Improving the Ca-assisted Urea-Glass Method for the
Synthesis of Zirconium Nitride as Potential Electrocatalyst
for the Nitrogen Reduction Reaction
Sebastian C. H. Bragulla,*[a, c] Aaron R. von Seggern,[a, d] Julian Lorenz,[a] Corinna Harms,[a]

Michael Wark,[d] and K. Andreas Friedrich*[b, c]

Zirconium mononitride (ZrN), among other nitrides, is of
interest as a catalyst in general and especially for the electro-
chemical nitrogen reduction reaction (NRR). Synthesis of nano-
particulate ZrN is challenging due to the formation of undesired
other phases and the need for hazardous precursors or high
temperatures. The urea glass method (UGM) was used in this
work to synthesise nanoparticulate zirconium nitride. A focused
parameter variation of the precursor composition and process
conditions lead to an optimised synthesis, greatly improving
material purity in regard to minor phase and residual carbon
content. The nitride starts to form at a pyrolysis temperature of
900 °C, likely by irreversible decomposition of previously-formed
oxynitride, which forms from amorphous ZrO2 and urea during
pyrolysis. An increased pyrolysis temperature improves product
purity but slightly lowers nitride purity. Calcium admixture

slows the urea decomposition, yielding a more phase-pure
product with less residual carbon at low amounts. A classical
carbothermal nitridation (CN) was used in comparison to the
UGM to synthesise ZrN. This was successful in synthesising
zirconium carbonitride Zr(N, C) without minor phase, but
significant carbon remains if added overstoichiometrically.
Using less carbon and a longer dwelling time yielded ZrN
without significant carbide (ZrC) incorporation or residual
carbon, but with some minor ZrO2 phase, proving a promising
synthesis route. A limited electrochemical investigation of a
synthesised ZrN material was carried out to scope the general
material behaviour and to qualitatively derive the potential
catalytic activity for the NRR. Higher reductive currents in
nitrogen-saturated electrolyte qualitatively point to NRR activity,
but further and more in-depth examination is required.

1. Introduction

Catalysts are indispensable for today’s society and industry.
They enable efficient, selective and high-yield economical
production of a plethora of products. The industrial scale
production of ammonia via the Haber-Bosch process using an
iron catalyst is a famous example, with an estimated world-
wide production of 150 Mt in 2021.[1,2] Many catalyst materials

are made from noble metals because of their beneficial material
properties. Transition metal carbides (TMC) and transition metal
nitrides (TMN) have similar electronic properties as noble
metals, making them promising alternative catalyst materials to
rare noble metals. TMNs are of interest as catalysts for several
electrochemical reactions such as hydrogen evolution reaction
(HER), oxygen evolution reaction (OER) and more.[3,4] Further-
more, TMNs are of special interest for the electrochemical
nitrogen reduction reaction (NRR), which potentially can enable
a small-scale on-demand electrochemical ammonia synthesis
(EAS) directly from water, nitrogen and renewable electricity. To
date, EAS suffers from insufficiently active and selective
catalysts, resulting in production rates and efficiencies more
than one order of magnitude too small for commercialisation.[5]

Most of these catalyst materials are fundamentally limited by
their reaction mechanism.[6] Certain TMNs are promising
alternatives, because they catalyse the NRR via a Mars-van-
Krevelen mechanism, possibly circumventing these limitations.[7]

Zirconium nitride (ZrN) is one of the promising candidates.
Theoretical investigations have identified the cubic (100) facet
of ZrN as active, selective and stable for NRR.[7] Recent
experimental investigation of sputtered ZrN thin films showed
ammonia production, but limited surface area and solubility of
nitrogen in the electrolyte were barriers to a proof of genuine
NRR activity.[8] For this reason, investigating nanoparticulate ZrN
as an electrocatalyst for NRR is highly interesting. However,
synthesising TMN nanoparticles with the desired properties is
complex. ZrN can be synthesised by a variety of physical and

[a] S. C. H. Bragulla, A. R. von Seggern, J. Lorenz, C. Harms
Institute of Engineering Thermodynamics, German Aerospace Center (DLR),
Carl-von Ossietzky-Str. 15, Oldenburg 26129, Germany
E-mail: Sebastian.Bragulla@dlr.de

[b] K. A. Friedrich
Institute of Engineering Thermodynamics, German Aerospace Center (DLR),
Pfaffenwaldring 38–40, Stuttgart 70569, Germany
E-mail: Andreas.Friedrich@dlr.de

[c] S. C. H. Bragulla, K. A. Friedrich
Institute for Building Energetics, Thermotechnology and Energy Storage
(IGTE), University of Stuttgart, Pfaffenwaldring 31, Stuttgart 70596,
Germany

[d] A. R. von Seggern, M. Wark
Institute of Chemistry, Chemical Technology 1, Carl von Ossietzky University
Oldenburg, Carl-von-Ossietzky-Str. 9–11, Oldenburg 26129, Germany

Supporting information for this article is available on the WWW under
https://doi.org/10.1002/cctc.202400613Supporting information for this ar-
ticle is available on the WWW under https://doi.org/10.1002/cctc.202400613

© 2024 The Authors. ChemCatChem published by Wiley-VCH GmbH. This is
an open access article under the terms of the Creative Commons Attribution
License, which permits use, distribution and reproduction in any medium,
provided the original work is properly cited.

Wiley VCH Montag, 14.10.2024

2420 / 365733 [S. 268/286] 1

ChemCatChem 2024, 16, e202400613 (1 of 19) © 2024 The Authors. ChemCatChem published by Wiley-VCH GmbH

ChemCatChem

www.chemcatchem.org

Research Article
doi.org/10.1002/cctc.202400613

http://orcid.org/0000-0002-3082-2174
http://orcid.org/0009-0003-0767-151X
http://orcid.org/0000-0002-9936-7667
http://orcid.org/0000-0001-5916-3224
http://orcid.org/0000-0002-8725-0103
http://orcid.org/0000-0002-2968-5029
https://doi.org/10.1002/cctc.202400613
https://doi.org/10.1002/cctc.202400613


chemical processes, such as physical or chemical vapour
deposition (PVD or CVD), direct nitridation of the metal or
oxide, ammonolysis of zirconium halides or carbothermal
nitridation.[9–12] Lerch et al. synthesised Zr3N4 by reaction of ZrCl4
with NH3 at 850–950 °C, which decomposes to high-purity ZrN
at 800 °C.[9] Parkinson and Nelson synthesised ZrN by carbother-
mic reduction nitridation from ZrO2 and graphite at temper-
atures of 1400–2000 °C.[13] These methods require either special
equipment (PVD, CVD), hazardous precursors (metallic Zr, NH3)
or high temperatures (>1400 °C). The achievable nitride purity,
phase purity (minor phases) and material purity (residual
carbon) vary depending on the synthesis method. Instead, urea
has been used as cheap, non-toxic, easy-to-handle in situ
nitrogen source to synthesise different TMN nanoparticles by a
non-aqueous sol-gel route at moderate temperatures.[14] This
route, termed ‘urea-glass method’ (UGM) by Giordano et al., is
accessible, versatile and easy to adapt.[14] A variety of TMNs
were successfully synthesised by UGM, for example chromium
and vanadium nitride.[14] However, precise control of the
occurring phases (oxide, oxynitride, nitride, carbide) and their
purity, as well as the overall sample composition (residual
carbon) is complex, because the reactive atmosphere is
generated in situ and is not under direct process control. Gao
et al. developed the Ca-assisted UGM by adding calcium ions,
which slows the urea decomposition down. This slowed-down
urea decomposition allows for more control over the occurring
nitride phases and enabled synthesis of the desired tantalum
nitride phase. A small calcium admixture also resulted in more
homogeneous and phase pure nitride nanoparticles. Larger
calcium admixture resulted in the formation of oxynitrides. An
additional advantage of the calcium admixture was the
reduction of remaining residual carbon from the decomposition
of urea during pyrolysis.[15] Investigating these effects of a
calcium admixture to optimise the synthesis of zirconium
nitride by UGM is attractive to achieve synthesis of high-purity
zirconium nanoparticles for use as electrocatalyst.

In this study, we aimed to synthesise nanoparticulate ZrN as
prospective electrocatalyst material using the UGM as acces-
sible and versatile synthesis route. In doing so, we varied the
pyrolysis mixture (urea content, calcium admixture), pyrolysis
temperature and dwelling time to achieve, in order, an ideally
pure nitride phase, minimal minor phases, homogeneous nano-
particles, and minimal residual carbon in the synthesis product.
In addition, we also employed a classical carbothermal
nitridation (CN) at high temperature as alternative method for
synthesizing zirconium nitride in a limited capacity, to illustrate
the use of both methods in comparison. The synthesised
powders were structurally studied by various physical character-
isation techniques to investigate the effect of the parameter
variations. A limited electrochemical investigation of synthes-
ised materials in regard to their electrocatalytic behaviour for
the NRR will also be presented.

2. Results

An overview of all samples synthesised can be found in the
supporting information Table S1. The samples have been
numbered consecutively with Roman numerals (I–XXV) for
cross-reference. These numbers are stated in the labels of
graphs to reference samples i. e. indicate one and the same
sample shown in more than one graph.

The following section starts with an overview of X-ray
diffraction (XRD) patterns of relevant zirconium phases taken
from literature and used for reference. This is followed by a
short description of the sol-gel synthesis as starting point of the
synthesis. The impact of the varied parameters, namely pyrolysis
temperature and calcium admixture, on the occurring phases
are discussed thereafter as initial main focus of this work. The
effect of a lower urea content and the impact of the choice of
pyrolysis vessel are discussed in the supporting information.
These results are concluded with a comparison of the urea-glass
method with the high-temperature synthesis of zirconium
nitride by carbothermal nitridation. Further characterization of
the UGM samples by X-ray photoelectron spectroscopy (XPS)
and transmission electron microscopy (TEM) is presented at the
end. The results close a with limited electrochemical inves-
tigation.

2.1. Ceramic Phases of Zirconium – Nitride, Oxynitride and
Oxide

There is some uncertainty in the exact value of the lattice
constant of zirconium nitride and zirconium carbide. Both
materials are easily contaminated by carbon and nitrogen
respectively, as well as oxygen. This contamination affects the
measured lattice constant to a different degree. Subsequently,
there is some variation in the given lattice constant for these
‘pure’ compounds depending on the literature source. Most
reliable are literature sources in which an accompanying
elemental analysis was carried out. An overview of the major
XRD patterns of relevant zirconium nitride, oxynitride and oxide
phases we used for reference in this work are shown in the
supporting information (Figure S1). The ICDD Powder Diffrac-
tion File (PDF) card and associated literature source of these
XRD patterns are given in Table 1 below. In general, PDF
patterns with available cited literature were chosen. The works

Table 1. Reference X-ray diffraction (XRD) patterns of zirconium nitride,
oxynitride, and oxide used in this work.

Phase ICDD PDF Literature

ZrN0.97(4) – [16]

ZrN 98-064-4895 [17]

γ (Zr2ON2) 98-008-8666 [18]

β (Zr7O8N4) 98-024-5751 [19]

β’ (Zr7O11N2) 98-015-7959 [20]

m-ZrO2 98-005-7158 [21]

t-ZrO2 98-008-5322 [22]

Wiley VCH Montag, 14.10.2024

2420 / 365733 [S. 269/286] 1

ChemCatChem 2024, 16, e202400613 (2 of 19) © 2024 The Authors. ChemCatChem published by Wiley-VCH GmbH

ChemCatChem
Research Article
doi.org/10.1002/cctc.202400613



of Lerch et al. and Constant et al. were selected, because both
give accompanying chemical compositions for the investigated
compounds.[16,17] Moreover, Constant et al. specifically investi-
gated the change in lattice constant of solid-solutions of
zirconium nitride and zirconium carbide.[17] The data given by
Constant et al. for zirconium mononitride (ZrN) and zirconium
carbide (ZrC) was used below to calculate the stoichiometries of
selected synthesised samples by Vegard’s Law (SI S13).[17]

The different oxynitride phases, β, β’ and γ, differ in nitrogen
and oxygen content and are difficult to distinguish by PXRD, as
can be seen in Figure S1b. This difficulty in assigning phases is
further compounded by the position of the main peak of
tetragonal zirconium dioxide (t-ZrO2). These circumstances
often prevent unequivocally assigning a pattern to an occurring
low intensity minor phase. Semiquantitative evaluation of
phases by their intensity as available in the used HighScore+

software version 4.1 with PDF4 database (2014) was thus not
used for this reason. For the sake of transparency and
comparability, all diffractograms show as-recorded data without
any background correction. The diffractograms have been
scaled to the highest intensity peak if not specified differently.
All angles given in this work refer to 2θ angles.

