The role of carbon in structural evolution during single step synthesis of nano tantalum carbide

Abstract

Cubic phase carbon-coated nano tantalum carbide (TaC) has been synthesized at 800 °C in a single step from tantalum oxide using the carbon and hydrogen produced in situ via decomposition of acetone in an autoclave. In the product phase(s) carbon exists: (a) inside the carbide, (b) on the surface of the carbide particles and (c) as free carbon (amorphous as well as graphitic). The effects of initial carbon concentration on the final carbon content inside as well as outside the TaC have been studied. The structural features of the final product have a complex dependency on the initial carbon concentration. The thermal behaviour of the final product clearly delineates the effects of internal and external carbon content. The soaking time studies show that the grain growth of TaC within the autoclave follows the simultaneous grain boundary migration and grain rotation model. The DSC/TG, XRD and microstructure analysis results along with thermal calculations have been used to predict the formation mechanism for the carbide particles. The reaction mechanism analysis brings forth the role of Mg in lowering the reaction temperature. In this process the carbon content of TaC, the size as well as the strain of the synthesized powders and the %free carbon content can be tailored as per the requirement for the given application.

---

1 Introduction

Wide stability of cubic phase tantalum carbide (TaC) over a wide range of carbon vacancies allows for tailoring its physical, chemical and mechanical properties for a large number of industrial applications.¹ The bonds in TaC show a unique combination of ionic, covalent and metallic contributions. The covalent contributions are the strongest but their precise nature and hence the material properties are carbon vacancy dependent.² Extreme hardness, high melting point, high electronic conductivity, high oxidation resistance as well as resistance to chemical attack and thermal shock are some of the properties of TaC being utilized in industrial applications.^1–7 The superior mechanical properties of the fine grained, sub-stoichiometric powders opened the demand for nano-TaC with tailored carbon content.⁸ The electrochemical energy storage, catalytic and catalyst support material applications of TaC in ammonia decomposition, hydrodenitrogenation and hydrogen dissociation reactions has further increased the interest in development of nano TaC powders having high specific surface area.^3,4,9–14 Hence, synthesis of pure phase nano-TaC with large specific surface area is of paramount importance for catalytic as well as for the traditional applications. Conventional synthesis methods of TaC powder are energy intensive, require careful monitoring and also result in coarsening of grains.^3,15,16

Over last few years many methods have emerged for the synthesis of TaC nano-particles. Ta₂O₅ is the most common low cost precursor for the synthesis of TaC. Ta₂O₅ along with various carburization sources have been used for the synthesis of nano-TaC either in vacuum or inert atmosphere where the single phase TaC nano-particles are obtained only at temperatures greater than 1000 °C.^4,15–22 Addition of catalysts such as Mg, Ni and/or halogenation agents such as NaF can reduce the reduction temperature.^7,17,21

Carbon content in the cubic carbides like TaC dictates the mechanical properties as well as the surface stoichiometric properties which ultimately affect the catalytic response of the material.^1,4 So the synthesized TaC powders need to be characterized for size, specific surface area as well as stoichiometry.²³ The final characteristics of the synthesized product depend largely on the initial carbon content in the reaction mixture as well as the homogeneity of mixing.²⁴ In most of the currently prevalent methods the reduction in reaction temperature and product size is achieved by multiple steps involving formation of nano – Ta₂O₅ followed by its mixing with carbon source and/or heating in reducing carbon rich atmosphere.^8,14,25,26 Slight variation in any one of the steps can change the properties of the final product drastically.^15,23,24

For particle size control as well as maintaining the purity of the synthesized transition metal carbide nano powder, the single step chemical reaction method in autoclave is well documented.^27–29 In the present study we report the role of carbon in structural evolution during synthesis. In this work the carbon concentration in the reaction mixture and soaking time were varied to obtain nano TaC_x powders of desired composition. The idea behind this work was to control x, the fraction of carbon in TaC_x, for different industrial applications. Commercial grade Ta₂O₅ has been used with acetone as carbon source. The in situ hydrogen and carbon, produced during the decomposition of acetone, have been utilized for the reduction and carburization in the presence of reducing agent Mg at 800 °C to get TaC nanopowder. To the best of our knowledge single step synthesis of such fine grained (5–8 nm) nano TaC from Ta₂O₅ has not been reported so far at such a low temperature. Only reported work with lower synthesis temperature which we have come across has been by Ma et al. in which TaC has been synthesized from TaCl₅ at 600 °C in 8 h with the average particle size is 40 nm.⁷

The evolution of the microstructure and properties of the synthesized powders with initial carbon concentrations and increasing soaking times have been studied. The initial carbon concentration emerges as a complex parameter for determining the final properties of the powders. It seems to control the concentration dependent diffusion of carbon into the particles which is actually the rate determining step. At the same time it also determines the amount of carbonaceous network outside the particles which in turn hinders the carbon diffusion and also strains the embedded particles. The carbon diffusion also plays an important role while determining the properties of the particles with increasing soaking time. One important result which emerged from the soaking time experiments is that the grain growth inside the autoclave follows the simultaneous grain boundary migration and grain rotation model.

2 Experimental section

2.1 Synthesis of TaC nanocrystals

For the synthesis of TaC nano powder, tantalum(V) oxide (Ta₂O₅, 99.99%, Sigma Aldrich), Mg (99%, Loba Chemie) and acetone (C₃H₆O, Loba Chemie) in molar ratios of 1 : 12 : y were sealed in a stainless steel autoclave of 20 ml capacity. The Ta : Mg ratio is very important in lowering of the synthesis temperature. We varied the ratio of Mg as: 9, 12, 22. For x = 9 the pure phase TaC is not formed even for initial y = 90, at 800 °C in 10 h (data not shown). The results for Mg = 12 and 22 showed the formation of pure phase TaC. We have performed this series of experiments for Mg = 12 as higher catalyst stoichiometry leads to larger agglomeration.²³ The value of y was varied from 1 to 90. The sealed autoclave was placed inside the furnace. The temperature of the furnace was raised from room temperature to desired temperature slowly (5° min⁻¹) and furnace was finally maintained at the extreme temperature for different durations (1–15 h). The autoclave was then allowed to cool slowly to room temperature within the furnace. The resultant product was collected and leached with diluted HCl (1 : 1) to remove magnesium oxide (MgO) and unreacted Mg. After leaching, the powder was washed several times with double distilled water to remove any traces of unreacted acid. Finally, the powder was washed with acetone and dried in vacuum at 100 °C.

Details of samples synthesized under different conditions along with sample names are given in Tables 1 and 2.

Table 1 Effect of initial carbon concentration (the soaking temperature for all the samples is 800 °C and the soaking time for all the samples is 2 h)

Sample name	Carbon conc.	χ^2	Rwp	a (nm)	x	Dv (nm)	Ds (nm)	RMSS (x 10^−3) at 5 nm
01C	1	0.510	4.20	0.44508	0.96	6.6	4.06	1
02C	2	0.540	3.96	0.44498	0.95	6.8	4.1	2.19
03C	3	0.565	4.09	0.44497	0.95	6.3	3.8	3.14
05p5C	5.5	0.577	4.43	0.44496	0.95	7.1	4.1	5.31
07p5C	7.5	0.744	4.75	0.44551	0.98	6.9	4.0	3.21
15C	15	0.945	5.51	0.44555	0.99	6.2	3.8	4.34
40C	40	1.55	7.10	0.44532	0.98	7.3	4.2	4.21
60C	60	1.31	6.40	0.44534	0.98	6.6	4.0	4.37
90C	90	1.46	6.88	0.44534	0.98	7.5	4.6	4.31

---

Table 2 Evolution of TaC with soaking time (the soaking temperature for all the samples is 800 °C and initial carbon concentration is 3)

Sample name	Soaking time (h)	χ^2	Rwp	a (nm)	x	Dv (nm)	Ds (nm)	RMSS (x 10^−3) at 5 nm
01Q	1	0.571	4.13	0.44445	0.92	6.5	4.1	6.22
01T	1	0.575	4.12	0.44458	0.93	9.2	5.2	7.39
02T	2	0.565	4.09	0.44497	0.95	6.3	3.8	3.14
07T	7	0.534	4.03	0.44547	0.99	5.1	2.6	14.45
10T	10	0.538	4.06	0.44485	0.95	7.4	4.4	1.34
15T	15	0.542	4.26	0.44551	0.99	6.2	3.8	4.10

---

2.2 Characterization

The X-ray diffraction (XRD) study of the synthesized products was done to identify and structurally characterize the crystalline phases present. The XRD of the samples was performed using PANalytical X'Pert HighScore Plus with CuK_α radiation (λ = 1.5406 Å) obtained from the copper target using an inbuilt Ni filter. The X-ray powder diffraction data for 20° ≤ 2θ ≤ 80° with a step size of 0.0130° (2θ) were collected for all samples at room temperature. Phase identification from the ICDD data base (using X'Pert HighScore Plus) was further confirmed by Rietveld refinement of all the XRD patterns (using Full Prof Suite). The Wyckoff positions used are 4a (0,0,0) for Ta and 4b (0.5, 0.5, 0.5) for C.

(a) X-ray line profile analysis. X-ray line profile analysis allows for meaningful determination of size and strain in the synthesized powders which can be used to correlate the microstructure with their properties. In the present case individual peaks have been fitted with pseudo-Voigt function which is a linear combination of Lorentzian (L(x)) and Gaussian (G(x)) functions:^30,31

F(x) = ηL(x) + (1 − η)G(x) (1)

where, η: mixing parameter.

Based on the better results for our earlier work²⁹ we use the Voigt double-line integral-breadth methods for the XRD line profile analysis. The double line analysis was carried out using Origin™ software.

(b) Voigt double-line integral-breadth method. The Lorentzian and Gaussian components of the size and strain integral breadths follow the convolution principle and the various components can be separated. Warren definition of mean-square strain requires that the Lorentz and Gauss distortion integral breadths depend differently on s and the multiple-line Voigt method by Langford becomes:^31,32where, β_SL, β_DL are modified Lorentzian components of size and strain integral breadth. β_SG, β_DG are modified Gaussian components of size and strain integral breadth.And,

This method is called “Double-Voigt” method (D-V method). β_L and β_G² are plotted w.r.t. s² and the corresponding strain and size components β_SL, β_DL(hkl), β_SG, β_DG(hkl) are obtained. The analytical expressions for the volume weighted domain size, 〈D_v〉, mean square strain, 〈ε²〉 and size coefficients, A_s(L) are:^30–32

A_s(L) = exp(−2Lβ_SL − πL²β_SG²) (7)

Eqn (6) is used to determine the m.s.s. strain distribution w.r.t. size (L) and direction. The data for one arbitrary size (5 nm) was used for comparison of strains (root mean square strain) between different samples. The 5 nm was chosen based on the particle sizes (surface and volume weighted) obtained for our samples and literature survey.³² The initial slope of the A_s(L) vs. L graph is used to determine the surface weighted domain size (〈D_s〉).³⁰ Surface weighted domain size determination of the powder is important for the applications where the active surface area is important e.g. catalysis.³³

DSC/TG (NETZSCH STA 449F3) was done at a heating rate of 5 °C min⁻¹ in air atmosphere to determine the phase transitions and thermal stability of the materials. The micro-structural features of synthesized TaC powders were analyzed with field-emission scanning electron microscope (FE-SEM) (SIGMA Carl Zeiss) operating at 5 kV and transmission electron microscope (TEM) (JEOL 2100F) operating at 200 kV. The N₂ sorption studies for surface analysis were conducted using a Tristar 3000 (Micromeritics) to determine the Brunauer–Emmett–Teller (BET) surface area, the pore size, and the pore volume.

3 Results and discussion

3.1 X-ray diffraction analysis (XRD) of the synthesized powders

Fig. 1 shows the XRD data for the starting Ta₂O₅ powder and the samples synthesized for the initial Ta : C ratio of 1 : 1 at various temperatures. For all the samples the autoclave was soaked for 1 hour at the maximum temperature. XRD data shows that till 650 °C the material is still Ta₂O₅ but the XRD peaks are broadened indicating a decrease in particle size. This is followed by formation of pure cubic phase Ta (ICDD – pattern 00-004-0788) at 700 °C. The Ta particles formed at 700 °C are small in size as well as highly strained as evidenced by the broad XRD peaks shifted towards lower angles as compared to standard reference.³⁴ At 750 °C TaC (ICDD pattern – 01-077-0205) and Ta₂C (ICDD – pattern 00-032-1280) coexist. The XRD data shows that the pure phase cubic TaC is formed only for 800 °C. Based on these results and the thermal calculations we have proposed the formation mechanism for TaC (discussed in Section 4).

Fig. 1 XRD results of the Ta₂O₅ powder and the acid leached samples synthesized from it with different soaking temperatures.

The final soaking temperature of 800 °C was thus chosen for the further studies: (a) effects of changing initial carbon concentration (1 to 90) on the final product (Fig. S1, (ESI†), Table 1) and (b) evolution of TaC with soaking time (1 h to 15 h) (Fig. S2, ESI,† Table 2).

Fig. 2(a) gives the Rietveld refinement plots for the 03C sample. The lattice parameter (a), R_wp and χ² – values obtained from the Rietveld refinement are listed in the Tables 1 and 2. The lattice parameter of the synthesized samples from the Rietveld refinement has been used to determine x in TaC_x:³⁵

a (Å) = 4.3007 + 0.1563x (8)

Fig. 2 (a) Rietveld refinement plots for the 03C sample. Tick marks indicate allowed peak positions. (b) The result of the pseudo-Voigt curve fit routine for the (111) peak of 03C sample.

Fig. 2(b) shows the results obtained from the fitting of (111) peak with the pseudo-Voigt function for the 03C sample. Fig. S3 in ESI† gives the D-V integral breadth method analysis graphs for the 03C sample. Similar graphs have been used to analyse all the samples. Fig. S4 in ESI† shows the isotropic nature of the strain and variation of Fourier transform coefficients as a function of column length for 03C sample. Tables 1 and 2 also give the details of the size and strain values (root mean square strain, RMSS) obtained for the D-V integral breadth methods.

(a) Effects of changing initial carbon (C) concentration. Increasing the initial C concentration (from 1 to 90) results in slight changes in the particle size (Fig. 3(a)). This variation lies within the error limits of the XRD analysis. More interesting picture about the role of initial C concentration emerges from the study of the variation of a (and hence the carbon inside the carbide phase) and RMSS (Fig. 3(b)). For the lower C concentrations (≤5.5) an increase in the initial C concentration results in a slight decrease in a and many fold increase in the strain. We believe this is due to an increase in the thickness of the graphitic carbon outside the carbide phase which hinders the concentration induced diffusion of carbon into the particles. XRD data (Fig. S1, ESI†) clearly shows that with increasing C concentration the graphitic carbon peak at 2θ ∼26° slowly becomes more prominent. This result is also supported by the TG/DSC, SEM and TEM data (discussed in following sections). Increase in C concentration beyond 5.5 results in a sudden increase in the value of a and a corresponding decrease in RMSS. This indicates that now the concentration gradient in the system is sufficient to overcome the hindrance to diffusion from the graphitic layer and resulting in increased carbon in the carbide phase and lower strain. Further increase in the C concentration results in saturation of a and RMSS to a constant value (Fig. 3(b)). The final saturation value of a is slightly lower than the maximum value since the graphitic carbon layer is increasing many-fold on the outside. The existence of critical initial carbon stoichiometry and its effect on the final powder properties has been discussed in literature earlier also for other synthesis methods.²³

Fig. 3 (a) Particle size variation with the initial carbon concentration. (b) Lattice constant (a) and root mean squared strain (RMSS) variation with the initial carbon concentration. The x-axis (carbon concentration) is plotted on logarithmic scale for clarity of data at the lower concentration.

Thus, the diffusion of carbon into the particle and hence the final carbon content in the TaC is a complex function of initial C concentration since on one side an increase in carbon concentration creates a concentration gradient which will enhance the process of diffusion but at the same time the extra carbon forms graphitic carbon network outside the particles which hinders the diffusion of carbon into the particle.

(b) Evolution of TaC with soaking time. For 01Q sample (sample is quench cooled outside the furnace after soaking) XRD pattern analysis shows the presence of single phase cubic TaC_x with x = 0.92. This means that at 800 °C the formation TaC, which is essentially a multistep reduction and carburization process, is very fast. This happens due to fast diffusion rate of C as well as the free energy considerations (the steps and the energy considerations are discussed in the formation mechanism section). So the carbon, which diffuses into the crystallites and reacts for the formation of TaC, either may not get sufficient time for diffusion or if diffused might not be in optimum locations. This is evidenced by the high strain in the sample. Even the slow cooling of the system (01T sample) allows only for more carbon to diffuse into the crystallites such that the particle volume weighted size as well as the strain increases. It is finally for the 2 h soaking time followed by slow cooling that the system attains a lower strain as well as the smaller crystallite size in spite of further increase in the carbon content. We believe this decrease in the crystallite size and strain happens since the larger soaking time has allowed for the carbon within the TaC to finally achieve its best possible configuration. As we further increase the soaking time the diffusion of carbon into the crystallites continues.

A closer look at the further time evolution of the size and strain data (Table 2) shows that the crystallites are now in turn becoming smaller in size and strained followed by the larger size and lower strain. We believe this to be a clear indication that once the carbon inside the TaC crystallites have achieved the lowest energy configuration the further heating results in grain growth which seems to follow the simultaneous grain rotation and grain boundary (GB) migration model.^36–38 In this mechanism initially the grain boundaries grow/migrate resulting in decreased grain size and increased strain in the system. This is followed by the grain rotation across the lowest angle grain boundary resulting in a larger grain as the GB between them vanishes. At the time of the grain shrinkage the GB becomes vastly enhanced so the samples show higher average strain and smaller grain size. An important consequence of this type of grain growth for TaC is that the manifold increase in strain just before rotation also results in expulsion of some of the carbon from the crystallites. So for these samples the carbon content of the carbide phase also alternates with soaking time and grain growth.

In the present system the evolution of the TaC nanopowder with increasing soaking time at 800 °C can essentially be divided into two regimes: (i) the formation of low strain TaC followed by (ii) grain growth via simultaneous grain rotation and GB movement mechanism. Fig. 4 show the proposed grain evolution mechanisms for TaC.

Fig. 4 (a) Initial growth mechanism leading to low strain samples of TaC (the unequal distance between the planes inside the crystallite represents the presence of strain). (b) Simultaneous GB migration/grain rotation mechanism of grain growth for the low strain crystallites (the arrow inside the crystallites represents the orientation).

The surface weighted size for all the samples is smaller than the volume weighted size but follows the same trend as the volume weighted size for the time and carbon variation. This we believe is an indicator that for the temperature and time scales used in present setup the grains do not have anisotropic growth.³³

3.2 Thermal analysis of synthesized powders

The thermal stability and oxidation resistance of the synthesized powders was analyzed by DSC/TG. Fig. 5 shows the DSC-TG curves for the 01C sample and this data is representative for all the synthesized samples. The main observations from the data are as follows: initial heating from room temperature onwards results in mass loss followed by gain in mass. This initial mass loss is signature of the surface adsorbed species on the surface of the particles and hence gives information about the surface area of the powder samples.²³ For our samples, before oxidation nearly 3 mass% of surface adsorbed species is removed indicating the availability of potentially large surface area for adsorption. The start of mass gain in TG curve is an indicator of the initial stability of the synthesized samples (and hence the handling temperatures). The maximum mass gain peak in TG is due to the conversion of TaC into Ta₂O₅.⁹ This conversion is a two step process: release of carbon from the TaC followed by oxidation. The temperatures for these are indicated by the two peaks in DSC. The high temperature inflection point (≥700 °C) in TG curve, with a corresponding peak/hump in DSC, occurs at nearly the same temperature for all the samples and is attributed to oxidation and combustion of the carbonaceous residue.⁸ Table 3 gives the data obtained from the detailed analysis of the DSC/TG data for the select samples.

Fig. 5 DSC-TG curves of 01C sample.

Table 3 The data obtained from the analysis of the DSC/TG curves for the synthesized samples

Sample name	Initial stability (°C)	C removal peak (°C)	Oxidation peak (°C)	Mass gain (%)	Final mass (%)	Free carbon (%)	Initial mass loss (%)
02T	255	305	434	5.9	91	20.4	5.3
10T	238	482	582	1.9	89	22	4.7
01C	212	330	507	7.6	102	10.7	2.7
02C	185	313	462	7.5	102	10.7	2.2
03C	255	305	434	5.9	91	20.4	5.3
20C	195	342	572	4.3	73	35.9	2.2
60C	211	325	556	1.06	46	59.7	2.3

---

If the sample consists of pure TaC, its oxidation will result in 14.5% increase in mass. In our samples this is never achieved. For our samples the maximum increase encountered is 2–8 mass% after the initial weight loss stops. The TG curves become stable for all the samples at ∼800 °C. The residual mass at these temperatures is used to calculate the free carbon content in the synthesized samples using the formula:³⁹

where, m_i: initial mass, m_f: final mass, M_Tac: molecular mass of Tac, M_Ta2O5: molar mass of Ta₂O₅.

The smaller increase in mass for our samples as compared with the expected value is due to the presence of carbonaceous content (graphitic as well as amorphous) in the synthesized samples.⁹ As the sample is heated the mass increase due to oxidation is being offset by the decrease in mass due to oxidation of carbonaceous residue. This implies that if we have more free carbonaceous content in the samples the %mass gain will be smaller. The data in Table 3 shows as the initial carbon content is increased the %free carbon in the synthesized samples increases with a corresponding decrease in %mass gain in TG curves.

As we increase the initial C the net stability of the synthesized samples (the location of the C removal and oxidation peaks in DSC) decreases. This can be attributed to smaller size and increased strain in the particles (Table 2). As we increase the initial carbon content from 2 (02C) to 3 (03C), the free carbon content in the samples nearly doubles whereas the carbon within the TaC remains the same (Table 2). Thus this thermal data supports our earlier conjecture that extra initial C forms a thick carbonaceous network around the particles which prevents the movement of carbon into the particles.

The large initial mass loss for 03C sample corresponds with the smallest surface weighted size for this sample (Table 1). When the soaking time for the sample is increased (02T to 10T) the locations of the DSC peaks indicate that the particles are more stable against oxidation. This is supported by the XRD data which shows that the sample consists of low strain large sized particles. The decrease in the initial stability for 10T sample is due to the increased free carbon content. This conjecture is further supported by the thermal data for the 20C and 60C data. For both the cases the initial stability of the sample decreases as the free carbon content increases.

Thus we conclude that for a nano-carbide sample initially the increased free carbon content increases the initial stability of the powders but on further increase (>20%) it decreases. Comparing 01C and 60C samples clearly shows that very high free carbon content in the system is detrimental to the quality of the sample by reducing not only the amount of the useful material but also deceasing the stability.

Complete analysis of XRD and thermal data from Tables 1–3 indicates that the initial stability and hence the handling temperature of the samples depends on the free carbon content whereas the oxidation stability depends on the carbon content in the carbide phase as well as the size of the particles.

3.3 Microstructure analysis

Fig. 6(a) shows the representative FE-SEM images of the final product synthesized for 01C sample. Agglomeration and carbon network are clearly visible, latter becoming denser for 03C sample (Fig. S5 in ESI†). The average size calculated from 01C sample images is in the range of 7–10 nm.

Fig. 6 (a) FE-SEM, (b) TEM and (c) HR-TEM images for the 01C sample.

Fig. 6(b) shows the representative TEM images of the synthesized TaC nanoparticles for 01C sample. The presence of carbon coating on the surface of each individual particle is clearly visible. Also visible is the carbonaceous network in which the particles are embedded. The morphology of carbon coated nanoparticles varies from faceted to spherical and they have a tendency to aggregate. Fig. 6(c) gives the High Resolution-TEM (HRTEM) image of a single crystalline TaC nanoparticle and shows the (111) facets of the particle. The distance between the adjacent lattice fringes is the interplanar distance of cubic TaC (111), which is 0.26 nm (ICDD pattern – 01-077-0205). This confirms that the synthesized TaC powder has cubic crystalline structure.

3.4 BET surface area analysis

N₂ sorption studies were done to obtain the BET surface area of the synthesized samples. The results are listed in Table 4. Analysis of the BET and XRD data shows that the largest surface area and the smallest pore diameter is for the samples with smallest size and it decreases with increasing size/soaking time.

Table 4 The data obtained from the analysis of the BET curves for the synthesized samples

Sample name	BET surface area (m^2 g^−1)	Average pore diameter (nm)	Pore volume (cm^3 g^−1)
01C	54.81	8.26	0.097
02C	56.4	10.52	0.083
03C	72.97	6.78	0.112
01Q	78.53	8.14	0.139
02T	72.97	6.78	0.112
10T	61.07	8.31	0.122

---

Fig. S6 in ESI† shows the N₂ adsorption–desorption isotherms for the samples with increasing soaking times. The adsorption isotherms exhibit the characteristics of a type-II isotherm according to IUPAC classification.^3,40 An empirical classification of the hysteresis loops which gives information about the texture of the adsorbent has also been given by IUPAC and the isotherms show H-4 hysteresis characteristics.^40,41 This implies that the powders are forming a complex structure with both micropores and mesopores. This behaviour is because of the agglomeration of particles in the synthesized product.

4 Proposed mechanism for TaC formation

During heating of the autoclave, Mg being highly reactive substance absorbs oxygen from the air present inside the autoclave and forms MgO. MgO, being one of the most active catalysts for reduction, results in reduction of the acetone to hydrogen and carbon above 200 °C.⁴² The decomposition of acetone leads to an excess of C and H₂ within the autoclave.⁴³ The carbon (C) and hydrogen (H₂) produced as a result of decomposition along with the Mg can act as reducing agents for Ta₂O₅. The first step in reduction of Ta₂O₅ occurs via conversion to TaO₂. Four possible reduction reactions can be expressed as:

2Ta₂O₅ + C → 4TaO₂ + CO₂ (10)

Ta₂O₅ + C → 2TaO₂ + CO (11)

Ta₂O₅ + H₂ → 2TaO₂ + H₂O (12)

Ta₂O₅ + Mg → 2TaO₂ + MgO (13)

The ΔG values for the above reactions at the standard pressure at different temperatures can be calculated by:

ΔG = ΔG⁰ + ΔH − TΔS (14)

C_p = Δa + ΔbT + ΔcT⁻² (17)

where, Δa, Δb, and Δc are the thermodynamic data of reactants and products at the standard state.

The variation of ΔG with temperature for the Ta₂O₅ reduction with C, H₂ and Mg is shown in Fig. 7(a). The curves show that although all the reactions are non-spontaneous (ΔG > 0) the reduction becomes more favorable as the temperature increases. The reaction involving Mg is more favourable at all temperatures. The reduction reactions involving Mg and C are a solid-solid reaction whereas that for H₂ is solid–gas reaction.⁴⁴

Fig. 7 Thermodynamic calculations for the possible reactions during reduction and carburization.

The H₂ being a small molecule has a large diffusion coefficient. So once formed (especially in excess) it is able to diffuse into the Ta₂O₅ particles whereas C coats the outside of the Ta₂O₅ particles. Even at lower temperatures (<650 °C) the reduction due to hydrogen (which is exothermic) is possible due to the size of the initial powder (130–220 nm) which allows for large contact and solid–gas nature of the reaction.⁴⁵ This reduction is extremely small in amount and only occurs for the diffused hydrogen because of low partial pressure of H₂O inside.^44,45 The exothermic reduction reaction leads to formation of steam which fragments Ta₂O₅ powders. This conjecture is supported by the broadened XRD peaks at lower temperatures and no signature of TaO₂. The small final size of the synthesized powders in our work is attributed to this initial reduction of size for Ta₂O₅.³⁴

The fragmented Ta₂O₅ gets coated immediately with carbon present in the autoclave. This carbon coating prevents the fragmented oxide particles from coalescing. The carbon coating also helps to increase the reaction rate for reduction and carburization by ensuring the close proximity of the reactants. For the fragmented Ta₂O₅ the reduction process proceeds at a faster rate due to reduced size and higher surface area of fragmented particles.⁴⁶ The decrease in the transformation temperature seen in our work is a combined effect of the reduced size of the Ta₂O₅ as well as addition of Mg which is energetically more favourable reductant. This is supported by existence of critical initial stoichiometry of Mg needed for complete transformation into TaC at 800 °C. We believe once the reaction starts the reduction proceeds via C and H₂ reduction pathways as well as Mg. This will be expected since the C and H₂ are in more intimate contact with the initial powders whereas Mg is in the form of solid powder.³⁴ Also all the reactions are exothermic (data not shown) so once the reaction starts the local temperature will increase making all reaction pathways possible. The detailed morphology of the Mg powder may also have an effect on its critical stoichiometry and needs to be studied further. As the temperature increases the reduction reactions become more favourable increasing the reaction rate.

For the further reduction of TaO₂ the following reactions are possible:

TaO₂ + C → TaO + CO (18)

TaO₂ + H₂ → TaO + H₂O (19)

TaO₂ + Mg → TaO + MgO (20)

TaO₂ + 2C → Ta + 2CO (21)

TaO₂ + 2H₂ → Ta + 2H₂O (22)

TaO₂ + 2Mg → Ta + 2MgO (23)

Fig. 7(b) clearly shows that reduction of TaO₂ to Ta in single step is a spontaneous process with reduction reactions involving C and H₂ being comparable but that for Mg being more probable from the energy point of view at the temperature of interest (≤700 °C). Since the TaO₂ particles are small in size, completely, encapsulated in carbon and reaction is spontaneous in nature so this reaction proceeds at a very fast rate and the Ta particles formed will be strained. This is supported by the XRD data which clearly shows the formation of pure cubic phase highly strained Ta nanoparticles at 700 °C.

The final possible carburization reactions are:

Ta + C → TaC (24)

Ta + C → Ta₂C (25)

Ta₂C + C → 2TaC (26)

Reactions (25) and (26) are competing reactions. Thermodynamic calculations show that the formation of Ta₂C is more favourable (Fig. 7(c)). So the Ta nanoparticles are carburized initially to Ta₂C and finally to TaC. Both these reaction steps will be very fast as both the reactions are spontaneous. This is supported by the XRD data where Ta₂C and TaC are formed simultaneously for the 750 °C sample. But the complete conversion to TaC takes place only at 800 °C. This indicates that the diffusion of carbon inside the particles is the reaction rate hindrance step. Increase in temperature increases the diffusion and results in complete final phase formation in 1 h.

So the complete reaction pathway for the synthesis of TaC from Ta₂O₅ is:

The formation mechanism is depicted in Fig. 8.

Fig. 8 The mechanism for transformation of Ta₂O₅ into TaC nanoparticles. The arrows within the stressed oxide particle represent the beginning of cracks.

5 Conclusions

High surface area carbon-coated nanosized TaC powders (5–8 nm) have been synthesized successfully at low temperature in single step by the chemical reaction method using Ta₂O₅ as the precursor with acetone as carbon source and Mg as catalyst. The pure phase cubic TaC is obtained within 1 hour at 800 °C. The lowering of the reaction temperature is critically dependent on the initial Mg stoichiometry. The effect of initial C concentration and soaking time on microstructure and property evolution of TaC have been studied.

Studying the evolution of TaC powders with increasing soaking time revealed that initially the powders are strained and require about 2 hours of soaking at 800 °C to achieve the low strain state with carbon diffusion being the rate limiting step. Soaking for higher times leads to increase in crystallite size which follows the simultaneous grain boundary migration and grain rotation model. We believe that the grain rotation effects are visible in our samples even at the comparatively low temperatures (for TaC) like 800 °C due to the small size of the initial powders (since the grain rotation rate varies inversely as 1/r⁻⁵ with the grain size) and the high pressure environment of the autoclave.^38,47

A detailed study of increasing the initial C concentration and soaking times brings forward complexity of the system dependence on these parameters. The initial C stoichiometry of 1 results in the formation of TaC_x with x = 0.96. As the C in the initial mixture is increased a strong carbon network (consisting of amorphous and graphitic components) is formed outside the particle which hinders the flow of carbon into the carbide. Quench cooled sample for 1 h results in TaC_x with x = 0.92. As the soaking time is increased the carbon diffusion inside the carbide increases along with grain size via the simultaneous grain boundary movement and grain rotation model. So the variation of initial C as well as soaking time can be used for tailoring the C concentration in carbide. The best TaC_x with x = 0.99 is achieved either for initial C ratio of 15 in 2 h or for C stoichiometry of 3 in 7 hours. With former having higher carbonaceous content on the outside but smaller strain. Depending on the requirement of the system/application the desired sample can be prepared. The overall stability of the powders against oxidation is determined by the particle size as well as the carbon content inside the TaC. The surface area and pore size/volume is determined by the particle size.

Thermodynamic calculations, thermal analysis and XRD analysis results have been used to predict the mechanism for the formation of the carbon coated nano-TaC particles from Ta₂O₅. It emerges from the analysis that the fine size in our synthesized powders is due to initial production of H₂ in the autoclave and, the lowering of the reaction temperature is due to the combined effect of the initial size reduction as well as addition of Mg. In this process the carbon content of TaC, size as well as the strain of the synthesized powders and the %free carbon content can be tailored as per the requirement for the given application. The microstructure evolution studies carried out here if extended can also help to predict the stabilization of dense nano-structures as expected from rapid densification techniques such as spark plasma sintering.^38,47

Acknowledgements

The authors are grateful to IIT Roorkee for providing FE-SEM, SAI Labs, Thapar University for providing XRD and AIRF, JNU, Delhi for providing TEM. All the authors want to thank Thapar University for providing financial support.

References

1. M. Ali, M. Ürgen, M. A. Atta, A. Kawashima and M. Nishijima, Mater. Chem. Phys., 2013, 138, 944–950.
2. C. L. Yeh and E. W. Liu, J. Alloys Compd., 2006, 415, 66–72.
3. Y.-J. Lee, S. H. Kim, T.-H. Lee, H. H. Nersisyan, K.-H. Lee, M.-H. Han, S.-U. Jeong, K.-S. Kang, K.-K. Bae and J.-H. Lee, Chem. Eng. Sci., 2014, 107, 227–234.
4. J.-G. Choi, J. Porous Mater., 2013, 20, 1059–1068.
5. L. Silvestroni, A. Bellosi, C. Melandri, D. Sciti, J. X. Liu and G. J. Zhang, J. Eur. Ceram. Soc., 2011, 31, 619–627.
6. J. P. Kelly, R. Kanakala and O. A. Graeve, J. Am. Ceram. Soc., 2010, 93, 3035–3038.
7. J. Ma, Y. Du, M. Wu and M. C. Pan, Mater. Lett., 2007, 61, 3658–3661.
8. D.-H. Kwon, S.-H. Hong and B.-K. Kim, Mater. Chem. Phys., 2005, 93, 1–5.
9. N. S. Alhajri, H. Yoshida, D. H. Anjum, A. T. Garcia-Esparza, J. Kubota, K. Domend and K. Takanabe, J. Mater. Chem. A, 2013, 1, 12606–12616.
10. J. Polonský, I. M. Petrushina, E. Christensen, K. Bouzek, C. B. Prag, J. E. T. Andersen and N. J. Bjerrum, Int. J. Hydrogen Energy, 2012, 37, 2173–2181.
11. A. L. Stottlemyer, T. G. Kelly, Q. Meng and J. G. Chen, Surf. Sci. Rep., 2012, 67, 201–232.
12. D. J. Ham and J. S. Lee, Energies, 2009, 2, 873–899.
13. A. Ishihara, M. Tamura, Y. Ohgi, M. Matsumoto, K. Matsuzawa, S. Mitsushima, H. Imai and K. Ota, J. Phys. Chem. C, 2013, 117, 18837–18844.
14. H. Zhang, J. Liu, Z. Tian, Y. Yea, Y. Cai, C. Liang and K. Terabe, Carbon, 2016, 100, 590–599.
15. M. Ma, W. Shen, P. Zhang, J. Zhang, Q. Wang and C. Ge, Mater. Lett., 2011, 65, 96–99.
16. V. G. Sevast'yanov, E. P. Simonenko, N. A. Ignatov, Y. S. Ezhov and N. T. Kuznetsov, Inorg. Mater., 2010, 46, 495–500.
17. H. I. Won, N. Hayk, C. W. Won and H. H. Lee, J. Mater. Sci., 2011, 46, 6000–6006.
18. P. G. Li, M. Lei and W. H. Tang, Mater. Res. Bull., 2008, 43, 3621–3626.
19. P. G. Li, M. Lei, Z. B. Sun, L. Z. Cao, Y. F. Guo, X. Guo and W. H. Tang, J. Alloys Compd., 2007, 430, 237–240.
20. A. M. Nartowski, I. P. Parkin, M. Mackenzie and A. J. Craven, J. Mater. Chem., 2001, 11, 3116–3119.
21. Y.-J. Chen, J.-B. Li, Q.-M. Wei and H.-Z. Zhai, Mater. Lett., 2002, 56, 279–283.
22. H. Xiang, Y. Xu, L. Zhang and L. Cheng, Scr. Mater., 2006, 55, 339–342.
23. J. P. Kelly and O. A. Graeve, J. Am. Ceram. Soc., 2011, 94, 1706–1715.
24. Y. Yosida and I. Oguro, Phys. C, 2006, 434, 138–140.
25. H. He, K. Zhou, X. Xiong and B. Huang, Mater. Lett., 2006, 60, 3409–3412.
26. H. He, K. Zhou and X. Xiong, J. Mater. Sci. Technol., 2005, 21, 381–385.
27. A. Kumar, K. Singh and O. P. Pandey, Phys. E, 2009, 41, 677–684.
28. G. Singla, K. Singh and O. P. Pandey, Ceram. Int., 2014, 40, 5157–5164.
29. L. K. Brar, G. Singla and O. P. Pandey, RSC Adv., 2015, 5, 1406–1416.
30. D. Balazar, Voigt-function model in diffraction line-broadening analysis, in Defect and Microstructure Analysis by Diffraction, ed. R. L. Snyder, J. Fiala and H. J. Bunge, Oxford University Press, Oxford, U.K., 1999, pp. 94–126.
31. K. Santra, P. Chatterjee and S. P. S. Gupta, Bull. Mater. Sci., 2002, 25, 251–257.
32. S. Vives, E. Gaffet and C. Meunier, Mater. Sci. Eng., A, 2004, 366, 229–238.
33. H. G. Merkus, Particle Size Measurements: Fundamentals, Practice, Quality, Springer Science & Business Media B.V, Dordrecht, Netherland, 2009.
34. L. K. Brar, G. Singla, N. Kaur and O. P. Pandey, J. Therm. Anal. Calorim., 2015, 119, 175–182.
35. A. L. Bowman, Phys. Chem., 1961, 65, 1596–1598.
36. A. J. Haslam, S. R. Phillpot, D. Wolf, D. Moldovan and H. Gleiter, Mater. Sci. Eng., A, 2001, 318, 293–312.
37. D. Moldovan, D. Wolf, S. R. Phillpot and A. J. Haslam, Acta Mater., 2002, 50, 3397–3414.
38. R. Chaim, Scr. Mater., 2012, 66, 269–271.
39. A. L. Tomas-Garcia, Q. Li, J. O. Jensen and N. J. Bjerrum, Int. J. Electrochem. Sci., 2014, 9, 1016–1032.
40. W. Wang, P. Liu, M. Zhang, J. Hu and F. Xing, Open J. Compos. Mater., 2012, 2, 104–112.
41. M. Thommes, Chem. Ing. Tech., 2010, 82, 1059–1073.
42. V. K. Díez, C. R. Apesteguía and J. I. Di Cosim, Lat. Am. Appl. Res., 2003, 33, 79–86.
43. A. Kumar, K. Singh and O. P. Pandey, J. Mater. Sci. Technol., 2014, 30, 112–116.
44. S. Luidold and H. Antrekowitsch, JOM, 2007, 59, 20–26.
45. T. H. Kim, J. H. Yu and J. S. Lee, Nanostruct. Mater., 1997, 9, 213–216.
46. H.-C. Huang and T.-E. Hsieh, J. Appl. Polym. Sci., 2010, 117, 1252–1259.
47. M. Upmanyu, D. J. Srolovitz, A. E. Lobkovsky, J. A. Warren and W. C. Carter, Acta Mater., 2006, 54, 1707–1719.

---

Footnote

† Electronic supplementary information (ESI) available: Supplementary material (XRD data for the carbon and time series, double-Voigt analysis graphs and BET graphs) is available in the online version. See DOI: 10.1039/c6ra24484j

---