Titanium Carbide Nanocrystals Synthesized from a Metatitanic Acid-Sucrose Precursor via a Carbothermal Reduction

A TiC powder is synthesized from amicron-sizedmesoporousmetatitanic acid-sucrose precursor (precursorM) by a carbothermal reduction process. Control specimens are also prepared using a nanosized TiO 2 -sucrose precursor (precursor T) with a higher cost. When synthesized at 1500C for 2 h in flowing Ar, the characteristics of the synthesized TiC from precursor M are similar to those of the counterpart from precursor T in terms of the crystal size (58.5 versus 57.4 nm), oxygen content (0.22wt% versus 0.25 wt%), and representative sizes of mesopores: approximately 2.5 and 19.7–25.0 nm in both specimens. The most salient differences of the two specimens are found in the TiC from precursor M demonstrating (i) a higher crystallinity based on the distinctive doublet peaks in the high-two-theta XRD regime and (ii) a lower specific surface area (79.4 versus 94.8m/g) with a smaller specific pore volume (0.1 versus 0.2 cm/g) than the counterpart from precursor T.


Introduction
Titanium carbide (TiC) is a representative material of transition metal carbides.It is a technologically important material primarily for cutting tools, abrasives, and polishing pastes.Because (i) its mechanical properties are similar to those of tungsten carbide (WC) that uses expensive cobalt as the sintering aid and (ii) it is sintered with a cheaper sintering aid (nickel), TiC is also considered as a partial substitute for WC.TiC has been synthesized via various processing routes including combustion reactions [1][2][3][4][5], sol-gel and solution processes [6][7][8][9][10][11], gas phase or thermal plasma reactions [12][13][14][15][16][17][18], mechanothermal reactions [19,20], and Mg-thermal reductions [21][22][23], while commercial production primarily uses a carbothermal reduction process  due to its low cost.The chemical reaction in the carbothermal reduction process is as follows: The Gibbs free energy change for reaction (1) was calculated at varying temperatures using the thermodynamic data given in [46] and the results are presented in Figure 1.As seen in this figure, the Gibbs free energy change for reaction (1) decreases as the partial pressure of CO gas ( CO ) decreases from the standard value (1 atm); therefore, synthesis of TiC via reaction (1) is thermodynamically favorable even at room temperature provided  CO is sufficiently low.The decrease of  CO can be achieved suitably via flowing an inert gas such as Ar over reaction sample: it removes the CO gas molecules from the reaction site.However, even under the flowing inert gas atmosphere, the synthesis of TiC via reaction (1) is usually conducted at the temperature range of 1700-2300 ∘ C [24] for 10-24 h due to the kinetic barrier.
From the perspective of reaction kinetics, intimate contact between TiO 2 and carbon is crucial to facilitate reaction (1) at high temperatures.In the conventional mechanical mixing of carbon powder and TiO 2 powder, it is difficult to achieve a thoroughly mixed state of the two constituents without agglomeration of the same kind.In this regard, the following studies were conducted.In Koc and Folmer [25], hydrocarbon gas was pyrolyzed, followed by the deposition of the pyrolyzed carbon onto the surfaces of TiO 2 (P-25) particles.While this method ensured intimate contact between the TiO 2 and carbon source, the pyrolysis of the hydrocarbon in a rotating chamber with the TiO 2 powder is not simple, and it imposes a nonnegligible cost.Shin et al. [26] Figure 1: Gibbs free energy changes with temperature for reaction (1) at varying partial pressures of CO gas.and Gotoh et al. [27] utilized cellulose as a carbon source but they used TiO 2 sources such as Tyzor-LA or TiO 2 hydrosol.However, these processes also impose a cost burden due to the expensive TiO 2 source.In 2009 [28], our research group reported an economical method that ensures intimate contact between the titanium source and carbon source: TiO 2 powder (P-25) was suspended in an aqueous sugar solution (sucrose, C 12 H 22 O 11 ) and dried to form a TiO 2 core-sucrose shell precursor.In this precursor, each TiO 2 particle was coated with the carbon source (sucrose molecules) without agglomeration of the TiO 2 particles.When this precursor was heat-treated at 1600 ∘ C in flowing Ar, the synthesized TiC exhibited a specific surface area of 137 m 2 /g with an oxygen content of 0.42 wt%.
In this paper, we report an even lower-cost method to carbothermally synthesize TiC from a micron-sized mesoporous metatitanic acid-sucrose precursor.Metatitanic acid (MTA) is an interim product in the process of producing TiO 2 by the sulfuric acid process; the price of MTA should be lower than TiO 2 .It is a hydrated TiO 2 with a mesoporous nature (an average pore size of 3.8 nm [47]) and a high specific surface area of approximately 337 m 2 /g [47].Due to its mesoporous nature, water-soluble carbon sources such as sucrose molecules can be suitably impregnated into MTA mesopores in an aqueous solution.This impregnation of sucrose into MTA mesopores is anticipated to ensure intimate contact between the sources of TiO 2 and carbon, as well as to reduce the distance of the material transport between the two sources for carbothermal reduction.This study investigates whether there is benefit in using the new MTA-sucrose precursor in synthesizing TiC by carbothermal reduction compared with using the TiO 2 -sucrose precursor that was reported in our previous study [28].

