You are on page 1of 12

Journal of the European Ceramic Society xxx (xxxx) xxx

Contents lists available at ScienceDirect

Journal of the European Ceramic Society


journal homepage: www.elsevier.com/locate/jeurceramsoc

Interplay between decarburization, oxide segregation, and densification


during sintering of nanocrystalline TaC and NbC
Arseniy Bokov *, Anna Shelyug, Alexey Kurlov
Institute of Solid State Chemistry, Ural Branch of the Russian Academy of Sciences, Yekaterinburg, 91 Pervomaiskaya st., 620990, Russia

A R T I C L E I N F O A B S T R A C T

Keywords: The present study shows that ball-milled nanosized powders of TaC and NbC can be successfully sintered to high
Sintering densities at 1300− 1400 ◦ C for 30 min under vacuum. Fabricated ceramics demonstrate hardness of 20− 25 GPa
Tantalum due to decrease in carbon stoichiometry and submicron grain size. The main techniques to investigate the un­
Niobium
derlying phenomena during processing of initial powders were XRD, SEM and TGA-DSC. After milling, carbide
Carbide ceramics
Decarburization
particles demonstrate a significant amount of oxygen impurities but no signs of oxidation. During sintering, these
impurities react with structural carbon and metal ions, which results in decarburization and segregation of
oxides. Above 1000 ◦ C, the oxide phases undergo partial reduction by structural carbon, promoting decarburi­
zation even further. Densification starts shortly after the reduction of oxides and provides dense microstructures.
The effects of decarburization and oxide segregation can be compensated by carbon excess, however it can be
difficult to control densification curve in such case.

1. Introduction results in a non-uniform microstructure due to segregation of the Ti3O5


or Ti4O7 phases [14,15]. In all these cases the observed phenomena
The group 5 transition metal carbides TaC and NbC are characterized were driven by oxygen impurities that existed in starting materials
by strong covalent bonding and high hardness [1], but slow diffusion prior processing. Non-oxide particles always have an oxidized layer at
makes it difficult to sinter dense bodies without wetting by metallic phase. the surface to be in equilibrium with the ambient air, therefore it can
Therefore these compounds nowadays are used only as additives for adversely affect properties of the product fabricated from nano­
improvement of mechanical properties in WC [2], Ti(C,N) [3], steel [4], or powders as the total amount of oxygen is directly proportional to the
Ni-based [5] alloys. However, there are also some energy applications surface area [16].
where the use of TaC and NbC ceramics might be beneficial. For example, In the context of nanomaterials, the TaC and NbC compounds are
tantalum carbide could be employed as solar receiver (absorber) in characterized by very useful features. These carbides remain as single
concentrated solar power applications [6,7], whereas niobium carbide phase under different carbon stoichiometry according to homogeneity
could be used in fabrication of microencapsulated fuels [8]. Nevertheless, range [1], whereas WC transforms into W2C upon carbon loss [10].
sinterability issues of TaC and NbC should be addressed in the first place The segregation of oxide phases could be expected similarly to the TiN
before introducing these materials to new application areas. case [14], however these oxides might be reduced by structural carbon
The refinement of grain size has been known as an effective way to from the carbide lattice [17]. Additionally, the oxides of Ta and Nb
fabricate highly dense oxide ceramics at relatively low temperatures have not been reported to trigger abnormal grain growth, whereas
due to increase in the thermodynamic driving force for sintering [9]. SiO2 has been proven to promote this effect [12]. Therefore, tantalum
However, literature data shows that this approach might not be suit­ and niobium carbides could be more suitable for grain refinement and
able for non-oxide materials. For instance, thermal treatment of sintering in comparison with other non-oxide materials. The goal of
nanocrystalline WC leads to formation of W2C phase [10,11], which this study is to verify such hypothesis and investigate whether nano­
causes embrittlement of tungsten-based hard alloys. Sintering of crystalline TaC and NbC could provide dense ceramics with reasonable
nanosized SiC particles often provides porous samples because of the characteristics.
abnormal grain growth [12,13]. Densification of nanocrystalline TiN

* Corresponding author.
E-mail address: bokov@ihim.uran.ru (A. Bokov).

https://doi.org/10.1016/j.jeurceramsoc.2021.05.007
Received 10 March 2021; Received in revised form 24 April 2021; Accepted 4 May 2021
Available online 7 May 2021
0955-2219/© 2021 Elsevier Ltd. All rights reserved.

Please cite this article as: Arseniy Bokov, Journal of the European Ceramic Society, https://doi.org/10.1016/j.jeurceramsoc.2021.05.007
A. Bokov et al. Journal of the European Ceramic Society xxx (xxxx) xxx

2. Materials and methods fractions of metal in oxide Me2O5 and carbon in carbide MeCx. For Ta,
fmo and fcc are 0.819 and 0.062, respectively. For Nb, fmo and fcc are 0.699
The production of nanocrystalline powders in this study was per­ and 0.114, respectively. In case of fully stoichiometric TaC and NbC with
formed by high energy ball milling of coarse carbide particles. This x = 1, the mass gain g is equal to 1.145 and 1.267, respectively. The
approach was selected based on its affordability, effectiveness of grain initial mass multiplied by the mass gain and weight fraction of metal in
refinement [18], and relevance for recycling of carbide materials in the oxide essentially shows the actual weight of metal in the oxidized
framework of circular economy [19]. The initial powders of TaC (4 μm, sample. Since the amount of metal is the same in the oxidized and initial
99.5 %, 14.49 g/cm3) and NbC (1 μm, 99.5 %, 7.58 g/cm3) were pro­ samples, it should be subtracted from the initial mass in order to
vided by a local supplier Kirovgrad Hard Alloys Plant (KZTS). The calculate the actual weight of carbon in the initial sample. The bottom
received powders were processed in a Retsch PM 200 planetary ball mill part of the equation is a normalization of carbon weight with respect to
for 5, 10, and 15 h at 500 rpm. Milling was performed with jars and balls fully stoichiometric compound. The top and bottom parts of the equa­
made of tungsten carbide. Each jar was loaded with 10 g of initial tion both contain Mi parameter so it can be reduced for an ease of
powder, 100 g of 3-mm balls, and filled with 10− 15 ml of isopropanol calculation.
(99.9 %). After milling, the powders were dried in a Binder VDL 23 For the experiments on carbon compensation, a desired amount of
drying oven at 85 ◦ C and 10 torr. After drying, all powdered samples extra carbon black was added to the milling jar together with initial
were stored in a Sanplatec vacuum desiccator. For any further use, the carbide powders. After milling, the produced powders with excess car­
powders were extracted from this desiccator for a minimal amount of bon were processed and analyzed in the same way as described in pre­
time and then placed back with a proper evacuation. vious paragraphs.
Specific surface area SBET of the produced powders was measured by The SEM imaging of fractured surfaces for sintered ceramics was
BET method using a Micromeritics Gemini VII 2390 t analyzer. Prior BET performed on a JEOL JSM-LA-6390 at 10 mm working distance and 15
testing, powders were degassed at 350 ◦ C and 10− 2 torr for 1 h. The kV gun voltage. Epoxy molding, polishing and hardness testing for
obtained values of surface areas allowed for calculation of the average ceramic samples was performed with a Buehler set of Pneumet, Motopol,
particle size dBET according to the relation dBET = 6/(ρ⋅SBET), where ρ is and Micromet instruments, respectively. Specimens were fixed in a mold
density of a powder. The oxygen content in produced powders was by exposing Buehler Epomet compound to 5 min hold at 150 ◦ C under 20
measured via fusion under inert atmosphere using a Horiba EMGA- MPa pressure in a Pneumet machine. Then, molded samples were
620W/C oxygen analyzer. Samples of carbide powders were encapsu­ mounted in a Motopol machine and polished with 30, 6, and 1 um
lated by metallic Sn and dropped into a degassed graphite crucible. The diamond suspensions. After that, polished ceramics were subjected to
crucible was rapidly heated up to 2500 ◦ C to initiate evolution of gases. indentations with Vickers pyramid at 0.2, 0.5, 1 and 2 kgf of load using a
The CO gas was then extracted from the CO/H2 stream and quantified Micromet indenter.
with a calibrated infrared detector in order to calculate the initial oxy­
gen content. 3. Results and discussion
The obtained ball-milled powders of TaC and NbC were cold-pressed
into 3 × 3 mm cylindrical pellets at 250 MPa. After that, compacted As a result of ball-mill processing, six powdered samples were produced
pellets were sintered in a Centorr LF 22-2000 furnace under high vac­ from the initial TaC and NbC powders, i.e. three for each compound. Ac­
uum of 10− 5 torr in the range of 1000− 1500 ◦ C with the heating rate of 5 cording to SEM (Fig. 1a and b), the initial powders demonstrated irregular

