Synthesis of Metastable Co4N, Co3N, Co2N, and CoO0.74N0.24from a Single Azide Precursor and Intermediates in CoBr2Ammonolysis

Article abstract of DOI:10.1002/ejic.201600684

Metastable cobalt nitrides Co₄N, Co₃N, Co₂N, and cobalt oxide nitride CoO_0.74N_0.24were synthesized by ammonolysis of one single metastable azide precursor, [Co(NH₃)₅N₃]Cl₂, in high purity by controlled variation of the reaction conditions (temperature, duration, gas atmosphere, and flow rate). The alternative route towards cobalt nitrides by ammonolysis of CoBr₂proceeds via [Co(NH₃)₆]Br₂and Co(NH₃)₂Br₂to exclusively yield Co₃N. For Co₃N_1.1a complex magnetic behavior is observed: dominating ferromagnetic ordering at low temperatures and a small coercive field of 10 mT. Co₂N orders antiferromagnetically below 10 K. Electronic structure calculations [DFT, generalized-gradient approximation (GGA)+U] on CoO_0.75N_0.25reveal that the experimentally observed rock-salt structure is less stable than the sphalerite-type polymorph by about 30 kJ mol^–1and possesses an antiferromagnetically ordered ground state.

Full text of DOI:10.1002/ejic.201600684

DOI: 10.1002/ejic.201600684
Full Paper
Cobalt Nitrides
Synthesis of Metastable Co4N, Co3N, Co2N, and CoO0.74N0.24
from a Single Azide Precursor and Intermediates in CoBr2
Ammonolysis
Marc Widenmeyer,[a,b] Larysa Shlyk,[a] Nils Becker,[c] Richard Dronskowski,[c] Elke Meissner,[d]
and Rainer Niewa*[a]
Abstract: Metastable cobalt nitrides Co4N, Co3N, Co2N, and co- dominating ferromagnetic ordering at low temperatures and a
balt oxide nitride CoO0.74N0.24 were synthesized by ammonoly- small coercive field of 10 mT. Co2N orders antiferromagnetically
sis of one single metastable azide precursor, [Co(NH3)5N3]Cl2, in below 10 K. Electronic structure calculations [DFT, generalized-
high purity by controlled variation of the reaction conditions gradient approximation (GGA)+U] on CoO0.75N0.25 reveal that
(temperature, duration, gas atmosphere, and flow rate). The al- the experimentally observed rock-salt structure is less stable
ternative route towards cobalt nitrides by ammonolysis of CoBr2 than the sphalerite-type polymorph by about 30 kJ mol–1 and
proceeds via [Co(NH3)6]Br2 and Co(NH3)2Br2 to exclusively yield possesses an antiferromagnetically ordered ground state.
Co3N. For Co3N1.1 a complex magnetic behavior is observed:
Introduction
around 22 nm, a saturation magnetization of 123.6 emu g–1,
and a low coercive field of 7.4 × 10–3 T were obtained.[14] Co4N
is a powerful catalytic material for hydrodenitrogenation of
crude oil,[15] decomposition of hydrazine,[16] reduction of
NO,[17,18] and oxidation of CO.[19]
In the temperature range up to 1200 °C cobalt dissolves only a
very small amount of nitrogen (up to 8.5 × 10–5 atom-%).[1] First
reports about the formation of a metastable cobalt nitride with
a proposed composition of Co3N2 were published at the begin-
ning of the 20th century.[2–4] Nowadays, the existence of four
cobalt nitrides is confirmed, namely Co4N,[5] Co3N,[4] Co2N,[4]
and CoN.[6]
Co4N was first synthesized in 1960 as 30–50 nm thick layers
by the nitridation of thin cobalt layers in an ammonia atmos-
phere between 350 and 450 °C within 1–7 h. Above 570 °C this
Co4N film decomposes to cubic ꢀ-Co.[5] Co4N crystallizes in the
perovskite-type structure of γ′-Fe4N with an approximately
5.5 % smaller unit cell parameter.[7] At lower temperatures be-
tween 200 and 350 °C the formation of Co3N and a still not
fully characterized hexagonal nitride, CoxN, was observed.[5]
Nowadays, there are various approaches for the controlled syn-
thesis of Co4N thin layers.[8–13] However, the synthesis of bulk
samples of Co4N remains challenging. The first reported pro-
duction of bulk Co4N was achieved by an urea-assisted synthe-
sis route in 2013. Ferromagnetic nanoparticles with sizes
Metastable Co3N crystallizes in the ε-Fe3N type structure with
only a narrow homogeneity range. A stability range from 7.6 to
8.0 wt.-% nitrogen was derived from X-ray diffraction data.[4]
From caloric measurements a standard formation enthalpy be-
tween 4[20] and 8 kJ mol–1[21] was determined, indicating meta-
stability with respect to decomposition into the elements. Cal-
culations of the thermodynamic properties delivered a consist-
ent value of 5.6 kJ mol–1.[22] The traditional synthesis route is
based on the thermal decomposition of Co2N at about 280 °C.
However, this reaction is difficult to control, and often the reac-
tion product is not single-phase.[4] In the last years, Co3N
moved into the focus of research as a potential electrode mate-
rial in lithium-ion batteries.[23–25] Additionally, nitrogen-rich
samples Co3N1+x exhibit a weak antiferromagnetic coupling be-
low 11 K.[26]
Co2N is an isotype of ꢁ-Fe2N with no significant deviation
from the ideal 2:1 composition. Different synthesis pathways
were reported: (i) ammonolysis of thermally decomposed co-
balt(II) oxalate (cf. pyrophoric iron), (ii) ammonolysis of the re-
duction product of Co3O4 in flowing hydrogen, (iii) ammonoly-
sis of CoBr2, and (iv) ammonolysis of CoF2·NH4F·Co(OH)2·
2H2O.[4] In the first two synthesis approaches, an intermediate
formation of cobalt powder containing a large quantity of
stacking faults was observed, leading to an enhanced reactivity
during the following nitridation with flowing ammonia at ap-
proximately 380 °C.[27,28] In the ammonolysis of CoBr2 an inter-
mediate formation of ammoniates was observed;[4] however, no
[a] Institute of Inorganic Chemistry, University of Stuttgart,
Pfaffenwaldring 55, 70569 Stuttgart, Germany
E-mail: rainer.niewa@iac.uni-stuttgart.de
http://www.iac.uni-stuttgart.de/arbeitskreise/akniewa
[b] Institute for Material Science, University of Stuttgart,
Heisenbergstraße 3, 70569 Stuttgart, Germany
[c] Institute of Inorganic Chemistry, RWTH Aachen University,
Landoltweg 1, 52074 Aachen, Germany
[d] Fraunhofer Institute for Integrated Systems and Device Technology,
Schottkystraße 10, 91058 Erlangen, Germany
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/ejic.201600684.
