PDF Archive

Easily share your PDF documents with your contacts, on the Web and Social Networks.

Share a file Manage my documents Convert Recover PDF Search Help Contact

10.1002@zaac.201300676 .pdf

Original filename: 10.1002@zaac.201300676.pdf
Title: Formation and Decomposition of Iron Nitrides Observed by in situ Powder Neutron Diffraction and Thermal Analysis

This PDF 1.3 document has been generated by SYSTEM400 Rev 18.04 / PDFlib PLOP 2.0.0p6 (SunOS)/Acrobat Distiller 7.0.5 (Windows), and has been sent on pdf-archive.com on 25/11/2016 at 16:29, from IP address 134.169.x.x. The current document download page has been viewed 652 times.
File size: 1 MB (10 pages).
Privacy: public file

Download original PDF file

Document preview

DOI: 10.1002/zaac.201300676

Formation and Decomposition of Iron Nitrides Observed by in situ Powder
Neutron Diffraction and Thermal Analysis
Marc Widenmeyer,[a] Thomas C. Hansen,[b] Elke Meissner,[c] and Rainer Niewa*[a]
Dedicated to Professor Gerd Meyer on the Occasion of His 65th Birthday
Keywords: IR spectroscopy; Iron; Neutron diffraction; Nitrides; X-ray diffraction
Abstract. In order to gain more information on the formation and
decomposition behavior of various iron nitrides from different starting
materials in situ neutron diffraction and thermal analysis under application of different gas atmospheres and heating rates were carried out.
The following phases were observed during these investigations:
crystalline α-Fe, γ-FeNz, γ⬘-Fe4Ny, ε-Fe3N1+x, ζ-Fe2N, FeCl2,
[Fe(NH3)6]Cl2, Fe(NH3)2Cl2, and amorphous Fe(NH3)Cl2. In situ neutron diffraction data were collected in high quality, due to an optimized
experimental setup with a time resolution of two minutes on D20 (Institut Laue-Langevin) allowing for detailed Rietveld analyses. For all
phase transitions, decomposition and formation temperatures a strong

dependency from the heating rate, thermal history of the sample, gas
flow conditions, and particle size exists. The nitrogen contents observed during thermal decomposition of ζ-Fe2N were related to the
binary phase diagram Fe-N. At low temperatures (⬍ 400 °C) ε- and
ζ-phase are non-equilibrated. However, through annealing equilibrium
state is reached. For γ⬘-Fe4Ny formed in situ at higher temperatures an
expansion of the homogeneity range towards lower nitrogen content is
observed above 600 °C. For the formation of ε-Fe3N1+x from FeCl2
and NH3 a previously proposed reaction sequence involving different
ammoniates was confirmed. This reaction occurs via formation of
amorphous Fe(NH3)Cl2. The therein observed in situ formed ammoniates were additionally characterized by IR spectroscopy.

1 Introduction

mental setups. Furthermore, with exception of the early refractory metal nitrides,[6] they suffer from high decomposition
pressures and low temperature stability.[7] As a consequence
the most common synthesis approach for (iron) nitrides is still
related to solid-gas-reactions yielding mostly microcrystalline
products. Single crystals of one binary iron nitride phase
ε-Fe3N1+x was achieved so far, but only by a high-pressure–
high-temperature approach.[8–11] In addition, the nitrogen has
only very low scattering power compared with most transition
metals in X-ray diffraction, this hinders an accurate determination of structural details.[12] In situ powder neutron diffraction as a bulk measurement technique, allowing for complex
sample environments to investigate solid-gas-reactions is ideally suited to resolve the above described analytical challenges.[12–15] Inherently, the problem of low scattering power
of e.g. nitrogen is eliminated, due to the large coherent neutron
scattering length bc of nitrogen different to those of other light
elements. Thus, nitrogen [bc = 9.36(2) fm] can easily be located in the crystal structure and distinguished from e.g. fluorine [bc = 5.65(1) fm], oxygen [bc = 5.803(4) fm] and carbon
[bc = 6.646(1) fm].[12] Therefore, possible intermediates in
chemical reactions can be observed by in situ diffraction studies and a reaction pathway may be proposed. The both information should help increasing the efficiency of single phase
transition metal nitride synthesis.
Before presenting the results concerning the formation and
decomposition of γ⬘-Fe4N, ε-Fe3N1+x, and ζ-Fe2N from different starting materials, we give a brief overview on the fun-

Iron nitrides are well-established examples for transition
metal nitrides usable in various technical applications. Especially γ⬘-Fe4Ny and ε-Fe3N1+x play an important role in the
industrial steel-surface hardening process, improving hardness
and tribological properties as well as corrosion and wear resistance of the work piece.[1,2] Furthermore, the ferromagnetic
iron nitrides are in research focus as magnetic data recording
materials according to their tuneable magnetic properties.[3,4]
Dealing with transition metal nitride synthesis and analysis
several obstacles have to be taken into account. In the first
place the synthesis of nitrides in an oxygen containing atmosphere is thermodynamically unfavored. Thus, in synthesis
oxygen has to be strictly excluded[5] via use of special experi* Prof. Dr. R. Niewa
Fax: +49-711-685-64 241
E-Mail: rainer.niewa@iac.uni-stuttgart.de
[a] Institute of Inorganic Chemistry
University of Stuttgart
Pfaffenwaldring 55
70569 Stuttgart, Germany
[b] Institut Laue-Langevin
6 rue Jules Horowitz
BP 156, 38052 Grenoble Cedex 9, France
[c] Fraunhofer Institute for Integrated Systems and Device Technology
Department of Crystal Growth
Schottkystr. 10
91058 Erlangen, Germany
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/zaac.201300676 or from the author.
Z. Anorg. Allg. Chem. 2014, 640, (7), 1265–1274

© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim


R. Niewa et al.

damental information of the interrelationships of the various
participating iron nitride phases and the previously known
facts to explain the achieved results.
During the initial investigations concerning the ammonia
synthesis in the Haber-Bosch process first attempts to determine the Fe–N phase diagram were carried out.[16–18] Six
thermodynamic stable iron nitride phases are described in the
binary phase diagram of Fe–N. A solid solution of nitrogen in
iron exists for each modification of elemental iron. According
to their bcc structures in α- and δ-iron only a very small
amount of nitrogen is soluble. For instance nitrogen ferrite (αFe) can take up to a maximum of 1.6 ⫻ 10–2 wt % nitrogen.[19]
In nitrogen austenite (fcc, γ-Fe) in contrast up to 10.3 at %
nitrogen can dissolve. Additionally, three stable iron nitride
phases are present: γ⬘-Fe4Ny with only a small homogeneity
range, ε-Fe3N1+x with a broad phase width, and ζ-Fe2N with
an again narrow range of compositions.[20] Furthermore, four
metastable phases are reported in literature: bct α⬘-Fe8N, α⬘⬘Fe16N2 as an order-variant thereof,[15,21] FeN in rock salt[22]
and in sphalerite structure types.[23]
Nitrogen austenite (γ-FeNz) is only stable above 592 °C.
The iron atoms are arranged in an fcc lattice and the nitrogen
atoms occupy octahedral interstitials (max. 10.3 %) in a disordered manner (space group Fm3¯m). The unit cell parameter
depends linearly on the nitrogen content.[20]
γ⬘-Fe4Ny crystallizes in space group Pm3¯m, which means
the iron atoms are arranged in a fcc motif and nitrogen occupies ¼ of the octahedral interstitials in a perfectly ordered
manner (inverse perovskite-type).[7] The strain energy in the
structure is minimized through the nitrogen order.[24] The
homogeneity range is very narrow at room temperature and
slightly broadened with increasing temperature.[14] The maximum value is reached at 600 °C.[20]
In the ε-phase iron atoms are arranged in an hcp motif and
nitrogen occupies octahedral interstitials. Due to partially ordered arrangements of nitrogen depending on composition, the
original hcp unit cell has to be expanded in a direction
according to the relation a⬘ = 冪3ahcp.[25–27] Through different
grades of filling of the octahedral voids a very large homogeneity range is realized, reaching from ε-Fe3N0.65 up to
ε-Fe3N1.47.[10,28] The unit cell parameters show a strong linear
dependency on the nitrogen content.[28,29] Diffraction data of
powderous samples were typically refined using space group
P6322,[30] however, a symmetry reduction to space group P312
decreases repulsion forces between nitrogen atoms according
to theoretical investigations.[9,31]
The ζ-phase has the hcp motif of iron atoms in common
with the ε-phase, but the nitrogen atoms show a different ordering scheme within the octahedral voids. Therefore, the crystal system is changed to orthorhombic and the space group
Pbcn is realized. The homogeneity range is narrow
again.[20,32,33] First thermodynamic considerations and measurements were carried out as early as 1901.[34] In 1952 it
was shown that polycrystalline ζ-Fe2N can be transformed into
microcrystalline ε-Fe3N1+x through annealing at temperatures
between 300 °C and 450 °C.[31] The thermodynamics of this
transformation were described later.[35] Furthermore, polycrys1266


talline ζ-Fe2N was transformed into single crystalline
ε-Fe3N1.47 at 15(2) GPa and 1600(200) K in a high-pressure–
high-temperature experiment.[10]

2. Results and Discussion
2.1 Investigations on the Decomposition Behavior of ζ-Fe2N
The decomposition behavior of ζ-Fe2N was investigated in
two independent in situ neutron diffraction experiments with
different heating rates and temperature programs in flowing
argon (100 sccm). The differentiation between the structurally
closely related phases ζ-Fe2N and ε-Fe3N1+x is difficult in Xray diffraction, because both phases share an almost identical
iron substructure. However, neutron diffraction easily allows
distinguishing both, due to their distinctly different nitrogen
substructure. From powder X-ray data both investigated samples were assumed to be single phase ζ-type, but neutron diffraction data revealed a contamination with ε-phase in both
In the first experiment (Figure 1) a sample consisting of
81(2) wt % ζ-phase and 19(2) wt % ε-phase according to Rietveld refinements on neutron diffraction data was heated with
5 K·min–1 to 400 °C and annealed at this temperature for 2 h
(argon atmosphere), because a transformation of ζ- into εphase was discussed in this temperature region.[31] This transformation was observed in both in situ neutron diffraction
studies (see below). Subsequently, the temperature was increased to 450 °C with 5 K·min–1 and hold constant for 1 h to
convert the formed ε-phase completely into γ⬘-Fe4Ny. We have
previously shown that ε-Fe3N1.37(2) transforms into γ⬘-Fe4Ny
above 427 °C under conditions of TG experiments.[14] To enhance the conversion rate from ε- to γ⬘-phase, the temperature
was further raised with 2.5 K·min–1, finally obtaining α-Fe after complete loss of nitrogen at T ⬎ 674 °C.

Figure 1. Temperature dependent (5 K·min–1 to 400 °C, 2 h,
5 K·min–1 to 450 °C, 1 h, 2.5 K·min–1 to 750 °C) neutron diffraction
data[36] of the decomposition of ζ-Fe2N (1) (ζ-Fe2N, ε-Fe3N1+x) in
flowing argon [λ = 186.802(5) pm]. The formation of ε-Fe3N1+x (2)–
(3) (start: T = 349 °C, complete: T = 400 °C) and γ⬘-Fe4Ny (5)
[T ⬎ 450 °C] is observed. Between 400 °C and 450 °C ε-phase looses
nitrogen (3)–(4). Both nitride phases are subsequently converted to αFe (6) (T ⬎ 569 °C) through further heating. (8) represents cooling
to 40 °C. During the experiment for 30 min no diffraction data were
collected, due to technical reasons (7). Measurement at D20, intensity
is given in false colors.

© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Z. Anorg. Allg. Chem. 2014, 1265–1274

Formation and Decomposition of Iron Nitrides

While increasing the temperature to 400 °C ζ-Fe2N very
slowly starts to convert into ε-Fe3N1+x above 349 °C (Figure 2). At 400 °C the conversion rate is considerably increased,
resulting in a 1:1 mixture of ε-Fe3N1.18(3) and ζ-Fe2N0.95(1)
after 8 min of annealing. At 400 °C this transformation to
ε-Fe3N1.25(3) is completed after another 65 min. During heating
to 450 °C ε-Fe3N1.25(3) slowly starts to release nitrogen and
the composition is changed to ε-Fe3N0.98(2). Above 450 °C the
in situ formed ε-phase is gradually converted into a nitrogenpoor γ⬘-phase with a composition of γ⬘-Fe4N0.86(9), while on
further heating the nitrogen content of that phase is increased
to γ⬘-Fe4N0.97(1) (Figure 3). Here again the rate of conversion
markedly increases with increasing temperature. Above 540 °C
the γ⬘-phase also starts to release nitrogen to a final composition of γ⬘-Fe4N0.919(7) at 599 °C and simultaneously residual
amounts of the ε-phase decompose. Additionally, α-Fe
emerges above 569 °C and the transformation of γ⬘-Fe4Ny into

α-Fe is completed at 674 °C (Figures S1–S4 and Table S1,
Supporting Information).
During the thermal decomposition of ζ-Fe2N with such a
rather small heating rate it is obvious that the transfer of nitrogen between the solid nitride phases is preferred over an alternative direct release of N2 from the nitrides up to about
540 °C. A similar behavior was previously observed in a lower
temperature region for the thermal decomposition of α⬘⬘Fe16N2.[15] Additionally, a second nitrogen transfer mechanism
is observed: Especially for ζ-Fe2N and ε-Fe3N1+x with decreasing amount of the respective nitride phase (Figure 2) the nitrogen content increases (Figure 3), which means most of the nitrogen is first enriched in the respective phase, before a gradual
conversion into the next stable phase proceeds. According to
the smaller heating rate applied in the in situ neutron diffraction experiment, all phase transitions, especially those related
to a release of N2, are observed at lower temperatures than in
previous TG experiments.[14] This observation can be taken
as strong evidence for kinetic control of the respective phase
In a second experiment (Figure 4) with a sample containing
66(5) wt % ζ-phase and 34(5) wt % ε-phase, the temperature
was raised in 100 sccm argon with 5 K·min–1 to 450 °C and
kept for 30 min. Afterwards, the reaction temperature was
raised with 2.5 K·min–1 to 700 °C.

Figure 2. Phase fractions of phases observed by in situ neutron diffraction during the decomposition of ζ-Fe2N (squares) in argon via εFe3N1+x (filled triangles), and γ⬘-Fe4Ny (open circles) to α-Fe (diamonds). The temperature is represented as black solid line.

Figure 4. Temperature dependent (5 K·min–1 to 450 °C, 0.5 h,
2.5 K·min–1 to 700 °C) neutron diffraction data[36] of the decomposition of ζ-Fe2N (1) (ζ-Fe2N, ε-Fe3N1+x) in flowing argon [λ =
186.690(7) pm]. The formation of ε-Fe3N1+x (2)–(3) (start: T = 450 °C,
complete: T ⬎ 516 °C) (with two different nitrogen contents), γ⬘-Fe4Ny
(4) (T ⬎ 526 °C) and γ-FeNz (5) (T ⬎ 614 °C) is observed. All nitride
phases are subsequently converted in α-Fe (6) and (7) (T ⬎ 639 °C for
γ⬘ and T ⬎ 679 °C for γ) through further heating. Measurement at
D20, intensity is given in false colors.

Figure 3. Nitrogen contents of phases observed by in situ neutron diffraction during the decomposition of ζ-Fe2Nw (squares) in argon via
ε-Fe3N1+x (filled triangles) and γ⬘-Fe4Ny (open circles) to α-Fe. The
temperature is represented as black solid line.
Z. Anorg. Allg. Chem. 2014, 1265–1274

Due to the different thermal treatment applied, ζ-Fe2N
showed distinctly different behavior with increasing temperature. The phase fraction of the ζ-phase is remarkable increased
from 66(5) wt % at room temperature to 85(5) wt % at 432 °C,
coincidently the amount of ε-phase is reduced to 15(5) wt %.
Ex situ experiments concerning the formation of ζ-Fe2N show
the occurrence of an Ostwald ripening process from ε-Fe3N1+x
to ζ-Fe2N (see Section 2.2). Through increasing the temperature, the activation energy for this process is provided and subsequently more of the ε-phase is converted into a nitrogen-

© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim



R. Niewa et al.

poor ζ-phase. This transformation is accompanied by a change
in the composition from ζ-Fe2N1.00(3) to ζ-Fe2N0.94(2) and from
ε-Fe3N1.22(7) to ε-Fe3N1.26(8). With decreasing amount of ζphase the nitrogen content of the remaining ζ-phase again increases to Fe2N1.00(8). After 30 min at 450 °C a large amount
of the ζ-phase is transformed into ε-phase. 19(4) wt % of
ζ-Fe2N1.00(6) and 81(4) wt % of ε-Fe3N1.27(4) are observed.
Above 516 °C the ζ-phase is completely converted into
ε-Fe3N1.36(8) and γ⬘-Fe4Ny starts to appear [2.7(5) wt %,
T ⬎ 526 °C]. At 536 °C already 13(1) wt % of γ⬘-Fe4N0.9(1) are
observed. Additionally, two ε-phases with different nitrogen
contents and therefore different unit cells are visible:
54(5) wt % of ε-Fe3N1.09(6) and 33(2) wt % of ε-Fe3N1.32(6).
With increasing temperature the amount of the second ε-phase
is strongly reduced, but with decreasing amount the nitrogen
content increases to ε-Fe3N1.44(7) at 550 °C [19(1) wt %]. This
process is followed by a very fast and sharp release of nitrogen
between 550 °C and 555 °C resulting in a composition of εFe3N1.11(5). However, the unit cell apparently is not able to
relax during this quick reaction (Table S2). At 570 °C one εphase with a composition of Fe3N1.26(7) is reformed again to
give a mixture of 58(2) wt % ε-Fe3N1.26(7) and 42(2) wt % γ⬘Fe4N0.92(2). Recently, it was observed by in situ neutron diffraction of the synthesis of Sr2MnO4–δ in flowing hydrogen
that two phases Sr2MnO4–δ with varying compositions, and in
this case, different structure types emerge during annealing at
500 °C. This conversion was incomplete even through extended annealing, but at higher temperatures (550 °C) the
transformation completes. This behavior was assumed to be
related to the large sample volume hindering a homogeneous
exchange with the surrounding gas atmosphere.[37] We suppose
that similar reasons are responsible for the temporary formation of two ε-phases with differing nitrogen contents. A typical
Rietveld plot as observed during the decomposition of ζ-Fe2Nw
is shown in Figure 7.
During further heating the nitrogen content of the γ⬘-phase
increases to a maximum value of Fe4N0.95(1) at 580 °C, followed by a slight release of nitrogen. At 614 °C most of the
ε-phase [remaining 3.0(2) wt %] is converted into 92(1) wt %
of γ⬘-Fe4N0.918(6) next to 4.7(5) wt % nitrogen austenite
(γ-FeNz) containing 4(3) at % nitrogen. At 634 °C all of the
ε-phase is consumed and the mixture consists of 68.8(9) wt %
γ⬘-Fe4N0.907(6) and 31.2(9) wt % of γ-FeN0.085(6). Above
639 °C formation of α-Fe slowly starts [1.01(1) wt %]. The
maximal amount of γ-FeNz with 73(3) wt % is reached at
654 °C and the maximum nitrogen content of γ-FeN0.091(6) at
644 °C. With further increase of the temperature the amount
of α-Fe increases rapidly to 68(1) wt % at 664 °C and all of
the γ⬘-phase is converted into α-Fe and 32(1) wt % γ-FeN0.07(1)
. Upon continued heating the amount of γ-FeNz rapidly decreases and most of the nitrogen is released, until at 679 °C all
nitrogen is released and only α-Fe remains (Figure 5, Figure 6,
and Figure 7, Figures S5–S7 and Table S2, Supporting Information).
Figure 8 relates the observed compositions with the accepted phase diagram Fe-N.[20] Below 450 °C a composition
of ζ-Fe2N0.94(2) is observed, which is well in agreement with


Figure 5. Phase fractions of phases observed by in situ neutron diffraction during the decomposition of ζ-Fe2N (squares) in argon via εFe3N1+x (filled and open triangles represent ε-phases with different
nitrogen content), γ⬘-Fe4Ny (open circles) and γ-FeNz (filled circles)
to α-Fe (diamonds). The temperature is represented as black solid line.
The drop in temperature curve is related to changing the heating mode
of the furnace.

Figure 6. Nitrogen contents of observed phases by in situ neutron diffraction during the decomposition of ζ-Fe2Nw (squares) in argon via
ε-Fe3N1+x (filled and open triangles represent ε-phases with different
nitrogen content), γ⬘-Fe4Ny (open circles) and γ-FeNz (filled circles)
to α-Fe. The temperature is represented as black solid line. The drop
in temperature curve is related to changing the heating mode of the

the value presented previously,[33] but not in accordance with
the homogeneity range of ζ-Fe2N. We assume that the equilibrium state is not yet reached according to the low temperature.
Through annealing at 450 °C the nitrogen content of ζ-Fe2N
reaches the nitrogen-rich phase boundary. Below 510 °C
ζ-Fe2N and ε-Fe3N1+x coexist and for the ε-phase nitrogen
contents near the upper phase boundary are observed. Above
510 °C ε-Fe3N1+x coexists with γ⬘-Fe4Ny and therefore nitrogen-poor ε-phases near the lower phase boundary are observed
The nitrogen content of the sample is conserved up to 530 °C
again. Above this temperature, a gradual release of nitrogen
occurs. Furthermore, especially for the two ε-phases, a transfer
of nitrogen between the solid phases rather then a release is

© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Z. Anorg. Allg. Chem. 2014, 1265–1274

Formation and Decomposition of Iron Nitrides

2.2 Investigations of the Reaction of Iron with NH3

Figure 7. Simultaneous Rietveld refinements of the crystal structures
of ζ-Fe2Nw next to ε-Fe3N1+x at T = 450 °C. Observed (grey circles),
calculated (solid black line) and difference (observed–calculated, grey
line) in situ powder neutron diffraction patterns [λ = 186.690(7) pm].
Vertical bars indicate the Bragg positions of ζ-Fe2Nw (light grey) and
ε-Fe3N1+x (dark grey). Rwp = 3.83 %, Rp = 2.57 %, RBragg(ζ-Fe2Nw) =
2.15 %, RBragg(ε-Fe3N1+x) = 1.23 %.

Recently, we presented in situ neutron diffraction data on
the synthesis of γ⬘-Fe4Ny and ε-Fe3N1+x from elemental iron
and ammonia.[14] Formation of ζ-Fe2N was not observed during such experiments so far. Due to experimental requirements
in this previous experiment a rather large gas flow of 510 sccm
ammonia was applied,[14] which is much higher as the one
normally used in ex situ synthesis of ζ-Fe2N with about
60 sccm NH3. To exclude that this fact hinders the formation
of ζ-Fe2N through a smaller nitridation potential,[40] new experiments were carried out using an experimental setup allowing for reduced ammonia gas flow of 60 sccm. Again only
γ⬘-Fe4Ny and ε-Fe3N1+x were observed (Figure 9). While raising the temperature with a rate of 5 K·min–1 at 392 °C first
weak reflections of γ⬘-phase [Fe4N0.96(6)] are visible. The temperature was further raised to 413 °C, which represents the ex
situ observed formation temperature of the ζ-phase. However,
only the amount of γ⬘-phase [Fe4N0.93(3)] increases up to
7.8(7) wt %. After 70 min of annealing at this temperature the
phase fraction of γ⬘-phase [Fe4N0.959(6)] increases to
64(1) wt %. Because no indication for the formation of the
ζ-phase was observed the temperature was raised with
5 K·min–1 to 463 °C, so that more iron was converted into
γ⬘-Fe4Ny and subsequently the nitrogen content of the γ⬘-phase
was slightly increased from γ⬘-Fe4N0.959(6) to γ⬘-Fe4N0.967(6).
This nitrogen uptake is accompanied with an expansion of the
unit cell parameter about five times larger as expected from
pure thermal expansion.[41] After 10 min at 463 °C first weak
reflections representing ε-Fe3N0.99 [0.45(9) wt %] appear. As
can be taken from Figure 10 another 35 min later elemental
iron was nearly completely consumed and converted into
γ⬘-phase [87(1) wt % Fe4N0.975(5)] and ε-phase [7.2(5) wt %
Fe3N1.1(1)]. With increased reaction time the nitrogen content
of both phases increases, but no reflections of the ζ-phase were
observed (Figure 11, Figures S8 and S9 and Table S3, Supporting Information). A representative Rietveld plot as observed