C/min and holding time of 30 min at the maximum temperature. At particle morphology with the average size of 4.5 μm for TaC and 0.9 μm for
least three specimens for each milling time and for each composition NbC. After milling, the size of carbide particles is much smaller and their
were sintered in the same experiment in order to collect some statistics, morphology is more uniform (Fig. 1c and d). The XRD patterns for the
i.e. about 18 different pellets per furnace run. initial and produced powders are shown in Fig. 1e and f. The initial TaC
The structure and phase composition of studied materials were and NbC are characterized by Fm-3m (225) space group according to their
characterized at different stages (before milling, after milling, crushed diffraction profiles. The milled powders maintain the same cubic structure,
pellets after sintering) by powder X-Ray Diffraction (XRD) on a Shi­ however a significant peak broadening can be observed due to reduction in
madzu XRD-7000. The XRD analysis was performed with Cu-Kα1,2 ra­ grain size and strain accumulation during milling. The results for full-
diation in Bragg-Brentano geometry with 0.03◦ per step and 2 s/point profile Rietveld refinement and BET analysis for all powders are shown
exposure. Rietveld refinement was completed in JADE software to in Table 1.
calculate lattice constant, lattice strain, and XRD particle size dXRD. The XRD particle size varies in the range of 18− 39 nm and 23− 48
Crushing of sintered pellets was performed in a custom-made titanium nm for TaC and NbC, respectively, which is quite similar to the BET
mortar in order to avoid oxide contaminations. analysis. The lattice constants for milled powders are a bit smaller with
Simultaneous Thermo-Gravimetric Analysis and Differential Scan­ respect to initial materials possibly due to the lattice strain induced by
ning Calorimetry (TGA-DSC) were performed on a Netzsch STA 449 F3 ball-milling. The strain increases from 0.75 to 1.04 for TaC and from
Jupiter. For the analysis of sintering processes, 3 × 2 mm (diameter × 0.60 to 0.98 for NbC as milling time increases from 5 to 15 h. A slightly
height) cold-pressed cylindrical pellets were placed in Al2O3 crucible larger strain and smaller particle size for tantalum with respect to
and heated under Ar flow with 10 ◦ C/min heating rate followed by some niobium can be explained by higher efficiency of milling. Equal weights
isothermal exposure at a desired temperature. For the oxidation exper­ of TaC and NbC were loaded into jars, but the volume of powders was
iments, sintered pellets were finely crushed, placed into Pt crucible, and not the same due to difference in molar mass. The WC impurity from
heated up to 1000 ◦ C with 20 ◦ C/min rate under Ar + 10 % of O2 flow. milling (33◦ and 48◦ peaks on XRD patterns) is more visible for NbC
The carbon content in fabricated ceramics was calculated from the powders (Fig. 1f) than for TaC (Fig. 1e) because atomic scattering factor
mass gain during oxidation. The relation between these values can be is much lower for niobium than for heavier elements such as tantalum
derived from the oxidation reaction for a transition metal carbide: and tungsten. Overall, characteristics of the produced powders are fairly
typical for ball-mill processing [20,21].
MeCx + (5/2 + 2x) O2 = Me2O5 + 2x CO2 (1)
The produced nanocrystalline powders were cold-pressed to about
x = (Mi - Mi⋅fmo g / (fcc⋅Mi) = (1 - g⋅fmo) / fcc (2) 45–55 % green density and then sintered at different temperatures. The
obtained densification curves are shown in Fig. 2. At 1000 ◦ C, the actual
Here, x is carbon content in TaCx or NbCx, Mi is initial mass of carbide density for samples was in the range of 8.1–8.5 g/cm3 for TaC and
phase MeCx (total mass minus oxide impurity), g is mass gain during 5.2–5.4 g/cm3 for NbC, which correspond to about 57 % and 68 % of
oxidation (ratio of final and initial sample mass), fmo and fcc are weight relative density if theoretical values are taken as 14.5 and 7.8 g/cm3,

2
A. Bokov et al. Journal of the European Ceramic Society xxx (xxxx) xxx

Fig. 1. Characterization of initial and milled powders used in this study. SEM images for initial TaC (a) and NbC (b), SEM images for milled TaC (c) and NbC (d), XRD
patterns for all TaC (e) and NbC (f) powders.

respectively [1]. However, sintering process significantly intensifies at from powders milled for 5, 10, and 15 h demonstrate pores trapped
higher temperatures and density values reach plateau at 1400 ◦ C for TaC within the volume of individual grains, pores of mixed character, and
and at 1300 ◦ C for NbC. The maximum densities for pellets sintered from pores existing at triple junctions in ceramics produced, respectively.
the 5-hs-milled powders were 14.0 and 7.6 g/cm3 for TaC and NbC, The microstructure of samples sintered from the 5-hs-milled powders
corresponding to 96 % and 97 % of theoretical densities. The seems to be consistent with calculated relative densities of
15-hs-milled powder provided only 13.5 and 7.4 g/cm3, being equal to 96–97 %. However, pellets sintered from the 15-hs-milled powders also
93 % and 94 %, respectively. look fairly dense, which is not consistent with their relative densities of
The SEM images of fractured surfaces for the TaC/NbC ceramics 93–94 %. The observed amount of porosity is small and cannot explain
sintered at 1400/1300 ◦ C are shown in Fig. 3. The microstructure is such discrepancy. Therefore it was hypothesized that phase composition
uniform for all types of pellets demonstrating sharp equiaxial grains. The of some samples changed during sintering toward phases with smaller
average grain size decreases from 0.7 to 0.4 μm for TaC and from 1 to 0.6 weight, similarly to oxide segregation in nanocrystalline TiN [14,15].
μm for NbC as milling time increases from 5 to 15 h. Although all sin­ To gain insight on the phase composition, all sintered pellets were
tered pellets have some residual porosity, its type varies as a function of crushed and analyzed by XRD. Fig. 4 shows that produced carbide ce­
starting particle size/milling time. The TaC and NbC ceramics produced ramics mainly consist of the same cubic Fm-3m (225) structure inherited

3
A. Bokov et al. Journal of the European Ceramic Society xxx (xxxx) xxx

Table 1
BET and XRD characterization for powders used in this study.
Particle size
Surface area Lattice constant Lattice strain
Powder description dBET dXRD
(m2/g) (Å ± 0.0005) (%)
(nm) (nm)

TaC
initial 0.09 ± 0.01 4503 ± 50 n/a 4.4519 n/a
milled-5 h 10.5 ± 0.2 39 ± 2 39 ± 3 4.4481 0.75 ± 0.05
milled-10 h 15.4 ± 0.3 27 ± 1 27 ± 2 4.4469 0.91 ± 0.06
milled 15 h 21.1 ± 0.4 19 ± 1 18 ± 1 4.4430 1.04 ± 0.06
NbC
initial 0.84 ± 0.01 907 ± 10 n/a 4.4655 n/a
milled-5 h 15.7 ± 0.2 49 ± 2 48 ± 4 4.4646 0.60 ± 0.04
milled-10 h 22.5 ± 0.3 34 ± 1 33 ± 3 4.4626 0.82 ± 0.05
milled 15 h 28.2 ± 0.4 27 ± 1 23 ± 2 4.4613 0.98 ± 0.05