Eur. J. Inorg. Chem. 2016, 4792–4801
4792
© 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
details on the composition or the structure of those ammoni- [Co(NH3)5N3]Cl2. Additionally, due to the reduced azide content,
ates are given. Similar to the reaction behavior of FeCl2,[29,30] at [Co(NH3)5N3]Cl2 was found to be a less explosive alternative
higher temperatures a reformation of a very fine (often amor- precursor compared to [Co(NH3)6](N3)3, [Co(NH3)5(N3)](N3)2, and
phous) powder of CoBr2 occurs, enabling an enhanced forma- [Co(NH3)4(N3)2]N3. Furthermore, the intermediates involved in
tion of Co2N.[4,31] The direct formation of Co2N starting from the ammonolysis of CoBr2 to form Co3N were investigated in
elemental cobalt seems possible if a low partial pressure of detail. The resulting reaction products were analyzed by Riet-
hydrogen is ensured.[32] Additionally, a high-pressure–high-tem- veld refinements of powder X-ray diffraction (PXRD) data, simul-
perature synthesis of Co2N starting from cobalt and nitrogen taneous thermal analysis in different gas atmospheres, and in-
was reported at 10 GPa and 1500 °C.[33]
frared vibrational spectroscopy. For Co2N and Co3N a characteri-
zation of the magnetic properties was carried out.
For thin layers of CoN, depending on the applied synthesis
method, either the rock-salt[34] or the sphalerite structure
type[35,36] was observed. For sphalerite-type CoN the thermal
decomposition behavior was investigated: Up to 231 °C CoN is
stable but decomposes above this temperature successively via
Co2N and Co3N to eventually be completely converted into α-
Co at 300 °C.[35] In the 1960s, different amino–azido complexes
of cobalt such as [Co(NH3)6](N3)3, [Co(NH3)5(N3)](N3)2, and
[Co(NH3)4(N3)2]N3 were used to synthesize powders of
CoN.[6,37–40] However, as a result of their high azide content,
those precursors tend to undergo an explosive thermal decom-
position between 100 and 140 °C depending on the particle
size.[39,40]
Previous calculations on CoO0.5N0.5 indicated the sphalerite
crystal structure to be significantly more stable than the rock-
salt structure because the latter structure type calls for stronger
antibonding Co–Co interactions;[42] nonetheless, this very stable
structure type is observed for CoO0.74N0.24. The electronic struc-
ture of this composition was therefore re-investigated by means
of DFT [generalized-gradient approximation (GGA)+U] including
both structure types and modeling ground states with diamag-
netic, ferromagnetic, and antiferromagnetic spin orderings.
Results and Discussion
Synthesis and Decomposition of [Co(NH3)5N3]Cl2
Oxide nitrides of cobalt CoO1–xNx were prepared in the form
of nanocrystalline particles and thin layers by thermal decom-
position of Co(CO)3(NO). The thin layers were grown by chemi-
cal vapor deposition and showed a strongly reduced electrical
resistivity, mainly due to the formation of mixed valence states
for Co by substituting N3– for O2–. X-ray diffraction data showed
the formation of CoO1–xNx in the sphalerite structure type with
x ≈ 0.4, as extracted from the unit cell expansion according to
a Vegard's law approximation. Chemical analysis in combination
with Rutherford backscattering spectrometry, in contrast, re-
sulted in an average composition of CoO0.9N0.5. However, a
large inhomogeneity was detected in the samples with compo-
sitions ranging from CoO0.78N0.12 to CoO1.3N0.67. Thermogravi-
metric (TG) measurements revealed thermal stability up to
390 °C.[41] DFT calculations on CoO0.5N0.5 delivered an energetic
preference of the sphalerite structure type over the rock-salt
structure type, in full agreement with the experimental results.
The calculated enthalpy indicates a phase transition from the
sphalerite to the rock-salt structure type around 30 GPa. From
supercell calculations, a preference for the formation of oxygen-
rich metastable oxide nitrides CoO1–xNx and no tendency for
anionic ordering was concluded. Additionally, all investigated
compositions showed vanishing magnetic moments.[42]
Single-phase metastable purple [Co(NH3)5N3]Cl2 was obtained
according to the reported procedure,[46] its structure being con-
firmed by PXRD and chemical analysis. In the following, the
term metastability for binary cobalt nitrides is used in the sense
that they tend to exothermally decompose to the elements by
providing sufficient (thermal) activation energy. The azide pre-
cursor exothermally decomposes under liberation of dinitrogen
gas. During a differential thermal analysis (DTA)/TG measure-
ment (see Figure 1) in an argon atmosphere, two not fully re-
solved steps in the TG curve with a total mass loss of 48.2 %
are observed above 144 and 233 °C. The mass loss represents
approximately the complete release of nitrogen and hydrogen
from the sample (Δmexp. = –49.4 %). In the DTA curve three
signals are detected, namely an exothermal signal at 155.5 °C
(onset) and two endothermal signals at 182.9 °C (onset) and
204.6 °C (onset). We conclude that the exothermal signal at
lowest temperature is caused by decomposition of azide anions.
Previously, a slow decomposition of [Co(NH3)5N3]Cl2 was ob-
served above 120 °C and an explosive decomposition at 195 °C
after approximately 50 s of annealing.[46] For amino–azido com-
plexes of cobalt with higher azide concentrations (see Introduc-
tion) explosive decomposition of the azide was detected be-
tween 110 and 140 °C.[37–40] In general, the results from our
DTA/TG measurement are in accordance with the decomposi-
tion behavior of [Co(NH3)5N3]Cl2 reported in the literature; how-
ever, sudden explosive behavior during our DTA/TG measure-
ments was prohibited by a low heating rate. The two endother-
mal signals observed for the decomposition of [Co(NH3)5N3]Cl2
can be related to the formation of firstly Co(NH3)2Cl2 and subse-
quently Co(NH3)Cl. In contrast to the reported decomposition
Information on ammoniates of CoBr2 is scarce in literature.
The first data about ammoniates of cobalt(II) halides CoX2 (X =
Cl, Br, I) were reported on the tensiometrically monitored de-
composition of the respective hexaammoniates [Co(NH3)6]X2. A
stepwise decomposition via diammoniate Co(NH3)2X2 and
monoammoniate Co(NH3)X2 (depending on the halide X) was
observed.[43] However, crystallographic data were only recorded
[44]
for [Co(NH3)6]Br2 and Co(NH3)Cl2.[45]
Herein, we present a new and easily controllable synthesis of [Co(NH3)6]Cl2, no clear steps in our TG curve of
approach for metastable cobalt nitrides Co4N, Co3N, and [Co(NH3)5N3]Cl2 were observed, but the obtained decomposi-
Co2N as well as the formation of an oxygen-rich cobalt oxide tion temperatures coincide well with the reported values 182
nitride CoO0.74N0.24 from
a
single metastable precursor and 230 °C,[43] respectively.
Eur. J. Inorg. Chem. 2016, 4792–4801
www.eurjic.org
4793 © 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Table 1. Unit cell parameters and residual values of Co4N, Co3N, Co2N, and
CoO0.74N0.24 as observed from Rietveld refinements (Mo-Kα1 radiation).
Parameter
Co4N
Co3N
Co2N
CoO0.74N0.24
Space group
a [pm]
b [pm]
Pm3m
374.84(1)
P6322
464.35(2)
Pbcn
Fm3m
425.51(3)
¯
¯
434.08(4)
569.68(5)
460.41(4)
113.85(2)
8.53
c [pm]
435.34(2)
81.292(5)
5.18
V [106 pm3]
Rp [%]
52.666(3)
7.46
77.045(9)
7.33
Rwp [%]
9.51
6.65
11.3
9.25
RBragg [%]
8.53
7.65
10.5
17.0
A stepwise decomposition of cobalt nitrides with higher
nitrogen content via Co4N to ꢀ-Co was reported previously.[5]
During a DTA/TG measurement of Co4N in an argon atmosphere
(see Figure S1) we observe the following reaction behavior:
Above 613 °C a mass loss of 6.1 % occurs. In parallel, an exother-
mal signal at 703.8 °C (onset) is observed as anticipated for a
metastable compound. The expected mass loss for the decom-
position of Co4N to ꢀ-Co is 5.6 %. Due to the metastable nature
of Co4N, it is clear that a strongly limited temperature range is
suitable for the synthesis of Co4N.