2.2. Sol-Gel Synthesis

The sol-gel synthesis yielding the pyrolysis mixture directly
affects the synthesis of the desired nitride. Dried ethanol
absolute was used as non-toxic, readily available solvent for the
sol-gel synthesis.[23] Zirconium tetrachloride reacts with ethanol
to form an alkoxide.[14] However, this reaction is incomplete. A
hydrogen chloride acceptor such as ammonia can be used to
complete the reaction.[24] The incompletely reacted zirconium
tetrachloride proved to be more stable against unwanted,
premature hydrolysis and polycondensation by traces of
moisture than commercial zirconium ethoxide. A 28 wt.%
ammonium hydroxide solution was used in this work for the
purpose of completing the reaction of the zirconium tetra-
chloride to the ethoxide, which caused the reaction to be
accompanied by rapid crosslinking. Rapid crosslinking likely
results in an amorphous zirconium dioxide phase, which is
beneficial for nitridation.[25,26] The powder X-ray diffractogram
(PXRD) of the unpyrolyzed mixture with a molar ratio of urea to
zirconium U of 15 (U15) without admixture of calcium
dichloride is shown in Figure S2a in the supporting information.
The mixture consists by sol-gel synthesis of mainly urea,
zirconium dioxide, and remaining ammonium chloride. This is
mirrored in the PXRD. The peak at 30.42° can be assigned to t-
ZrO2, but otherwise does not match the pattern of t-ZrO2 well.
The other marked peaks (22.20°, 24.58°, 29.28°, 35.46°, 37.06°)
match urea more or less well in their position. Thus, the
unpyrolyzed mixture likely consists of amorphous zirconium
dioxide with well dispersed urea. The commercial nanoparticu-
late zirconium dioxide used for carbothermal nitridation is
shown in Figure S2b. It matches m-ZrO2 well and was used as
supplied. Purposely crosslinking the sol by polycondensation
after addition of urea and calcium chloride, followed by drying

at 60 °C, results in a pourable, easy and safe to handle powder.
The rapid crosslinking might also preserve the state of
molecularly mixed components, which would be beneficial for
the pyrolysis. The powder can be stored at room temperature
and atmosphere without apparent detrimental effect on the
pyrolysis result.

2.3. Effect of Pyrolysis Temperature

The following section shows the effect of the pyrolysis temper-
ature on the occurring phases by X-ray diffraction (XRD). The
pyrolysis temperature is a critical process parameter, because a
high enough temperature is necessary, but not sufficient in
itself. A sufficiently high urea content is also needed. A molar
ratio of urea to metal U of 15 (U15) was established in prior
experiments to synthesise ZrN at 900 °C.

Figure 1 shows powder X-ray diffractograms (XPRD) of
samples with a molar ratio of 15 (U15), pyrolyzed at different
temperatures of 800 °C, 900 °C, 1150 °C and 1400 °C with a
dwelling time of 3 h, with exception of the highest temper-
ature.

The PXRD pattern at 800 °C (Figure 1a) is complex, consist-
ing of a diverse mixture of likely oxide and oxynitride. As
expected, no ZrN-phase is obtained at a pyrolysis temperature
of 800 °C.[9] It has to be noted, that this pyrolysis was done using
a glassy carbon boat with a lid in contrast to the 900 °C and
1150 °C sample shown in Figure 1. The major peak at 34.01° can
be attributed to ZrN but should be accompanied by a peak of
similar intensity at 39.4°, which is absent (see 900 °C, Figure 1b).
Still, a successful synthesis of ZrN at 800 °C using UGM is
reported in literature, using a lower urea content and higher
heating rate.[27] Nevertheless, synthesising ZrN at 800 °C follow-
ing these reported parameters did not succeed in our work, in
contrast to the reference.

A major cubic ZrN-phase (33.94°, 39.38°, 56.92°, 67.94°,
71.4°, 84.75°) emerges at 900 °C (Figure 1b) along with minor
phases. The oxynitride phases (γ, β, β’) are difficult to distinguish
by PXRD (Figure S1b). There is also pattern overlap with
monoclinic (m) and tetragonal (t) zirconium dioxide (Fig-
ure S1c). The clearly visible peak at 30.41° can be assigned to
either oxynitride or oxide, allowing for a small shift. However,
the small peaks visible upon close inspection of the background
at 21.47° and 45.62° match the γ-phase without any pattern
overlap with the other regarded phases shown in Figure S1. The
peak at 26.5° does not fit any of the mentioned dioxide or
oxynitride patterns well. The peak at 28.27° can tentatively be
assigned to m-ZrO2 allowing for a small shift. The remaining
peaks (28.27°, 35.18°, 50.71°, 60.33°) again can be assigned to
either one of the oxynitride or dioxide phases. The amorphous
region spanning from 20° to 35° can in part be caused by the
used glass substrate, which does span a similar region.[28]

However, an amorphous region is visible in the same range for
the crystallisation of amorphous zirconium hydroxide Zr(OH)4,
which points to similar processes.[29] Regardless of the minor
phases, synthesis of ZrN was successful at 900 °C. A minimum
temperature of 900 °C concurs with the quasi-binary ZrO2-Zr3N4

Wiley VCH Montag, 14.10.2024

2420 / 365733 [S. 270/286] 1

ChemCatChem 2024, 16, e202400613 (3 of 19) © 2024 The Authors. ChemCatChem published by Wiley-VCH GmbH

ChemCatChem
Research Article
doi.org/10.1002/cctc.202400613



phase diagram given by M. Lerch, in which nitrogen-rich
zirconium oxynitride phases irreversibly decompose to ZrN,
nitrogen and oxide or oxynitride, depending on the decompos-
ing oxynitride phase and temperature.[9] Nitrogen diffusion
seems to be the rate controlling factor for the conversion. The
phase diagram by M. Lerch was constructed from sintering
defined mixtures of ZrO2 and Zr3N4, resulting in the described
phase diagram.[9] Although the pyrolysis mixture does not
consist of such a mix, the conversion to the nitride seems to
occur similarly via an oxynitride phase starting from the dioxide.
These conversion processes could also be the cause for some
peak shifts as a result of non-stochiometric phases or defects.

Increasing the pyrolysis temperature to 1150 °C (Figure 1c),
results in smaller peaks of the minor phases, indicating lower
residual amounts of likely zirconium dioxide. The peak at 28.19°
again fits m-ZrO2. The remaining small peaks (30.31°, 35.16°,
50.59°, 60.11°) can be assigned to either oxynitride phase.
However, the characteristic γ-peaks at 21.47° and 45.62° have
vanished. Also, there is a small shift of all ZrN peaks to smaller
2θ angles by 0.04° to 0.09°, pointing to carbon incorporation.

Regardless, an increased pyrolysis temperature of 1150 °C is
beneficial to reduce undesired minor phases.

Increasing the temperature further to 1400 °C (Figure 1d)
results in minor peaks barely above the background alongside
the ZrN pattern, the hitherto cleanest ZrN pattern. The
diminishingly small peaks at 28.18°, 30.31°, 50.65° and 60.18°
are the same as before. This does not exclude some remaining
oxygen in the ZrN crystal lattice. Furthermore, there is no visible
amorphous region anymore, which can also be due to the
scaling to the highest intensity peak. Regardless, the amount of
minor phase is much reduced. The increased pyrolysis temper-
ature from 900 °C to 1400 °C is accompanied by a small shift of
the ZrN peaks to lower angles of 0.06° 2θ (33.94° to 33.88°),
which is greater than estimated deviations (SI S12). This shift to
lower angles is indicative of incorporation of carbon into the
ZrN lattice, increasing the lattice parameter. It is well known
that zirconium nitride is easily contaminated by oxygen and
carbon.[16,17,30] This solid solution is often abbreviated as Zr(N, O,
C). While the effect of oxygen on the lattice parameter is
negligible depending on magnitude, carbon incorporation
results in a noticeable shift.[17] This shift caused by the solid

Figure 1. PXRDs of sample coatings on glass pyrolyzed at 800 °C (a) (III), 900 °C (b) (VI), 1150 °C (c) (X) and 1400 °C (d) (XII). A glazed alumina combustion boat
(B) with an unglazed alumina lid (L) open at the ends was used at 900 °C and 1150 °C. A different kind of crucible, a glassy carbon (GC) boat, was used at
800 °C and 1400 °C.

Wiley VCH Montag, 14.10.2024

2420 / 365733 [S. 271/286] 1

ChemCatChem 2024, 16, e202400613 (4 of 19) © 2024 The Authors. ChemCatChem published by Wiley-VCH GmbH

ChemCatChem
Research Article
doi.org/10.1002/cctc.202400613



solution of ZrN and ZrC can be quantitatively described by
Vegard’s Law. However, this calculation is very sensitive to the
determined lattice parameter of the solid solution and chosen
reference lattice parameters. Vegard’s Law was used exemplarily
in the comparison of synthesis methods (Table 2). The urea
decomposition during the pyrolysis results in residual carbon in
the pyrolysis mixture. The more urea, the more residual carbon
may remain after pyrolysis, which may have been incorporated
into the lattice to a greater extend as well.

Synthesis of zirconium nitride by UGM is possible above a
pyrolysis temperature of 800 °C. The minor oxynitride and oxide
phase, occurring alongside the nitride at 900 °C, are diminished
with increasing pyrolysis temperature. There is no noticeable
minor phase at 1400 °C anymore. However, the increased
pyrolysis temperature is accompanied by a small shift of the
peaks to lower angles, indicating undesired carbon incorpo-
ration into the crystal lattice i. e. a reduced purity of the nitride.

2.4. Effect of ‘Calcium’ Admixture

The following section shows the effect of adding calcium ions
to the sol-gel on the synthesis of zirconium nitride as described
by Gao et al. for tantalum nitride.[15] To this end, calcium
dichloride was added in a molar ratio to the zirconium
tetrachloride. This admixture of calcium dichloride is simply
referred to as ‘calcium admixture’ in the following text and is
denoted by ‘Ca’ followed by the ratio as a number. The added
calcium ions slow the urea decomposition and in turn the
release of reactive nitrogen species. The effect thereof on the
conversion of the likely amorphous oxide precursor to the
desired nitride phase was investigated by varying the calcium
admixture. The effect of added calcium ions on the thermal
decomposition of pure urea was investigated by thermogravi-
metric analysis (TGA), showing a slowed down decomposition
(SI S4). Shown below are PXRDs of samples with a calcium
admixture of Ca0.25, Ca0.5, Ca1.0, and Ca2.0.

A calcium admixture as low as Ca0.25 (Figure 2a) results in
the visible occurrence of oxynitride (and oxide) as minor phases
alongside a major ZrN-phase (33.98°, 39.44°, 56.99°, 68.04°,
71.51°, 84.66°). The different oxynitrides are difficult to distin-
guish by PXRD due to their similar lattice parameters.[31] There is
also pattern overlap of the oxynitrides with t-ZrO2 and m-ZrO2

(Figure S1c). The noticeable peak at 10.97°, indicating a large
lattice distance, is likely caused by a layered structure such as
layered double hydroxides.[32,33] Large thin plate-like structures
with small homogeneously embedded nanoparticles were
observed in TEM images for samples with calcium admixture
(Figure 4), pointing at this explanation. However, the formation
of some compound of zirconium with calcium is also a possible
explanation. Some form of nucleation seems to originate within
these structures (Figure 4b). The peak at 21.44° is matched only
by the γ-phase, Zr2ON2. Although the angles are slightly too
large, the peaks at 30.53° and 35.41° can be assigned best to
the γ-phase as well. The amorphous region occurring for this
composition without calcium is absent.