from the initial powders, however traces of oxide phases can also be observations with the results of present work and other studies [10,12],
noticed. Moreover, the intensity of oxide reflections gradually increases it can be concluded that oxygen is not visible in carbide nanopowders
with the increase in milling time of starting powders. Another important from the XRD/TEM point of view.
note is that oxidation state is not identical in the Ta- and Nb-based ox­ In the case of transition metal carbides like TaC or NbC, difficulties in
ides. The segregation of Ta2O5 with C2mm (38) space group is observed detecting oxygen impurities are probably related to uniform dissolution
in the TaC ceramic, whereas NbO2 with I41/a (88) forms in the NbC of O in the MeCx lattice via formation of the oxycarbide phase [24]. The
case. The full-profile Rietveld refinement estimated the oxide content as existence of oxycarbide implies that nanocrystalline particles should
4− 10 wt.%. have TaOy and NbOy composition in the surface layers and TaCxOy and
Segregation of oxide impurities along the grain boundaries could NbCxOy as overall composition. Such a surficial oxycarbide is indistin­
explain the decrease in grain size for ceramics produced from the guishable from the parent carbide by XRD or TEM because oxygen atoms
powders with longer milling time (Fig. 3) because large amount of are located at the carbon sites. In the TiN case, oxynitride also maintains
secondary phase can inhibit grain growth. Also, such oxide phases might parent rock-salt structure as long as O/N ratio is below 0.55 [25]. The
be a reason of why actual density of sintered pellets noticeably deviates SiC case is different from the transition metals because oxygen is stored
from theoretical values (the XRD density is around 8.4 g/cm3 for Ta2O5 in amorphous SiO2 coating and thus can be detected by microscopy [13].
and 5.6 g/cm3 for NbO2). To understand why sintering of TaC and NbC was successful despite
Since all ceramics in this study were sintered under high vacuum, the presence of oxygen impurities, the milled powders were thoroughly
segregation of oxide phase was possible only due to preexisting oxygen. investigated by TGA-DSC analysis. Fig. 5 shows temperature profiles of
For further verification of this hypothesis, all powdered samples were the experiments as well as associated mass changes and heat signals as a
analyzed in terms of their oxygen content by the inert gas fusion method. function of time. The samples of tantalum and niobium carbides were
As it can be seen in Table 2, the milled TaC and NbC have 3− 5 and 5− 8 subjected to isothermal hold at 1400 ◦ C and 1300 ◦ C (Fig. 5a and b),
wt.% of oxygen, respectively. These values are much larger than for the respectively, because sintering experiments provided the highest den­
initial powders, indicating that oxygen uptake occurred either during sities under these conditions.
milling (from isopropanol) or after milling (from ambient air). Although The TGA curves (Fig. 5c and d) show that all samples demonstrate two
such amount of extra oxygen should imply noticeable oxidation, the stages of mass loss. The first stage, evaporation of water, can be seen from
diffraction profiles for starting powders in Fig. 1e and f do not show any 0 to about 40 min of the experiment. It extends up to 400 ◦ C because
traces of oxides (to exclude time factor, XRD patterns and oxygen con­ analysis is performed on densely compacted powders. The second stage,
tent were obtained simultaneously shortly after milling). which is the evolution of oxygen in the form of CO gas [16], starts at about
The presence of XPS peaks for Ta-O and Nb-O bonding have been 80 min (700 ◦ C) and ends a little before isotherm around 125− 135 min.
reported for nanosized TaC and NbC [22,23], but no oxide film has been These two stages were assigned to release of H2O and CO based on
detected by TEM at the surface of such particles. Combining these mass-spectroscopy data from a previous study, where ball-milled TaC and

Fig. 2. Density of pellets as a function of sintering temperature for TaC (a) and NbC (b). The X-axis refers to the maximum temperature reached during sintering
followed by 30 min of isothermal hold. The legends refer to different starting particle size/milling time.

4
A. Bokov et al. Journal of the European Ceramic Society xxx (xxxx) xxx

Fig. 3. SEM images of fractured surfaces for TaC (a, b, c) and NbC (d, e, f) ceramics sintered from powders with different milling time.

Fig. 4. XRD patterns for initial powders and sintered/crushed pellets of TaC (a) and NbC (b).

NbC demonstrated identical TGA behavior [20]. Such a process of carbon deflection in the TGA curves). The fourth effect (IV) is an exothermic
loss via CO release, often called as “decarburization” [26,27], has been process that starts at 110 min (1100 ◦ C) and finishes at isotherm around
also detected by the decrease of lattice constant after annealing of TaC 135− 145 min.
nanopowders [17]. The effect II is assigned to segregation of oxide phase due to a
The analysis of the DSC signals (Fig. 5e and f) allows for a more characteristic temperature of this process observed during step-wise
comprehensive understanding of the observed phenomena. The endo­ annealing [17]. The effect IV is assigned to sintering because its
thermic peak in the beginning of curves corresponds to the first stage of onset corresponds to a steep increase in density (Fig. 2). The effects I
mass loss due to the desorption of water. The second stage of mass loss, i.e. and III are both assigned to reduction and CO evolution [20]. The ef­
decarburization, results in a series of overlapping peaks. The first effect (I) fect I is probably related to reduction of oxycarbide phase at the sur­
starts at about 80− 90 min (700− 800 ◦ C) indicating the onset of endo­ face of nanoparticles, whereas the effect III is a reduction of segregated
thermic CO release. The second effect (II) is exothermic and takes place in oxides [28,29] (see Fig. S1 in Supplementary Information).
the 90− 110 min range (900− 1100 ◦ C). The third effect (III) occurs within A suggested sequence of the heat effects can be interpreted as follows.
110− 135 min (1100− 1300 ◦ C) and represents the continuation of endo­ The oxycarbide phases TaCxOy and NbCxOy at the surface of carbide
thermic CO release (distinguished from the first effect due to a slight slope particles are stable up to about 700− 800 ◦ C. At higher temperatures,

5
A. Bokov et al. Journal of the European Ceramic Society xxx (xxxx) xxx

Table 2 sintering is proportional to the amount of lattice strain and grain size in
Analysis of TGA curves in Fig. 5 for TaC and NbC with respect to oxygen content starting powders (elimination of interfaces and relaxation of defects are
in these samples before and after experiments. exothermic processes). Since the TGA-DSC data was recorded per mg of a
H2O CO O Oxide left material, the NbC samples have a more pronounced mass loss and stronger
Starting
Powder type O
loss loss left (wt. %) endothermic effects with respect to tantalum due to a lower molar mass.
(wt. (wt. (wt. If one compares the total TGA mass loss for all tested powders with
(wt.%)* from TGA XRD**
%) %) %)
respect to the starting oxygen content (Table 2), it can be noticed that
Ta2O5 oxygen evolution is not complete. Some portion of oxygen still remains
initial TaC 0.1 0.02 0.10 0.03 0.1 –
in the sample and forms the oxide phase (effect II), which is then
milled-5 h 2.7 1.0 1.8 0.8 4.4 1.0
milled-10 h 3.9 1.5 2.2 1.3 7.3 5.8 reduced (effect III) but not entirely. The amount of leftover oxide can be
milled-15 h 5.2 2.1 2.6 1.8 9.9 8.7 calculated based on the amount of leftover oxygen, assuming that
NbO2 weight of metal is 4.5 times higher than oxygen in Ta2O5 and only 2.9
initial NbC 1.3 0.4 1.1 0.3 1.3 – times in NbO2. The resulting oxide content from TGA for the most part is
milled-5 h 5.0 1.8 3.2 1.6 6.1 4.5
milled-10 h 6.9 2.3 3.8 2.6 10.3 8.5
similar to the values estimated by XRD.
milled-15 h 8.2 3.0 4.7 2.8 11.0 12.4 As the oxide content in sintered TaC and NbC ceramics is known at
*
this point, it is therefore possible to determine how much carbon is left
Starting oxygen content was measured by inert gas fusion.
** in the carbide phase after CO release. For this purpose, crushed pellets
Estimation by Rietveld refinement of XRD profiles for crushed samples.
were oxidized during TGA-DSC experiments under Ar + 10 % of O2 flow
in order to evaluate the mass gain (Fig. 6). The oxide impurities in
sintered ceramics do not gain any weight (the NbO2 phase is stable in the
oxycarbide starts to slowly decompose into MeCx and O, and exsolved studied range of temperatures), and the weight gain of WC impurity is
oxygen reacts with structural carbon from the carbide lattice producing negligibly small.
CO gas (effect I). In the 900− 1100 ◦ C range, exsolved oxygen starts to As it can be seen in Fig. 6c and d, the initial TaC and NbC demonstrate
react with metal ions leading to segregation of oxide phases (effect II). The the mass gain around 1.149 (14.9 %) and 1.274 (27.4 %), meaning that
reduction of segregated oxides by structural carbon from carbide begins initial stoichiometry was 0.95 and 0.96, respectively. Sintered pellets
around 1100/1000 ◦ C for TaC/NbC (effect III), but this phase is not show a much larger degree of non-stoichiometry with x equals to 0.75 and
reduced entirely (Fig. 4). The onset of densification is observed around 0.75− 0.79 for TaC and NbC (Table 3). The shape of associated heat sig­
1200/1100 ◦ C so the heat of sintering (effect IV) considerably overlaps nals is similar within the group of samples for the same compound (Fig. 6a
with the reduction of oxides. and b). The intensity of heat effects looks larger in the NbC case due to its
The deconvolution of different effects is shown on the right of DSC lower molar mass (heat signals are shown as mW/mg). The onset of heat
curves in Fig. 5e and f. The heat effects of reduction and oxide segregation release starts at 500 ◦ C in NbC and at 600 ◦ C in TaC, meaning that niobium
are directly proportional to the starting oxygen content. The heat of carbide is less resistant to oxidation.