Figure 1. Simultaneous DTA (red) and TG (black) measurement of
[Co(NH3)5N3]Cl2 in Ar, heating rate 0.25 K min–1
.
Synthesis and Characterization of Co4N
Metastable Co4N can be synthesized from [Co(NH3)5N3]Cl2 in
flowing ammonia only in a small temperature interval between
400 and 450 °C to avoid formation of ꢀ-Co. During Rietveld
refinements of the collected PXRD data of a sample synthesized
at 425 °C within 8 h (see Figure 2 and Table 1) a very small
residue of ꢀ-Co was discovered. From our experimental results
we conclude that the reaction of [Co(NH3)5N3]Cl2 to Co4N is not
completed at 400 °C, and decomposition of Co4N to ꢀ-Co al-
ready occurs at 450 °C. This is supported by our experiments:
besides Co4N significant amounts of ꢀ-Co were found even after
12 h at 400 °C, and at 450 °C only small quantities of Co4N were
observed next to ꢀ-Co.
Synthesis and Characterization of Co3N
For the synthesis of Co3N three different strategies were stud-
ied: (i) ammonolysis of [Co(NH3)5N3]Cl2 in a pure ammonia at-
mosphere, (ii) ammonolysis of [Co(NH3)5N3]Cl2 in a diluted am-
monia atmosphere, and (iii) ammonolysis of CoBr2.
Ammonolysis of [Co(NH3)5N3]Cl2
During ammonolysis of [Co(NH3)5N3]Cl2 at 400 °C in flowing
ammonia within 8 h a sample of Co3N containing a significant
amount of ꢀ-Co was obtained. To achieve a single-phase sam-
ple of Co3N, a change in the gas atmosphere, the temperature
program, and a reduction of the amount of precursor was nec-
essary. Single-phase Co3N, according to X-ray diffraction, is ob-
tained in a diluted ammonia gas atmosphere (21 vol.-% NH3 in
Ar) by annealing [Co(NH3)5N3]Cl2 first at 350 °C (5 h) and then
at 400 °C (5 h).
Ammonolysis of CoBr2
In contrast to the pioneering investigations presented by Juza
and Sachße,[4] during the ammonolysis of CoBr2 no formation
of Co2N was observed upon application of a low ammonia gas
flow (100 ml min–1). Instead, the formation of hexagonal Co3N
occurs, similar to the ammonolysis of FeCl2 leading to ε-Fe3N.[30]
Directly after the contact between ammonia and CoBr2
a
swelling of the starting material is observed, which is accompa-
nied by a color change from green to blue and is caused
by the exothermal formation of the hexaammoniate
[Co(NH3)6]Br2.[43,44] The hexaammoniate can be recovered as a
single-phase microcrystalline powder. The results of the Riet-
veld refinements of the PXRD data are gathered in Figure S2
and Table 2 and are in good agreement with reported data.[44]
Figure 2. Rietveld refinements of the crystal structure of Co4N (Mo-Kα1 radia-
tion). The measured (red circles), calculated (black line), and difference (ob-
served–calculated, blue line) patterns as well as the Bragg positions of Co4N
(vertical bars) are shown. The asterisk marks the position of the most intense
reflection of ꢀ-Co. The most intense superstructure reflection due to primitive
Co4N as compared to simple fcc Co is clearly visible at 2Θ = 15.38°. Cobalt
occupies the Wyckoff sites 1a and 3c [Biso(Co) = 0.22(2) × 104 pm2], while
nitrogen is located at 1b [Biso(N) = 0.04(4) × 104 pm2].
Eur. J. Inorg. Chem. 2016, 4792–4801
www.eurjic.org
4794
© 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
The infrared vibrational spectrum can be found in Figure S3, ration temperature is decreased by about 30 K by the treatment
and the respective wavelengths of the vibrational modes are of CoBr2 with NH3 compared to that of pristine CoBr2,[50] indi-
presented in Table 4.
cating an increasing reactivity during the formation of Co3N in
the ammonolysis reaction.
Table 2. Structural data from Rietveld refinements for [Co(NH3)6]Br2 and
Co(NH3)2Br2 (Mo-Kα1 radiation). In [Co(NH3)6]Br2 cobalt is located at Wyckoff
site 4a, bromine at 8c, nitrogen at 24e, and hydrogen at 96k. In Co(NH3)2Br2
cobalt is located at 2a, bromine at 4h, nitrogen at 4g, and hydrogen at 8i.
Parameters fixed during the refinements are marked by *.
The diammoniate Co(NH3)2Br2 is synthesized in a controlled
way by the reaction of CoBr2 in flowing ammonia at 160 °C and
cooling under dynamic vacuum. The resulting powder is single-
phase according to Rietveld refinements (see Table 2 and
Figure 3) and IR vibrational spectroscopy (see Table 4 and
Figure S3).
Parameter
[Co(NH3)6]Br2
Co(NH3)2Br2
Space group
a [pm]
b [pm]
Fm3m
1039.2(1)
Pbam
¯
587.36(9)
1173.7(2)
392.58(5)
270.64(7)
4.6(3)
c [pm]
V [106 pm3]
1122.3(2)
2.0(2)
B
iso(Co) [104 pm2]
x(Br)
y(Br)
B
0.261(9)
0.3900(4)
1.3(2)
iso(Br) [104 pm2]
2.9(2)
x(N)
y(N)
B
0.340(6)
0.309(4)
0.124(3)
0.8(8)
iso(N) [104 pm2]
0.5*
x[H(1)]
y[H(1)]
z[H(1)]
0.077(3)
0.38(1)
0.11*
0.14* = z[H(2)]
8(1) = Biso[H(2)]
0.23*
0.18*
0.75*
2.59
3.30
0.265(3)
0.5*
B
iso[H(1)] [104 pm2]
x[H(2)]
y[H(2)]
P(H)
Rp [%]
Rwp [%]
RBragg [%]
0.75*
2.21
2.79
8.36
12.1
Figure 3. Rietveld refinements of the crystal structure of Co(NH3)2Br2 (Mo-Kα1
radiation). The measured (red circles), calculated (black line), and difference
(observed–calculated, blue line) patterns as well as the Bragg positions of
Co(NH3)2Br2 (vertical bars) are shown.
To investigate the thermal decomposition behavior of
[Co(NH3)6]Br2 a DTA/TG/evolved gas analysis (EGA) measure-
ment in a helium atmosphere up to 900 °C was carried out (see
Figure S4), (Table 3). Three steps are observed in the TG curve.
In the first step above 78 °C a mass loss of 16.2 % is detected.
