × T
(6)
2.2.6. Na2FeO2(s) and other higher order oxides
It is reasonable to exclude Na2FeO2(s) from the
present calculation since experimental attempts to
produce this phase failed and theoretic analysis sus-
pected its stability [4–6]. Though some higher or-
der Na–Fe oxides, such as Na3Fe5O9, Na4Fe6O11,
Na10Fe16O29 and Na34Fe8O29 had been reported [3],
they were observed neither in the present laboratory
nor in Sridharan’s. So, these compounds were not
considered in the present calculation too.
ꢀfH◦(298) = ꢀfG◦(298) + T ꢀfS◦(298)
(3)
Thermal analysis on this compound by DSC carried
out in the present laboratory shows that the melting
point is around 1008 15 K. No other phase transitions
were found up to its melting point. Property of its
liquid phase is not determined so the whole calculation
was done below its melting point.
2.2.3. Na3FeO3(s)
Experimental measurement on this compound was
very scarce. The vapor pressure measurements by the
present authors provided its Gibbs energy of formation
[13], as expressed in the following:
3. Calculated phase diagrams and discussion
By means of Thermo-Calc code, new chemical po-
tential diagram and ternary Na–Fe–O phase diagram
were constructed up to about 1000 K. Isothermal sec-
tions of the ternary phase diagram were illustrated in
Figs. 1–6.
The partial phase diagram in the region of Na(l)–
Na4FeO3(s)–Na3FeO3(s)–Fe(s) over 693 K is identi-
cal with the schematic diagram drawn by Sridharan
et al. [5,6] as shown in Figs. 4–6. It indicates that the
present theoretic study agrees well with their experi-
mental results. Outside the above zone, formation of
Na8Fe2O7(s) over about 637 K is thermodynamically
favorable according to this calculation. This prediction
should be consistent with that of Lindemer’s at higher
temperatures if Na2FeO2(s) were excluded from their
study.
ꢀfG◦(Na3FeO3) = −1168629 + 338.34 × T
(4)
The expression should be valid until about 1000 K
because no phase transition was observed till 1033 K
by DSC and XRD analysis. Similar treatments were
made to estimate S◦(298), Cp(T) and ꢀfH◦(298) of
Na3FeO3(s) as expressed above.
2.2.4. Na5FeO4(s)
Thermal analysis in the present laboratory shows
that there are no phase transitions for this compound
from room temperature to 1000 K. Up to date, ex-
perimentally measured results of ꢀfG◦(Na5FeO4)
have been seldom reported in publications. Ther-
modynamic data for this compound, ꢀfH◦(298) =
−1596 kJ mol−1, S◦(298) = 246.3 J mol−1 K−1 were
employed according to Lindemer’s estimation while
heat capacity was estimated in the similar way as
described above.
Due to the importance in nuclear industry, spe-
cial attention was paid to low oxygen potentials.
Na4FeO3(s) is considered as one of the main cor-
rosion products in the sodium-leak incident of the
MONJU FBR. In this calculation, it is found that
Na(l), Fe(s) and Na4FeO3(s) coexist over 694 K and
ꢀfG◦(Na5FeO4) = −1602430 + 467.3 × T
(5)

70
J. Huang et al. / Thermochimica Acta 405 (2003) 67–72
Fig. 3. Isothermal cross-sections of the Na–Fe–O ternary phase
diagram in 637–694 K. (A) Na2O; (B) Na2O2; (C) Na5FeO4;
(D) Na4FeO3; (E) Na8Fe2O7; (F) Na3FeO3; (G) NaFeO2; (H)
Hematite; (M) Magnetite.
Fig. 1. Isothermal cross-sections of the Na–Fe–O ternary phase
diagram in 298–536 K. (A) Na2O; (B) Na2O2; (C) Na5FeO4;
(D) Na4FeO3; (F) Na3FeO3; (G) NaFeO2; (H) Hematite; (M)
Magnetite.
Predominance diagram on coordinates of log P(O2)
versus log P(Na) at 800 K was shown in Fig. 7. It
should be noted that the phase Na4FeO3(s) only exists
at low oxygen potentials compared to the other Na–Fe
complex oxides. The calculated highest oxygen po-
tential at 800 K in which Na4FeO3(s) can stably exist
Na(l)–Na2O(s)–Fe(s) is more stable at lower temper-
atures. The calculated transition temperature is much
higher than 629 K found by Sridharan et al. [15]
but quite close to the 723 K reported by Bhat and
Borgstedt [9].
Fig. 2. Isothermal cross-sections of the Na–Fe–O ternary phase
diagram in 536–637 K. (A) Na2O; (B) Na2O2; (C) Na5FeO4;
(D) Na4FeO3; (F) Na3FeO3; (G) NaFeO2; (H) Hematite; (M)
Magnetite.
Fig. 4. Isothermal cross-sections of the Na–Fe–O ternary phase
diagram in 694–838 K. (A) Na2O; (B) Na2O2; (C) Na5FeO4;
(D) Na4FeO3; (E) Na8Fe2O7; (F) Na3FeO3; (G) NaFeO2; (H)
Hematite; (M) Magnetite.