Fig. 5. TGA-DSC analysis for initial and milled TaC and NbC powders shown as temperature profiles (a, b), mass changes (c, d), and heats signal (e, f). The heat
signals of different samples are shifted by -0.4 with respect to each other for visual clarity. Deconvoluted peaks are plotted on the same scale as original experi­
mental data.

6
A. Bokov et al. Journal of the European Ceramic Society xxx (xxxx) xxx

Fig. 6. Oxidation of initial and sintered/crushed TaC (a,c) and NbC (b,d) samples performed during TGA-DSC analysis.

Table 3
Characteristics of TaC and NbC ceramics produced in this study.
Density HV 1 kgf Grain size Impurities
Ceramic type x in MeCx
(g/cm3) r.o.m. (GPa) (μm) (wt. %)

TaC
m-5 h s-1400C 13.9 ± 0.1 97 % of 14.4 0.75 ± 0.01 21.9 ± 0.3 0.74 ± 0.03 4 Ta2O5 + 3 WC
m-10 h s-1400C 13.7 ± 0.1 97 % of 14.2 0.75 ± 0.01 20.3 ± 0.4 0.51 ± 0.02 7 Ta2O5 + 3 WC
m-15 h s-1400C 13.5 ± 0.1 96 % of 14.1 0.75 ± 0.01 18.8 ± 0.3 0.41 ± 0.02 10 Ta2O5 + 4 WC
1.1c m-5 h s-1300C 14.1 ± 0.1 97 % of 14.5 0.93 ± 0.01 14.0 ± 0.3 1.85 ± 0.08 3 WC

NbC
m-5 h s-1300C 7.57 ± 0.06 96 % of 7.9 0.79 ± 0.01 20.3 ± 0.4 1.02 ± 0.05 6 NbO2 + 4 WC
m-10 h s-1300C 7.55 ± 0.07 96 % of 7.9 0.78 ± 0.01 19.2 ± 0.4 0.77 ± 0.04 10 NbO2 + 5 WC
m-15 h s-1300C 7.36 ± 0.04 94 % of 7.9 0.75 ± 0.01 16.6 ± 0.4 0.63 ± 0.03 11 NbO2 + 5 WC
1.6c m-5 h s-1350C 7.27 ± 0.07 90 % of 8.0 0.90 ± 0.01 15.5 ± 0.4 0.96 ± 0.05 3 WC

“r.o.m” – rule-of-mixture density that includes all impurity phases, “m” – refers to milling time for initial powder, “s” – refers to sintering temperature for milled
powder, “c” – refers to addition of extra carbon prior milling.

It is rather unexpected to see that stoichiometry of all sintered TaC and oxide phase rather than reduce it.
NbC ceramics stabilizes roughly around x = 0.75. From a kinetical The decarburization and oxide segregation during sintering could be
perspective, sintering overlaps with reduction of oxides and thus it could compensated by the addition of excess carbon. A desired amount of
slow down the production of CO gas due to elimination of porosity. This excess carbon can be calculated based on the stoichiometry and amount
explanation would be consistent with the fact that decarburization in of oxide impurities in sintered pellets. For the case of ceramics produced
loose nanopowders continues up to the point of Me2C formation [17]. The from the 5-hs-milled powders, 1.5 wt.% and 2.7 wt.% of extra carbon
issue with such argument is that it could not explain why samples with black are necessary to completely eliminate oxides and maintain x =
different densification kinetics have very similar stoichiometry of the 0.99 in TaC and NbC. Therefore, a required amount of carbon was added
carbide phase. Alternatively, stabilization of stoichiometry in sintered TaC to the initial powders and the mixtures were milled for 5 h. The obtained
and NbC can be interpreted by a thermodynamic balance. The oxides of powders were analyzed by TGA-DSC up to 1450 ◦ C with 10 ◦ C/min
tantalum and niobium generally have lower symmetry with respect to heating rate under Ar flow. The isothermal hold was not used (in
carbide counterparts [1] and thus have higher surface energy. However, contrast with experiments in Fig. 5) because optimal sintering temper­
the existence of such oxide phase is energetically less expensive in dense ature for the samples with excess carbon was not known at that point.
ceramics because grain boundaries have lower energy than surface of the As follows from Fig. 7a and c, TaC and NbC powders with 1.5 wt.%
particles. On the other hand, the energy cost of vacancy formation in MeCx and 2.7 wt.% of extra C, noted as 1.5c and 2.7c respectively, demon­
increases with decreasing x parameter [30]. Therefore, it might not be strate a more pronounced reduction (effect III) and mass loss, and higher
difficult to consume carbon from a nearly-stoichiometric sample, but as heat of sintering (effect IV) in comparison with milled powders without
soon as x approaches 0.75, it could become more favorable to coexist with excess carbon. This is consistent with differences in densification

7
A. Bokov et al. Journal of the European Ceramic Society xxx (xxxx) xxx

Fig. 7. TGA-DSC curves (a, c) and densities for TaC (b) and NbC (d) powders mixed with different amounts of excess carbon and milled for 5 h. The NbC curves are
shifted by -2% and -0.3 mW/mg with respect to TaC for visual clarity.

kinetics, which can be seen in Fig. 7b and d. At earlier stages of sintering, achieved because the rate of gas release did not decrease despite
excess carbon probably isolates carbide grains and thus slows down lowering of carbon excess.
densification. As soon as the reduction of segregated oxides is initiated, The XRD analysis of samples with lowered excess carbon also did not
density starts to increase rapidly but does not exceed 90 %. detect oxide impurities (Fig. S2). The stoichiometry in TaC-1.1c and
According to XRD, the TaC-1.5c sintered at 1300 ◦ C and NbC-2.7c NbC-1.6c was measured as 0.93 and 0.90 (Fig. S3), which was very
sintered at 1350 ◦ C demonstrate peaks only for Fm-3m structure (Fig. surprising considering lowered content of excess carbon. Dissolution of
S2). It means that excess carbon does allow for a complete removal of carbon is also reflected in a high intensity of the effect IV (Fig. 7c),
oxide impurities. However, the reduction of oxides (effect III) has a very similarly to TaC-1.5c and NbC-2.7c samples. The only reason why stoi­
characteristic temperature and occurs right before the onset of sintering. chiometry could still be regained even if the carbon content was lowered
A massive release of CO pushes particles apart and hinders subsequent is that required carbon excess was overestimated, i.e. there was enough
densification so that TaC-1.5c and NbC-2.7c demonstrate relatively low carbon in all cases. A slightly higher stoichiometry in TaC-1.1c with
densities in comparison with samples without excess carbon. respect to TaC-1.5c is simply an indication that high density is more
Analysis of stoichiometry in TaC-1.5c and NbC-2.7c provided x in important for dissolution of C than concentration of this carbon along
MeCx as 0.91 and 0.94 (Fig. S3), respectively. Although these values are a grain boundaries. Difficulty in estimating carbon content is probably
bit lower than initially planned, it is still an indication for successful related to the fact that presence of extra C (free carbon) in nanopowders
dissolution of extra carbon in the carbide, which lost its own structural affects the behavior of oxygen impurities. For instance, it could shift the
carbon due to decarburization. The dissolution of carbon in MeCx is rather reaction of gas formation toward CO2 rather than CO (especially during
exothermic and thus provides a much higher intensity of the effect IV due the effect I prior 1000 ◦ C), which would decrease the relative amount of
to summation with the heat of sintering. carbon needed for removal of oxides and restoration of stoichiometry.
The next step of the study was to lower the amount of excess carbon The effect of excess carbon can be also visualized by SEM images of
in order to decrease the volume of evolving gas and mitigate densifi­ fracture surfaces. The presence of excess carbon in TaC-1.1c provides
cation issues. The desired amount of excess was calculated as 43 % of the dense and coarse microstructure at temperature as low as 1300 ◦ C
2nd stage mass loss in TaC-1.5c and NbC-2.7c, which simply reflects the (Fig. 8a). If no extra carbon is present, high density is obtained only at
weight of carbon lost via CO evolution. Therefore, TaC/NbC powders 1400 ◦ C, but grain size is in the submicron range (Fig. 8c). Comparing
were mixed with 1.1/1.6 wt.% of extra C (noted as 1.1c and 1.6c, these cases, it can be concluded that dissolution of carbon can in fact
respectively), milled for 5 h, sintered, and then analyzed by TGA-DSC. promote a considerable grain growth [10].
The decrease in excess carbon resulted in a considerable improve­ The NbC case is different from TaC in a sense that elected values of
ment for the density of TaC pellets, but had a negligible impact on NbC carbon excess did not provide a plausible outcome (Fig. 8b and d).
(Fig. 7b and d). The mass loss and reduction (effect III) are less pro­ Although NbC-1.6c at 1350 ◦ C has similar grain size as the sample
nounced in TaC-1.1c than in TaC-1.5c, and high density can be obtained without extra C, the amount of residual porosity is still too high. If
if isothermal hold is performed at 1300 ◦ C (the rate of densification at sintering temperature is increased even further, the density significantly
this temperature is slow enough to allow for CO evolution). In the NbC- drops due to activation of grain growth, similarly to the TaC case. The
1.6c case, although the overall mass loss is as small as in the powder shape of DSC curve around 1200 ◦ C in Fig. 7c demonstrates that overlap
without extra C, a large portion of CO is released within a short range of between reduction/sintering is larger for TaC than for NbC. Therefore,
temperatures (1100− 1300 ◦ C). This means that high density was not the evolution of gas is not that much detrimental in TaC because it is