This is related to the loss of four molecules of NH3 per formula
unit. However, for the release of four molecules of NH3 a mass
loss of 21.2 % is expected. Deviations to lower measured values
in such experiments are well known for the similar hexaammo-
Table 4. Wave numbers ν [cm–1] of the NH3 vibrational modes from IR spec-
˜
troscopy of [Co(NH3)6]Br2 and Co(NH3)2Br2 compared to reference values are
marked by #.[51,52]
#
#
Vibrational [Co(NH3)6]Br2
mode
[Co(NH3)6]Br2
Co(NH3)2Br2
Ni(NH3)2Br2
[49]
[29,30]
niates [Mn(NH3)6]Cl2
and [Fe(NH3)6]Cl2
due to their low
νas (N–H)
νs (N–H)
2δas (HNH) 3191
δas (HNH)
δs (HNH)
δs (HNH)
δs (HNH)
ρ (NH3)
3341
3250
3343
3230
3156
1600
3310
3233
3147
1590
1406
n/s
1244
642
591
3347
3264
3171
1615
1405
1258
1247
674
decomposition temperatures and thus partial ammonia loss
during sample preparation. The loss of four molecules of NH3
results in the formation of diammoniate Co(NH3)2Br2, which
crystallizes isotypically to Fe(NH3)2Br2.[29] Above 168.5 °C a sec-
ond mass loss of 12.6 % occurs; this is related to the loss of the
residual two NH3 molecules per formula unit and the reforma-
tion of CoBr2. Above 635 °C the evaporation of CoBr2 starts. For
the first two steps only signals for m/z = 15 (NH+), 16 (NH2+),
and 17 (NH3+) are observed in the mass spectra. In perfect ac-
cordance with the observation for FeCl2,[30] for CoBr2 the evapo-
1596
1170
643
1172
645
ρ (NH3)
ρ (NH3)
607
442
420
The reaction behavior changes significantly if the gas atmos-
phere is changed to ammonia. A DSC/TG experiment on
[Co(NH3)6]Br2 (see Figure S5) was carried out in 21 vol.-% NH3
in Ar up to 445 °C. After 3 h the reaction mixture was cooled
to ambient temperature. A PXRD pattern recorded afterwards
shows the formation of single-phase Co3N (see Figure 4 and
Table 5). In general, the same decomposition sequence is ob-
served for [Co(NH3)6]Br2 in an inert atmosphere as well as for
[Fe(NH3)6]Cl2.[30] Due to the higher ammonia partial pressure in
Table 3. DTA/TG/EGA measurement of [Co(NH3)6]Br2 in He. The respective
onset temperatures are given.
TG
EGA
T [°C]
Δmobs. [%] Δmexp. [%]
Compound
m/z
78
168.5
635
–16.2
–12.6
–62.6
–21.23
–10.61
n/s
Co(NH3)2Br2
CoBr2
evaporation
15, 16, 17
15, 16, 17
n/s
Eur. J. Inorg. Chem. 2016, 4792–4801
www.eurjic.org
4795
© 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
the differential scanning calorimetry (DSC)/TG experiment com- complete nitrogen loss) starts above 419 °C, which is 60 K lower
pared to that in the DTA/TG/EGA experiment, the respective than the decomposition temperature for ε-Fe3N1.37(2).
[54] In gen-
decomposition temperatures are shifted to higher values. The eral, a decrease in stability with an increasing number of 3d
TG curve shows two steps above 160 °C with a mass loss of electrons is observed for 3d transition metal nitrides.[22] The
20.6 and 58.5 % above 243 °C. The first step is related to the observed mass loss indicates the presence of a nitrogen-rich
release of four molecules of NH3 per formula unit from Co3N1+x (x ≈ 0.1). Additionally, in the DTA curve an exothermal
[Co(NH3)6]Br2 with formation of Co(NH3)2Br2. The latter step is signal is observed at 593.5 °C (onset), revealing the metastable
caused by the decomposition of Co(NH3)2Br2 and the starting nature of Co3N. Co3N shows only limited stability towards oxid-
formation of Co3N accompanied by a release of ammonium ation in a DTA/TG experiment in synthetic air. The formation of
bromide. For the reaction of [Co(NH3)6]Br2 to Co3N a mass loss Co3O4 starts slowly above 163 °C and is remarkably enhanced
of 80.2 % is expected, compared to the experimental 79.1 %. In above 300 °C. In parallel at 389 °C (onset) an exothermal signal
the DSC curve two endothermal signals at 178.9 °C (onset) and is detected in the DTA curve.
268.4 °C (onset) are obtained. In contrast to the reported tensio-
[43]
metric decomposition measurements on [Co(NH3)6]Br2
and the thermal decomposition of [Fe(NH3)6]Cl2
[30]
Magnetic Characterization of Co3N
and
Cu(NH3)2F2,[53] we observed no formation of a monoammoni-
ate; however, for [Fe(NH3)6]Br2 also no formation of a mono-
ammoniate was found.[29] A synthesis of Co3N starting from
Co(NH3)2Br2 is equally possible.
The variation of the magnetic susceptibility of Co3N with tem-
perature indicates two major anomalies: a pronounced maxi-
mum at T = 27 K and a broad shoulder in the vicinity of T ≈
65 K (see Figure 5). To identify the nature of these transitions
we have performed magnetization measurements at T = 5 K
and T = 40 K (see Figure 6). By collecting a hysteresis loop at
Figure 4. Rietveld refinements of the crystal structure of Co3N (Mo-Kα1 radia-
tion). The measured (red circles), calculated (black line), and difference (ob-
served – calculated, blue line) patterns as well as the Bragg positions of Co3N
(vertical bars) are shown.
Figure 5. Temperature dependence of (a) the magnetic susceptibility and (b)
the reciprocal susceptibility of Co3N. The solid lines are the fits of the data
to a Curie–Weiss relation.
Table 5. Structural data from Rietveld refinements of the crystal structures of
Co3N and Co2N (Mo-Kα1 radiation). In Co3N cobalt is located at Wyckoff site
6g and nitrogen at 2c, and in Co2N cobalt is at 2b and nitrogen at 8d and 4c.
Parameters fixed during the refinements are marked by *.
Parameter
Co3N
Co2N
x(Co)
y(Co)
z(Co)
0.336(1)
0.253(2)
0.1326(7)
0.0746(3)
0.29(5)
B
iso(Co) [104 pm2]
0.37(2)
y(N)
B
P[N(1)]
P[N(2)]
0.340(6)
0.3(6)
1.00(6)
iso[N(1)] [104 pm2]
0.5* = Biso[N(2)]
1.00*
0.031*
Thermal Behavior of Co3N
The thermal decomposition of Co3N in an argon atmosphere
(see Figure S6) with a mass loss of 8.5 % (expected 7.3 % for
Figure 6. (a) Field-dependent magnetization of Co3N at T = 5 K and T = 40 K.
(b) Low-field magnetization curves.
Eur. J. Inorg. Chem. 2016, 4792–4801
www.eurjic.org
4796
© 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
5 K, a ferromagnetic behavior with a coercive field of Hc ≈ 0.01 T curve is only linear in the small temperature range from 250 to
was observed (see inset in Figure 6). At 40 K a weak ferromag- 300 K. A fitting of the data according to a Curie–Weiss law in
netic component with a coercive field of Hc ≈ 0.0015 T superim- this temperature range results in a Curie–Weiss temperature of
posed to the paramagnetic phase was still evident.