J. Huang et al. / Thermochimica Acta 405 (2003) 67–72
71
Fig. 5. Isothermal cross-sections of the Na–Fe–O ternary phase
diagram in 838–944 K. (A) Na2O; (B) Na2O2; (C) Na5FeO4;
(D) Na4FeO3; (E) Na8Fe2O7; (F) Na3FeO3; (G) NaFeO2; (H)
Hematite; (M) Magnetite; (W) Wustite.
Fig. 7. Predominance diagram of the Na–Fe–O system at 800 K.
peratures are a little hard to be determined precisely
because of possible errors in estimating the heat ca-
pacities of the corresponding complex oxides. Further
experiments were done and the following equilibrium
reactions were confirmed by vapor pressure measure-
ments and XRD analysis in the present laboratory.
is as low as about −536 kJ mol−1. On the other hand,
stable area for NaFeO2(s) and Na3FeO3(s) can be rela-
tively very wide. Na8Fe2O7 appears over about 637 K
and tends to replace Na5FeO4 when temperature is
higher than about 944 K. The related transition tem-
1
NaFeO2(s) → 1 Fe3O4(s) + Na(g) + O2(g)
3
3
(1288–1380 K) (7)
1
2
Na3FeO3(s) → NaFeO2(s) + 2Na(g) + O2(g)
(823–1023 K) (8)
1
Na5FeO4(s) → 1 Na8Fe2O7(s) + Na(g) + O2(g)
2
4
(603–673 K) (9)
Na4FeO3(s) → Na3FeO3(s) + Na(g) (590–717 K)
(10)
It was a pity that the partial pressures of oxygen in
above conditions were too low to be measured by the
high temperature mass spectrometer. However, phase
identifications, measurements of temperature and par-
tial vapor pressures of sodium provided some experi-
mental evidence that is consistent with the phase dia-
grams constructed in the present study.
Fig. 6. Isothermal cross-sections of the Na–Fe–O ternary phase
diagram in 944–1000 K. (A) Na2O; (B) Na2O2; (C) Na5FeO4;
(D) Na4FeO3; (E) Na8Fe2O7; (F) Na3FeO3; (G) NaFeO2; (H)
Hematite; (M) Magnetite; (W) Wustite.

72
J. Huang et al. / Thermochimica Acta 405 (2003) 67–72
During the phase diagram calculation, it is also
phase diagrams of the Na–Fe–O system have been
constructed from room temperature to 1000 K. Envi-
ronmental effects on stability of the ternary oxides
were presented in oxygen–sodium potential diagram
at 800 K.
found that the phase diagrams are very sensitive to
the Gibbs energy of formation of the main ternary
Na–Fe oxides. So the assessment of thermodynamic
data becomes very important to obtain correct phase
diagrams. For example, even a 0.7 kJ mol−1 positive
shift in ꢀfG◦(Na4FeO3) and ꢀfG◦(Na3FeO3) may
refuse coexistence of the two phases and greatly
changed the ternary phase diagram. Usually, an exper-
iment error of a few kJ mol−1 seems quite reasonable.
It reflects the difficulty of construction of Na–Fe–O
phase diagram. This might be the main reason why
there exists large discrepancy in the Na–Fe–O phase
diagrams published by some pioneers. Fortunately,
thermodynamic evaluations of Na3FeO3(s) in the
present study were directly based on the relationship
between Na4FeO3(s) and Na3FeO3(s) [13]. Thus, the
user database can be considered as self-consistent so
that calculated phase diagrams constructed seem quite
consistent with the experiment results for the time
being.
References
[1] W. Dai, S. Seetharaman, L.-I. Staffansson, Scand. J. Metall.
13 (1984) 32.
[2] W. Dai, S. Seetharaman, L.-I. Staffansson, Metall. Trans. B
15 (1984) 319.
[3] T.B. Lindemer, T.M. Besmann, C.E. Jonhson, J. Nucl. Mater.
100 (1981) 178.
[4] S. Seetharaman, S. Du, High Temp. Mater. Proc. 12 (1993)
145.
[5] R. Sridharan, T. Gnanasekaran, C.K. Mathews, J. Alloys
Compd. 191 (1993) 9.
[6] R. Sridharan, T. Gnanasekaran, G. Periaswami, C.K.
Mathews, in: J. Borgstedt, G. Fress (Eds.), Liquid Metal
Systems, Plenum Press, New York, 1995, p. 269.
[7] B. Sundman, B. Jansson, J.-O. Andersson, Calphad 9 (1985)
153.
[8] T.B. Massalski, Binary Alloy Phase Diagram, second ed.,
1990, ASM International, Metals Park, OH.
[9] N.P. Bhat, H.U. Borgstedt, J. Nucl. Mater. 158 (1988) 7.
[10] P. Gross, G.L. Wilson, J. Chem. Soc. A (1970) 1913.
[11] B.J. Shaiu, P.C.S. Wu, P. Chiotti, J. Nucl. Mater. 67 (1977)
12.
4. Conclusion
Ternary phase diagrams of the Na–Fe–O sys-
tem are studied from the thermodynamic point of
view. By reviewing literature data together with re-
cent DSC and vapor pressure measurements by the
present authors, thermodynamic data of main ternary
Na–Fe oxides Na4FeO3(s), Na3FeO3(s), Na5FeO4(s)
and Na8Fe2O7(s) have been evaluated. New ternary
[12] H. Yokokawa, S. Yamauchi, T. Matsumoto, Themochim. Acta
245 (1994) 45.
[13] J. Huang, T. Furukawa, K. Aoto, Thermochim. Acta, in press.
[14] J.M. Stuve, L.B. Pankratz, D.W. Richardson, Report of
investigations 7535, US Department of the Interior, Bureau
of Mines, 1971.
[15] R. Sridharan, D. Krishnamurthy, C.K. Mathews, J. Nucl.
Mater. 167 (1989) 265.