An increased calcium admixture of Ca0.5 results in the
highest relative intensity of the ZrN phase for all tested calcium
contents. Further increasing the calcium admixture resulted in
less ZrN by intensity in relation to the other phases. The nitride
becomes a minor phase at Ca2.0, where the γ-phase becomes
dominant. An increasing peak intensity at 10.97° coincides with
reduced intensity of ZrN. The same trend occurs for the γ-phase
peak at 30.53°. These similar trends diverge at Ca2.0, where the
γ-phase is dominant, but the peak at 10.97° is at a slightly lower
intensity than in Ca0.25. This trend and divergence can be
explained by a decomposition of γ-phase to nitride.[9] The
conversion seems to be accompanied by the layered structure
causing the peak at 10.97°. The less γ-phase decomposes, the
less nitride forms, the less layered structure occurs, in which
possibly some form of nucleation happens. It is uncertain,
whether this structure forms from decomposition the γ-phase
or itself forms the nitride. However, the peak at 10.97° is not
visible at a lower urea content of U5 (Figure S9–S11), neither
with calcium, when no nitride forms, nor without calcium, when
some nitride forms. The γ-phase is dominant in both cases.
Noteworthy as well is a shift of the ZrN peaks to slightly higher
angles with increasing calcium content, which could be caused
by higher nitride purity i. e. closer to a perfect stoichiometry
(Figure S6). Although the effect of oxygen on the lattice
parameter of ZrN is low, it results in a shift to lower Bragg
angles.[17] The slight shift to higher angles observed with
calcium admixture might indicate less oxygen in the ZrN lattice.
Lerch and Wrba give a slightly lower lattice parameter for ZrN,
generated from decomposition of Zr3N4, of 0.45569(4) nm than
Constant et al. with 0.4571 nm (� 0.3%), which yields Bragg
angles at the observed higher values with calcium admixture
(Figure S1a).[16,17] Ammonolysis of ZrCl4 in flowing ammonia at
850–950 °C results in Zr3N4,

[34] but only if there is no oxygen
present, otherwise oxynitrides form.[31] In consequence, ZrN
formed from decomposition of Zr3N4 should contain one of the
least amounts of oxygen in the crystal lattice. Lattice parame-
ters derived from this ZrN should be close to the true value.

Based on this interpretation, the conversion of amorphous
ZrO2 to ZrN happens by successive conversion of the oxide to
oxynitrides with increasing nitrogen content. These oxynitrides
in turn decompose into nitride, oxynitride and nitrogen (see
scheme below).

Table 2. Stoichiometry by Vegard’s law and residual carbon content by
oxidation for selected samples synthesised by urea-glass method (UGM)
and high temperature carbothermal nitridation (CN).

Sample Vegard’s Law Residual carbon
wC

UGM: Ca0.0, 1400 °C, 2 h (XII) ZrN0.93�0.05C0.07�0.05 13.8�1.2%

UGM: Ca0.5, 1150 °C, 3 h
(XVII)

ZrN0.95�0.05C0.05�0.05 6.2�1.4%

CN: C4, 1400 °C, 3 h (XXIII) ZrN0.63�0.17C0.37�0.17 16.3�3.7%

CN: C4, 1400 °C, 4.5 h (XXV) ZrN0.90�0.05C0.10�0.05 20.3�1.1%

CN: C2, 1400 °C, 4.5 h (XXIV) ZrN0.91�0.05C0.09�0.05 � 0.2�2.1%

Wiley VCH Montag, 14.10.2024

2420 / 365733 [S. 272/286] 1

ChemCatChem 2024, 16, e202400613 (5 of 19) © 2024 The Authors. ChemCatChem published by Wiley-VCH GmbH

ChemCatChem
Research Article
doi.org/10.1002/cctc.202400613



ZrO2 ! Zr7O11N2 ðb
0

Þ ! Zr7O8N4 bð Þ !

Zr2ON2 gð Þ ! ZrNþ bþ N2

Following the quasi-binary phase diagram given by Lerch, γ
decomposes to nitride, β and nitrogen above approximately
840 °C and below 1000 °C.[9] Assuming γ-phase decomposes
stoichiometrically (Equation (1)) and disregarding other compo-
nents such as residual carbon, the achievable ZrN mass fraction
via this conversion is only 54% (Equation (2)).

16 Zr2ON2 ! 2 Zr7O8N4 þ 18 ZrNþ 3 N2 (1)

18 MZrN

18 MZrN þ 2 MZr7O8N4

¼
18 � 105:2307

18 � 105:2307þ 2 � 822:5868 � 54 % (2)

Though, if the gradual nitridation of ZrO2 takes place continu-
ously by a retained local atmosphere, and not solely via
decomposition of an early evolved glassy intermediate, then
the oxynitride by-product is nitrided again and in turn
decomposes to form nitride and oxynitride (see scheme below).

b! g! ZrNþ bþ N2

The amount of ZrN was determined exemplarily for one sample
using XRD by standard addition of commercial ZrO2, giving a
mass fraction of about 90%. This much higher mass fraction
than the theoretical value supports the assumption of a
continuous cycle as long as there are reactive nitrogen species.
An increased dwelling time should result in more nitride under
these circumstances, but only if this reservoir of nitriding
atmosphere is retained, not already depleted and the phases
are not yet in equilibrium. However, an increased dwelling time
of 4.5 h (50% increase) had no significant effect at 900 °C with
or without added calcium dichloride, with the exception of an
approximately halved peak intensity at 30.53° for Ca0.5 (SI S8).
Calcium admixture slows the urea decomposition and release of
reactive nitrogen species based on its effect on the phase
composition, which depends on the amount of calcium ions
added. The slowed release results in an oxynitride instead of
oxide minor phase occurring alongside the major ZrN phase.
The cause is either a better utilisation of the available, more
slowly released reactive nitrogen species, converting more

Figure 2. PXRDs of sample coatings on glass pyrolyzed at 900 °C for 3 hours with calcium admixtures of Ca0.25 (a) (XIII), Ca0.5 (b) (XIV), Ca1.0 (c) (XIX) and
Ca2.0 (d) (XXI) respectively.

Wiley VCH Montag, 14.10.2024

2420 / 365733 [S. 273/286] 1

ChemCatChem 2024, 16, e202400613 (6 of 19) © 2024 The Authors. ChemCatChem published by Wiley-VCH GmbH

ChemCatChem
Research Article
doi.org/10.1002/cctc.202400613



oxide, or it is as yet undecomposed oxynitride. A large
admixture inhibits the conversion of the oxide to the nitride.

2.5. High–Temperature Synthesis – Carbothermal Nitridation

A carbothermal nitridation (CN) synthesis of zirconium nitride
(Equation (3)) becomes possible at a temperature of 1400 °C, at
which zirconium dioxide can be nitrided directly by a nitrogen
atmosphere.[16] A classical CN of ZrN was tested in comparison
to the UGM using a high-temperature oven and a mixture of
commercial ZrO2 and carbon black. The added carbon acts as
oxygen getter reducing the ZrO2, for which the theoretical
stoichiometric ratio is 2 (Equation (3)).

2 ZrO2 sð Þ þ N2 gð Þ þ 4 C sð Þ

! 2 ZrN sð Þ þ 4 CO gð Þ
(3)

A molar ratio of C to ZrO2 of 4 (C4) resulted in a single
Zr(N,C) phase with a noticeable shift of �0.38° to lower angles
at a dwelling time of 3 h, which is indicative of a large ZrC
fraction in the solid solution of ZrN and ZrC (Figure S16). This
was alleviated by an increased dwelling time of 4.5 h (Fig-
ure S16). The large amount of remaining added carbon after
pyrolysis (Table 2) was remedied by reducing the molar ratio to
the theoretical stochiometric ratio of C2. Although C2 prevents
residual carbon, it also results in a small minor oxide phase. This
optimised CN product as well as UGM 1400 °C and UGM 1150 °C
Ca0.5 are shown in Figure 3 for comparison. All three PXRDs
shown a major ZrN phase, with a varying shift to lower angles,
which is caused by carbon in the ZrN lattice. The stoichiometry
of this solid solution of ZrN and ZrC was evaluated by Vegard’s
law. However, the lattice parameters of the pure phases, ZrN
and ZrC, are required. These data are optimally generated on
the same setup with the identical systematic measurement
error. In absence of that, suitable reference data can be used.
The lattice parameters of ZrN and ZrC as well as mixtures
thereof given by Constant et al. were linearly fitted (SI S13).[17]

Lattice parameters of aZrN=0.4571 nm and aZrC=0.4695 nm
were used for calculations. All peaks shown were included in
the calculation, weighted by their peak intensity and averaged.
A linear error propagation was used to calculate a standard
deviation. The stoichiometries calculated this way are ZrN0.91�

0.05C0.09�0.05 (CN), ZrN0.93�0.05C0.07�0.05 (UGM 1400 °C) and Ca0.5
ZrN0.95�0.05C0.05�0.05 (UGM 1150 °C) (Table 2). Although all three
calculated stoichiometries are similarly pure in regard to the
nitride, they show that a higher synthesis temperature results in
a larger amount of undesired carbon incorporated in the crystal
lattice. Higher temperatures enhance carbon incorporation, the
carbide is more stable than the nitride at high temperatures.[35]

Yet a minimum temperature of 1400 °C is needed for direct
nitridation of ZrO2 in nitrogen.[16] Likely cause of the larger ZrC
fraction of the CN sample lies within the reaction mechanism.
CN works by conversion of the ZrO2 to ZrC, which in turn is
converted into ZrN.[36,37] A further increased dwelling time might
remove both the ZrC fraction and minor oxide phase. In
contrast, the nitrides synthesised by UGM at 1400 °C and

1150 °C with Ca0.5 do not contain significant carbon in the
crystal lattice as determined by Vegard’s Law, which is the
desired outcome, a pure ZrN. Both UGM samples contain small
amounts of oxynitride, less so at 1400 °C than at 1150 °C. All
three samples were oxidized in pure oxygen using a TGA to

Figure 3. PXRDs of sample coatings on glass pyrolyzed at 1400 °C via CN (a)
(XXIV), UGM (b) (XII) and at 1150 °C with Ca0.5 (c) (XVII).

Wiley VCH Montag, 14.10.2024

2420 / 365733 [S. 274/286] 1

ChemCatChem 2024, 16, e202400613 (7 of 19) © 2024 The Authors. ChemCatChem published by Wiley-VCH GmbH

ChemCatChem
Research Article
doi.org/10.1002/cctc.202400613



derive the amount of residual carbon remaining after pyrolysis
and washing (SI S14). The carbon mass fraction in the
unoxidized sample was derived from the change in mass
assuming complete oxidation and conservation of the amount
of zirconium. The stoichiometry derived by Vegard’s Law was
used to calculate the molar mass. Other impurities such as
oxynitrides were discounted. The carbon mass fraction calcu-
lated this way were in order � 0.2�2.1%, 13.8�1.2% and 6.2�
1.4% (Table 2). The CN sample does not contain residual
carbon, which is desired. The UGM 1150 °C Ca0.5 sample
contains less residual carbon than UGM 1400 °C, as expected.
One effect of calcium admixture is reduction of residual
carbon.[15]

Both synthesis methods allow to synthesise relatively pure
nanocrystalline ZrN. UGM achieves a higher ZrN purity in regard
to ZrC, but at a much lower synthesis yield with some
remaining residual carbon and general more complex synthesis
mixture. CN results in slightly lower nitride purity, but at higher
synthesis yield with no residual carbon, albeit with a minor
oxide phase at the current synthesis parameters, and always
requires a higher pyrolysis temperature.