Θ = –853 K and an effective magnetic moment μeff = 4.9 μB,
which is close to the spin-only value for Co3+(HS). It is interest-
ing to note that some cobaltates such as La2–xSrxCoO4 show a
similar deviation from Curie–Weiss behavior.[56] An increase of
susceptibility in this material at high temperatures was attrib-
uted to a thermally induced spin-state transition of Co3+. This
scenario might be also applied to Co3N, but bulk magnetic
measurements cannot address this question. It is worth to note
that a similar magnetic behavior was already observed between
200 and 300 K for the Co3N1.55 sample mentioned before, which
follows a Curie–Weiss law only in this temperature interval. A
large negative Curie–Weiss temperature of Θ = –500 K and an
effective magnetic moment equal to 3 0.3 μB were ob-
served.[26] To clarify the nature of the spin states of Co ions,
other methods like X-ray absorption spectroscopy or neutron
scattering have to be utilized. However, the present sample is
still inhomogeneous, comprising at least two phases. This result
underlines the necessity of further experiments to verify the
ground state of Co3N.
This behavior suggests the presence of at least two contribu-
tions in the sample: a ferromagnetic and a paramagnetic one.
Indeed, at low temperatures, a paramagnetic tail is visible due
to a small fraction of unpaired Co spins. These arise most proba-
bly from a very small fraction of elemental Co impurity not
detectable by PXRD and behave as localized free spins. This
produces a magnetic susceptibility following a Curie–Weiss law.
A fit of the data below 14 K results in a Curie–Weiss tempera-
ture Θ = –2.6 K and an effective magnetic moment μeff = 1.2 μB.
The effective magnetic moment of Co is clearly reduced as com-
pared to the spin-only values typically observed for low-spin
(LS) or high-spin (HS) states of Co: μeff(Co2+,LS) = 3.87 μB,
μ
μ
eff(Co2+,HS) = 1.87 μB, μeff(Co3+,LS) = 0 μB, μeff(Co3+,HS) = 4.9 μB,
eff(Co4+,LS) = 1.73 μB, μeff(Co4+,HS) = 5.82 μB. The reported
magnetic behavior of Co3N1+x in the literature is enigmatic. For
a sample identified as hexagonal Co3N1+x by PXRD, primarily a
paramagnetic to antiferromagnetic transition below 11 K was
reported. However, the results of chemical analysis delivered a
[26]
composition of Co3N1.55
,
which is more reason-
able to be described as Co2N1.03. This behavior strongly differs
Synthesis and Characterization of Co2N
from our observations and from the presented magnetic behav-
[55]
In contrast to the results of the pioneering work by Juza and
Sachße,[4] synthesis of Co2N from elemental cobalt as well as
from Co3O4 was unsuccessful during our studies. A successful
pathway to Co2N is found by ammonolysis of [Co(NH3)5N3]Cl2
at 450 °C (12 h) followed by annealing in ammonia at 350 °C
(18 h). According to the reported synthesis conditions for Co4N
and Co3N, it seems reasonable to assume the following reaction
sequence: In the first nitridation step at 450 °C Co4N is pro-
duced possibly in a mixture with Co3N. Those reaction products
are completely nitridated to Co2N in the second step at lower
temperature (see Figure 7 and Table 2).
ior of mixtures of Co4+yN and Co3N1+x
.
Nevertheless, the re-
sults presented previously[26] fit our observation on Co2N (see
below) exactly. Magnetization measurements were performed
[55]
on different mixtures of Co4+yN and Co3N1+x
.
From the DFT
calculations, an energetic preference of a ferromagnetic order-
ing compared to a paramagnetic configuration was confirmed.
Similarities in the behavior of the saturation magnetization of
[55]
Co3N1+x
and ε-Fe3N1+x were observed.[57] With increasing
nitrogen content, represented by an increase in the phase frac-
tion of Co3N1+x compared to that of Co4+yN, a decrease in the
observed saturation magnetization was observed.[55] For ε-
Fe3N1+x a decrease in the Curie temperature TC and conse-
quently a reduction of the saturation magnetization was ob-
served with increasing nitrogen content. When x approaches
0.5, paramagnetic behavior down to 4 K was observed.[57]
The amount of unpaired spins has been estimated from the
experimental value of the Curie constant C according to Equa-
tion (1):
(1)
where kB represents the Boltzmann constant, N the number of
magnetic spins per unit volume, g the Landé g-factor (g = 2
was taken for the free spins), μB the Bohr magneton, and S the
spin quantum number. Since the spin state of Co was not obvi-
ous, we only estimate the approximate range of the paramag-
netic impurity fraction to be in the order of 1022 mol–1.
The inverse susceptibility of Co3N is shown in the inset of
Figure 5. A significant deviation from the Curie–Weiss law can
be observed at temperatures below approximately 250 K; the
Figure 7. Rietveld refinements of the crystal structure of Co2N (Mo-Kα1 radia-
tion). The measured (red circles), calculated (black line), and difference (ob-
served–calculated, blue line) patterns as well as the Bragg positions of Co2N
(vertical bars) are shown.
Eur. J. Inorg. Chem. 2016, 4792–4801
www.eurjic.org
4797
© 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Under the conditions of a DTA/TG measurement in argon oxygen-rich composition detected by chemical analysis is sup-
(see Figure S7), Co2N exhibits a mass loss of 8.5 % above 344 °C. ported by the evaluation of the diffraction data, since the meas-
An exothermal signal at 429.6 °C (onset) reveals the metastable ured unit cell parameter of the oxide nitride [a(CoO1–xNx) =
nature of Co2N. The expected mass loss for the decomposition 425.51(3) pm] is much closer to the value of rock-salt type CoO
of Co2N to ꢀ-Co is 10.7 %. The large difference between meas- [a(CoO) = 425.18(2) pm][47] than to that of CoN [a(CoN) =
ured and expected mass loss is due to the presence of about 427.86 pm].[48] Additionally, according to electronic structure
20 % ꢀ-Co in the particular investigated sample. In general, a calculations, the largest stabilization of CoO1–xNx was observed
decreasing thermal stability with increasing nitrogen content is for x ≈ 0.2,[42] and experimentally an oxide nitride with rock-
observed for the investigated metastable cobalt nitrides.
salt structure type with a very similar composition of
CoO0.71N0.27 was reported previously to form during the ammo-
nolysis of Co3O4 at 390 °C.[4]
Magnetic Characterization of Co2N
In the magnetization curve of Co2N (single-phase according to
PXRD) antiferromagnetic behavior was observed. The maxi-
mum, which equals the Néel temperature, was obtained at
TN = 10 K (see Figure 8) and perfectly fits the data previously
reported for Co3N1.55 (≡ Co2N1.03).[26] However, the mass suscep-
tibility we obtained at T = 295 K (ꢀ= 148 × 10–6 emu g–1)
deviates significantly from the values reported earlier (ꢀ =
23 8 × 10–6 emu g–1).[26] The collection of the field-dependent
magnetization curves at T = 5 K and T = 100 K reveal the pres-
ence of a small fraction of a ferromagnetic impurity, not detect-
able by PXRD, with a very small coercive field. This can be re-
lated to the presence of either Co or Co4N. Additionally, the
presence of a small fraction of unpaired Co spins, as observed
before in the case of Co3N, cannot be ruled out.