8
A. Bokov et al. Journal of the European Ceramic Society xxx (xxxx) xxx

Fig. 8. SEM images of fractured surfaces for ceramics sintered from milled powders with (a, b) and without excess carbon (c, d).

counterbalanced by relatively earlier onset of densification. It is likely the carbide lattice [31] and submicron grain size [32]. As previously
that further adjustments of carbon content and fine tuning of tempera­ noted (Figs. 3 and 4), fine microstructure in produced ceramics is
ture profile would eventually provide denser microstructure for NbC, associated with presence of oxide impurities inhibiting grain growth
but it is not performed in this study. [33], and samples with larger content of segregated oxides demonstrate
The next step was the analysis of hardness for sintered TaC and NbC smaller grain sizes. However, when the oxide content becomes too high,
as this property is relevant for the most common application of carbides it could promote plastic deformation via shear of grains due to weakness
(development of mechanically stable alloys). Prior hardness measure­ of the oxide-carbide interface (Fig. S4).
ments, the tantalum and niobium ceramics were mounted into epoxy A relatively low hardness of the NbC-1.6c ceramic is not surprising as
resin and polished. After that, Vickers hardness was measured at four its microstructure is rather porous, but the behavior of TaC-1.1c is not
different loads. that intuitive. Although large grains should not impede dislocation glide
As it can be seen in Fig. 9, the highest hardness of 23− 25 GPa and as much as in the submicron case, such a significant decrease in hardness
20− 23 GPa is observed for TaC and NbC ceramics sintered from regular does not follow the Hall-Petch relationship. A more feasible explanation
5-hs-milled powders. Such high values arise from non-stoichiometry of is that TaC-1.1c ceramic might contain some unreacted excess carbon,

Fig. 9. Hardness as a function of load for TaC (a) and NbC (b) ceramics produced in this study. The legend refers to milling and sintering conditions used for sample
fabrication.

9
A. Bokov et al. Journal of the European Ceramic Society xxx (xxxx) xxx

which promotes plastic deformation via grain sliding. Further optimi­ references typically provide less experimental details, mainly focusing only
zation of carbon content would probably solve this issue, but it was not on essential properties like hardness. Later works thoroughly describe
performed in this study. experiments but pay less attention to carbon content. Nonetheless, some
The characteristics for the most notable samples in this study are general conclusion can still be drawn, i.e. hardness values are higher when
summarized in Table 3. The maximum possible density was calculated measured at lower indentation load and when carbon content considerably
based on the rule-of-mixture between carbide phases and impurities deviates from stoichiometry.
(effect of non-stoichiometry on the density of MeCx was also accounted). The effect of indentation load is related to rigidity of carbides. A
The resulting relative densities are more consistent with SEM images in recoverable elastic portion of deformation increases with respect to un­
Figs. 3 and 8 than values calculated only from theoretical densities. recoverable plastic response when the load is decreased [34], which makes
The tungsten impurity was not discussed in previous paragraphs the indenter impression look smaller. Therefore, it is of a great importance
because its amount was very similar in all ceramics and assumed to be a to report hardness values at least for several loads. The increase in hard­
somewhat fixed variable. After ball-milling, all powders contained about ness upon decrease in carbon content is not a general trend and can be
3− 5 wt. % of WC due to grinding of the milling media as calculated from observed only in the group 5 carbides such as TaC and NbC [31]. The
its weight before and after experiments. However, the tungsten carbide carbides of the group 4 (TiC, ZrC, HfC) demonstrate the opposite behavior
phase was detected by XRD only in the starting powders but not in sin­ as their hardness decreases if carbon content is decreased. The group 6
tered ceramics, implying dissolution of such impurity in the host carbide carbides are not applicable to any of those effects as they typically favor
during sintering. Therefore, the density of cubic WC (15.4 g/cm3) is taken only stoichiometric compositions such as Cr7C3/Cr3C2, Mo3C2/Mo2C and
for the rule-of-mixture calculations. It is possible that tungsten carbide W2C/WC.
impurity does not dissolve entirely and its traces exist as W2C or W, but it The samples produced in this study demonstrate relatively high
was not considered for the sake of simplicity. hardness which is comparable to the best numbers in the literature, i.e.
Before concluding this study, it is very important to discuss properties Samsonov’s [38] and Bychkov’s [52] works for TaC and NbC, respec­
of the obtained ceramics with respect to previous studies. As it can be seen tively. However, the novelty of the present study is that such ceramics
in Table 4, the results on TaC and NbC ceramics significantly vary are fabricated at relatively low sintering temperatures, by 500− 1000 ◦ C
depending on the year of publication and methodological aspects. Older lower with respect to previous works. The TaC and NbC have been

Table 4
Literature data on fabrication and hardness of TaC and NbC ceramics.
x in MeCx Relative density Hardness HV HV load Grain size Fabrication Source
(GPa) (N) (μm)

TaC
s high 17.8 2.9 coarse conventional sintering at 1900 ◦ C for 30 h Miyoshi 1965 [35]
s high 17.6 microHV coarse conventional sintering at ~2400 ◦ C Atkins 1966 [36]
s 92 15.8 1.47 coarse conventional sintering at ~2300 ◦ C Skuratovskii 1969 [37]
0.99 91 17.1 microHV 20 hot pressing 15 MPa at 2700 ◦ C for 10 min Samsonov 1974 [38]
0.91 92 21.1 microHV 20
0.82 94 22.5 microHV 20
0.98 high 18.5 microHV coarse conventional sintering at 2400 ◦ C for 60 min Shvab 1983 [39]
s 94.3 14.1 4.9 2.4 hot pressing at 2300 ◦ C for 45 min Zhang 2009 [40]
0.99 97.1 13.5 9.8 7.7 hot pressing 30 MPa at 1800 ◦ C for 120 min Hackett 2009 [41]
0.80 high 16.8 9.8 6.4
0.70 high 20.1 9.8 5.8
s high 22.0 196 1 high-frequency induction heating up to 1350 ◦ C Kim 2009 [42]
0.86 high 21.7 49 2 spark plasma sintering 30 MPa 1600 ◦ C 5 min Limeng 2010 [43]
s 94.6 18.9 25 4.5 spark plasma sintering 100 MPa 1800 ◦ C 10 min Nieto 2013 [44]
s 94.3 21.7 0.0045 0.5 hot consolidation 7.7 GPa 1800 ◦ C 10 min Lahiri 2013 [45]
s 98.5 14.3 98 16 spark plasma sintering 50 MPa 1800 ◦ C 10 min Nino 2015 [46]
s 93.1 15.5 2 0.8 spark plasma sintering 40 MPa 1850 ◦ C 5 min Nisar 2016 [47]
s 94.2 13.7 49 11 spark plasma sintering 30 MPa 1700 ◦ C 5 min Geng 2017 [48]
0.98 97.8 15.7 49 30 hot pressing 40 MPa at 2000 ◦ C for 45 min Rezaei 2017 [49]
s 98 19.9 0.1 12 hot pressing 200 MPa at 2000 ◦ C for 60 min Smith 2018 [50]
0.99 high 20.9 19.6 0.7 hot consolidation 5.5 GPa 1400 ◦ C 1 min Sun 2020 [51]
NbC
s high 18.8 2.9 coarse conventional sintering at 1900 ◦ C for 30 h Miyoshi 1965 [35]
s high 24.5 micro HV coarse conventional sintering at ~2200 ◦ C Atkins 1966 [36]
s 93 18.5 0.68 coarse conventional sintering at ~2300 ◦ C Skuratovskii 1969 [37]
0.94 97 20 0.98 coarse conventional sintering at ~2100 ◦ C Bychkov 1974 [52]
0.84 97 25.6 0.98 coarse
0.81 98 26.6 0.98 coarse
0.74 98 27.8 0.98 coarse
0.91 high 22 0.98 30 conventional sintering at ~2600 ◦ C Spivak 1975 [53]
0.95 high 20 0.98 30
0.98 high 19.9 0.49 coarse hot pressing 20 MPa at 2350 ◦ C for 5 min Samsonov 1975 [54]
0.91 high 24.7 0.49 coarse
s high 21.6 98 2.5 high-frequency induction heating up to 1350 C ◦
Kim 2009 [55]
s 99 18.5 0.98 coarse spark plasma sintering 50 MPa 1800 ◦ C 5 min Nino 2010 [56]
0.99 98 17.3 9.8 120 hot pressing 50 MPa at 2150 ◦ C for 4 h Woydt 2013 [57]
0.99 high 16.4 9.8 coarse spark plasma sintering 100 MPa at 2000 ◦ C 12 min Woydt 2014 [58]
s 99 22.1 9.8 3 hot consolidation 5.5 GPa 1400 ◦ C 15 min Liu 2017 [59]
s 98 15.1 9.8 10 spark plasma sintering 35 MPa at 2200 ◦ C 30 min Balko 2017 [60]

“s” – composition reported as stoichiometric, “high” – density was assessed qualitatively, “coarse” – large grains but exact size was not reported, “microHV” –
microhardness indenter was used but exact load was not reported.

10
A. Bokov et al. Journal of the European Ceramic Society xxx (xxxx) xxx

always thought as ultra-high temperature materials, so sintering of such Declaration of Competing Interest
ceramics to a high density at 1300− 1400 ◦ C is a rather unprecedented
result. Such an improvement is likely associated with mechanical acti­ The authors report no declarations of interest.
vation due to high energy ball milling, similarly to some other studies on
milled W [61], ZrC [62], and HfB2 [63]. The ball-mill processing pro­ Acknowledgments
vides a powdered material with a high surface area and a large con­
centration of defects which increase the driving force for sintering (it is The present work was performed in accordance with the state
energetically favorable to eliminate interfaces and relax distorted sites assignment no. AAAA-A19-119031890029-7 from the Ministry of Sci­
via grain growth and densification). ence and Higher Education of the Russian Federation. The authors are
The exotic methods of fabrication such as ultra-high pressure grateful to D. A. Danilov from the Ural Federal University for mea­
consolidation [51,59] or high-frequency induction heating [42,55] can surements of oxygen content in nanocrystalline powders.
also provide dense ceramics at relatively low temperature but rarely
available for researchers and manufacturers. In contrast, conventional Appendix A. Supplementary data
vacuum furnaces and planetary mills like ones used in this study are
readily available on the market at a reasonable price. Another important Supplementary material related to this article can be found, in the
note is that sintering of nanocrystalline carbides requires slow heating online version, at doi:https://doi.org/10.1016/j.jeurceramsoc.2021.0
profile due to extensive gas evolution. Therefore, the spark plasma 5.007.
sintering approach, which is used in the majority of recent references,
might not be suitable for such materials due to high heating rates. References

4. Conclusions [1] I.L. Shabalin, Ultra-High Temperature Materials I.I., Springer Netherlands,
Dordrecht, 2019 https://doi.org/10.1007/978-94-024-1302-1.
[2] S.G. Huang, R.L. Liu, L. Li, O. Van der Biest, J. Vleugels, NbC as grain growth
The nanocrystalline powders of TaC and NbC with particle sizes of inhibitor and carbide in WC–Co hardmetals, Int. J. Refract. Met. Hard Mater. 26
18− 48 nm were produced by high energy ball milling. The produced (2008) 389–395, https://doi.org/10.1016/j.ijrmhm.2007.09.003.
[3] P. Wu, Y. Zheng, Y. Zhao, H. Yu, Effect of TaC addition on the microstructures and
powders contained about 3− 8 wt.% of oxygen, which was probably mechanical properties of Ti(C,N)-based cermets, Mater. Des. 31 (2010) 3537–3541,
accumulated in the surficial oxycarbide phase. https://doi.org/10.1016/j.matdes.2010.01.047.
Sintering of such nanopowders at 1300− 1400 ◦ C for 30 min under [4] W.H. Kan, G. Proust, V. Bhatia, L. Chang, K. Dolman, T. Lucey, X. Tang, J. Cairney,
Slurry erosion, sliding wear and corrosion behavior of martensitic stainless steel
vacuum provided highly dense ceramics with the 0.4− 1 μm grains even composites reinforced in-situ with NbC particles, Wear 420–421 (2019) 149–162,
though actual densities were just 13.5–14.0 g/cm3 and 7.4–7.6 g/cm3 https://doi.org/10.1016/j.wear.2018.09.013.
for TaC and NbC, respectively. [5] M. Chao, W. Wang, E. Liang, D. Ouyang, Microstructure and wear resistance of TaC
reinforced Ni-based coating by laser cladding, Surf. Coatings Technol. 202 (2008)
The TGA-DSC method was used to analyze sintering process in great
1918–1922, https://doi.org/10.1016/j.surfcoat.2007.08.021.
detail. It was observed that compacted nanopowders experienced a [6] D. Sciti, D.M. Trucchi, A. Bellucci, S. Orlando, L. Zoli, E. Sani, Effect of surface
significant mass loss starting from 700 ◦ C due to release of CO, i.e. texturing by femtosecond laser on tantalum carbide ceramics for solar receiver
preexisting oxygen evolved at the expense of carbon from the carbide applications, Sol. Energy Mater. Sol. Cells 161 (2017) 1–6, https://doi.org/
10.1016/j.solmat.2016.10.054.
lattice. At about 1000 ◦ C, oxygen impurities also started to react with [7] P. Kondaiah, K. Niranjan, S. John, H.C. Barshilia, Tantalum carbide based
metal ions, forming separate oxide phase. The mass loss became more spectrally selective coatings for solar thermal absorber applications, Sol. Energy
pronounced starting from 1100 ◦ C due to reduction of segregated oxides Mater. Sol. Cells 198 (2019) 26–34, https://doi.org/10.1016/j.
solmat.2019.04.016.
at the expense of carbon from the carbide phase. The onset of densifi­ [8] C. Ang, L. Snead, K. Benensky, Niobium carbide as a technology demonstrator of
cation was around 1200 ◦ C. ultra-high temperature ceramics for fully ceramic microencapsulated fuels, Int. J.
The analysis of phase composition revealed that reduction of segre­ Ceram. Eng. Sci. 1 (2019) 92–102, https://doi.org/10.1002/ces2.10014.
[9] R.H.R. Castro, On the thermodynamic stability of nanocrystalline ceramics, Mater.
gated oxide was partial and sintered pellets had about 4− 11 wt.% of Lett. 96 (2013) 45–56, https://doi.org/10.1016/j.matlet.2013.01.007.
Ta2O5 or NbO2 impurities. Due to a significant loss of carbon in the [10] A.S. Kurlov, A.I. Gusev, Peculiarities of vacuum annealing of nanocrystalline WC
course of heat treatment (decarburization), the stoichiometry in fabri­ powders, Int. J. Refract. Met. Hard Mater. 32 (2012) 51–60, https://doi.org/
10.1016/j.ijrmhm.2012.01.009.
cated ceramics decreased down to 0.75− 0.79 from the initial values of [11] E. Lantsev, N. Malekhonova, A. Nokhrin, V. Chuvil’deev, M. Boldin,
0.96− 0.95. Y. Blagoveshchenskiy, P. Andreev, K. Smetanina, N. Isaeva, S. Shotin, Influence of
The additional set of experiments was performed to compensate the oxygen on densification kinetics of WC nanopowders during SPS, Ceram. Int. 47
(2021) 4294–4309, https://doi.org/10.1016/j.ceramint.2020.09.272.
loss of carbon. The initial powders of TaC and NbC were mixed with
[12] B. Lanfant, Y. Leconte, G. Bonnefont, V. Garnier, Y. Jorand, S. Le Gallet, M. Pinault,
excess amount of carbon black and then processed and analyzed in the N. Herlin-Boime, F. Bernard, G. Fantozzi, Effects of carbon and oxygen on the spark
same way as regular samples. Presence of excess carbon helped to plasma sintering additive-free densification and on the mechanical properties of
eliminate segregated oxides and restore stoichiometry, but it became nanostructured SiC ceramics, J. Eur. Ceram. Soc. 35 (2015) 3369–3379, https://
doi.org/10.1016/j.jeurceramsoc.2015.05.014.
difficult to control densification curve due to extensive release of CO gas [13] D. Feng, Q. Ren, H. Ru, W. Wang, Y. Jiang, S. Ren, C. Zhang, Effect of oxygen
and promotion of grain growth. For the TaC case, 1.1 wt.% of extra content on the sintering behaviour and mechanical properties of SiC ceramics,
carbon provided very dense and coarse microstructure as early as at Ceram. Int. 45 (2019) 23984–23992, https://doi.org/10.1016/j.
ceramint.2019.08.100.
1300 ◦ C. For the NbC case, proper parameters were not found as it [14] J.R. Groza, J.D. Curtis, M. Krämer, Field-Assisted Sintering of Nanocrystalline
required much finer tuning of experimental conditions. Titanium Nitride, J. Am. Ceram. Soc. 83 (2000) 1281–1283, https://doi.org/
Overall, the present study demonstrates that highly dense ceramics 10.1111/j.1151-2916.2000.tb01369.x.
[15] B.M. Moshtaghioun, D. Gómez-García, A. Domínguez-Rodríguez, Spark plasma
of TaC and NbC can be sintered at temperatures as low as 1300− 1400 ◦ C sintering of titanium nitride in nitrogen: does it affect the sinterability and the
(by 500− 1000 ◦ C lower than previously reported). Enhanced sinter­ mechanical properties? J. Eur. Ceram. Soc. 38 (2018) 1190–1196, https://doi.org/
ability was associated with a high surface area and large concentration 10.1016/j.jeurceramsoc.2017.12.029.
[16] A.S. Kurlov, N.D. Yumasheva, D.A. Danilov, Concentration of oxygen and forms of
of defects in starting nanopowders which provided a strong thermody­ it in TaC nanopowders, Russ. J. Phys. Chem. A. 93 (2019) 501–508, https://doi.
namic driving force for sintering. Hardness of produced TaC/NbC ce­ org/10.1134/S0036024419030117.
ramics was 22/20 GPa at 9.8 N of test load, which is consistent with the [17] A.S. Kurlov, N.D. Yumasheva, D.A. Danilov, Vacuum annealing of TaC
nanopowders, Russ. J. Phys. Chem. A. 94 (2020) 1447–1455, https://doi.org/
best results from the literature.
10.1134/S0036024420070183.
[18] A.M. Balagurov, I.A. Bobrikov, G.D. Bokuchava, R.N. Vasin, A.I. Gusev, A.
S. Kurlov, M. Leoni, High-resolution neutron diffraction study of microstructural

11
A. Bokov et al. Journal of the European Ceramic Society xxx (xxxx) xxx

changes in nanocrystalline ball-milled niobium carbide NbC0.93, Mater. Charact. [41] K. Hackett, S. Verhoef, R.A. Cutler, D.K. Shetty, Phase constitution and mechanical
109 (2015) 173–180, https://doi.org/10.1016/j.matchar.2015.09.025. properties of carbides in the Ta-C system, J. Am. Ceram. Soc. 92 (2009)
[19] A. Shemi, A. Magumise, S. Ndlovu, N. Sacks, Recycling of tungsten carbide scrap 2404–2407, https://doi.org/10.1111/j.1551-2916.2009.03201.x.
metal: a review of recycling methods and future prospects, Miner. Eng. 122 (2018) [42] B.R. Kim, K. Do Woo, J.M. Doh, J.K. Yoon, I.J. Shon, Mechanical properties and
195–205, https://doi.org/10.1016/j.mineng.2018.03.036. rapid consolidation of binderless nanostructured tantalum carbide, Ceram. Int. 35
[20] J. Sautereau, A. Mocellin, Sintering behaviour of ultrafine NbC and TaC powders, (2009) 3395–3400, https://doi.org/10.1016/j.ceramint.2009.06.012.
J. Mater. Sci. 9 (1974) 761–771, https://doi.org/10.1007/BF00761796. [43] L. Limeng, Y. Feng, Z. Yu, Z. Zhiguo, Microstructure and mechanical properties of
[21] A.S. Kurlov, A.I. Gusev, Effect of nonstoichiometry of NbCy and TaCy powders on spark plasma sintered TaC0.7 ceramics, J. Am. Ceram. Soc. 93 (2010) 2945–2947,
their high-energy ball milling, Int. J. Refract. Met. Hard Mater. 46 (2014) 125–136, https://doi.org/10.1111/j.1551-2916.2010.03908.x.
https://doi.org/10.1016/j.ijrmhm.2014.05.018. [44] A. Nieto, D. Lahiri, A. Agarwal, Graphene NanoPlatelets reinforced tantalum
[22] Z. Kou, K. Xi, Z. Pu, S. Mu, Constructing carbon-cohered high-index (222) faceted carbide consolidated by spark plasma sintering, Mater. Sci. Eng. A. 582 (2013)
tantalum carbide nanocrystals as a robust hydrogen evolution catalyst, Nano 338–346, https://doi.org/10.1016/j.ijrmhm.2015.06.015.
Energy 36 (2017) 374–380, https://doi.org/10.1016/j.nanoen.2017.04.057. [45] D. Lahiri, V. Singh, G.R. Rodrigues, T.M.H. Costa, M.R. Gallas, S.R. Bakshi, S. Seal,
[23] W. Cai, G. Li, K. Zhang, G. Xiao, C. Wang, K. Ye, Z. Chen, Y. Zhu, Y. Qian, A. Agarwal, Ultrahigh-pressure consolidation and deformation of tantalum carbide
Conductive Nanocrystalline Niobium Carbide as High-Efficiency Polysulfides at ambient and high temperatures, Acta Mater. 61 (2013) 4001–4009, https://doi.
Tamer for Lithium-Sulfur Batteries, Adv. Funct. Mater. 28 (2018), https://doi.org/ org/10.1016/j.actamat.2013.03.014.
10.1002/adfm.201704865, 1704865. [46] A. Nino, T. Hirabara, S. Sugiyama, H. Taimatsu, Preparation and characterization
[24] C. Kunkel, F. Viñes, F. Illas, Surface activity of early transition-metal oxycarbides: of tantalum carbide (TaC) ceramics, Int. J. Refract. Met. Hard Mater. 52 (2015)
CO2 adsorption case study, J. Phys. Chem. C. 123 (2019) 3664–3671, https://doi. 203–208, https://doi.org/10.1016/j.ijrmhm.2015.06.015.
org/10.1021/acs.jpcc.8b11942. [47] A. Nisar, A. S, T. Venkateswaran, N. Sreenivas, K. Balani, Oxidation studies on TaC
[25] J. Graciani, S. Hamad, J.F. Sanz, Changing the physical and chemical properties of based ultra-high temperature ceramic composites under plasma arc jet exposure,
titanium oxynitrides TiN1-xOx by changing the composition, Phys. Rev. B 80 Corros. Sci. 109 (2016) 50–61, https://doi.org/10.1016/j.corsci.2016.03.013.
(2009), https://doi.org/10.1103/PhysRevB.80.184112, 184112. [48] G. Geng, L. Liu, Y. Wang, W. Hai, W. Sun, Y. Chen, L. Wu, Microstructure and
[26] S. Jung, M.S. Kwon, S. Bin Kim, K.S. Shin, Characterization of chemical mechanical properties of TaC ceramics with 1-7.5 mol% Si as sintering aid, J. Am.
information and morphology for in-depth oxide layers in decarburized electrical Ceram. Soc. 100 (2017) 2461–2470, https://doi.org/10.1111/jace.14806.
steel with glow discharge sputtering, Surf. Interface Anal. 45 (2013) 1119–1128, [49] F. Rezaei, M.G. Kakroudi, V. Shahedifar, N.P. Vafa, M. Golrokhsari, Densification,
https://doi.org/10.1002/sia.5237. microstructure and mechanical properties of hot pressed tantalum carbide, Ceram.
[27] V. Katranidis, S. Gu, D.C. Cox, M.J. Whiting, S. Kamnis, FIB-SEM sectioning study Int. 43 (2017) 3489–3494, https://doi.org/10.1016/j.ceramint.2016.10.067.
of decarburization products in the microstructure of HVOF-Sprayed WC-Co [50] C.J. Smith, X.-X. Yu, Q. Guo, C.R. Weinberger, G.B. Thompson, Phase, hardness,
coatings, J. Therm. Spray Technol. 27 (2018) 898–908, https://doi.org/10.1007/ and deformation slip behavior in mixed HfxTa1-xC, Acta Mater. 145 (2018)
s11666-018-0721-3. 142–153, https://doi.org/10.1016/j.actamat.2017.11.038.
[28] H. Preiss, D. Schultze, E. Schierhorn, Preparation of NbC, TaC and Mo2C fibres and [51] W. Sun, X. Kuang, H. Liang, X. Xia, Z. Zhang, C. Lu, A. Hermann, Mechanical
films from polymeric precursors, Synthesis (Stuttg). 33 (1998) 4687–4696, https:// properties of tantalum carbide from high-pressure/high-temperature synthesis and
doi.org/10.1023/A:1004476701618. first-principles calculations, Phys. Chem. Chem. Phys. 22 (2020) 5018–5023,
[29] I.D. Simonov-Emel’yanov, N.L. Shembel’, E.E. Nikishina, E.N. Lebedeva, A. https://doi.org/10.1039/C9CP06819H.
V. Nikitina, D.V. Drobot, E.P. Simonenko, N.P. Simonenko, V.G. Sevast’yanov, N. [52] A.K. Bychkov, V.S. Neshpor, S.S. Ordan’yan, Effect of plastic deformation on some