2.6. Influence of the Synthesis on observed Particle
Morphologies

Particle size and morphology were qualitatively assessed by
TEM, a representative selection of samples and particles is
shown in Figure 4. The samples generally showed aggregates
and agglomerates of small and large particles as shown in
Figures 4a, c, d and f. Large particles were in the range of 50–
100 nm, while small particles were in the range of 10–20 nm.
Particles, large and small, separate or in aggregates, often
locally showed a core-shell structure visible in the contrast
(Figure 4d). This shell likely consists of zirconium dioxide or
residual carbon, albeit an oxide shell is more likely, either by
surface oxidation or in consequence of the reaction mechanism.
Admixture of calcium (Figures 4b, f) resulted in the additional
occurrence of large plate-like particles as shown in Figure 4b. It
is suspected that these structures are responsible for the PXRD
peak at 10.9°. Small nanoparticles in the range of 2–5 nm were
evenly embedded in some of these plate-like structures (Fig-
ure 4b). Further investigation of these particles via HR-TEM and
EDS was not possible, because they disintegrated upon
irradiation at higher acceleration voltages typical for HR-TEM
(200 kV). Calcium admixture also resulted in some much larger
cubic particles in the range of 500–1000 nm (Figure 5), indicat-
ing higher crystallinity. This observation matches the higher
intensity of PXRD patterns and Kα1-Kα2-splitting seen for
samples with calcium admixture (Figure S15).

HR-TEM brightfield and selected area electron diffraction
(SAED) images were used to determine d-spacings (SI S16).
Reference d-spacings were taken from the used ICDD PDF
cards. While both brightfield images and SAED yield d-spacings
which can be assigned to ZrN and ZrO2 for samples with and
without calcium admixture, SAED also yields d-spacings that
can be in part tentatively assigned to oxynitride phases,

affirming the generally observed phases by PXRD. The elemen-
tal composition was analysed by EDS for the same two samples.
Figure 5 exemplarily shows the EDS measurement of the sample
with calcium admixture Ca0.5 pyrolyzed at 900 °C. All EDS
measurements showed a largely homogeneous elemental
distribution in regard to the examined elements (Zr, N, C, O).
Detected Zr is spatially localized to particles, most notably for
the approximately 500 nm particle in the top of the image.
High N concentrations coincide well with Zr and thus the
particles, indicating ZrN particles. However, O shows a similar
local distribution. This points to either a surface oxidation, oxide
or oxynitride shell. Both N and O show a higher concentration
towards the particles fringe. This also points to a surface
oxidation or assumed conversion of an oxynitride to a nitride.
The diffusion rate of oxygen is higher than that of nitrogen,
resulting in a N gradient towards the centre, which was
observed by Zhao et al. for ZrN synthesized by CN.[9,37] Particle
geometry, thickness and edge orientation can cause a higher
local concentration as well e.g. more detected N and O at the
edge of a particle. Having said this, N and more so O also
coincide locally with C outside of particles. The Formvar grid
coating as well as the residual carbon can be the cause,
contributing to the background. Residual carbon probably
precipitates around particles during the synthesis and accumu-
lates, as a carbon shell forms in case of chromium nitride
nanoparticle formation by UGM.[14] Calcium and chloride were
also detected with some localisation at the particles (Figur-
es S21 and S22). Chloride ions likely remain from the used
zirconium tetrachloride precursor, added calcium chloride or
acid-washing with hydrochloric acid. Because both elements are
more or less evenly dispersed throughout the sample, it is
assumed both are residues. The elements quantified by EDS are
given in the Table S7. The surface composition determined by
XPS has been included for sake of comparison. The XPS results
will be discussed in detail in a separate section. Both samples
show a higher carbon content by EDS than XPS. Oxygen shows
the opposite trend, a much higher oxygen content by XPS than
EDS, which is attributed to the expected surface oxidation.
Whether this suspected oxide shell forms during synthesis, after
pyrolysis or during handling in normal atmosphere is uncertain.
XPS in contrast to EDS is a surface sensitive spectroscopy with a
penetration depth of a few nanometres. The ratio of oxygen
content by XPS to EDS is of similar magnitude for both samples,
with and without calcium admixture. Calculating the composi-
tion from EDS as a rough estimate results in a stoichiometry of
ZrO0.34N1.05 with and ZrO1.02N1.34 without calcium admixture
discounting carbon, indicating Zr(N, O) phases. The thus
calculated stoichiometry of the sample with calcium admixture
is close to the γ-phase, Zr2ON2, which occurs with calcium
admixture among other oxynitride phases as determined by
PXRD. Although the effect of dissolved oxygen on the lattice
parameter is small, the stoichiometry of the sample without
calcium determined from EDS, ZrO1.02N1.34, does not fit well with
the determined PXRD pattern, which is matched to ZrN as main
phase.[17] Oxygen in the lattice has been linked to vacancies in
the metal sublattice.[30] Comparing the X-ray density 1x calcu-
lated from the unit cell, assuming complete occupancy of the

Wiley VCH Montag, 14.10.2024

2420 / 365733 [S. 275/286] 1

ChemCatChem 2024, 16, e202400613 (8 of 19) © 2024 The Authors. ChemCatChem published by Wiley-VCH GmbH

ChemCatChem
Research Article
doi.org/10.1002/cctc.202400613



metal sites, to the actual powder skeletal density 1p, disregard-
ing closed pores, allows to calculate the vacancy density, which
is likely caused by dissolved oxygen. Determining the powder

skeletal density by gas pycnometry is in principle a simple task.
However, this is hampered by insufficient sample volume for
accurate measurements and, beyond that, fails due to the

Figure 4. TEM and HR-TEM images representative of the particle size and morphology encountered for different synthesis compositions and pyrolysis
conditions: (a, c, e) Ca0.0, 900 °C (VI), (b) Ca0.5, 900 °C (XIV), (d) Ca0.0, 1150 °C (X), (f) Ca0.5, 1150 °C (XVII).

Wiley VCH Montag, 14.10.2024

2420 / 365733 [S. 276/286] 1

ChemCatChem 2024, 16, e202400613 (9 of 19) © 2024 The Authors. ChemCatChem published by Wiley-VCH GmbH

ChemCatChem
Research Article
doi.org/10.1002/cctc.202400613



residual carbon, which has a much lower, unknown density. The
measured density would have to be corrected for this unknown
density. A powder skeletal density of 5.334�0.04 gcm� 3 was
determined for the sample pyrolyzed at 1150 °C with Ca0.5.
Assuming a density range of less than 1 gcm� 3 for amorphous
carbon to 2.3 gcm� 3 for graphite results in a nitrogen vacancy
concentration of 0.03 to 0.22. The composition determined by
Vegard’s law of ZrN0.95�0.05C0.05�0.05 should correspond to a
nitrogen vacancy concentration of approximately 0.1 in pub-
lished data, which roughly fits the estimated range.[30] Taken
together, this indicates the synthesised material is ZrN as
determined by XRD, but the surface is oxidized and there is
some dissolved oxygen in the lattice.

2.7. Surface Composition of Zirconium (oxy)nitride Materials

XPS measurements were performed to analyse the surface
composition of the prepared ZrN-based materials in respect to
the varied parameters and synthesis routes. TMNs are prone to
surface oxidation. Thus, a surface oxide layer is expected on
successfully synthesized ZrN particles. Even though sputter

cleaning of the pristine samples can remove the surface oxide
layer, it was not performed, because of preferential sputtering
of certain elements. This would result in artificial variation of
the atomic concentrations or cause cascade mixing, where one
element is smeared along deeper levels than in the pristine
sample. Tose effects also obscure an exact assignment of core
level binding energy (BE) values.[38,39]

First, samples synthesised by UGM pyrolyzed at 900 °C are
discussed according to the calcium admixture. Survey spectra
are presented in the supporting information (Figure S14),
showing the expected elements Zr, N, O and C alongside
contaminations from the synthesis and sample processing,
mainly Ca and Cl. The peak-deconvoluted high-resolution Zr 3d,
N 1s and O 1s spectra are shown in separate columns in
Figure 6. Each row pertains to one sample pyrolyzed at 900 °C
with varying calcium admixture (Ca0.0, Ca0.5, Ca1.0, Ca2.0),
starting at the top with no calcium admixture (Ca0.0). The Zr 3d
spectra consist of spin-orbit doublet peaks (Zr 3d5/2 and Zr 3d3/
2), which were fitted with three components corresponding to
zirconium nitride (z1), oxynitride (z2) and oxide (z3) phases.[39,40]

The high-resolution N 1s spectra were fitted with four

Figure 5. EDS measurement of the TEM sample Ca0.5, 900 °C (XIV) showing the elemental composition for Zr, N, O and combined with C. The colour balance
has been adjusted for the light elements N, O, C.

Wiley VCH Montag, 14.10.2024

2420 / 365733 [S. 277/286] 1

ChemCatChem 2024, 16, e202400613 (10 of 19) © 2024 The Authors. ChemCatChem published by Wiley-VCH GmbH

ChemCatChem
Research Article
doi.org/10.1002/cctc.202400613



components corresponding to oxynitride (n1), nitride (n2) and
adsorbed nitrogen species (n3), as well as adsorbed molecular
N2 (n4).[39,41,42] It has to be noted that a low nitrogen
concentration was measured by XPS due to supposed surface
oxidation of the major nitride phase, resulting in a low signal to
noise ratio of the N 1s spectrum. The N 1s spectra were
nonetheless fitted to illustrate changing contributions of the
components among the synthesized materials. The O 1s spectra
were fitted with three components corresponding to oxide (o1),
adsorbed oxygen/hydroxide species (o2) and adsorbed water
(o3).[39,42] Correction of the recorded binding energies was
accomplished by using the shift of the aliphatic carbon signal
(component c1 in C 1s spectra, Figure S15) to a BE of
248.8 eV.[43] Further components (c2–c5) belong to oxidized
carbon species from adventitious carbon, if not otherwise
discussed below.

The calcium-free sample (Figure 6, a–c) is exhaustively
oxidized at the surface as apparent from the zirconium dioxide
peak, component z3, at a (BE of 182.8 eV (Figure 6a). This
correlates with the low signal intensity of the N 1s spectrum
(Figure 6b) and a high intensity of the oxide peak in the O 1s
spectrum, component o1, at 530.4 eV corresponding to zirco-
nium oxide (Figure 6c).[39] Slightly visible (oxy)nitride signals
could originate from an underlying major nitride phase as
determined by XRD (Figure 1b). Calcium admixture resulted in a
gradual decrease of the nitride phase and increasing oxynitride
phases as revealed by XRD. This trend is supported by XPS
measurements. The samples with a calcium admixture of Ca0.5
and Ca1.0 (Figure 6d–i) markedly show a mixture of nitride,
component z1 (180.1 eV and 180.5 eV), and oxynitride, compo-
nent z2 (181.2 eV and 181.1 eV), in their respective Zr 3d
spectrum (Figure 6d and g). This agrees with pronounced
contributions of the oxynitride, component n1 (396.2 eV and

Figure 6. XP spectra of samples with varying calcium admixture (Ca0.0, Ca0.5, Ca1.0, Ca2.0) pyrolyzed at 900 °C for 3 hours. The peak-deconvoluted high-
resolution Zr 3d, N 1s and O 1s spectra are shown by column: (a–c) Ca0.0, (d–f) Ca0.5, (g–i) Ca1.0, (j–l) Ca2.0

Wiley VCH Montag, 14.10.2024

2420 / 365733 [S. 278/286] 1

ChemCatChem 2024, 16, e202400613 (11 of 19) © 2024 The Authors. ChemCatChem published by Wiley-VCH GmbH

ChemCatChem
Research Article
doi.org/10.1002/cctc.202400613



395.5 eV), and nitride, component n2 (398.3 eV and 397.7 eV),
to the N 1s spectra (Figure 6e and h) as well as a decreased
contribution of the oxide (o1, 530.2 eV and 530 eV) to the O 1s
spectra (Figure 6f and i).[39,42] This trend of larger contribution of
nitride and oxynitride components to the XP spectra, and at the
same time diminished contribution of the oxide, with increasing
calcium admixture is coherent with the observed changes in
phase composition by PXRD i. e. increasing amount of oxy-
nitride with increasing calcium admixture. In summary, the
more oxynitride present in the sample, the less surface
oxidation occurs in relation to the sample without calcium. An
explanation for this observation is the decreasing oxidation
enthalpy of oxynitrides with decreasing nitrogen content.[44] The
oxynitrides do not oxidize as readily as the nitride.