Figure 9. Rietveld refinements of the crystal structure of CoO0.74N0.24 (Mo-Kα1
radiation). The measured (red circles), calculated (black line), and difference
(observed–calculated, blue line) patterns as well as the Bragg positions of
CoO0.74N0.24 (vertical bars) are shown. The composition was fixed during the
refinements to the value observed by chemical analysis. The inset shows the
simulated intensities of CoO0.74N0.24 in sphalerite (red) and rock-salt (black)
structure type.
New DFT calculations were performed with the explicitly cor-
related GGA+U method to shed some light on the supposed
discrepancy between experimental observation (rock-salt type)
and previous calculations (sphalerite type suggested). Because
the rock-salt and sphalerite unit cells both contain four formula
units, this allows for an easy theoretical modeling of the experi-
mental composition CoO0.75N0.25. After relaxation, the cell pa-
rameters of the optimized structures were varied by scaling
them, and the resulting energy versus volume data were fitted
to the Birch–Murnaghan equation of state (Figure 10). As ex-
pected, the most stable polymorph is the sphalerite structure
(blue), independent of the magnetic state; among them, how-
Figure 8. Temperature-dependent magnetic susceptibility of Co2N at H =
0.05 T.
Synthesis and Characterization of CoO1–xNx
During experiments aiming for the synthesis of CoN ever, the antiferromagnetic ordering (AFM) is lowest, slightly
from [Co(NH3)5N3]Cl2 a cobalt oxide nitride was obtained by higher in energy is the ferromagnetic (FM) ordering, and then
thermal decomposition of the precursor at 400 °C under the diamagnetic one comes far above. In contrast, the rock-salt
dynamic vacuum. From chemical analysis, a composition of structure (red) is always higher in energy but with the same
CoO0.74 0.03N0.24
was determined. Rietveld refinements of order of the magnetic states: AFM, then FM (somewhat unsta-
0.11
the collected PXRD data revealed the presence of a rock-salt ble), then diamagnetic. As a result, the AFM rock-salt polymorph
structure type cobalt oxide nitride (see Figure 9) in contrast to is less stable than the AFM sphalerite type by more than
earlier experimental results[41] and theoretical predictions.[42] 30 kJ mol–1. Therefore, the previous calculations are fully cor-
The refined isotropic displacement parameters are Biso(Co) = roborated. It is also worth mentioning that the experimental
0.59(7)×104 pm2 and Biso(O) = Biso(N) = 0.8(2) × 104 pm2. The cell volume (dashed line) of 19.26 Å3 per formula unit agrees
Eur. J. Inorg. Chem. 2016, 4792–4801
www.eurjic.org
4798
© 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
perfectly with the calculated AFM-ordered rock-salt structure.
An alternative synthesis of Co3N starts from CoBr2 as precur-
Regarding the perfect coincidence, there is good reason to as- sor. In the first step, at temperatures lower than 200 °C, the blue
sert that the experimentally found structure is a metastable and hexaammoniate [Co(NH3)6]Br2 forms in ammonia. With increas-
antiferromagnetically ordered polymorph of CoO0.75N0.25
.
ing temperature the formation of the diammonate Co(NH3)2Br2
occurs prior to a conversion to Co3N at 450 °C.
Experimental Section
Syntheses
All manipulations of starting materials and samples were carried
out in an argon-filled glove box [MBraun, p(O2) < 0.1 ppm].
Metastable Co4N was synthesized from metastable [Co(NH3)5N3]Cl2
within 8 h at 425 °C in a controlled ammonia (>99.999 %, Linde)
flow of 60 mL min–1
.
Metastable Co3N was synthesized from two different precursors,
namely [Co(NH3)5N3]Cl2 and CoBr2 (97 %, Alfa Aesar). [Co(NH3)5N3]-
Cl2 was heated in a gas atmosphere consisting of 21 vol.-% NH3 in
argon with a heating rate of 2 K min–1 to 350 °C. After 5 h, the
temperature was increased to 400 °C with 2 K min–1. Cooling to
ambient conditions was done after 5 h at 400 °C. Starting from
green CoBr2 at ambient temperature in an ammonia gas flow of
150 mL min–1, within 2 h the blue hexaammoniate [Co(NH3)6]Br2 is
formed. Heating [Co(NH3)6]Br2 at 160 °C in flowing ammonia results
in the formation of the diammoniate Co(NH3)2Br2 within 2 h. Cool-
ing was carried out under dynamic vacuum (pmin ≈ 2 × 10–3 mbar).
Single-phase Co3N was synthesized by heating CoBr2 in an ammo-
nia gas flow of 150 mL min–1 at 445 °C for 5 h.
Figure 10. Density functional energy versus volume phase diagram of
CoO0.75N0.25. The rock-salt [NaCl] (in red) and the sphalerite [ZnS] polymorphs
(in blue) are each represented by nonmagnetic, ferromagnetic (FM), and anti-
ferromagnetic (AFM) spin orderings. The experimental volume of 19.26 Å3
per formula unit has been indicated by a dashed black line and matches the
calculated volume for [NaCl] AFM-type CoO0.75N0.25
.
Metastable Co2N was obtained by ammonolysis of [Co(NH3)5N3]Cl2
at 450 °C for 12 h in a gas flow of 60 mL min–1 NH3. Afterwards,
the reaction mixture was cooled to 350 °C, and the nitridation was
continued for additional 18 h.
Conclusions
A new approach for the synthesis of metastable cobalt nitrides
Co4N, Co3N, and Co2N starting from a single metastable azide
precursor [Co(NH3)5N3]Cl2 was established. The respective
nitrides can be obtained in a controlled way by varying the
The thermal decomposition of metastable [Co(NH3)5N3]Cl2 at 400 °C
for 24 h under dynamic vacuum (pmin ≈ 2 × 10–2 mbar) resulted in
the formation of CoO0.74 0.03N0.24
.
0.11
temperature program and/or the gas atmosphere. The metasta- [Co(NH3)5N3]Cl2 was synthesized as described earlier.[46] CoCl2·6H2O
(reinst, Merck) was dissolved in a small quantity of deionized water.
To this solution was added a second solution consisting of NaN3
(>99 %, Merck), NH4Cl (suprapur, Merck), and NH3 (aq., 25 wt.-%,
chem. pure, Merck) under vigorous stirring at 60 °C. After 2 h, the
reaction mixture was cooled with ice water to ensure complete pre-
cipitation. The precipitate was collected and washed three times
with diluted hydrochloric acid and ethanol (100 %). For recrystalliza-
tion the precipitate was dissolved in deionized water and after-
wards precipitated by the careful addition of concentrated hydro-
chloric acid. The collected precipitate was washed with ethanol and
ble nature of the cobalt nitrides was confirmed by the detection
of an exothermal decomposition during DTA measurements in
Ar. Measurements of the magnetic properties of our Co3N1.1
revealed a clear difference from the magnetic behavior previ-
ously observed on a sample with higher nitrogen content
(Co3N1.55). For Co3N1.1 a ferromagnetic behavior with a small
coercive field superimposed to a paramagnetic impurity phase
was observed. Former DFT calculations presented in the litera-
ture reveal an energetic preference of a ferromagnetic ordering
for Co3N1+x over a paramagnetic assembly. For Co3N1.55 an anti- dried at 50 °C for 4 d. The detected weight fractions of nitrogen
and hydrogen by chemical analysis were w(N) = 43.6 0.2 wt.-%
and w(H) = 5.89 0.05 wt.-%. The expected values are wexp.(N) =
43.55 wt.-% and wexp.(H) = 5.88 wt.-%.
ferromagnetic behavior was observed experimentally below
11 K in the literature. However, this resembles exactly the mag-
netic behavior we have obtained for Co2N. Regarding the fact
that Co3N1.55 can also be described as Co2N1.03 this observation
becomes plausible.