Figure 8. Binary phase diagram of Fe–N after reference[20] with addition of respective nitrogen contents of γ-FeNz (filled circles), γ⬘Fe4Ny (open circles), ε-Fe3N1+x (filled and open triangles; different
nitrogen content) and ζ-Fe2Nw (squares) as observed by in situ neutron
diffraction data. The symbols are larger as the experimental errors of
temperatures and nitrogen contents.

observed. In accordance with our previously presented results
on the formation of γ⬘-Fe4Ny[14] the nitrogen contents observed
herein suggest an expansion of the homogeneity range of
γ⬘-Fe4Ny towards lower nitrogen contents above 600 °C. In
general for the lower phase boundary of γ⬘-Fe4Ny great variations ranging from γ⬘-Fe4N0.87(3) to γ⬘-Fe4N0.93(2) at 600 °C
were observed.[38,39] Nitrogen austenite first coexists with γ⬘Fe4Ny. During the decomposition of γ⬘-Fe4Ny the nitrogen content of γ-FeNz increases to a maximum value at 644 °C. Above
this temperature the nitrogen content decreases again through
decomposition of γ-FeNz in α-Fe.
Z. Anorg. Allg. Chem. 2014, 1265–1274

Figure 9. Temperature dependent (5 K·min–1 to 413 °C, 1 h,
5 K·min–1 to 463 °C, 1 h) neutron diffraction data[36] of the reaction
of α-Fe in flowing ammonia [λ = 186.690(7) pm]. The formation of
γ⬘-Fe4Ny (1) (T ⬎ 392 °C) and ε-Fe3N1+x (2) (T ⬎ 463 °C) is observed.
Measurement at D20, intensity is given in false colors on a logarithmic

© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim



R. Niewa et al.

during the reaction of iron powder with ammonia is shown in
Figure 12.

Figure 10. Phase fractions of phases observed by in situ neutron diffraction during the reaction of α-Fe (diamonds) with ammonia producing γ⬘-Fe4Ny (open circles) and ε-Fe3N1+x (triangles). The temperature
is represented as black solid line.

Figure 11. Nitrogen contents of phases observed by in situ neutron
diffraction during the reaction of α-Fe with ammonia producing
γ⬘-Fe4Ny (open circles) and ε-Fe3N1+x (triangles). The temperature is
represented as black solid line.

Apparently, the gas flow rate has hardly any influence on
the formation of γ⬘-Fe4Ny, but has a significant influence on
the formation rate of ε-Fe3N1+x. Previously, we observed the
occurrence of γ⬘-Fe4Ny at 387 °C, which is in good agreement
to the value of 392 °C presented herein. For the ε-phase the
formation temperature is strikingly increased from 397 °C to
463 °C compared to the previous experiment with a larger gas
flow applied.[14] Interestingly the nitrogen content of those two
ε-phases is identical within experimental error, namely
ε-Fe3N1.15(8)[14] and ε-Fe3N1.1(1).
To understand why no ζ-phase was formed during the in
situ experiment, different ex situ experiments were carried out,
quenching samples after different annealing durations at various temperatures (Table S4, Supporting Information). Clearly,
the formation of ζ-Fe2N always proceeds via the formation of
γ⬘-Fe4Ny and ε-Fe3N1+x. Apparently, ζ-Fe2N precipitates in


Figure 12. Simultaneous Rietveld refinements of the crystal structures
of α-Fe next to γ⬘-Fe4Ny and ε-Fe3N1+x at T = 463 °C. Observed (grey
circles), calculated (solid black line) and difference (observed–calculated, grey line) in situ powder neutron diffraction patterns [λ =
186.690(7) pm]. Vertical bars indicate the Bragg positions of α-Fe
(light grey), γ⬘-Fe4Ny (grey), and ε-Fe3N1+x (dark grey). Rwp = 4.39 %,
Rp = 2.61 %, RBragg (α-Fe) = 1.27 %, RBragg(γ⬘-Fe4Ny) = 2.30 %,
RBragg(ε-Fe3N1+x) = 4.24 %.

some kind of Ostwald ripening process from ε-Fe3N1+x, with
a rather slow reaction rate. At 353 °C, beside a large amount
of unreacted α-Fe, some γ⬘-Fe4Ny and traces of ε-Fe3N1+x are
formed. Raising the temperature to 383 °C, the amount of
γ⬘-Fe4Ny formed well increases. Additionally, more of the
α-Fe is converted into ε-Fe3N1+x. Further increase of the temperature to 413 °C converts most of the α-Fe into nitride
phases, now mainly ε-Fe3N1+x with x ≈ 0.2 and γ⬘-Fe4Ny.
While increasing the reaction time up to 6 h at 413 °C, most
of the γ⬘-phase is converted into ε-Fe3N1.26(5). After 12 h of
reaction time the γ⬘-phase is completely consumed and first
weak reflections of ζ-Fe2N are observed.
A significant higher formation temperature of ε-phases is
observed in situ. We assume this to be related to different sample volumes (ca. 10 cm3 in situ vs. 0.3 cm3 ex situ) and gassolid contact areas. Therefore, a homogeneous nitridation potential might not be realized for the whole sample in the in situ
experiments. Nevertheless, in both studies the same reaction
sequence was observed.
2.3 Investigations of the Reaction of FeCl2 with NH3
Another approach for the preparation of transition metal nitrides, beside direct conversion of the metal with ammonia,
was first described in the 1940⬘s by Juza et al. for the synthesis
of Co3N and Ni3N from the respective metal(II) halides and
gaseous ammonia. It was proposed that the formation of the
nitrides proceeds via intermediate ammoniates, subsequently
decomposing to a very fine and consequently more reactive
powder of the metal(II) halide at higher temperatures, but no
experimental proof was presented and no further characterization of those ammoniates was carried out.[42,43] For the trans-

© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Z. Anorg. Allg. Chem. 2014, 1265–1274

Formation and Decomposition of Iron Nitrides
Table 1. Data of TG/DTA analysis of [Fe(NH3)6]Cl2 in argon in combination with an independent TG/DTA/EGA analysis in helium. The
respective onset temperatures are given.
T /°C

Δmobs. / %

Δmexp. / %

Obtained compound






formation of iron halides into iron nitride (ε-Fe3N1+x) the first
results were given in 2003.[44] We have reinvestigated this alternative synthesis route in order to form ε-Fe3N1+x using thermal analysis (TA) and in situ neutron diffraction.
In a DTA/TG experiment in combination with exhaust gas
analysis (EGA) [Fe(NH3)6]Cl2 was heated in flowing helium
up to 900 °C. Only signals for m/z = 15 (NH+), 16 (NH2+), 17
(NH3+), 35 (35Cl+) and 37 (37Cl+) were detected (Table 1, Figure S10, Supporting Information).
From ex situ decomposition experiments of [Fe(NH3)6]Cl2
in flowing argon at T = 150 °C we observe the formation of
Fe(NH3)2Cl2 within 1 h. Therefore, the first step in the TG
curve is related to the decomposition of [Fe(NH3)6]Cl2 to
Fe(NH3)2Cl2. However, for this decomposition step, a mass
loss of 29.6 % is expected. The difference in expected and observed mass loss can be explained by a partial release of ammonia from [Fe(NH3)6]Cl2 while flushing the DTA machine
with inert gas prior to the start of the measurement. A similar
behavior was previously observed for [Mn(NH3)6]Cl2.[45] The
residual two ammonia molecules per formula unit are released
in a two step process via Fe(NH3)Cl2, each related with an
expected mass loss of 7.4 %, as assumed previously.[44] In an
independent DTA/TG run we interrupted the decomposition
process precisely at the step of Fe(NH3)Cl2 formation. Subsequent powder X-ray diffraction revealed very small amounts
of remaining Fe(NH3)2Cl2 next to a large hump, indicating the
amorphous nature of the Fe(NH3)Cl2 formed (see below). As
shown in Table 1, the first two steps in the TG curve were
exclusively related to the release of ammonia. In the third and
fourth step liberation of ammonia and a chlorine containing
compound were observed by EGA. In the last step only a chlorine-containing compound was observed in the gas phase,
which was obviously related to the evaporation of FeCl2. The
endothermal signal at 660.6 °C represents the melting of residual FeCl2. This powder of FeCl2 in accordance with the prediction of Juza et al.[42,43] we believe to be very fine, because in
our thermal analysis experiments evaporation temperature and
the thermal signal related to melting of FeCl2 was decreased
by about 35 K and 15 K, respectively, through the prior treatment of ammonia compared with pristine FeCl2.
A DSC/TG experiment (Figure S11) in flowing ammonia
(heating rate: 5 K·min–1 to 445 °C, 3 h annealed, subsequent
cooling: rate, 5 K·min–1) revealed slightly different results
(Table 2) The first three signals in the DTA curves occur in
both measurements, but are clearly shifted to higher temperatures in ammonia atmosphere as expected due to the increased
Z. Anorg. Allg. Chem. 2014, 1265–1274

T /°C





16, 17
16, 17, 35, 37
16, 17, 35, 37

ammonia partial pressure. The first step in the TG curve is
clearly related to the release of four molecules ammonia. The
following steps are mainly generated from the release of the
further two ammonia molecules, but must be accompanied
with loss of some chlorine containing compound, because
mass loss is too high to be explained only by emission of ammonia. We think that high filling grade of the crucible and the
liberation of large amounts of ammonia gas in combination
with the low bulk density of the formed ammoniates lead to a
loss of ammoniate material and therefore is responsible for the
difference in observed and expected mass loss. For the last two
steps we assume exclusively the release of ammonium chloride
and the beginning formation of ε-Fe3N1+x. Earlier a noticeable
sublimation of ammonium chloride above 200 °C was described,[46] which is in good agreement with our observations.
Powder X-ray diffraction data collected afterwards revealed
the formation of ε-Fe3N1.4(1) according to unit cell expansion
and some traces of Fe3O4, which were assumed to be related
with a small amount of residual humidity in the measurement
Table 2. Data of TG/DSC analysis of [Fe(NH3)6]Cl2 in flowing ammonia. The respective onset temperatures are given.
T /°C
445 (isothermal)

Δmobs ./ % Δmexp. / % Obtained

T /°C






The ammoniates [Fe(NH3)6]Cl2 and Fe(NH3)2Cl2 were additionally characterized by infrared spectroscopy (Figure S12)
in an inert gas atmosphere. The obtained results agree
well (Table 3 and Table 4) with values of [Fe(NH3)6]Br2,
[Ni(NH3)6]Cl2, and Ni(NH3)2Cl2 from the literature.[47–49]
In an in situ neutron diffraction experiment FeCl2 was reacted with 130 sccm of flowing ammonia. While increasing
the temperature only, the formation of [Fe(NH3)6]Cl2 and
Fe(NH3)2Cl2 besides FeCl2 was observed (Figure 13). Between 89 °C and 128 °C some intermediate amount of
[Fe(NH3)6]Cl2 is detected, however, decomposition of
[Fe(NH3)6]Cl2 to Fe(NH3)2Cl2 already occurs. At 252 °C the
maximum amount of Fe(NH3)2Cl2 [73(2) wt %] is present
(Figure S13). On further increase of the temperature the inten-

© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim



R. Niewa et al.

Table 3. Wave numbers ν˜ /cm–1 of NH3 vibrational modes of
[Fe(NH3)6]Cl2 and literature data.