In contrast to this trend, there are certain oxide components
visible again in the Zr 3d (z3, 182.1 eV) and O 1s (o1, 540.5 eV)
spectrum at the highest calcium admixture of Ca2.0 (Figure 6j
and l). The higher the calcium admixture, the slower the urea
decomposition driving the conversion of the oxide to the
nitride. In consequence, calcium admixture causes intermediate
oxynitrides of varying nitrogen content to remain. If the
conversion is slowed to an extent that the oxides are partially
not transformed to oxynitride nor nitride during the synthesis
in the first place, the oxidic precursor will partially remain and
its oxide components are detected at the high calcium
admixture. However, the N 1s spectrum (Figure 6k) still contains

a pronounced nitride signal (n2, 398.5 eV), even though there is
no corresponding signal in the Zr 3d spectrum (Figure 6l). It has
to be noted that the correction of all XP spectra by the C 1s
signal resulted in a large applied shift of the binding energies
here, which might have caused the missing nitride signal in the
Zr 3d signal. Regardless, this N 1s nitride signal can also be
related to carbon nitride components of the amorphous carbon
phase emerging in the UGM synthesis.[45] Pronounced C 1s and
N 1s signals at 288.2 eV and 398.7 eV respectively have been
reported for crystalline carbon nitrides e.g. C3N4. However,
graphitic C3N4 is not readily apparent in our diffractograms.
Furthermore, the C 1s spectrum (Figure S23d) also contains a
pronounced signal at 287.6 eV (c3) that can originate from
amorphous carbon nitride species in this sample. The larger
ratio of n2 to corresponding z1 component at Ca2.0 in relation
to Ca0.5 and Ca1.0 points to amorphous carbon as cause as
well.

The effect of the pyrolysis temperature on the surface
composition in the UGM and CN synthesis route was evaluated
by XPS next (Figure 7). An increased pyrolysis temperature of
1150 °C and 1400 °C facilitated the nitride formation independ-
ently of calcium admixture (Figure S15) and synthesis route
(Figure 1 and Figure 3). A pyrolysis temperature of 1150 °C
without calcium admixture resulted in a similar indication of
surface oxidation (Figure S25) as discussed for the analogue
900 °C sample. A small calcium admixture of Ca0.5 at 1150 °C

Figure 7. XP spectra of samples pyrolyzed at 1150 °C and 1400 °C via UGM and CN. The peak-deconvoluted high-resolution Zr 3d, N 1s and O 1s spectra are
shown by column: (a–c) UGM, Ca0.5, 1150 °C, (d–f) UGM, 1400 °C, (g–i) CN, C2, 4.5 h

Wiley VCH Montag, 14.10.2024

2420 / 365733 [S. 279/286] 1

ChemCatChem 2024, 16, e202400613 (12 of 19) © 2024 The Authors. ChemCatChem published by Wiley-VCH GmbH

ChemCatChem
Research Article
doi.org/10.1002/cctc.202400613



shows an increased intensity of the nitride component n2
(397.7 eV) (Figure 7b), whereas both the Zr 3d (z3, 182.4 eV) and
the O 1s spectrum (o1, 530.3 eV) reveal the presence of a
surface oxide layer. However, the in comparison higher intensity
n2 component points to a thinner surface oxide layer, which
yields more information from the underlying nitride phase.
Additionally, the high-resolution C 1s spectra at a pyrolysis
temperature of 1150 °C (Figures S25 and S26) show two addi-
tional components, c6 and c7, which correspond to graphite
and its π to π* satellite, which might originate from graphitic
carbon at the higher pyrolysis temperature of 1150 °C. A further
increased pyrolysis temperature of 1400 °C results in very similar
XP spectra (Figure 7d–f) but for the higher contributions of the
nitride components (z1, 179.9 eV and n2, 397.4 eV), indicating a
thinner oxide layer than at 1150 °C.

Finally, all three CN samples consist of a major ZrN phase by
PXRD (Table 2), which shows an expected surface oxidation and
thus similar XPS signals. However, the two samples with a
higher carbon ratio of C4 show a larger contribution from
residual graphitic carbon (c6, Figures S28b and S29b), both C4
samples contain significant carbon after the pyrolysis. The
significant shift to lower angles apparent in the diffractogram at
the lower dwelling time of 3 h (Figure S16) was attributed to a
solid solution of nitride and carbide as evaluated by Vegard’s
Law (Table 2). The lack of distinct zirconium carbide signals in
the XP spectra of this C4 sample substantiate the assumed solid
solution of nitride and carbide i. e. partial substitution of
nitrogen atoms by carbon instead of two distinct phases.[45] A
higher dwelling time of 4.5 h rectified this undesired carbide
fraction, and only a minor shift likely characteristic for the high
synthesis temperature of 1400 °C remained. The last CN sample
was synthesized with a halved carbon ratio of C2, because
significant carbon remained before at C4 and C2 is the
stoichiometric ratio. The reduced carbon ratio resulted in
carbon free ZrN, albeit with a minor oxide phase by PXRD. The
C2 sample also shows an oxidized surface (Figure 7) as apparent
in the Zr 3d (z3, 182.2 eV) and O 1s (o1, 530.0 eV) spectrum
respectively. Interestingly, the contribution of the nitride phase
is clearly visible in the Zr 3d (z1, 180.1 eV) and N 1s (n2,
397.5 eV) spectrum and is also stronger compared to the other
ZrN materials synthesized by UGM at 1150 °C and 1400 °C.
Either the minor zirconium dioxide phase in the UGM samples
contributes majorly to the XPS signals or a thinner surface oxide
layer of the CN sample allows for a larger contribution of the
underlying ZrN phase. In summary, the recorded XP spectra
show the expected surface oxidation of the synthesized ZrN
materials. The nitride phase (component n2) contributes to a
different extent to the signals depending on the sample. This
varying contribution, although all samples consist of ZrN as
determined by XRD, is believed to be caused by a different
surface oxide layer thickness. This hypothesis will have to be
investigated further by HR-TEM measurements of suitable
specimens. A reduced surface oxide formation was found for
samples synthesized by Ca-assisted UGM, which has been
related to the shift to less easily oxidized oxynitride materials by
calcium admixture.

2.8. Electrochemical Characterisation – Nitrogen Reduction
Reaction Activity

TMNs are potentially promising NRR catalysts because it is
hypothesised that the NRR follows a Mars-van-Krevelen (MvK)
mechanism on these catalyst surfaces. A MvK mechanism is
characterised by the participation and incorporation of surface
atoms in the reaction and product, which are replenished
subsequently, concluding the catalytic cycle.[7] In case of the
NRR, lattice nitrogen is directly protonated forming ammonia,
which is released. The lattice nitrogen vacancy is replenished by
gaseous nitrogen. The assumption is that replenishment of
these lattice nitrogen vacancies, either as monomer or dimer, is
energetically favourable compared to the sole adsorption of
nitrogen on the surface of catalysts following the associative
Heyrovsky-mechanism.[6] Therefore, TMN NRR catalysts should
be more selective and active in theory. However, the primary
competing HER reaction has a similar theoretical onset potential
as the NRR and is kinetically favoured.

The electrochemical behaviour of selected synthesised
zirconium nitride materials was exemplarily investigated to
qualitatively deduce potential activity for the electrochemical
NRR. A quantitative investigation of the electrochemical NRR
activity, following current state-of-the-art protocol including
ammonia quantification after prolonged turnover experiments
and isotope-labelled experiments, was not within the scope of
this work.[5] Instead of this conducted potentiodynamic experi-
ments give insights to the general electrochemical behaviour
and possible contributions of the material’s NRR activity to the
measured HER current density. These results are important
prerequisite for subsequent quantitative investigations in the
future, demonstrating their NRR potential, but can up to here
not answer if a material is active for the NRR or not. To this end,
cyclic voltammograms (CV) and linear sweep voltammograms
(LSV) in argon, inert to NRR, and nitrogen, educt gas, were
performed in a relevant potential region. All potentials are in
reference to the used reversible hydrogen electrode (RHE). The
upper potential was limited to +0.6 V vs. RHE in all measure-
ments, which corresponds approximately to the initial open
circuit potential, to avoid potential oxidation of the nitride
material. Electrochemical (EC) cleaning of nitride catalysts has
been described in literature to remove the surface oxide layer
prior to investigations of the potential NRR activity.[8] EC
cleaning was performed, because the synthesised materials
showed a surface oxide layer as determined by XPS. EC cleaning
was done by performing several fast CV scans (250 mVs� 1,
~100 cycles) in argon-saturated electrolyte and supplying the
GDE with inert argon gas as well. Figure 8 illustrates the effect
of the EC cleaning, showing the initial state before the cleaning
(red line) and, in comparison, the steady state after cleaning
(blue line). The reductive currents at low (more negative)
potentials are related to the occurring HER, with a maximum of
� 2 mAcm� 2 at � 0.6 V in the initial state. No further pro-
nounced signals are visible besides a small hump at ~0.0 V in
the cathodic and at ~0.2 V in anodic potential scan, respec-
tively. The EC cleaning resulted in an increased HER onset
potential, generally narrowed HER region (see arrows in Fig-

Wiley VCH Montag, 14.10.2024

2420 / 365733 [S. 280/286] 1

ChemCatChem 2024, 16, e202400613 (13 of 19) © 2024 The Authors. ChemCatChem published by Wiley-VCH GmbH

ChemCatChem
Research Article
doi.org/10.1002/cctc.202400613



ure 8) and a lower maximum reductive current of � 1.5 mAcm� 2,
indicating a diminished HER activity with consecutive scans.
There is no significant change between scans after EC cleaning
in the positive potential region from 0 V to the chosen upper
limit of +0.6 V, showing a stable state. The continuous
reduction of the surface oxide species during the EC cleaning
seems to have reduced the HER activity, explaining the
observed quickly-diminished reductive currents within the
initial consecutive scans. Furthermore, NRR should initially occur
in inert argon gas, due to the assumed MvK mechanism, until
active lattice nitrogen is depleted, as there is no nitrogen gas
supplied for replenishment. To qualitatively deduce potential
activity for the electrochemical NRR, CV and LSV analysis were
compared by supplying reactive nitrogen to the GDE in
separate measurements of the same sample, while the electro-

lyte was furthermore saturated with argon (Figure 9). Only the
backwards scan of the third cycle of each CV is shown for the
sake of clarity.

The first material, 1150 °C Ca0.5 (Figure 9a and d), was
chosen for its high nitride and phase purity, low residual carbon
content and high crystallinity. The material pyrolyzed at 900 °C
with a calcium admixture of Ca0.5 (Figure 9b and e) was chosen
in comparison, because it contains more minor phase, is less
crystalline and contains more residual carbon. The last chosen
material was pyrolyzed at 900 °C, but had the highest calcium
admixture of Ca2.0. This material consists mainly of oxynitride
phases and likely contains the least amount of carbon. The
generally observed behaviour will be exemplarily described for
the first measured sample of the 1150 °C Ca0.5 material
(Figure 9d, GDE� A). The maximum reductive current was about
� 3.5 mAcm� 1 in inert argon (red line) after EC cleaning. This
current is likely caused solely by HER as discussed before.
Switching the supplied gas to reactive nitrogen (blue line), the
reductive current increased to approximately � 7 mAcm� 2,
which is on the order of magnitude seen before. In addition,
there was a crossover of currents at approximately � 0.3 V in
the complete CVs (not shown). This current crossover, a higher
current at the same potential in the anodic scan than the
cathodic scan, is characteristic for a higher activity of the
reduced state than the oxidised state for both measurements.
An electrochemical NRR occurring simultaneously to the HER
supplying reactive nitrogen can explain the lower onset
potential and increased reductive currents compared to inert
argon. A similar behaviour was observed by Hanifpour et al.
investigating a polycrystalline ZrN film and other TMNs.[8,46–48] A
higher reductive current in reactive nitrogen than inert argon
was observed for all tested samples and materials. The 900 °C
samples with Ca0.5 and Ca2.0, which contain oxynitrides to
differing amounts, showed higher current density in LSV

Figure 8. Electrochemical cleaning of ZrN-GDE (UGM 1150 °C, Ca0.5) in 0.1 M
sulfuric acid electrolyte: Cyclic voltammetry (100 mVs� 1) in argon-fed electro-
lyte before (red line) and after electrochemical (EC) cleaning (blue line).