Additionally, the decomposition of the precursor in vacuo
allows access to a rock-salt type cobalt oxide nitride with a
composition of CoO0.74N0.24. DFT calculations revealed that this
oxide nitride is a metastable polymorph with an antiferromag-
netic ordering. Despite the fact that the likewise AFM sphalerite
polymorph is preferred by more than 30 kJ mol–1, the rock-salt
polymorph forms under experimental conditions, possibly be-
cause of smaller activation barriers for crystallization.
Thermal Analysis: Thermal analysis experiments in ammonia/
argon mixtures were carried out with a Netzsch STA449C thermal
analyzer equipped with a corrosive gas kit. The DTA/TG/EGA meas-
urement was performed with a Netzsch thermal analyzer STA409CD
with a skimmer coupling and a Quadstar 422 (Pfeiffer Vacuum) mass
spectrometer in a helium atmosphere (>99.999 %, AirLiquide). Fur-
ther DTA/TG measurements in an argon atmosphere (>99.999 %,
AirLiquide) were carried out with Netzsch thermal analyzers STA409
and STA449C. Before the start of the respective measurements the
sample chamber was flushed at least three times with inert gas. All
TG measurements were corrected for buoyancy effects.
Eur. J. Inorg. Chem. 2016, 4792–4801
www.eurjic.org
4799
© 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
[1] A. Sieverts, H. Hagen, Z. Phys. Chem. Abt. A 1934, 169, 237–240.
[2] A. C. Vournasos, Compt. Rend. 1919, 168, 889–891.
[3] F. W. Bergstrom, J. Am. Chem. Soc. 1924, 46, 2631–2636.
[4] R. Juza, W. Sachße, Z. Anorg. Allg. Chem. 1945, 253, 95–108.
[5] N. Terao, Mem. Sci. Rev. Metall. 1960, 57, 96–100.
[6] T. B. Joyner, F. H. Verhoek, J. Am. Chem. Soc. 1961, 83, 1069–1072.
[7] S. F. Matar, A. Houari, M. A. Belkhir, Phys. Rev. B 2007, 75, 245109.
[8] K. Oda, T. Yoshio, K. Oda, J. Mater. Sci. 1987, 22, 2729–2733.
[9] J.-S. Fang, L.-C. Yang, C.-S. Hsu, G.-S. Chen, Y.-W. Lin, G.-S. Chen, J. Vac.
Sci. Technol. A 2004, 22, 698–704.
[10] H. Jia, X. Wang, W. Zhao, Y. Cahen, S. Feng, Mater. Sci. Eng. B 2008, 150,
121–124.
[11] X. Wang, H. Jia, W. T. Zheng, Y. Chen, S. Feng, Thin Solid Films 2009, 517,
4419–4424.
[12] K. K. Shih, J. Karasinski, J. Appl. Phys. 1993, 73, 8377–8379.
[13] H. B. Bhandari, J. Yang, H. Kim, Y. Lin, R. G. Gordon, Q. M. Wang, J.-S. M.
Lehn, H. Li, D. Shenai, J. Solid State Sci. Technol. 2012, 1, N79–N84.
[14] J. Theerthagiri, S. B. Dalavi, M. M. Raja, R. N. Panda, Mater. Res. Bull. 2013,
48, 4444–4448.
[15] I. K. Milad, K. J. Smith, P. C. Wong, K. A. R. Mitchell, Catal. Lett. 1998, 52,
113–119.
[16] H. Cheng, Y. Huang, A. Wang, X. Wang, T. Zhang, Top. Catal. 2009, 52,
1535–1540.
[17] Z. Yao, A. Zhang, Y. Li, Y. Zhang, X. Cheng, C. Shi, J. Alloys Compd. 2008,
464, 488–496.
[18] Z. Yao, A. Zhu, C. T. Au, C. Shi, Catal. Lett. 2009, 130, 63–71.
[19] Z. Yao, X. Zhang, F. Peng, H. Yu, H. Wang, J. Yang, Int. J. Hydrogen Energy
2011, 36, 1955–1959.
[20] H. Hahn, A. Konrad, Z. Anorg. Allg. Chem. 1951, 264, 181–183.
[21] F. R. De Boer, R. Boom, W. C. M. Mattens, A. R. Miedema, A. K. Niessen,
Cohesion in Metals, North-Holland, Amsterdam, 1988.
[22] A. F. Guillermet, S. Jonsson, Z. Metallkd. 1992, 83, 21–31.
[23] Z.-W. Fu, Y. Wang, X.-L. Yue, S.-L. Zhao, Q.-Z. Qin, J. Phys. Chem. B 2004,
108, 2236–2244.
[24] J. Ma, L. Yu, Z.-W. Fu, Electrochim. Acta 2006, 51, 4802–4814.
[25] S. Dong, X. Chen, X. Zhang, G. Cui, Coord. Chem. Rev. 2013, 257, 1946–
1956.
[26] K.-H. Mader, F. Thieme, A. Knappwost, Z. Anorg. Allg. Chem. 1969, 366,
274–279.
[27] H. Nowotny, R. Juza, Z. Anorg. Allg. Chem. 1945, 253, 109–112.
[28] R. Puza, Angew. Chem. 1945, 58, 25.
[29] S. Bremm, G. Meyer, Z. Anorg. Allg. Chem. 2003, 629, 1875–1880.
[30] M. Widenmeyer, T. C. Hansen, E. Meissner, R. Niewa, Z. Anorg. Allg. Chem.
2014, 640, 1265–1274.
[31] M. Widenmeyer, Dissertation, Univ. Stuttgart, 2014.
[32] J. Clarke, K. H. Jack, Chem. Ind. 1951, 46, 1004–1005.
[33] M. Hasegawa, T. Yagi, J. Alloys Compd. 2005, 403, 131–142.
[34] X. Liu, H. Lu, M. He, K. Jin, G. Yang, H. Ni, K. Zhao, J. Alloys Compd. 2014,
582, 75–78.
[35] K. Suzuki, T. Kaneko, H. Yoshida, M. Morita, H. Fujimori, J. Alloys Compd.
1995, 224, 232–236.
[36] B. Das, M. V. Reddy, P. Malar, T. Osipowicz, G. V. Subba Rao, B. V. R.
Chowdari, Solid State Ionics 2009, 180, 1061–1068.
[37] T. B. Joyner, D. S. Stewart, L. A. Burkhardt, Anal. Chem. 1958, 30, 194–
196.
[38] T. B. Joyner, F. H. Verhoek, Inorg. Chem. 1963, 2, 334–336.
[39] T. B. Joyner, J. Phys. Chem. 1970, 74, 1552–1557.
[40] T. B. Joyner, J. Phys. Chem. 1970, 74, 1563–1567.
[41] N. R. M. Crawford, J. S. Knutsen, K.-A. Yang, G. Hugstad, S. McKernan, F. B.
McCormick, W. L. Gladfelter, Chem. Vap. Deposition 1998, 4, 181–183.