T. Kuznetsov, Preparation of highly porous NbxTa1–xC ceramics from polymer- properties of niobium monocarbide in its homogeneity range, Sov. Powder Metall.
matrix composite materials based on a phenol-formaldehyde binder and low Met. Ceram. 13 (1974) 560–565, https://doi.org/10.1007/BF00792563.
hydrated hydroxide of niobium and tantalum, Inorg. Mater. Appl. Res. 51 (2015) [53] I.I. Spivak, V.M. Baranov, V.I. Knyazev, V.N. Rystsov, M.A. Fedotov,
1066–1072, https://doi.org/10.1134/S0020168515100143. Physicomechanical properties of zirconium nitride and niobium carbide within the
[30] S.H. Jhi, S.G. Louie, M.L. Cohen, J. Ihm, Vacancy hardening and softening in range of homogeneity, Strength Mater. 7 (1975) 1097–1100, https://doi.org/
transition metal carbides and nitrides, Phys. Rev. Lett. 86 (2001) 3348–3351, 10.1007/BF01522581.
https://doi.org/10.1103/PhysRevLett.86.3348. [54] G.V. Samsonov, V.Y. Naumenko, L.P. Ryabokon’, A.D. Verkhoturov, Some
[31] X.X. Yu, G.B. Thompson, C.R. Weinberger, Influence of carbon vacancy formation properties of alloys of zirconium and niobium carbides in their homogeneity
on the elastic constants and hardening mechanisms in transition metal carbides, ranges, Sov. Powder Metall. Met. Ceram. 14 (1975) 44–46, https://doi.org/
J. Eur. Ceram. Soc. 35 (2015) 95–103, https://doi.org/10.1016/j. 10.1007/BF00804070.
jeurceramsoc.2014.08.021. [55] B.R. Kim, K. Do Woo, J.K. Yoon, J.M. Doh, I.J. Shon, Mechanical properties and
[32] B. Roebuck, Extrapolating hardness-structure property maps in WC/Co hardmetals, rapid consolidation of binderless niobium carbide, J. Alloys. Compd. 481 (2009)
Int. J. Refract. Met. Hard Mater. 24 (2006) 101–108, https://doi.org/10.1016/j. 573–576, https://doi.org/10.1016/j.jallcom.2009.03.036.
ijrmhm.2005.04.021. [56] A. Nino, A. Tanaka, S. Sugiyama, H. Taimatsu, Indentation size effect for the
[33] S.-J. Oh, B.-S. Kim, J.-K. Yoon, K.-T. Hong, I.-J. Shon, Enhanced mechanical hardness of refractory carbides, Mater. Trans. 51 (2010) 1621–1626, https://doi.
properties and consolidation of the ultra-fine WC–Al2O3 composites using pulsed org/10.2320/matertrans.M2010110.
current activated heating, Ceram. Int. 42 (2016) 9304–9310, https://doi.org/ [57] M. Woydt, H. Mohrbacher, Friction and wear of binder-less niobium carbide, Wear.
10.1016/j.ceramint.2016.02.113. 306 (2013) 126–130, https://doi.org/10.1016/j.wear.2013.07.013.
[34] S.J. Bull, T.F. Page, E.H. Yoffe, An explanation of the indentation size effect in [58] M. Woydt, H. Mohrbacher, The tribological and mechanical properties of niobium
ceramics, Philos. Mag. Lett. 59 (1989) 281–288, https://doi.org/10.1080/ carbides (NbC) bonded with cobalt or Fe3Al, Wear 321 (2014) 1–7, https://doi.
09500838908206356. org/10.1016/j.wear.2014.09.007.
[35] A. Miyoshi, A. Hara, High temperature hardness of WC, TiC, TaC, NbC and their [59] F. Liu, P. Liu, F. Peng, J. Liu, D. He, Hardness and compression behavior of niobium
mixed carbides, J. Japan Soc. Powder Powder Metall. 12 (1965) 78–84, https:// carbide, High Press. Res. 37 (2017) 244–255, https://doi.org/10.1080/
doi.org/10.2497/jjspm.12.78. 08957959.2017.1297810.
[36] A.G. Atkins, D. Tabor, Hardness and deformation properties of solids at very high [60] J. Balko, T. Csanádi, R. Sedlák, M. Vojtko, A. Kovalčíková, K. Koval, P. Wyzga,
temperatures, Proc. R. Soc. London. Ser. A. Math. Phys. Sci. 292 (1966) 441–459, A. Naughton-Duszová, Nanoindentation and tribology of VC, NbC and ZrC
https://doi.org/10.1098/rspa.1966.0146. refractory carbides, J. Eur. Ceram. Soc. 37 (2017) 4371–4377, https://doi.org/
[37] V.N. Skuratovskii, Y.G. Tkachenko, V.A. Borisenko, A technique for investigating 10.1016/j.jeurceramsoc.2017.04.064.
the microhardness of refractory compounds within a wide temperature range, [61] Y. Han, J. Fan, T. Liu, H. Cheng, J. Tian, The effects of ball-milling treatment on the
Strength Mater. 1 (1969) 393–396, https://doi.org/10.1007/BF01526953. densification behavior of ultra-fine tungsten powder, Int. J. Refract. Met. Hard
[38] G.V. Samsonov, V.S. Fomenko, I.A. Podchernyaeva, L.N. Okhremchuk, Thermionic Mater. 29 (2011) 743–750, https://doi.org/10.1016/j.ijrmhm.2011.06.010.
emission properties of refractory compounds and materials based on them (a [62] B. Núñez-González, A.L. Ortiz, F. Guiberteau, M. Nygren, Improvement of the
review), Sov. Powder Metall. Met. Ceram. 13 (1974) 836–842, https://doi.org/ spark-plasma-Sintering kinetics of ZrC by high-energy ball-milling, J. Am. Ceram.
10.1007/BF01177935. Soc. 95 (2012) 453–456, https://doi.org/10.1111/j.1551-2916.2011.05007.x.
[39] S.A. Shvab, F.F. Egorov, Sintering kinetics, structure, and some properties of [63] P. Zhao, J. Zhu, H. Wang, X. Zhao, M. Li, W. Liu, B. Fan, G. Shao, H. Lu, H. Xu,
heterophase materials of the titanium diboride-tantalum carbide system, Sov. R. Zhang, S.-H. Lee, Dense HfB2 ceramics fabricated by high-energy ball milling
Powder Metall. Met. Ceram. 22 (1983) 261–264, https://doi.org/10.1007/ and spark plasma sintering, Mater. Chem. Phys. 258 (2021), https://doi.org/
BF00795597. 10.1016/j.matchemphys.2020.123845, 123845.
[40] X. Zhang, G.E. Hilmas, W.G. Fahrenholtz, Densification and mechanical properties
of TaC-based ceramics, Mater. Sci. Eng. A. 501 (2009) 37–43, https://doi.org/
10.1016/j.msea.2008.09.024.

12

You might also like