[Fe(NH3)6]Br2[47] [Ni(NH3)6]Cl2[48]






Table 4. Wave numbers ν˜ /cm–1 of NH3 vibrational modes of
Fe(NH3)2Cl2 and literature data.
Vibration mode






sities of the reflections vanish to completely form an amorphous product above 277 °C. We assume the formation of
Fe(NH3)Cl2 as observed in TA, because the sample still contained hydrogen, which can be seen from the high background
due to the large incoherent scattering of hydrogen.[12] At
416 °C the sample is completely amorphous and all ammonia
is released, followed by crystallization of FeCl2 above 437 °C.
In contrast to the TA measurement in ammonia and ex situ
preparation experiments no indications for the formation of
ε-Fe3N1+x can be found, even after a rise of temperature to
500 °C (Figures S13–S16 and Table S5, Supporting Information). Apparently, the formation of ε-Fe3N1+x in the in situ
neutron diffraction experiment is suppressed, because of the
large sample volume (ca. 10 cm3) hindering a fast and homogeneous exchange between the solid and the gas atmosphere.
Especially, the removal of the by-product NH4Cl is believed
to be strongly hindered in this experimental setup. The NH4Cl
formation is the driving force for the nitride formation from a
thermodynamic point of view.[50] In general, the phase transition temperatures observed in DSC/TG and neutron diffraction
experiments in ammonia are in good agreement.
Bremm and Meyer carried out thermal analysis and temperature dependent in situ X-ray diffraction on the reaction of
FeCl2 with flowing ammonia.[44] A spontaneous exothermal
reaction related to the formation of microcrystalline
[Fe(NH3)6]Cl2 was observed already at ambient temperature.
Upon heating (TA: ⬎ 115 °C, X-ray: ⬎ 225 °C) decomposition to Fe(NH3)2Cl2 was observed, followed by the formation
of Fe(NH3)Cl2 (TA: ⬎ 290 °C), FeCl2 (TA: ⬎ 350 °C), and εFe3N1+x (TA: ⬎ 360 °C) or direct conversion into ε-Fe3N1+x
(X-ray: ⬎ 350 °C), respectively. In the X-ray diffraction experiment decomposition of [Fe(NH3)6]Cl2 was shifted to a
higher temperature and the reaction sequence from diammoni1272


Figure 13. Temperature dependent (2.5 K·min–1 to 450 °C, 1 h,
2.5 K·min–1 to 500 °C, 0.5 h) neutron diffraction data[36] of the reaction of FeCl2 (1) in flowing ammonia [λ = 186.690(7) pm]. The formation of [Fe(NH3)6]Cl2 (2) (89 °C ⱕ T ⱕ 128 °C) and Fe(NH3)2Cl2 (3)
(89 °C ⱕ T ⱕ 277 °C) is observed. The sample is amorphous (4) between 416 °C and 437 °C, above crystallization of FeCl2 occur (5).
Measurement at D20, intensity is given in false colors.

ate via monoammoniate and iron(II) chloride to ε-Fe3N1+x was
not observed. This was traced back to different gas flow conditions and a too fast reaction sequence to be resolved on the Xray time scale.[44] In the light of our in situ neutron diffraction
study, the formation of ε-Fe3N1+x occurs via an amorphous
intermediate with the composition Fe(NH3)Cl2.
In combination the results presented in reference[44] and
herein confirm the reaction path predicted by Juza et al.[42,43]
Several ammoniates are formed, namely [Fe(NH3)6]Cl2,
Fe(NH3)2Cl2, and amorphous Fe(NH3)Cl2, which were detected through combination of various analytical methods. In
situ neutron diffraction additionally showed the intermediate
formation of amorphous FeCl2 powder from the ammoniates
after the release of all ammonia.

3 Conclusions
Through combination of thermal analysis and in situ neutron
diffraction the influence of heating rate and gas flow conditions on the formation and decomposition of γ-FeNz, γ⬘-Fe4Ny,
ε-Fe3N1+x, and ζ-Fe2N was investigated. It was shown that all
those transformations are under kinetic control. In the rather
slow thermal decomposition of ζ-Fe2N nitrogen-poorer iron nitrides subsequently occur and the observed respective nitrogen
contents with exception of γ⬘-Fe4Ny agree well with the phase
diagram. For γ⬘-Fe4Ny an expansion of the phase width above
600 °C towards smaller nitrogen contents is observed. Additionally, for the synthesis of ε-phase from iron(II) halides
like FeCl2, in literature a reaction sequence involving ammoniates, was supposed. This assumption was confirmed by combining the results of in situ neutron diffraction and thermal
analysis. The involved ammoniates are [Fe(NH3)6]Cl2,
Fe(NH3)2Cl2, and amorphous Fe(NH3)Cl2.

4 Experimental Section
4.1 Synthesis of Iron Nitrides
ζ-Fe2N was produced by reaction of iron powder (99.9 %, 325 mesh,
Alfa Aesar) at 413 °C for 48 h in a gas flow of 60 sccm ammonia

© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Z. Anorg. Allg. Chem. 2014, 1265–1274

Formation and Decomposition of Iron Nitrides
(99.999 %, Linde).[33] Afterwards a controlled cooling to ambient temperature within 6 h was applied.[14] According to powder X-ray diffraction patterns the samples were single phase.

account. Therefore, for most measurements involving ε-phases as well
as γ⬘-phases the applied experimental conditions are above TC and no
magnetic contribution is expected.

The reaction of iron(II) chloride (dry, 99.99 %, Chempur) in an ammonia flow[44] of 130 sccm at 450 °C for 8 h was used to synthesize

We have recently presented different types of measurement cells suitable for in situ neutron diffraction with gas pressures up to 16.0 MPa
or in gas flow configuration. Sample heating is realized with a contactless laser heating system.[14] For reactions with light colored 3d
metal(II) halides such a laser heating system does not transfer heat
efficiently to reach the desired reaction temperature throughout the
whole sample, due to poor thermal conductivity of those compounds.
Therefore, we prefer for those experiments presented herein a high
temperature resistive furnace (HTF). A configuration similar to one
previously described was chosen. As sample container a bottom-closed
tube of silica glass (12 mm outer diameter) was used for in situ neutron
diffraction.[15,37,53,54] Additionally, the HTF provides the possibility of
using predissociated ammonia through the modified gas supply. This
is also advantageous for reactions of metals and ammonia, because the
reactivity of the ammonia is increased, which is easily realized by
preheating the ammonia before it comes into contact with the sample.[50] Temperature control was realized with a thermocouple type
K[55] inserted into the reaction tube and positioned off-beam close
above the sample. The reactions were performed in well defined gas
flows of ammonia and argon controlled with mass flow controllers
(MF1, MKS Instruments Germany).

[Fe(NH3)6]Cl2 was generated from FeCl2, while treated for 2 h at ambient temperature in flowing ammonia. For the synthesis of
Fe(NH3)2Cl2 as-prepared [Fe(NH3)6]Cl2 was heated to 175 °C in flowing ammonia for 3 h and afterwards cooled down to ambient temperature in dynamic vacuum (p ≈ 5 ⫻ 10–3 mbar).
Filling of the reactors was done in an argon filled glove box
[p(O2) ⬍ 0.1 ppm, MBraun] to avoid oxygen and water contamination
of the starting materials.

4.2 Powder X-ray Diffraction
Powder diffraction patterns were recorded with a STOE STADI P
equipped with a Mythen1K micro-strip detector in transmission geometry using Mo-Kα1 radiation (λ = 70.93 pm). Samples were fixed on
an adhesive tape.

4.3 In situ Powder Neutron Diffraction
In situ powder neutron diffraction data were collected with the high
flux two-axis powder diffractometer D20 (Institut Laue-Langevin,
Grenoble, France) in the range 0° ⱕ 2Θ ⱕ 151° (step size Δ2Θ =
0.1°) with a time resolution of Δtmin = 2 min. A high take-off-angle of
118° from the (115) atomic plane of the Ge-monochromator was chosen, resulting in a high average resolution of Δd/d ≈ 3 ⫻ 10–3, a high
neutron flux at the sample of about 1 ⫻ 107 s–1·cm–2 and a nominal
wavelength of λ = 188 pm.[13] The exact determination of the wavelength was performed using a silicon standard (NIST 640b) with corrected unit cell parameters[51] filled into a thin wall vanadium container
(outer diameter 6 mm). A wavelength of λ1 = 186.802(5) pm was determined for the first measurement of the decomposition of ζ-Fe2N.
For all other in situ experiments a wavelength of λ2 = 186.690(7)
pm was obtained. These values were used during all crystal structure
refinements from neutron diffraction data.
The nitrides γ⬘-Fe4Ny and ε-Fe3N1+x show ferromagnetic order and the
Curie temperatures TC [γ⬘: 480 °C ⱕ TC ⱕ 508 °C, ε: –273 °C ⱕ TC
ⱕ 294 °C] depend on the nitrogen content of the respective phase.[20]
Therefore, in neutron diffraction experiments a contribution of the
magnetic structure to the intensity of the Bragg reflections is expected.
However, all reflection intensities in Rietveld refinements were described by the pure nuclear structure model and no hints for such a
contribution were observed in our in situ collected neutron diffraction
data for the following reasons: According to the chosen setup of the
neutron diffractometer (high resolution) a rather large wavelength was
employed and in combination with the small unit cells of nitride phases
even low indexed reflections occur at rather high diffraction angles.
Therefore, it is difficult to observe any magnetic contribution, because
of the strong Q = sinΘ·λ–1 dependency of the magnetic moment. For
example first magnetic intensities of the ε-phase arise in a Q range,
where the magnetic from factor is decreased to about 0.6.[52] In combination with the low counting statistics according to good time resolution necessary to observe fast phase transitions the detection of ferromagnetic contribution is hampered. Especially, for the ε-phase the
strong coupling of TC to the nitrogen content has to be taken into
Z. Anorg. Allg. Chem. 2014, 1265–1274

All reaction processes observed by in situ neutron diffraction (Figure 1, Figure 4, Figure 8, and Figure 11) were visualized with the
LAMP program[36] available from ILL.

4.4 Rietveld Refinement
Rietveld refinements[56,57] of the crystal structures on powder neutron
diffraction data were performed using the program FULLPROF 2.k[58]
and pseudo-Voigt functions for the reflection profiles. The following
parameters were allowed to vary during refinements: The zero point
of the 2Θ scale, one scale factor per phase, three reflection widths
(Caglioti formula, U, V, and W), one mixing (η), two asymmetry
parameters, the lattice parameters, the atomic site parameters, and the
isotropic thermal displacement parameters (Biso). Additionally, an
angle dispersive correction (χ) of the mixing parameters was applied.
Due to the influence of the gas flow measurement cell the background
was treated by interpolation between chosen background points with
refinable heights. In case of ex situ collected neutron diffraction data
in vanadium cylinders the background was described by a polynomial.

4.5 Thermal Analysis
Thermal analysis experiments on the synthesis of ε-Fe3N1+x from
FeCl2 were applied with a Netzsch STA 449C thermal analyser with
corrosive gas equipment in ammonia/argon atmosphere (up to 50 vol %
NH3). The DTA/EGA measurement was carried out with a Netzsch
STA 409 thermal analyser with skimmer coupling and a Quadstar 422
(Pfeiffer Vacuum) mass spectrometer in helium (99.999 %, AirLiquide)
atmosphere. Before the start of the measurement the sample chamber
was flushed at least three times with inert gas. All TG measurements
were corrected for buoyancy effects.

4.6 Vibrational Spectroscopy
Vibrational spectroscopy measurements on [Fe(NH3)6]Cl2 and
Fe(NH3)2Cl2 were carried out in inert gas atmosphere of a glove box

© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim



Related documents

10 1002 zaac 201300676
j environ sci heal a 49 2014 753 762
5070 w09 er
untitled pdf document 5
5070 w06 ms 2

Related keywords