[42] M.-W. Lumey, R. Dronskowski, Adv. Funct. Mater. 2004, 14, 371–376.
[43] W. Biltz, B. Fetkenheuer, Z. Anorg. Allg. Chem. 1914, 89, 97–133.
[44] R. Eßmann, G. Kreiner, A. Niemann, D. Rechenbach, A. Schmieding, T.
Sichla, U. Zachwieja, H. Jacobs, Z. Anorg. Allg. Chem. 1996, 622, 1161–
1166.
Powder Diffraction and Rietveld Refinements: PXRD patterns of
the reaction products were recorded with a STOE STADI P equipped
with a Mythen1K micro-strip detector in transmission geometry
with Mo-Kα1 radiation (λ = 70.93 pm).
Rietveld refinements[58,59] of the crystal structures on PXRD data
were carried out with use of the FULLPROF 2.k[60] program and
pseudo-Voigt functions to describe the reflection profile. Addition-
ally, the following parameters were allowed to vary: The zero point
of the 2Θ scale, one scale factor per phase, three reflections widths
(Caglioti formula. U, V, and W), two asymmetry parameters, one
mixing (η) parameter and its angle-dispersive correction (ꢀ), the
unit cell parameters, the atomic site parameters, and the isotropic
thermal displacement parameters (Biso). The background was
treated either by a polynomial or by interpolation between chosen
background points with refinable heights. The structural models
used for the Rietveld refinements of Co4N,[14] Co3N,[54] Co2N,[54] and
[47,48]
CoO0.74N0.24
are based on the indicated references.
Chemical Analysis: Quantitative analysis of the hydrogen and
nitrogen contents of [Co(NH3)5N3]Cl2 was carried out by the hot gas
extraction method with a LECO TC300/EF300. Quantitative analysis
of the oxygen and nitrogen contents of the cobalt oxide nitride was
carried out by the hot gas extraction method with a LECO ONH836.
Infrared Vibrational Spectroscopy: For collection of the vibra-
tional spectra of [Co(NH3)6]Br2 and Co(NH3)2Br2 a Thermo Electron
Scientific Instruments FTIR Nicolet iS5 in ATR mode installed inside
a glove box was used.
Magnetization Measurements: The samples were sealed under an
argon atmosphere into a polycarbonate capsule, which was glued
into a small polyvinylchloride tube. Magnetization measurements
were performed with a SQUID magnetometer (Quantum Design)
for Co2N and a MPMS7 magnetometer (Quantum Design) for Co3N
in the temperature range of 2–300 K and magnetic fields up to 6 T.
Electronic Structure Calculations: The unexpected rock-salt struc-
ture type of cobalt oxide nitride as well as its sphalerite counterpart
were investigated with periodic DFT as implemented in the Vienna
ab initio simulation package (VASP).[61] Projector-augmented waves
(PAW)[62] were used, and the contributions of exchange and correla-
tion were treated within the GGA as described by Perdew, Burke,
and Ernzerhof.[63,64] To minimize the self-interaction error, the
GGA+U method according to Dudarev et al.[65] was used. An effec-
tive U of 3.3 eV[66,67] was used as it has previously shown to provide
good results for CoO. An energy-cutoff of 500 eV and an adapted
k-point sampling ensured well-converged structures. During the op-
timization process the atomic positions and the lattice parameters
were allowed to relax until the convergence criterion of 10–6 eV was
reached. The cell parameters of the optimized structures were then
scaled from 94 to 104 %, and the resulting energy versus volume
data were fitted to the Birch–Murnaghan equation of state.[68]
Acknowledgments
The authors thank Oliver Puntigam for the collection of vibra-
tional spectra data and Barbara Förtsch and Dr. Christoph Ney
for chemical analysis. M. W., N. B., R. D., and R. N. gratefully
acknowledge financial support by the Deutsche Forschungs-
gemeinschaft within the priority program 1415 (SPP 1415).
[45] A. Leineweber, H. Jacobs, P. Allensbach, F. Fauth, P. Fischer, Z. Anorg. Allg.
Chem. 2001, 627, 2063–2069.
Keywords: Cobalt · Nitrides · Magnetic properties ·
Metastable compounds · Density functional calculations
[46] M. Linhard, H. Flygare, Z. Anorg. Allg. Chem. 1950, 262, 328–343.
[47] D. Taylor, Trans. J. Br. Ceram. Soc. 1984, 83, 5–9.
[48] O. Schmitz-Dumont, N. Kron, Angew. Chem. 1955, 67, 231.
Eur. J. Inorg. Chem. 2016, 4792–4801
www.eurjic.org
4800
© 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
[49] H. Reardon, J. M. Hanlon, M. Grant, I. Fullbrook, D. H. Gregory, Crystals
2012, 2, 193–212.
[50] Material safety data sheet, Cobalt(II) bromide, anhydrous, Alfa Aesar
GmbH & Co KG, Karlsruhe, 05.07.2012.
[51] R. Eßmann, J. Mol. Struct. 1995, 351, 91–94.
[52] R. Eßmann, J. Mol. Struct. 1995, 351, 87–90.
[53] P. Woidy, A. J. Karttunen, M. Widenmeyer, R. Niewa, F. Kraus, Chem. Eur.
J. 2015, 21, 3290–3303.
[54] M. Widenmeyer, R. Niewa, T. C. Hansen, H. Kohlmann, Z. Anorg. Allg.
Chem. 2013, 639, 285–295.
[59] H. M. Rietveld, J. Appl. Crystallogr. 1969, 2, 65–71.
[60] J. Rodriguez-Carvajal, FULLPROF 2.k, version 5.30, March2012-ILL-JRC,
2012.
[61] G. Kresse, J. Furthmüller, Comput. Mater. Sci. 1996, 6, 15–50.
[62] P. E. Blöchl, Phys. Rev. B 1994, 50, 17953–17979.
[63] J. P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 1997, 78, 1396.
[64] J. P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 1996, 77, 3865–3868.
[65] S. L. Dudarev, G. A. Botton, S. Y. Savrasov, C. J. Humphreys, A. P. Sutton,
Phys. Rev. B 1998, 57, 1505.
[66] L. Wang, T. Maxisch, G. Ceder, Phys. Rev. B 2006, 73, 195107.
[67] A. Jain, S. P. Ong, G. Hautier, W. Chen, W. D. Richards, S. Dacek, S. Cholia,
D. Gunter, D. Skinner, G. Ceder, K. A. Persson, APL Materials 2013, 1,
011002.
[55] M. B. Lourenço, M. D. Carvalho, P. Fonseca, T. Gasche, G. Evans, M. God-
inho, J. Alloys Compd. 2014, 612, 176–182.
[56] N. Hollmann, M. W. Haverkort, M. Benomar, M. Cwik, M. Braden, T. Lorenz,
Phys. Rev. B 2011, 83, 174435.
[68] F. Birch, Phys. Rev. 1947, 71, 809–824.
[57] H. A. Wriedt, N. A. Gokcen, R. H. Nafziger, Bull. Alloy Phase Diagr. 1987,
8, 355–377.
Received: June 7, 2016
[58] H. M. Rietveld, Acta Crystallogr. 1967, 22, 151–152.
Published Online: September 13, 2016
Eur. J. Inorg. Chem. 2016, 4792–4801
www.eurjic.org
4801
© 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim