《复合材料 Composites》课程教学资源（学习资料）第二章 增强体_On the carbothermal vapour-liquid-solid（VLS）mechanism for TaC, TiC, and TaxTi1?xC whisker growth

E驅≈3S Journal of the European Ceramic Society 20(2000)2607-2618 On the carbothermal vapour-liquid-solid (Vls) mechanism for Tac, TiC, and Ta_c whisker growth Niklas ahlen. Mats Johnsson a, Ann-Kristin larsson a. Bo Sundman b a Department of Inorganic Chemistry, Stockholm University, S-10691 Stockholm. Swede dEpartment of Material Science and Engineering, Royal Institute of Technology, S-100 44 Stockholm, Sweden Received 9 December 1999: received in revised form 10 April 2000; accepted 13 April 2000 Abstract The growth of TaC, TiC and TaxTil-C whiskers has been studied in some detail. The whiskers were synthesised via a vapour- liquid-solid (VLS)growth mechanism in the temperature range 1220-1400oC. The starting materials were Ta2Os, TiO2, C, NaCl and a catalyst metal (Ni, Co, Fe, and Cu were tested). The main reaction during synthesis was a carbothermal reduction of Ta O and TiO2, and NaCl was added to form the oxochlorides and chlorides of Ta and Ti that account for the transport to the catalyst metal. The syntheses were made in a protecting Ar atmosphere. From experiments interrupted after different times at the synthesis temperature it is clear that sodium tantalates form as intermediate products, whereas sodium titanates cannot be identified. Only metals that are able to dissolve the elements building up the whiskers work as catalysts. Whisker growth starts either from a catalys droplet in contact with carbon or from an oxide particle in contact with both catalyst metal and carbon. For Tac and TaxTil-xC the only growth direction observed is [100], while TiC may grow either along [100] or along [lll]. 2000 Elsevier Science Ltd. All rights reserved Keywords: Carbides; TaC; TiC; VLS process; Whiskers 1. ntroduction can be favoured by optimising the molar ratios, the particle sizes of the precursor materials and the reaction This article presents a study of the carbothermal temperature. The temperature should be raised as higl vapour-liquid-solid (VLS) mechanism for whisker as possible in order to reduce the residual amount of owth. The whisker materials we have studied are TaC, oxygen. There is a compromise to be achieved, however TiC and Ta Ti-C. The Vls growth mechanism because at a too high synthesis temperature the solid involves all three aggregation states: a vapour phase state reaction between the oxide precursor and carbon transport of one or more of the whisker components to may dominate over the VLs growth process. Further- a melted catalyst droplet where the desired whisker more, at high temperatures the gaseous chloride and grows. The catalyst must be able to dissolve the whisker oxochloride species that are responsible for transport of components and the vls mechanism is only operative Ta and/ or Ti show a tendency to escape from the reac- at temperatures above the melting point of the catalyst tion chamber before reacting at the catalyst. Ar nother (i.e. the eutectic temperature of the multi-component problem at too high synthesis temperatures is that system catalyst-whisker), which therefore constitutes the whiskers may start to sinter together, forming agglom lower temperature limit for whisker synthesis via this erates growth mechanism. The catalyst metal must also be able wo main types growth mechanisms can reactions takia Components. There are two defined in general: (i)all mass transport takes place in perature: the VLs growth of whiskers and the direct acteristic for growth of whiskers via chemical vapour reaction between carbon and the transition metal oxides deposition, CVD), and(i) as in the present case solid resulting in carbide particles. The VLS growth mechanism carbon dissolves into the catalyst at the droplet inter face, while the other whisker component are transported in the vapour phase to the catalysts(characteristic for the carbothermal VLS mechanism).2,3 0955-2219/00/S-see front matter C 2000 Elsevier Science Ltd. All rights reserved PII:S0955-2219(00)00121-7