Materials and Methods
A mesoporous MTA (Maanshan Starplanet Chemical Co. Ltd., Ma'anshan, Anhui, China) with an average particle size  1).The heat-treated MTA was transferred to an aqueous sucrose solution.In this stage, the weight ratio of sucrose to MTA was 3.8 based on separate trials from the perspective of finding the minimum required ratio to produce the single crystalline phase of TiC without titanium suboxides when synthesized at 1300 ∘ C.Then, the MTA slurry was ultrasonicated for 2 h in order to impregnate the aqueous sucrose molecules into the pore channels of the heat-treated MTA.Then, the sucroseimpregnated MTA slurry was dried at 110 ∘ C for 6 h in an oven, which resulted in a cake form.This sucrose-impregnated MTA is referred to as precursor M. As a control specimen, a sucrose-coated TiO 2 (P-25, nanopowder) precursor was prepared following [28], and it is referred to as precursor T.
For the subsequent carbothermal reduction process, either precursor M or T was transferred into an alumina boat, which was loaded to an alumina tube with a 60 mm inner diameter.The temperature of the tube furnace was increased to 1250 ∘ C-1500 ∘ C for 2 h in a flowing argon atmosphere (1 L/min).The type B thermocouple was sealed with an alumina protector, which was in contact with the outer surface of the tube.
The MTA powder was characterized by thermogravimetry (TG-DTA; Model STA PT1000, Linseis Messgeräte GmbH, Selb, Germany) at a heating rate of 10 ∘ C/min in air.Crystal phases of the reaction products and reactants were identified by X-ray diffraction (XRD) analysis (Model X'PertPro MPD, PANalytical BV, EA Almelo, Netherlands) using Cu K radiation with a scan speed of 0.4 ∘ /min and a step size of approximately 0.03 ∘ .The lattice parameter  was calculated from each diffraction angle () from the (111), (200), ( 220), (311), and (222) planes, followed by plotting  as a function of cos 2 , based on the relationship for cubic materials [48], as follows: where  o (the determined lattice parameter of the specimen) is the intercept of the ordinate in the  versus cos 2  plot and  is the slope.The size of the crystals () was quantified using the degree of broadening of XRD peaks based on the Scherrer equation [48]: where  is the full width at half maximum of the diffraction peak (in radian) and  is the wavelength of the X-ray.The  values from all available diffraction peaks were averaged in order to report the crystal size.The oxygen content in the synthesized samples was characterized using an elemental analyzer (Model EA1110, ThermoQuest, Rodano, Italy).The morphology of the powders was observed by scanning electron microscopy (SEM; Model EM-20, COXEM, Daejeon, Korea).The specific surface area was determined by the Brunauer-Emmett-Teller method (BET; Model Macsorb HM-1210, Mountech Co. Ltd., Tokyo, Japan).The pore size distribution was examined by the Barrett-Joyner-Halenda method (BJH; Model ASAP 2010, Micromeritics Instrument Co., Norcross, GA, USA) based on the N 2 adsorption and desorption at 77 K.The presence of the mesopores was checked using field emission transmission electron microscopy (FE-TEM; Model JEM-2100F, JEOL, Tokyo, Japan).

Results and Discussion
MTA, a hydrated form of TiO 2 , loses water as the temperature increases as seen in Figure 2. The adsorbed water is removed at 100 ∘ C (approximately 5 wt% of the total weight), the weakly bound crystal water at 100-550 ∘ C (approximately 11 wt%), and the strongly bound crystal water at 550-950 ∘ C (approximately 5 wt%).The average pore diameter and specific surface area of the as-received MTA and MTA heat-treated at 600 ∘ C (for 1 h in air) are presented in Table 1.The pore diameter increases with temperature, which naturally decreases the specific surface area.In Table 1, we acknowledge that the value of the specific surface area might be overestimated due to the presence of micropores; the filling of the micropores by N 2 molecules is not considered in the BET theory.The morphologies of the MTA heat-treated for 1 h in air at 100 and 600 ∘ C are presented in Figure 3.The 100 ∘ C heat-treated MTA represents the as-received MTA without adsorbed water.From Figure 3(b), the primary particles of the 600 ∘ C heat-treated MTA, which was used for the preparation of precursor M, are micron-sized: most primary particles were in the range of 1-3 m.The particle size of MTA did not decrease apparently by the heat treatment at 600 ∘ C compared with the 100 ∘ C heat-treated MTA (Figure 3(a)).This finding indicates that the elimination of weakly bound crystal water primarily results in increases in the size of the internal pores.As seen in Figure 3(c), P-25 TiO 2 , which was used to prepare precursor T, is nanosized.The size of its primary particles is known to be approximately 21 nm [8].
The crystal phases of the as-received MTA, heat-treated MTA at 600 ∘ C, and as-received TiO 2 (P-25) were identified by XRD, and the results are presented in Figure 4.The crystal form of the as-received MTA and the heat-treated MTA was anatase, while the as-received TiO 2 was the mixture of anatase and rutile as well known.Crystal sizes quantified by  the Scherrer equation were 5.8, 16.5, and 21.1 nm, for the asreceived MTA, heat-treated MTA (at 600 ∘ C), and as-received P-25 TiO 2 , respectively.Through combining these results with TG (Figure 2), the MTA is interpreted to be a mixture of anatase nanocrystals and crystal water.In contrast, P-25 TiO 2 is water-free, which is well known [49].
The crystal phases of the synthesized product from precursor M and those from precursor T were identified by XRD and the results are presented in Figure 5.As seen in the figure, at a synthesis temperature of 1300 ∘ C and higher, products from both precursors only exhibited the TiC phase without any titanium suboxides such as Ti 3 O 5 .The diffraction peaks in a high-two-theta regime (for the (110), (210), and (200) planes) are presented in Figure 6.As clearly seen in the figure, the synthesized TiC from precursor M exhibited two distinct diffraction peaks for each atomic plane.However, the counterpart from precursor T exhibited a single broad peak for the respective planes.As explained in the Appendix, the peak separation (the observation of two distinct peaks) originally results from two different wavelengths (K 1 and K 2 ) in the incoming X-ray and is augmented as the diffraction angle increases.In nanocrystals, because peak broadening results from both (i) imperfect crystallinity and (ii) the size effect, the two distinct diffraction peaks are barely observed even at a high diffraction angle, unless its crystallinity is very high.In this regard, crystallinity of TiC nanocrystals from precursor M is very high.In this study, the crystal sizes of TiC from precursors M and T are very similar.Therefore, the observed phenomenon of the peak separation in only TiC from precursor M (Figure 6) means that its crystallinity is certainly higher than that of the counterpart from precursor T. The origin of such higher crystallinity of TiC from precursor M may arise from the intimate contact of the reducing agent (residual carbon) with the TiO 2 source due to the impregnation of the waterdissolved sucrose molecules into the mesopores of MTA; the distance of the material transport may decrease due to the sucrose impregnation, which facilitates reaction (1).The morphology of the synthesized TiC powder at 1500 ∘ C from both precursors is shown in Figure 7.Although it is not easy to quantify the size of the primary particles in the SEM images with micron-scale bars, it is clear that the size of the primary particles of the TiC from both precursors (Figure 7) was significantly smaller than that of MTA (Figure 3(b)).The crystal size of TiC was quantified through applying the Scherrer equation [48] to the XRD patterns shown in Figure 5.The results are presented in Figure 8.The crystal size of the synthesized specimens is in the range of approximately 20-60 nm; it increased with temperature due to the grain growth.The crystal sizes of the synthesized TiC from both precursors were similar at a given temperature.Furthermore, their trend to increase with temperature increases was similar.
It is known that the nominal TiC phase that initially forms via carbothermal reduction is not pure TiC but TiC x O y , in which the presence of oxygen lowers the lattice parameter [4].The determined lattice parameters of the synthesized products from both precursors increased with the synthesis temperature (Figure 9(a)), which was accompanied by the decrease of the oxygen content with the synthesis temperature (Figure 9(b)).These observations indicate that the purification process of the TiC x O y phase toward pure TiC progresses as temperature increases.Based on the oxygen content (Figure 9(b)), when synthesized at 1500 ∘ C, the purity of TiC from precursor M was higher than that from precursor T and vice versa at lower temperatures.This result is inconsistent with the trend of the lattice parameter (Figure 9(a)) that exhibited similar lattice parameter values from 1300 ∘ C.However, in this case, greater value is placed on the oxygen content trend (Figure 9(b)), because, in general, there is high data scatter in the determination process of the lattice parameter for each sample from the XRD patterns using (2); for the investigated specimens, the coefficient of determination (the  2 value) was in the range of 0.4-0.7.When synthesized at 1500 ∘ C for 2 h, the lattice parameter  and oxygen content of the TiC synthesized from precursor M were 4.323 Å and 0.22 wt%, respectively, while those from precursor T were 4.323 Å and 0.25 wt%, respectively.
The specific surface area of the synthesized TiC from both precursors is illustrated in Figure 10 as a function of the synthesis temperature.The decreasing trend with the temperature might result from grain growth, particle agglomeration, and possible changes in the pore diameter with temperature (note that there are mesopores in the synthesized TiC, as seen in Figure 11).The TiC from precursor M exhibited a lower specific surface area at the respective temperature than that from precursor T. This observation is consistent with the specific pore volume (in cm 3 /g unit) of the former being significantly smaller than the latter: its value was only half of the latter when synthesized at 1500 ∘ C (Table 1).When synthesized at 1500 ∘ C, the specific surface areas were 79.41 and 94.78 m 2 /g, for the TiC from precursor M and precursor T, respectively.As for the reason that TiC from precursor M exhibits a lower specific pore volume and a lower specific surface area, this phenomenon may be associated with the sucrose impregnation into the mesopores of MTA, but a further study is needed to clarify this issue.
Figure 11(a) presents the nitrogen adsorption isotherm at 77 K and the pore volume versus pore diameter relationship for the TiC synthesized at 1500 ∘ C from precursor M, based on BJH analysis.The corresponding data for the counterpart from precursor T are presented in Figure 12(a).The BJH method for calculating the pore size distributions is based on a model of the adsorbent treated as a collection of cylindrical pores.The theory considers capillary condensation in the pores using the classical Kelvin equation [9], which assumes a hemispherical liquid-vapor meniscus and a well-defined surface tension.The nitrogen adsorption isotherms with pressure (Figures 11(a) and 12(a)) exhibited a type IV isotherm according to the classification of the International Union of Pure and Applied Chemistry (IUPAC), which is typical of mesoporous materials [50].Furthermore, the pore volume analyses (Figures 11(b) and 12(b)) indicated two representative pore sizes: one with a diameter of approximately 19.7-25.0nm and the other with a diameter of approximately 2.5 nm (Table 1).Regarding the pore size, there is no significant difference between the TiC from precursor M and that from precursor T.However, as stated, the specific pore volume of the former (0.1 cm 3 /g) was approximately half that of the latter (0.2 cm 3 /g; Table 1) when synthesized at 1500 ∘ C.These specific pore volumes indicate open porosity values of 0.02 and 0.04, respectively, based on a density of 4.93 g/cm 3 .
The presence of mesopores in the synthesized TiC was also verified by FE-TEM.In Figure 13(a), the synthesized TiC from precursor M shows a number of distinctive areas marked as P (approximately 3-4 nm in size).These areas are believed to be mesopores.In the result of the FE-TEM for the TiC from precursor T (Figure 13(b)), the images of the mesopore areas were not as distinctive as those in Figure 13(a), but there were many areas with different contrast (marked as P), which are believed to be mesopores.A clearer image of the mesopores in the TiC specimen from precursor T is available in [28].
The specific surface area and oxygen content of the TiC from precursor T in this study were slightly lower than those values in [28].When synthesized at 1500 ∘ C, these values were 94.8 m 2 /g and 0.25 wt%, respectively, in this study, while they were 147.5 m 2 /g and 0.49 wt% in [28].These differences might result from differences in experimental conditions, such as the thermocouple positions, different specimen environment in the furnace due to differences in the shape and size of the sample boat, different sizes in the alumina tube, and different analyses instruments.This study conducted control experiments using precursor T despite the availability of the data in [28] in order to eliminate these experimental inconsistencies.Therefore, a clear comparison reference (TiC from precursor T) was prepared in this study for the TiC samples from precursor M.