Figure 9. Electrochemical characterization of GDEs of selected synthesized zirconium nitride materials. The top row shows the CVs in argon after activation: (a)
1150 °C, Ca0.5, (b) 900 °C, Ca0.5, (c) 900 °C, Ca2.0. The bottom row shows the material behavior in argon (red) and nitrogen (blue) as qualitative comparative
measure of the NRR activity: (d) 1150 °C, Ca0.5, (e) 900 °C, Ca0.5, (f) 900 °C, Ca2.0.

Wiley VCH Montag, 14.10.2024

2420 / 365733 [S. 281/286] 1

ChemCatChem 2024, 16, e202400613 (14 of 19) © 2024 The Authors. ChemCatChem published by Wiley-VCH GmbH

ChemCatChem
Research Article
doi.org/10.1002/cctc.202400613



investigation (Figure 9e and f). Speculatively, oxygen neigh-
bouring nitrogen atoms or vacancies in the nitride lattice could
influence single mechanistic reaction steps resulting in an
energetically beneficial NRR via MvK, which was postulated for
vanadium oxynitride in literature.[49] However, there is a large
variance within and between materials. The current increased
on average by 75�13%, 252�194% and 228�120% for the
three materials. This increase in current is substantial and
promising if it is caused by NRR. Still, further understanding of
the material, reaction and electrochemical system is required,
especially in light of the different EC behaviour of one sample
to the next within one batch.

Furthermore, this qualitative comparison of reductive
currents is limited in its significance. A proof of genuine
electrochemical NRR to date requires a quantitative determi-
nation of formed ammonia by trace analysis, as well as isotope-
labelled experiments, proving the supplied gas as nitrogen
source of the ammonia, which is both beyond the scope of this
study. In addition, further pursuit of robust electrochemical
protocols that avoid false positive results due to contaminations
and other effects is required.[5,50] Moreover, TMN are also
discussed to undergo non-catalytic decomposition resulting in
ammonia formation instead of genuine NRR.[51] Thus, the state-
of-the-art experimental data on genuine NRR activity of TMNs is
as yet inconclusive, but they remain promising due to the great
catalytic potential for the NRR via the MvK mechanism.[52]

Consequently, the theoretically predicted activity and onset
potential for ZrN, depending on crystal structure and facet, are
yet to be confirmed in literature.[7] More in-depth electro-
chemical characterisation coupled with quantitative trace
analysis of the electrolyte (and gases) for ammonia is needed to
further understand and elucidate possible discrepancies be-
tween theory and practice, which will be the scope of further
studies. Equally important, reproduction and reproducibility of
determined NRR activity must be a cornerstone of the
continued electrochemical NRR research, especially on TMNs.
The synthesised ZrN materials are interesting, but remain to be
proven active for a catalytic electrochemical ammonia produc-
tion.

3. Discussion

3.1. Zirconium Nitride Synthesis

The UGM allows for easy and accessible synthesis of nano-
particulate ZrN at moderate temperatures. The formation of
zirconium nitride requires a minimum pyrolysis temperature of
900 °C (literature �840 °C) to form from an intermediate oxy-
nitride phase (Figure 1).[9] Oxide and oxynitride phases occur
alongside the desired ZrN depending on the composition and
pyrolysis conditions i. e. pyrolysis temperature (Figure 1), cal-
cium admixture (Figure 2), urea content (Figures S9–S11) and
pyrolysis crucible (Figures S12–S15). Presumably, amorphous
zirconium dioxide is progressively nitrided by reactive nitrogen
species, which are generated during the decomposition of urea.
This conversion results in oxynitrides with increasing nitrogen

content (β’, β, γ). The irreversible decomposition of this nitro-
gen-rich oxynitride γ-phase yields nitride and a less nitrogen-
rich oxynitride following the quasi-binary phase diagram
determined by M. Lerch et al.[9] This resulting oxynitride is in
turn nitrided under the right conditions, until it also decom-
poses into nitride and oxynitride. This cycle continues until the
nitridation reaction ceases or the oxide has been completely
converted. This process explains the observed minor and major
oxide and oxynitride phases and their trends with added
calcium ions (Figure 2), which slows urea decomposition.[15]

Urea decomposition must generate a critical local concentration
of reactive nitrogen species for nitride to occur, which was
shown by the occurrence of nitride at the lower urea content of
U5. Nitride did form, but only as a minor phase (Figure S9).
Moreover, this nitriding atmosphere must be sufficient in
concentration over time, the generation neither too fast nor too
slow for extensive conversion. A small calcium admixture results
in a higher average concentration in time, reducing the loss of
reactive nitrogen species, resulting in more extensive conver-
sion to the nitride. Already-formed particles grow for a longer
time, forming larger more crystalline particles. A large calcium
admixture slows the generation of reactive nitrogen species so
much that the necessary critical concentration is not reached
locally. No nitride, and only less nitrogen-rich oxynitrides, form
in consequence, depending on the magnitude of calcium
admixture (Figure 2). The beneficial effect of using crucibles
with a closed lid works in the same way as a slightly-slowed
urea decomposition. The lid retains the nitriding atmosphere
for a longer time, resulting in more nitride-formation by better
utilisation of the atmosphere (Figure S13). A pyrolysis temper-
ature above 900 °C also has a beneficial effect on the conversion
of the oxide to the desired nitride, due to two effects. The first
effect is a different decomposition path of the γ-phase. The γ-
phase decomposes to nitride, nitrogen and oxide rather than
oxynitride above approximately 1000 °C, following the quasi-
binary phase diagram of M. Lerch (Equation (4)).[9] The theoret-
ical yield is larger by a third, again assuming a stoichiometric
decomposition (Equation (5)). The other effect at an even higher
temperature of 1400 °C is the partial nitridation of zirconium
dioxide directly by the nitrogen atmosphere.[26] This source of
nitrogen is continuous over time, assisting the synthesis by the
UGM. However, the improved diffusion rates at elevated
temperatures must not be disregarded. The possibility of
directly (partially) nitriding zirconium dioxide by nitrogen at
temperature of �1400 °C also enables the carbothermal
nitridation (CN), which then becomes an alternative synthesis
route.

4 Zr2ON2 ! 2 ZrO2 þ 6 ZrNþ N2 (4)

6 MZrN

6 MZrN þ 2 MZrO2

¼
6 � 105:2307

6 � 105:2307þ 2 � 123:222 � 72% (5)

Carbothermal nitridation of zirconium was successful at a
temperature of 1400 °C in nitrogen using a molar ratio of
carbon black to zirconium dioxide of 4. A significant amount of
ZrC determined by Vegard’s Law (ZrN0.63�0.17C0.37�0.17) remained

Wiley VCH Montag, 14.10.2024

2420 / 365733 [S. 282/286] 1

ChemCatChem 2024, 16, e202400613 (15 of 19) © 2024 The Authors. ChemCatChem published by Wiley-VCH GmbH

ChemCatChem
Research Article
doi.org/10.1002/cctc.202400613



at a dwelling time of 3 hours. Zirconium carbide is an
intermediate phase in the carbothermal nitridation of
zirconium.[37] This undesired phase was greatly reduced by
increasing the dwelling time to 4.5 hours (ZrN0.90�0.05C0.10�0.05),
which confirms it is indeed an intermediate to the nitride.
Adding carbon in excess of 4 to the theoretical molar ratio of 2
(Equation (3)) caused undesired carbon black to remain in the
product with a mass fraction of roughly 16 to 20% (Table 2).
This corresponds to approximately half of the carbon added to
the zirconium dioxide, indicating that the theoretical molar
ratio of 2 should suffice. A molar ratio of 2 at a dwelling time of
4.5 hours consequently resulted ZrN (ZrN0.91�0.05C0.09�0.05) free of
residual carbon, albeit with a minor zirconium dioxide fraction.
A slight adjustment of the carbon ratio above 2 could lead to
phase-pure ZrN without residual carbon.

Comparing the synthesised material properties to literature
is hampered by differing synthesis routes, different synthesis
conditions, choice of suitable comparator and available liter-
ature. However, Das et al. used a synthesis route that slots in
between UGM and CN by directly pyrolyzing mixtures of
zirconium tetrachloride and urea at high temperature and
evaluating the nitride stoichiometry by Vegard’s Law as well.[35]

They found trends in occurring phases and nitride purity which
match our own observations. A sole major zirconium nitride
phase contains some carbon as zirconium carbide in the crystal
lattice as determined by Vegard’s law. The achieved stoichiom-
etry of ZrN0.91C0.09 at 1400 °C by Das et al. fits our findings well,
although derived by use of different reference lattice constants.
An occurring minor nitride phase alongside other major phases
is purer than a (sole) major nitride phase. Both a critical urea
content and pyrolysis temperature are required for nitride to
form. The minor phases consist of tetragonal and monoclinic
zirconium dioxide. The higher the pyrolysis temperature, the
more carbide forms. They also found similar amounts of residual
carbon remaining in the product, which is located around
nitride particles in TEM images. Overall, the material properties
are comparable, although the synthesis routes differ greatly.

3.2. Comparison of Synthesis Routes

Both synthesis routes, UGM and CN, are successful in synthesis-
ing nanoparticulate zirconium nitride with certain advantages
and disadvantages. UGM enables a synthesis of zirconium
nitride starting at a moderate temperature of 900 °C using
cheap, safe to handle, non-toxic urea. The method also allows
to synthesise γ-phase oxynitride by adjusting process parame-
ters. However, a high urea content is necessary for high nitride
yield, leading to a small absolute yield and undesired residual
carbon in the pyrolysis product. Residual minor phases are a
complex, likely oxide-oxynitride mixture. A small calcium
admixture improves the synthesis. It reduces the amount of
minor phase and residual carbon as well as improving
crystallinity. Having said this, the admixture requires sample
purification, while some calcium likely remains in the sample.
The higher crystallinity is a consequence of likely larger
particles, which is undesired. Similarly, the use of zirconium

tetrachloride as zirconium precursor results in some residual
chloride. Higher pyrolysis temperatures reduce minor phases as
well, but decrease nitride purity, increase ZrC incorporation,
and CN becomes a feasible alternative method at high temper-
atures (�1400 °C). Synthesising high-purity ZrN without any
minor phases or residual carbon may not be possible using
UGM alone. CN enables synthesising zirconium nitride at
1400 °C in a nitrogen atmosphere using readily available nano-
particulate zirconium dioxide and a suitable carbon source such
as carbon black. Tuning the amount of added carbon as oxygen
getter and dwelling time might enable a synthesis of pure ZrN
without either minor phase or residual carbon. However, the
required high temperature, and reaction mechanism, likely
results in larger nanoparticles and a slightly lower purity than
using UGM.

Optimising the zirconium nitride synthesis for high purity of
the nitride, phases and product generally has the same
constraints for both used methods, a perfectly pure nitride was
hitherto not achievable in this work without any minor phases
or some residual carbon. Further optimisation will be carried
out to this end.

4. Conclusions

The urea-glass method (UGM) was systematically investigated
in this study as a simple and accessible synthesise route for
nanoparticulate zirconium nitride (ZrN) for the prospective use
as an electrocatalyst. The synthesis of ZrN by UGM follows the
quasi-binary phase diagram of ZrO2–Zr3N4, occurring via irrever-
sible decomposition of the γ-phase oxynitride Zr2ON2 starting
above 800 °C. An increased pyrolysis temperature reduces the
amount of minor phase, but results in reduced nitride purity
and likely larger particle size. Addition of calcium ions affects
the urea decomposition, improving the nitride conversion in
small amounts, reducing the amount of residual carbon and
improving crystallinity, but likely due to larger particles. A
simple carbothermal nitridation (CN) of commercial nanoparti-
culate zirconium dioxide with carbon black was successfully
used in comparison to UGM. An adjusted pyrolysis composition
resulted in nanoparticulate ZrN of similar purity without
residual carbon, but with some remaining minor oxide phase.
Regardless of the synthesis method, an increased synthesis
temperature reduces the amount of minor phase but increases
the particle size and seems to be accompanied by lower nitride
purity. Both synthesis methods resulted in similarly pure nitrides
but the utilized analysis methods are insufficient to exhaustively
determine the material composition in regard to the elemental
composition, as nitrides are easily contaminated by carbon and
oxygen. An elemental analysis suitable for these refractory
materials is needed to determine the absolute content of
nitrogen, oxygen and carbon to account for their effect. The
initial qualitative investigation of the ZrN nanoparticles synthe-
sized by UGM for NRR was promising, there is substantially
higher reductive current in nitrogen than argon, but further
quantitative measurements of the electrochemically-produced
ammonia must follow.

Wiley VCH Montag, 14.10.2024

2420 / 365733 [S. 283/286] 1

ChemCatChem 2024, 16, e202400613 (16 of 19) © 2024 The Authors. ChemCatChem published by Wiley-VCH GmbH

ChemCatChem
Research Article
doi.org/10.1002/cctc.202400613



Methods

Synthesis – Urea-Glass Method (UGM)

Zirconium tetrachloride (ABCR, AB122360, 99.5% Zr) powder is
weighed into a three-neck round-bottom flask in a glovebox. The
sealed flask is transferred to a Schlenk line. Dried ethanol (Sigma-
Aldrich, 24102, 99.8%) is added (45 mLg� 1ZrCl4) and stirred at room
temperature until complete dissolution of the powder.[23] Then urea
(Sigma-Aldrich, 33247–250G, �99.5%) is added to the solution in a
set molar ratio to the zirconium of 5 or 15 (U5, U15). The urea
containing solution was stirred for 2 h to allow sufficient time for
dissolution and complexation. Calcium chloride dihydrate (Sigma-
Aldrich, 223506, �99%) is added in a set molar ratio to the
zirconium tetrachloride (Ca0.25, Ca0.5, Ca1.0, Ca2.0) after complete
dissolution of the added urea. This admixture of calcium chloride
dihydrate is simply referred to as ‘calcium admixture’ in the main
text. The formed solution is transparent, colourless and slightly
hazy. The final step is addition of 28 wt.% ammonium hydroxide
solution (Alfa Aesar, L13168, 28 wt.%) (2.1 mLg� 1ZrCl4) to induce
rapid crosslinking. The pulpy mixture is spread in a crystallization
bowl and dried over night at 60 °C. The dry mixture is scraped out
of the bowl and squashed into a fine, white, pourable powder. This
powdery pyrolysis mixture is stored in closed rolled-rim glass vials
until use.

Pyrolysis – UGM

The UGM pyrolysis mixture is poured into a crucible (glazed or
unglazed alumina boat, glassy carbon crucible, with or without lid)
and placed in a tube furnace (Nabertherm, RHTC 80–230/15,
alumina tube C799, alumina 99.7%, max. 1500 °C). Nitrogen (Linde,
5.0) is flushed through the tube at 100 Lmin� 1 (NTP) for 15 min
before the oven program is started. The sample is heated at
5 °Cmin� 1 to the final pyrolysis temperature (800, 900, 1150,
1400 °C) and kept there for the set dwelling time (2, 3, 4.5 h). The
cooling rate is 5 °Cmin� 1 until limited by natural cooling at lower
temperatures. The sample is left in the oven until room temper-
ature has been reached, only then the gas flow is switched off.

The pyrolysis product is scraped from the crucible and ground by
hand to a fine powder. This powder is washed with 10 mL of 1 M
hydrochloric acid in a 15 mL centrifuge tube using a roll mixer for 5
to 6 h to remove residual calcium species. The powder is separated
by centrifugation and washed three times with ultrapure water,
supernatant is decanted. Residual liquid is removed by drying at
60 °C overnight. This powder is used without further purification.

High-Temperature Synthesis – Carbothermal nitridation (CN)

Commercial nanoparticulate zirconium dioxide i. e. zirconia powder
(Sigma-Aldrich, 544760–5G, Zirconium(IV)oxide, nanopowder,
<60 nm particle size (TEM)) is weighed into a rolled rim glass vial.
Carbon black (Cabot, VULCAN® XC72) is added in a molar ratio to
the zirconium of 2 or 4. Three zirconia grinding balls (5 mm
diameter) are added and the mixture is homogenized overnight
using a roll mixer. The homogenized powder is used as pyrolysis
mixture after removal of the zirconia grinding balls.

The pyrolysis mixture is transferred into a glassy carbon (GC)
crucible on which a GC plate is placed as lid. The crucible is placed
into the high temperature oven (Carbolite Gero, LHTG 100–200/30,
max. 3000 °C). The oven is heated at 10 °Cmin� 1 to 1400 °C under a
nitrogen gas flow of 150 Lh� 1 (NTP) and kept there for a set
dwelling time (3, 4.5 h) under a nitrogen gas flow of 200 Lh� 1 (NTP).

The sample cooled down to room temperature with 10 °Cmin� 1

under this atmosphere.

Thermogravimetric Analysis (TGA)

Thermogravimetric analysis (TGA) was performed using a Perki-
nElmer TGA4000 with 180 μL aluminium oxide crucibles using
either a nitrogen (Linde, 5.0) or oxygen (Linde 6.0) gas flow of
40 mLmin� 1. The temperature programme is given in the support-
ing information (SI S4, S10).

X-ray Diffraction (XRD)

Powder X-ray diffraction (XRD) measurements were performed on a
Malvern Panalytical Empyrean Series 2 diffractometer in Bragg-
Brentano configuration with a Cu X-ray tube and Pixcel1D detector
(240 mm, Soller slits 0.04 rad, divergence slit 1/8°, anti-scatter slit
1/2°, PIXcel1D-Medipix 3 with FASS, Cu Kα1 0.1540980 nm, anti-
scatter slit AS slit 7.5 mm, 40 kV, 40 mA). Samples were measured
as coating on circular glass slides (32 mm diameter, 3 mm thick-
ness) if not specified differently. To this end, 20 mg of powder was
dispersed in 200 μL isopropanol using an ultrasonic bath. The
whole volume was aspirated using an Eppendorf pipette and drop-
cast onto the centre of the glass slide. The coating was left to dry at
room temperature. Diffractograms were recorded from 10° to 90°
2θ in 0.013° steps with an accumulation time of 800 s per step
spinning at 2 Us� 1 if not specified differently (total time per sample
5 h). The general aim was to achieve �50.000 counts for peaks with
�50% intensity and �5.000 counts for peaks of �5% intensity
inspired by the recommendations of the International Centre for
Diffraction Data (ICDD).

Transmission Electron Microscopy (TEM)

Samples were prepared using TEM grids (Plano, TEM SF 162, HR-
TEM S 162) with a carbon and, in case of HR-TEM, formvar coating.
1 mg of sample powder was dispersed in 4 mL ethanol absolute
using an ultrasonic bath. 8 μL of homogeneous dispersion was
dropped onto a grid and dried at room temperature. TEM measure-
ments were conducted using a Zeiss EM902 A with tungsten
electrode at an acceleration voltage of 80 kV. High resolution (HR)
TEM measurements were performed using a Joel JEM2100 F with a
Schottky-emitter at an acceleration voltage of 200 kV. Energy-
dispersive X-ray spectroscopy (EDS) was done using an Oxford INCA
Energy TEM250 EDS-system with a silicon drift X-ray detector
MAX80.

X-ray Photoelectron Spectroscopy (XPS)

X-ray photoelectron spectroscopy (XPS) was performed using a
Thermo Fisher Scientific ESCALAB 250Xi with monochromatic Al Kα
photon radiation (hv=1486.6 eV). Pristine powder samples were
pressed into the cavity of an aluminium sample holder. Survey
scans were collected in triplicate (100 eV, 1 eV step size, 10 ms).
High resolution spectra of Zr 3d, N 1s, O 1s and C 1s were collected
separately (10 eV, 0.02 eV step size, 50 ms) with 5 scans per Element
with exception of N, for which 10 scans performed. The flood gun
was used to compensate charging effects of potentially oxidized
surfaces. Samples were measured without prior sputter-cleaning.
All spectra were charge corrected to the aliphatic carbon signals
(component c1 in respective C 1s spectra) at a binding energy of
248.8 eV. Peak deconvolution analysis was performed by a product
of Gaussian and Lorentzian Peak shapes with a Shirley background
using the Avantage software (version 5.9925).

Wiley VCH Montag, 14.10.2024

2420 / 365733 [S. 284/286] 1

ChemCatChem 2024, 16, e202400613 (17 of 19) © 2024 The Authors. ChemCatChem published by Wiley-VCH GmbH

ChemCatChem
Research Article
doi.org/10.1002/cctc.202400613



Electrochemical Analysis (EC)

Electrochemical analysis was performed using a commercial
Gaskatel FlexCell-PTFE (SKU: 83100) measurement half-cell. A
classical three-electrode setup was employed using self-made gas
diffusion electrodes (GDE) as working electrode. GDEs of ZrN
material (U15, Ca0.5, 1150 °C ()) were spray-coated by hand using a
professional airbrush-pistol. A measured amount of ink (isopropyl
alcohol, ultrapure water, 1 : 1 V :V) was deposited on a vertically
fixed piece of gas diffusion layer (9×9 cm[2]) using a mask leaving an
area of 6×6 cm[2] uncovered. Target loading was 1 mgcm� 2 catalyst
material with 30 wt.% Nafion in the coating. More details regarding
the manual spray-coating and electrochemical setup are available
in the supporting information of a prior publication.[53] A PtIr-spiral
supplied with the cell was used as counter electrode. The reference
electrode consisted of a Gaskatel Mini-HydroFlex (SKU: 81020)
reversible hydrogen electrode (RHE). All potentials are reported
versus RHE scale. The experiments were controlled by a Metrohm
Autolab PG128 potentiostat using the manufacturer supplied
Nova® software version 2.1. The electrolyte consisted of 0.1 molar
high-purity sulfuric acid (Merck, 1.01516.0250, sulfuric acid 96%,
ultrapure) freshly prepared. The supplied N2 and Ar gases were
purged through 5 mM sulfuric acid electrolyte and ultrapure water
as basic cleaning step for potential NH3 contaminations before
entering the electrochemical cell. The experiments were performed
at room temperature.

The electrochemical test protocol consists of (1) cyclic voltammetry
(CV) measurements with a scan rate of 100 mVs� 1, (2) electro-
chemical (EC) cleaning by several fast CV cycles (250 mVs� 1, ~100
scans), (3) CV measurements (100 mVs� 1), (4) double layer capaci-
tance measurements and (5) slow CV measurements (10 mVs� 1).
The double layer capacitance measurements were performed in
faradaic current regime by CV. It was recorded at +0.2 V in a small
potential window (�0.1 V) at several scan rates (500, 200, 100, 50,
20, 10 mVs� 1). The initial measurement sequence (1–5) was
performed in Ar-saturated electrolyte to evaluate the electro-
chemical background signal. This was followed by repeating steps
(3) and (5) in N2-saturated electrolyte to evaluate potentially
increased currents due to ongoing NRR in addition to background
HER. Thus, potential NRR activity is qualitatively investigated as
initial step of a more sophisticated testing protocol, which is
required to prove genuine NRR activity and is beyond the scope of
this study.

Supporting Information Summary

Supporting information is available from the Wiley Online
Library or from the author.

Acknowledgements

This work was partially funded by the German Research
Foundation (Deutsche Forschungsgemeinschaft, DFG) priority
programme 2370, Project SUNRed (project number 460921994).

The powder X-ray diffraction and X-ray photoelectron
spectroscopy measurements were made possible by DFG
funding through grant number 276839650 and 251668893
respectively. Furthermore, the authors thank Prof. Wittstock for
access to the X-ray photoelectron spectrometer.

The authors acknowledge the Electron and Light Micro-
scopy Service Unit, Carl von Ossietzky University of Oldenburg,

for the use of the imaging facilities. The authors further thank
Heinrich Vocke (Carl von Ossietzky University of Oldenburg) for
conducting HR-TEM measurements and Jasmin Schmeling (DLR)
for conducting TEM measurements as well as evaluation of the
SAED images. The authors also thank Jana Ewert for her
assistance in operating and using the high temperature oven.
Open Access funding enabled and organized by Projekt DEAL.

Conflict of Interests

The authors declare no conflict of interest.

Data Availability Statement

The data that support the findings of this study are available
from the corresponding author upon reasonable request.

Keywords: Transition metal nitride · Zirconium nitrides ·
Nanoparticles · Calcium-assisted urea-glass method ·
Carbothermal nitridation · Gas diffusion electrodes ·
Electrochemical ammonia synthesis

[1] R. Schlögl, in The Chemistry and Physics of Solid Surfaces and
Heterogeneous Catalysis (Ed: J. R. Jennings), Springer Germany, 1995.

[2] L. E. Apodaca, Nitrogen (FIXED)–Ammonia, in Mineral Commodity
Summaries, U.S. Department of the Interior, U.S. Geological Survey,
2022, DOI: 10.3133/mcs2022.

[3] D. J. Ham, J. S. Lee, Energies (Basel, Switz.) 2009, 2(4), 873–899.
[4] H. Wang, J. Li, K. Li, Y. Lin, J. Chen, L. Gao, V. Nicolosi, X. Xiao, J.-M. Lee,

Chem. Soc. Rev. 2021, 50(2), 1354–1390.
[5] H. Iriawan, S. Z. Andersen, X. Zhang, B. M. Comer, J. Barrio, P. Chen, A. J.

Medford, I. E. L. Stephens, I. Chorkendorff, Y. Shao-Horn, Nat. Rev.
Methods Primers 2021, 1(1), 56.

[6] J. H. Montoya, C. Tsai, A. Vojvodic, J. K. Norskov, ChemSusChem 2015,
8(13), 2180–2186.

[7] Y. Abghoui, A. L. Garden, J. G. Howalt, T. Vegge, E. Skúlason, ACS Catal.
2015, 6(2), 635–646.

[8] F. Hanifpour, C. P. Canales, E. G. Fridriksson, A. Sveinbjörnsson, T. K.
Tryggvason, J. Yang, C. Arthur, S. Jónsdóttir, A. L. Garden, S. Ólafsson, K.
Leósson, L. Árnadóttir, E. Lewin, Y. Abghoui, Á. S. Ingason, F. Magnus,
H. D. Flosadóttir, E. Skúlason, J. Catal. 2022, 413, 956–967.

[9] M. Lerch, J. Mater. Sci. Lett. 1998, 17(6), 441–443.
[10] M. Banerjee, N. B. Srinivasan, H. Zhu, S. J. Kim, K. Xu, M. Winter, H.-W.

Becker, D. Rogalla, T. de los Arcos, D. Bekermann, D. Barreca, R. A.
Fischer, A. Devi, Cryst. Growth Des. 2012, 12(10), 5079–5089.

[11] H. Yamamura, S. Emoto, T. Mori, J. Ceram. Soc. Japan 1998, 106(1235),
650–653.

[12] S. Zhao, J. Ma, R. Xu, X. Lin, X. Cheng, S. Hao, X. Zhao, C. Deng, B. Liu, Sci
Rep. 2019, 9(1), 19199.

[13] A. J. Parkison, A. T. Nelson, J. Am. Ceram. Soc. 2016, 99(5), 1525–1533.
[14] C. Giordano, C. Erpen, W. Yao, B. Milke, M. Antonietti, Chem. Mater.

2009, 21(21), 5136–5144.
[15] Q. Gao, C. Giordano, M. Antonietti, Small 2011, 7(23), 3334–3340.
[16] M. Lerch, J. Wrba, J. Mater. Sci. Lett. 1996, 15(5), 378–380.
[17] K. Constant, R. Kieffer, P. Ettmayer, Monatsh. Chem./Chemical Monthly

1975, 106, 823–832.
[18] S. J. Clarke, C. W. Michie, M. J. Rosseinsky, J. Solid State Chem. 1999,

146(2), 399–405.
[19] T. Bredow, M. Lerch, Z. Anorg. Allg. Chem. 2007, 633(15), 2598–2602.
[20] T. Locherer, L. Dubrovinsky, H. Fuess, Solid State Commun. 2007, 143(8–

9), 408–411.
[21] R. J. Hill, L. M. D. Cranswick, J. Appl. Crystallogr. 1994, 27(5), 802–844.
[22] J. Málek, L. Beneš, T. Mitsuhashi, Powder Diffr. 1997, 12(2), 96–98.
[23] D. B. Williams, M. Lawton, J. Org. Chem. 2010, 75(24), 4.

Wiley VCH Montag, 14.10.2024

2420 / 365733 [S. 285/286] 1

ChemCatChem 2024, 16, e202400613 (18 of 19) © 2024 The Authors. ChemCatChem published by Wiley-VCH GmbH

ChemCatChem
Research Article
doi.org/10.1002/cctc.202400613

<REFPPF>10.3133</REFPPF>/mcs<REFPPF>2022</REFPPF>


[24] R. C. Mehrotra, Inorg. Chim. Acta, Rev. 1967, 1, 99–112.
[25] R. Srinivasan, B. H. Davis, O. B. Cavin, C. R. Hubbard, J. Am. Ceram. Soc.

1992, 75(5), 1217–1222.
[26] M. Lerch, J. Am. Ceram. Soc. 1996, 79(10), 2641–2644.
[27] Y. Yuan, J. Wang, S. Adimi, H. Shen, T. Thomas, R. Ma, J. P. Attfield, M.

Yang, Nat. Mater. 2019, 19(3), 282–286.
[28] R. Allmann, in Powder Diffraction: Theory, Practice, (Eds: R. E. Dinnebier,

S. J. L. Billinge), RSC Publishing, UK, 2008.
[29] A. A. M. Ali, M. I. Zaki, Thermochim. Acta 2002, 387(1), 29–38.
[30] J. Desmaison, Nitrogen diffusion in zirconium nitride (Doctoral disserta-

tion), 1973 (https://scholar.google.com/scholar?hl=de&as_sdt=0%
2C5&q=J.+Desmaison+1973&btnG=).

[31] M. Lerch, F. Krumeich, R. Hock, Solid State Ion. 1997, 95(1), 87–93.
[32] M. K. Motandi, Z. Zhang, S. Inkoua, L. Yan, Environ. Prog. Sustainable

Energy 2022, 41(2), e13744.
[33] J. Qu, L. Sha, Z. Xu, Z. He, M. Wu, C. Wu, Q. Zhang, Appl. Clay Sci. 2019,

173, 29–34.
[34] M. Lerch, E. Füglein, J. Wrba, Z. Anorg. Allg. Chem. 1996, 622(2), 367–

372.
[35] S. Das, D. Dubois, M. S. I. Sozal, Y. Emirov, B. Jafarizadeh, C. Wang, V.

Drozd, A. Durygin, Z. Cheng, J. Am. Ceram. Soc. 2022, 105(6), 3925–3936.
[36] L. M. Berger, W. Gruner, E. Langholf, S. Stolle, Int. J. Refract. Met. Hard

Mater. 1999, 17(1), 235–243.
[37] S. Zhao, R. Xu, J. Ma, C. Deng, X. Cheng, X. Zhao, S. Hao, J. Li, Mater.

Today Commun. 2021, 29, 102968.
[38] G. Greczynski, D. Primetzhofer, J. Lu, L. Hultman, Appl. Surf. Sci. 2017,

396, 347–358.
[39] T. Muneshwar, K. Cadien, Appl. Surf. Sci. 2018, 435, 367–376.
[40] P. Prieto, L. Galán, J. M. Sanz, Surf. Interface Anal. 1994, 21(6–7), 395–

399.

[41] M. Matsuoka, S. Isotani, W. Sucasaire, N. Kuratani, K. Ogata, Surf. Coat.
Technol. 2008, 202(13), 3129–3135.

[42] I. Milošev, H. H. Strehblow, M. Gaberšček, B. Navinšek, Surf. Interface
Anal. 1996, 24(7), 448–458.

[43] M. C. Biesinger, Appl. Surf. Sci. 2022, 597, 153681.
[44] I. Molodetsky, A. Navrotsky, F. DiSalvo, M. Lerch, J. Mater. Res. 2000,

15(11), 2558–2570.
[45] S. N. Katea, G. Westin, Ceram. Int. 2021, 47(8), 10828–10847.
[46] L. Zhang, X. Ji, X. Ren, Y. Luo, X. Shi, A. M. Asiri, B. Zheng, X. Sun, ACS

Sustainable Chem. Eng. 2018, 6(8), 9550–9554.
[47] F. Hanifpour, C. P. Canales, E. G. Fridriksson, A. Sveinbjörnsson, T. K.

Tryggvason, E. Lewin, F. Magnus, Á. S. Ingason, E. Skúlason, H. D.
Flosadóttir, Electrochim. Acta 2022, 403, 139551.

[48] R. Zhang, Y. Zhang, X. Ren, G. Cui, A. M. Asiri, B. Zheng, X. Sun, ACS
Sustainable Chem. Eng. 2018, 6(8), 9545–9549.

[49] X. Yang, S. Kattel, J. Nash, X. Chang, J. H. Lee, Y. Yan, J. G. Chen, B. Xu,
Angew. Chem., Int. Ed. 2019, 58(39), 13768–13772.

[50] J. Choi, B. H. R. Suryanto, D. Wang, H.-L. Du, R. Y. Hodgetts, F. M.
Ferrero Vallana, D. R. MacFarlane, A. N. Simonov, Nat Commun. 2020,
11(1), 5546.

[51] H.-L. Du, T. R. Gengenbach, R. Hodgetts, D. R. MacFarlane, A. N.
Simonov, ACS Sustainable Chem. Eng. 2019, 7(7), 6839–6850.

[52] Z. Qiao, D. Johnson, A. Djire, Science 2021, 2(5), 100438.
[53] S. C. H. Bragulla, J. Lorenz, C. Harms, M. Wark, K. A. Friedrich,

ChemSusChem 2023, 16(13), e202202211.

Manuscript received: April 19, 2024
Accepted manuscript online: June 20, 2024
Version of record online: August 23, 2024

Wiley VCH Montag, 14.10.2024

2420 / 365733 [S. 286/286] 1

ChemCatChem 2024, 16, e202400613 (19 of 19) © 2024 The Authors. ChemCatChem published by Wiley-VCH GmbH

ChemCatChem
Research Article
doi.org/10.1002/cctc.202400613

https://scholar.google.com/scholar?hl=de&as_sdt=0%2C5&q=J.+Desmaison+1973&btnG=
https://scholar.google.com/scholar?hl=de&as_sdt=0%2C5&q=J.+Desmaison+1973&btnG=

	Improving the Ca-assisted Urea-Glass Method for the Synthesis of Zirconium Nitride as Potential Electrocatalyst for the Nitrogen Reduction Reaction
	1. Introduction
	2. Results
	2.1. Ceramic Phases of Zirconium – Nitride, Oxynitride and Oxide
	2.2. Sol-Gel Synthesis
	2.3. Effect of Pyrolysis Temperature
	2.4. Effect of ‘Calcium’ Admixture
	2.5. High–Temperature Synthesis – Carbothermal Nitridation
	2.6. Influence of the Synthesis on observed Particle Morphologies
	2.7. Surface Composition of Zirconium (oxy)nitride Materials
	2.8. Electrochemical Characterisation – Nitrogen Reduction Reaction Activity

	3. Discussion
	3.1. Zirconium Nitride Synthesis
	3.2. Comparison of Synthesis Routes

	4. Conclusions
	Methods
	Synthesis – Urea-Glass Method (UGM)
	Pyrolysis – UGM
	High-Temperature Synthesis – Carbothermal nitridation (CN)
	Thermogravimetric Analysis (TGA)
	X-ray Diffraction (XRD)
	Transmission Electron Microscopy (TEM)
	X-ray Photoelectron Spectroscopy (XPS)
	Electrochemical Analysis (EC)

	Supporting Information Summary
	Acknowledgements
	Conflict of Interests
	Data Availability Statement