Conclusion
A TiC powder was prepared from a micron-sized mesoporous metatitanic acid-sucrose precursor (precursor M) by the carbothermal reduction process, and it was characterized.Reference specimens were also prepared using a nanosized TiO 2 -sucrose precursor (precursor T) with a higher cost.The TiC prepared using precursor M at 1500 ∘ C for 2 h in flowing Ar exhibited similar properties to those of the counterpart from precursor T in terms of the crystal size (58.5 versus 57.4 nm), oxygen content (0.22 wt% versus 0.25 wt%), and representative sizes of mesopores: approximately 2.5 and 19.7-25.0nm in both specimens.The most salient differences of the two specimens were found in that the TiC from precursor M demonstrated (i) a higher crystallinity based on the distinctive doublet peaks in the high-two-theta XRD regime and (ii) a lower specific surface area (79.4 versus 94.8 m 2 /g) with a smaller specific pore volume (0.1 versus 0.2 cm 3 /g) than that from precursor T.

Appendix
Consider ideal polycrystals with a perfectly identical atomic spacing ().The wider peak separation (the observation of two distinct diffraction peaks) at a high-two-theta regime that is observed in these crystals can be understood from (i) Bragg's law and (ii) the sine function characteristics, as follows.Bragg's law is where  is the wavelength of the X-ray,  is the spacing of the atomic plane (ℎ) diffracting the X-ray, and 2 is the diffraction angle.According to this law, the difference of Because the value of Δ is constant (0.003828 Å), Δ sin  is constant for a given -spacing.As apparent in the sin  versus  relationship (Figure 14), a constant value of Δ sin  results in a wider separation of  (Δ) in higher  regimes.In Figure 14, the terms Δ sin  and Δ denote the degree of separation of the two diffraction peaks from K 1 and K 2 rays.According to (2), constant Δ results in a higher value of Δ sin  as  increases due to the diminished  value (see Table 2).Then, the Δ value increases further with  due to the characteristics of the sine function.from the two different wavelengths (K doublet) widens as the diffraction angle  increases.
Consider a real crystal with a range of -spacing values for a given atomic plane due to imperfect crystallinity.Assume that  has a single value (the weight-averaged value of 1.541874 Å).Then, variations of the -spacing (Δ) yield variations of Δ sin  (Δ): 2Δ = /Δ sin .The terms Δ sin  and Δ here now denote the degree of broadening of a given peak when an X-ray with a fixed value of  diffracts.The lower the crystallinity (i.e., a larger Δ), the broader the diffraction peaks (a larger Δ).The broadening of a peak due to imperfect crystallinity screens the peak separation phenomenon, which contributes to yielding a single broad peak even at a high diffraction angle as seen in Figure 6(b).
For nanocrystals, there is another source of peak broadening apart from imperfect crystallinity.The small size of the nanocrystals is another source for peak broadening as described in (3).In nanocrystals, the peak separation is barely observed, even at a high diffraction angle, because the two sources of peak broadening described above (i.e., the small size and imperfect crystallinity) are operating simultaneously.Therefore, for a nanocrystal, the peak separation

p
CO = 0.001 atm p CO = 0.01 atm p CO = 0.1 atm

Figure 2 :
Figure 2: Thermogravimetry (TG) and differential TG of MTA.The temperature ranges are marked for the removal of (1) adsorbed water, (2) weakly bound crystal water, and (3) strongly bound crystal water.

Figure 3 :
Figure 3: Results of the secondary electron microscopy (SEM) for the MTA powder heat-treated at (a) 100 ∘ C and (b) 600 ∘ C. (c) The SEM result for the P-25 TiO 2 powder.

Figure 4 :
Figure 4: X-ray diffraction patterns of the (a) as-received MTA and (b) MTA heat-treated at 600 ∘ C for 1 h in air and (c) P-25 TiO 2 .

Figure 5 :Figure 6 :Figure 7 :
Figure 5: X-ray diffraction (XRD) patterns of the synthesized products from (a) precursor M and (b) precursor T.

Figure 8 :
Figure 8: Crystal size of the TiC synthesized at varying temperatures from precursors M and T.

Figure 9 :
Figure 9: (a) Lattice parameter and (b) oxygen content of TiC synthesized at varying temperatures from precursors M and T.

Figure 10 :
Figure 10: Specific surface area of the TiC synthesized at varying temperatures from precursors M and T.

Figure 11 :Figure 12 :
Figure 11: Results of the BJH analyses for the TiC synthesized from precursor M at 1500 ∘ C: (a) the nitrogen adsorption isotherm at 77 K as a function of the relative pressure and (b) the pore volume versus pore diameter relationship.

Table 1 :
Specific surface area, specific pore volume, and average pore diameter of the (a) as-received MTA, (b) MTA heat-treated at 600 ∘ C for 1 h in air, (c) TiC synthesized from precursor T at 1500 ∘ C, and (d) TiC synthesized from precursor M at 1500 ∘ C.

Table 2
presents the calculated Δ (peak separation) using (A.2) for varying values of -spacing.In this table,  was determined by (A.1) based on the weightaveraged  value of 1.541874 Å (a weight factor of 2 was applied for the K 1 wavelength in order to reflect the intensity ratio of the diffraction peak from K 1 to that from K 2 ).As apparent in Table2, the peak separation (Δ) originating