On the carbothermal vapour±liquid±solid (VLS) mechanism for TaC, TiC, and TaxTi1ÿxC whisker growth Niklas AhleÂn a , Mats Johnsson a,*, Ann-Kristin Larsson a , Bo Sundman b a Department of Inorganic Chemistry, Stockholm University, S-106 91 Stockholm, Sweden bDepartment of Material Science and Engineering, Royal Institute of Technology, S-100 44 Stockholm, Sweden Received 9 December 1999; received in revised form 10 April 2000; accepted 13 April 2000 Abstract The growth of TaC, TiC and TaxTi1ÿxC whiskers has been studied in some detail. The whiskers were synthesised via a vapour± liquid±solid (VLS) growth mechanism in the temperature range 1220±1400C. The starting materials were Ta2O5, TiO2, C, NaCl, and a catalyst metal (Ni, Co, Fe, and Cu were tested). The main reaction during synthesis was a carbothermal reduction of Ta2O5 and TiO2, and NaCl was added to form the oxochlorides and chlorides of Ta and Ti that account for the transport to the catalyst metal. The syntheses were made in a protecting Ar atmosphere. From experiments interrupted after di€erent times at the synthesis temperature it is clear that sodium tantalates form as intermediate products, whereas sodium titanates cannot be identi®ed. Only metals that are able to dissolve the elements building up the whiskers work as catalysts. Whisker growth starts either from a catalyst droplet in contact with carbon or from an oxide particle in contact with both catalyst metal and carbon. For TaC and TaxTi1ÿxC the only growth direction observed is [100], while TiC may grow either along [100] or along [111]. # 2000 Elsevier Science Ltd. All rights reserved. Keywords: Carbides; TaC; TiC; VLS process; Whiskers 1. Introduction This article presents a study of the carbothermal vapour±liquid±solid (VLS) mechanism for whisker growth. The whisker materials we have studied are TaC, TiC and TaxTi1ÿxC. The VLS growth mechanism involves all three aggregation states: a vapour phase transport of one or more of the whisker components to a melted catalyst droplet where the desired whisker grows. The catalyst must be able to dissolve the whisker components and the VLS mechanism is only operative at temperatures above the melting point of the catalyst (i.e. the eutectic temperature of the multi-component system catalyst-whisker), which therefore constitutes the lower temperature limit for whisker synthesis via this growth mechanism. The catalyst metal must also be able to dissolve the whisker components. There are two competing reactions taking place at the synthesis tem￾perature: the VLS growth of whiskers and the direct reaction between carbon and the transition metal oxides resulting in carbide particles. The VLS growth mechanism can be favoured by optimising the molar ratios, the particle sizes of the precursor materials and the reaction temperature. The temperature should be raised as high as possible in order to reduce the residual amount of oxygen. There is a compromise to be achieved, however, because at a too high synthesis temperature the solid￾state reaction between the oxide precursor and carbon may dominate over the VLS growth process. Further￾more, at high temperatures the gaseous chloride and oxochloride species that are responsible for transport of Ta and/or Ti show a tendency to escape from the reac￾tion chamber before reacting at the catalyst. Another problem at too high synthesis temperatures is that whiskers may start to sinter together, forming agglom￾erates. Two main types of VLS growth mechanisms can be de®ned in general: (i) all mass transport takes place in the vapour phase to the vapour±liquid interface (char￾acteristic for growth of whiskers via chemical vapour deposition, CVD),1 and (ii) as in the present case solid carbon dissolves into the catalyst at the droplet inter￾face, while the other whisker component are transported in the vapour phase to the catalysts (characteristic for the carbothermal VLS mechanism).2,3 0955-2219/00/$ - see front matter # 2000 Elsevier Science Ltd. All rights reserved. PII: S0955-2219(00)00121-7 Journal of the European Ceramic Society 20 (2000) 2607±2618 * Corresponding author. E-mail address: matsj@inorg.su.se (M. Johnsson)

2608 N. Ahlen et al. Journal of the European Ceramic Society 20(2000)2607-2618 The carbothermal vls mechanism thus involves four Within a certain temperature range, however, this main steps:(1)dissolution of carbon at the interface direct carbothermal reaction is not kinetically favoured between the catalyst droplet and solid carbon, (2)mass in the presence of NaCl and Ni. The reactions that transport of whisker elements in the vapour phase to the actually take place in the carbothermal VLS mechanism vapour-liquid interface (3)chemical reaction at the are complicated to stud vapour-liquid interface, (4) dissolution into and diffu There must be a gas phase transport of Ta and Ti to sion through the liquid alloy phase, and (5) precipita- the catalyst metal. Thermodynamic calculations with tion at the liquid-solid interface he computer program HSC that uses the Gibbs energy The overall chemical reaction for the synthesis of minimisation method suggest the following reactions to TaxTil-C from TiO2 and Ta2O5 is a straightforward take place at the synthesis temperature carbothermal reduction of the oxides 0.5xTa2O5(s)+(1-x)TiO2(s)+(3+0.5x)C(s) Raw materials used in the synthesis ->TaTil-.C(S)+(2+0.5x)co(g) Substance Purity Manufacturer Particle Comment (wt.%) 99.9 Crac 99.9 Aldrich Anatase Degussa ng Carbon black Fw200) Ta, c agglomerates 21 wt. volatiles) Nacl 99.5 Akzo 99.9 Cerad <325 mesh Co 99.9 Cerad <325 mesh 9 Aldrich <325 mesh 99.5 Aldrich <150 mesh P(red) 99 Kebo Table 2 5 The molar ratios in the starting mixtures that were found to give the Fig. 1. A simplified sequence for the carbothermal VLs growth highest whisker yield for TaC, TiC and TaosTio. Ca mechanism. The reactions at the catalyst droplet can be described as TaC TIC allows:(I) the catalyst particle(Ni) in contact with carbon start to dissolve the whisker constituents: C, Ta, and Ti. (2)The droplet is TarOs 1(60.00g) supersaturated and a Ta,Ti-C whisker is nucleated. (3)The surface TiOz 1(20.00g) tension balance at the interface between the whisker and the catalyst 13(14.24g) droplet determines the whisker diameter. (4)As the reactants are con- Nacl 0.503.97g) tinuously being dissolved into the catalyst, the whisker grows, and the Ni 0.05(0.40g) 0.075(1.13g) arbon particles are consumed. (5)When all carbon in contact with the catalyst droplet has been consumed, the whisker growth terminates. a Weighed- in amounts are given within parentheses. Gas Out Reactor g Element Gas In Fig. 2. An SEM micrograph of Tao.s Tio sC whiskers. The yield is estimated to be 80 vol % The whisker diameter is 0.2-0.6 um and the length 10-30 Fig 3. The furnace set-up with the gas flow around the reacto

The carbothermal VLS mechanism thus involves four main steps: (1) dissolution of carbon at the interface between the catalyst droplet and solid carbon, (2) mass transport of whisker elements in the vapour phase to the vapour±liquid interface (3) chemical reaction at the vapour±liquid interface, (4) dissolution into and di€u￾sion through the liquid alloy phase, and (5) precipita￾tion at the liquid±solid interface. The overall chemical reaction for the synthesis of TaxTi1ÿxC from TiO2 and Ta2O5 is a straightforward carbothermal reduction of the oxides. 0:5xTa2O5… †‡ s … † 1 ÿ x TiO2… †‡ s … † 3 ‡ 0:5x C s… † ! TaxTi1ÿxC s… †‡ … † 2 ‡ 0:5x CO g…† …1† Within a certain temperature range, however, this direct carbothermal reaction is not kinetically favoured in the presence of NaCl and Ni. The reactions that actually take place in the carbothermal VLS mechanism are complicated to study. There must be a gas phase transport of Ta and Ti to the catalyst metal. Thermodynamic calculations with the computer program HSC4 that uses the Gibbs energy minimisation method suggest the following reactions to take place at the synthesis temperature: Fig. 1. A simpli®ed sequence for the carbothermal VLS growth mechanism. The reactions at the catalyst droplet can be described as follows: (1) the catalyst particle (Ni) in contact with carbon start to dissolve the whisker constituents: C, Ta, and Ti. (2) The droplet is supersaturated and a TaxTi1ÿxC whisker is nucleated. (3) The surface tension balance at the interface between the whisker and the catalyst droplet determines the whisker diameter. (4) As the reactants are con￾tinuously being dissolved into the catalyst, the whisker grows, and the carbon particles are consumed. (5) When all carbon in contact with the catalyst droplet has been consumed, the whisker growth terminates. Fig. 2. An SEM micrograph of Ta0:5Ti0:5C whiskers. The yield is estimated to be 80 vol.%. The whisker diameter is 0.2±0.6 mm and the length 10±30 mm. Table 1 Raw materials used in the synthesis Substance Purity (wt.%) Manufacturer Particle size Comment Ta2O5 99.9 Cerac ÿ325 mesh Ceramic grade TiO2 99.9 Aldrich 0.2±0.6 mm Anatase C ± Degussa (FW200) 13 nm forming ¯occulent agglomerates Carbon black (containing 21 wt.% volatiles) NaCl 99.5 Akzo <5 mm Ni 99.9 Cerac <325 mesh Co 99.9 Cerac <325 mesh Fe 97 Aldrich <325 mesh Cu 99.5 Aldrich <150 mesh P(red) 99 Kebo Table 2 The molar ratios in the starting mixtures that were found to give the highest whisker yield for TaC, TiC and Ta0:5Ti0:5Ca TaC TiC Ta0:5Ti0:5C Ta2O5 1 (60.00 g) ± 1 (30.00 g) TiO2 ± 1 (20.00 g) 2 (10.85 g) C 7.2 (14.86 g) 3.3 (12.56 g) 13 (14.24 g) NaCl 0.5 (3.97 g) 0.5 (7.31 g) 1.5 (5.95 g) Ni 0.05 (0.40 g) 0.075 (1.13 g) 0.15 (0.60 g) a Weighed-in amounts are given within parentheses. Fig. 3. The furnace set-up with the gas ¯ow around the reactor. 2608 N. AhleÂn et al. / Journal of the European Ceramic Society 20 (2000) 2607±2618

N. Ahlen et al. / Journal of the European Ceramic Society 20(2000)2607-2618 2Ta2O5(s)+6CI(g)+3C(s)-2TaOCl3(g)+3CO(g) (2) TaOCl3 (g)+C(s)+ Ni-Ti-C()- Ni-Ta-Ti-CO +Co(g)+3cl(g) TiO2(s)+3Cl(g)+2C(s)- TiCl3 (g)+ 2C0(g) (3) Ni-Ta-Ti-C()- TaxTil-C(s)+NiO) Other gas-phase species that also form by similar A simplified sequence for the carbothermal VLS eactions but in lower concentrations are TaO, CI(g), growth mechanism is outlined in Fig. I TaCls(g), and TiCl(g), and these may also to some For some years, we have been studying growth of extent be responsible for the transport of Ta and Ti different transition-metal carbide and carbonitride The melting temperature of the catalyst metal must whiskers 2.3, 5-7 We have achieved control of the synth match the temperature range where TiCl3(g) and esis so as to gain a high whisker yield(see Fig. 2), but TaOCl3(g)form in high concentrations. If a catalyst we have not obtained a detailed understanding of the metal with a too high melting point is chosen, then growth mechanism. In the present article we have per direct carbothermal reaction, resulting in particles, may formed a detailed study of the growth mechanisms of dominate over the whisker formation by the VLS Tax Til-xC whiskers with 0Ni-Ti +3cl(g) (5) Ta Ti-C whiskers are listed in Table 1. The carbon Table 3 Chemical analysis data and measured weight loss for the series with interrupted TaC synthesis experiments" eight loss(wt % O(wt % C(wt % Na(wt % Ni(wt % N(wt % Ta+Cl (wt % nalys sis method LECO LECO ng mixture 6.8 23.5 30.6 00000 33 33.0 6.3 33.3 6.4 Table 4 Phases present ding to powder X-ray diffraction data(XRD) after different reaction times for the series of interrupted Tac-whisker syntheses Starting mixture 15 90 20 240 a Na]Ta,O 5xxxx 4xxxx not the experiments were interrupted after different holding times, from 0 to 240 min, at the synthesis temperature 1220oC.NaCl,Ni,and Ccould

2Ta2O5… †‡s 6Cl g… †‡3C s…†! 2TaOCl3… †‡g 3CO g…† …2† TiO2… †‡ s 3Cl g… †‡ 2C s…†! TiCl3… †‡ g 2CO g…† …3† Other gas-phase species that also form by similar reactions but in lower concentrations are TaO2Cl(g), TaCl5(g), and TiCl4(g), and these may also to some extent be responsible for the transport of Ta and Ti. The melting temperature of the catalyst metal must match the temperature range where TiCl3(g) and TaOCl3(g) form in high concentrations. If a catalyst metal with a too high melting point is chosen, then a direct carbothermal reaction, resulting in particles, may dominate over the whisker formation by the VLS mechanism. The following reactions are expected to occur at the catalyst (Ni±C, etc., denote that C, Ta, and Ti are dissolved in the Ni catalyst). C s… †‡ Ni l…† ! NiÿC l…† …4† TiCl3… †‡ g NiÿC l…† ! NiÿTiÿC I… †‡ 3Cl g…† …5† TaOCl3… †‡ g C s… †‡ NiÿTiÿC l…† ! NiÿTaÿTiÿC l… † ‡ CO g… †‡ 3Cl g…† …6† NiÿTaÿTiÿC l…† ! TaxTi1ÿxC s… †‡ Ni l…† …7† A simpli®ed sequence for the carbothermal VLS growth mechanism is outlined in Fig. 1. For some years, we have been studying growth of di€erent transition-metal carbide and carbonitride whiskers.2,3,5ÿ7 We have achieved control of the synth￾esis so as to gain a high whisker yield (see Fig. 2), but we have not obtained a detailed understanding of the growth mechanism. In the present article we have per￾formed a detailed study of the growth mechanisms of TaxTi1ÿxC whiskers with 04x41 designed to explore some essential features of the VLS mechanism. 2. Experimental 2.1. Synthesis The starting materials used in the preparation of TaxTi1ÿxC whiskers are listed in Table 1. The carbon Table 3 Chemical analysis data and measured weight loss for the series with interrupted TaC synthesis experimentsa Weight loss (wt.%) O (wt.%) C (wt.%) Na (wt.%) Ni (wt.%) N (wt.%) Ta+Cl (wt.%) Analysis method LECO LECO AAS AAS LECO Balance Starting mixture 0 16.2 15.0 2.4 0.8 0.2 65.1 0 6.8 13.7 14.3 1.9 0.8 0.2 69.1 5 8.6 13.2 14.0 1.6 0.9 0.1 70.2 15 12.2 12.3 13.4 1.3 0.9 0.1 72.0 25 16.7 10.8 12.6 0.8 0.7 0.1 74.9 45 21.2 8.4 11.3 0.3 1.3 0.2 78.5 60 23.5 7.3 10.7 0.2 1.1 0.1 80.6 90 25.8 5.5 9.7 0.2 1.0 0.2 83.5 120 28.1 3.7 8.6 0.2 1.1 0.3 86.2 150 30.6 2.0 8.6 0.2 1.1 0.2 88.0 180 33.0 0.4 6.5 0.2 1.1 0.2 91.6 210 33.0 0.09 6.3 0.02 1.2 0.2 92.2 240 33.3 0.1 6.4 0.01 1.2 0.2 92.1 a The experiments were quenched after di€erent holding times, from 0 to 240 min, at the synthesis temperature 1220C. Ta and Cl were not analysed. No information on the number of vacancies is available. Table 4 Phases present according to powder X-ray di€raction data (XRD) after di€erent reaction times for the series of interrupted TaC-whisker synthesesa Starting mixture 0 5 15 25 45 60 90 120 150 180 210 240 Ta2O5 ± ± Na2Ta4O11 ± ±± ±±± NaTaO3 ± ±±± ± ± TaC ± a The experiments were interrupted after di€erent holding times, from 0 to 240 min, at the synthesis temperature 1220C. NaCl, Ni, and C could not be detected. N. AhleÂn et al. / Journal of the European Ceramic Society 20 (2000) 2607±2618 2609

10 N. Ahlen et al. Journal of the European Ceramic Society 20 (2000)2607-2618 source contained 21 wt. volatile components that nace(Thermal Technology) at a rate of 1000C/h to were burned off during the heating-up period. The choice the synthesis temperature in an Ar atmosphere. All of carbon source has a strong influence on the whisker experiments were conducted at atmospheric pressure ield. It has been shown in earlier studies that carbon and at temperatures in the range 1220-1400C Previous black with a volatile part retains its fluffy consistency after studies have revealed the optimum temperature for tac heat treatment, in contrast to carbon powder without and Tax Til- C whisker growth to be 1220-1250C3, 5 volatiles, and we believe that this is essential in improv- and 1400C for TiC whiskers. The whis sers o ng the ease with which the volatile Ta and Ti species were 0. 2-0. 6 um in diameter and 5-30 um in length. The reach the catalyst.2,3 The weighed-in TiO2/Ta2Os molar ratio can be varied in order to yield whiskers with dif- ferent x-values. The molar ratios of the different pre- Table 5 cursor materials used in this study for preparation of TaC, Chemical analysis data for the series of interrupted Tic synthesis TiC and Tao. Tios C are listed in Table 2 experiments The starting materials were mixed to homogeneity in Total o c Na Ni a blender. Portions of about 10-20 g were placed in weight (wt %)(wt %)(wt %)(wt %)(wt %)(wt % graphite crucible covered with a lid with a number of holes to allow controlled gas exchange between the (wt%) reactor chamber and the surrounding atmosphere(see LECO LECO AAs AAS Fig. 3). The mixture was then heated in a graphite fur 597.52.9 12.519.324.77.63.2 333.1 23418.32506.23.69938.1 34.216.72534.14.06944. 43.114.225.82.24.8 48.112.226.00.45.3 9725.30.085.30.0158.9 52.58.224.60.055.5 G0.4 a The experiments were quenched after different holding times, from 0 to 960 min, at the synthesis temperature 1250 C. The compo- sition of the starting mixture is not analysed it is calculated from the 总02 weight in amounts of the different starting materials. Chemical analysis data for the series of interrupted TiC synthesis Total o C Ni weight (wt %)(wt %)(wt %)(wt %)(wt %)(wt % 08 (wt.%) :2 LECO LECO AAS AAS Balance 0.925.97.2 11.631.3 谷04 23.717.723.86.53.510.937.5 47.31102501.7 51.39.224.80.4 53.27624.30.085.6 968 58.8 61.2 00250 3.86.623.80.026.2 1062.5 55.1 5.3 .700164 1.164.0 time(min) 14.623.60.016.0 240 3 0.0096.2 0.01658 Fig 4.(a) Observed weight change as a fraction of the expected vs. 960 57.5 2.1 23. 1 0.006 6.2 0 reaction time for the series of interrupted Tac whisker synthesis experiments at 1220.C;(b)observed weight change of the different The experiments were quenched after different holding times, Na,Cl, and Ni disappear from the reactor during synthesi that all O, sition of the starting mixture is not analysed it is calculated from o- elements as a fraction of the expected vs reaction time. Based on the from 0 to 960 min, at the synthesis temperature 1400.C. The com assumption that the starting mixture transforms to Tac and weight in amounts of the different starting materials

source contained 21 wt.% volatile components that were burned o€ during the heating-up period. The choice of carbon source has a strong in¯uence on the whisker yield. It has been shown in earlier studies that carbon black with a volatile part retains its ¯u€y consistency after heat treatment, in contrast to carbon powder without volatiles, and we believe that this is essential in improv￾ing the ease with which the volatile Ta and Ti species reach the catalyst.2,3 The weighed-in TiO2=Ta2O5 molar ratio can be varied in order to yield whiskers with dif￾ferent x-values. The molar ratios of the di€erent pre￾cursor materials used in this study for preparation of TaC, TiC and Ta0:5Ti0:5C are listed in Table 2. The starting materials were mixed to homogeneity in a blender. Portions of about 10±20 g were placed in a graphite crucible covered with a lid with a number of holes to allow controlled gas exchange between the reactor chamber and the surrounding atmosphere (see Fig. 3). The mixture was then heated in a graphite fur￾nace (Thermal Technology) at a rate of 1000C/h to the synthesis temperature in an Ar atmosphere. All experiments were conducted at atmospheric pressure and at temperatures in the range 1220±1400C. Previous studies have revealed the optimum temperature for TaC and TaxTi1ÿxC whisker growth to be 1220±1250C3,5 and 1400C for TiC whiskers.2 The whiskers obtained were 0.2±0.6 mm in diameter and 5±30 mm in length. The Fig. 4. (a) Observed weight change as a fraction of the expected vs. reaction time for the series of interrupted TaC whisker synthesis experiments at 1220C; (b) observed weight change of the di€erent elements as a fraction of the expected vs. reaction time. Based on the assumption that the starting mixture transforms to TaC and that all O, Na, Cl, and Ni disappear from the reactor during synthesis. Table 5 Chemical analysis data for the series of interrupted TiC synthesis experimentsa Total weight loss (wt.%) O (wt.%) C (wt.%) Na (wt.%) Ni (wt.%) Cl (wt.%) Ti (wt.%) Analysis method LECO LECO AAS AAS Balance Starting mixture 20.9 25.9 7.5 2.9 11.6 31.3 0 12.5 19.3 24.7 7.6 3.2 12.3 33.1 30 23.4 18.3 25.0 6.2 3.6 9.9 38.1 60 34.2 16.7 25.3 4.1 4.0 6.9 44.3 90 43.1 14.2 25.8 2.2 4.8 3.8 50.8 120 48.1 12.2 26.0 0.4 5.3 2.2 55.0 180 50.8 9.7 25.3 0.08 5.3 0.01 58.9 240 52.5 8.2 24.6 0.05 5.5 0 60.7 960 55.7 5.0 22.8 0 6.5 0 64.4 a The experiments were quenched after di€erent holding times, from 0 to 960 min, at the synthesis temperature 1250C. The compo￾sition of the starting mixture is not analysed it is calculated from the weight in amounts of the di€erent starting materials. Table 6 Chemical analysis data for the series of interrupted TiC synthesis experimentsa Total weight loss (wt.%) O (wt.%) C (wt.%) Na (wt.%) Ni (wt.%) Cl (wt.%) Ti (wt.%) Analysis method LECO LECO AAS AAS Balance Starting mixture ± 20.9 25.9 7.5 2.9 11.6 31.3 0 23.7 17.7 23.8 6.5 3.5 10.9 37.5 10 39.5 13.8 24.5 4.3 4.5 8.7 47.0 20 47.3 11.0 25.0 1.7 5.2 3.9 55.0 30 51.3 9.2 24.8 0.4 5.5 1.6 58.8 40 53.2 7.6 24.3 0.08 5.6 1.8 61.2 60 53.8 6.6 23.8 0.02 6.2 1.0 62.5 120 55.1 5.3 23.7 0.01 6.4 1.1 64.0 180 56.1 4.6 23.6 0.01 6.0 0.02 65.2 240 56.3 3.7 23.1 0.009 6.2 0.01 65.8 960 57.5 2.1 23.1 0.006 6.2 0 68.3 a The experiments were quenched after di€erent holding times, from 0 to 960 min, at the synthesis temperature 1400C. The compo￾sition of the starting mixture is not analysed it is calculated from the weight in amounts of the di€erent starting materials. 2610 N. AhleÂn et al. / Journal of the European Ceramic Society 20 (2000) 2607±2618

N. Ahlen et al. Journal of the European Ceramic Society 20(2000)2607-2618 reactions were completed after 4 h at the respective composition and the phase composition changes during optimum synthesis temperature. Three series with synthesis; one series with Tac whisker synthesis at experiments interrupted after different times from o to 960 1220C and two series with Tic whisker synthesis at min were performed in order to study how the chemical 1250 and 1400 C, respectively. Only a few whiskers terminated with a catalyst dro- (a)1.0 plet after completed reaction, indicating that there is a rapid redistribution of Ni in the reactor after a whisker has grown to full length. The search for droplets in experiments interrupted before completed whisker growth was a time-consuming task. It was observed that 1400°C small additions of red phosphorus to the starting mix- --1250C ture increased the occurrence of nickel droplets. After 0.4 solidification of the droplets small nickel phosphide particles were found to have precipitated on the droplet surface. No effect on the whisker yield was observed however. Therefore, experiments were done where red 0200060800010 phosphorous was added to the starting mixture in order to increase the possibility to find whiskers terminated lyst dre 22. characterisation The products were characterised by their X-ray powder diffraction patterns(XRD), obtained with a Guinier 806 Hagg focusing camera. CuKaI radiation (=1.54060 A) sed. and fine ed silicon a=5. 430880(35)AI were evaluated in an automatic film scanner Orded films was added as an internal standard. The reco The morphology and composition of the whiskers and catalyst droplets were investigated using a high-resolu tion scanning electron microscope(SEM, JEOL 880) equipped with an energy-dispersive spectrometer 200 (LINK ISIS), which allows detection of boron and heavier elements. A high-resolution transmission elec tron microscope (TEM, JEOL 3010)and a TEM JEOL 2000FX) equipped with EDS (LINK AN10000) were used for the electron diffraction work to determine whisker growth directions and for studies of the catalyst droplets 806/ 0.4 Table 7 0.2 Phases present, according to powder X-ray diffraction data (XRD) after different reaction times for the 1250C series of interrupted TiC whisker synthesis experiments 0 Starting03060120180240 mixture time(min) Fig. 5.(a)Observed weight change as a fraction of the expected vs TiO,(rutile) eaction time for the two temperature series(1250 and 1400oC)of TiO,(anatase) nterrupted TiC whisker synthesis experiments; (b)observed weight ange of the different elements as a fraction of the expected vs reac. ion time for TiC whisker synthesis at 1250C. Based on the assump- ion that the starting mixture transforms to TiC and that all O, Na, CL. and Ni disappear from the reactor during synthesis; (c)same as(b)but The experiments were interrupted after different holding times for TiC whisker synthesis at 1400oC. from o to 240 min. Ni and c could not be detected from Xrd data

reactions were completed after 4 h at the respective optimum synthesis temperature. Three series with experiments interrupted after di€erent times from 0 to 960 min were performed in order to study how the chemical composition and the phase composition changes during synthesis; one series with TaC whisker synthesis at 1220C and two series with TiC whisker synthesis at 1250 and 1400C, respectively. Only a few whiskers terminated with a catalyst dro￾plet after completed reaction, indicating that there is a rapid redistribution of Ni in the reactor after a whisker has grown to full length. The search for droplets in experiments interrupted before completed whisker growth was a time-consuming task. It was observed that small additions of red phosphorus to the starting mix￾ture increased the occurrence of nickel droplets. After solidi®cation of the droplets small nickel phosphide particles were found to have precipitated on the droplet surface. No e€ect on the whisker yield was observed, however. Therefore, experiments were done where red phosphorous was added to the starting mixture in order to increase the possibility to ®nd whiskers terminated with a catalyst droplet. 2.2. Characterisation The products were characterised by their X-ray powder di€raction patterns (XRD), obtained with a Guinier￾HaÈgg focusing camera. CuKa1 radiation (l=1.54060 AÊ ) was used, and ®nely powdered silicon [a=5.430880(35) AÊ ] was added as an internal standard. The recorded ®lms were evaluated in an automatic ®lm scanner.8 The morphology and composition of the whiskers and catalyst droplets were investigated using a high-resolu￾tion scanning electron microscope (SEM, JEOL 880) equipped with an energy-dispersive spectrometer (LINK ISIS), which allows detection of boron and heavier elements. A high-resolution transmission elec￾tron microscope (TEM, JEOL 3010) and a TEM (JEOL 2000FX) equipped with EDS (LINK AN10000) were used for the electron di€raction work to determine whisker growth directions and for studies of the catalyst droplets. Fig. 5. (a) Observed weight change as a fraction of the expected vs. reaction time for the two temperature series (1250 and 1400C) of interrupted TiC whisker synthesis experiments; (b) observed weight change of the di€erent elements as a fraction of the expected vs. reac￾tion time for TiC whisker synthesis at 1250C. Based on the assump￾tion that the starting mixture transforms to TiC and that all O, Na, Cl, and Ni disappear from the reactor during synthesis; (c) same as (b) but for TiC whisker synthesis at 1400C. Table 7 Phases present, according to powder X-ray di€raction data (XRD), after di€erent reaction times for the 1250C series of interrupted TiC whisker synthesis experimentsa Starting mixture 0 30 60 120 180 240 TiC ± TiO2 (rutile) ± ± TiO2 (anatase) ± Ti2O3 ± ±±± Ti3O5 ± ±± ± ± NaCl ±±± a The experiments were interrupted after di€erent holding times from 0 to 240 min. Ni and C could not be detected from XRD data. N. AhleÂn et al. / Journal of the European Ceramic Society 20 (2000) 2607±2618 2611

N. Ahlen et al. Journal of the European Ceramic Society 20(2000)2607-2618 3. Results and Discussion it grows to full length very rapidly. The rate-determining step should then be the formation of supersaturated 3.1. Interrupted experiments catalyst droplets from which whiskers can grow. The time dependence of the TaC formation is extracted from 3.. TaC the data given in Tables 3 and 4. The fraction of expec- We followed the reaction forming TaC whiskers at the ted weight loss is thus plotted versus reaction time(see optimum synthesis temperature, 1220C. TaC whiskers Fig. 4a). These data show that the reaction is complete were found as soon as the synthesis temperature had after 210 min and that Tac is then the only phase that been reached, with the same length and diameter as can be identified. The change in chemical composition hose present after completed carbothermal Vls during synthesis can be followed in a graph showing the growth. This indicates that, once a whisker has nucleated, observed weight change as a fraction of expected vs. reac- tion time for the different elements, assuming that the starting mixture will transform to Tac during the reaction (see Fig. 4b). The sodium level rapidly dropped to about 10% of the initial amount, but stayed at this level during the remaining tac whisker growth period, just until the reaction came to completion after 210 min. According to XRD analysis, this is due to that sodium tantalates (NaTaO3 and Na Ta4O11 are formed as intermediate phases during synthesis(see Table 4). As the last reduc tion step those sodium tantalates react. The main impurity in the synthesis product, found from chemical analysis, is remnants of the Ni catalyst that, however, is too low to be detected by use of XRD 100nm 3.1.2.TiC Two series of experiments interrupted after different reaction times were performed, one at 1250C and one at 1400 C. At 1400C the TiC whisker yield is slightl higher and the residual level of oxygen is lower than at 1250C. Primary chemical analysis data are given 200nm Tables 5 and 6 The reaction time required to form TiC whiskers at b) 1250C is about the same as for TaC. deduced both from weight loss curves and from chemical analysi data(Fig. 5a-c). Just as for TaC synthesis, a few TiC whiskers had already formed when the synthesis tem perature was reached. No sodium titanates were found as intermediates during the reaction, according to the XRD studies, and neither did chemical analysis show plateau in the ent(see Fig 5b and c), wh was the case for TaC synthesis. Traces of reduced tita- nium oxides(e.g. Ti3Os and Ti,O3)were, however, found in some of the interrupted experiments(see Table 7) 3. 2. The catalyst metal Ni has proved to be the best catalyst metal for synthesis of TiC and TaC whiskers. 3 After completed whisker 200nm 100nm growth, however, only a minor fraction of the whiskers terminate with a Ni droplet (<5%), and even those droplets are found at different levels of erosion(see Fig Fig. 6. (a)TEM micrograph of a Tic whisker terminated by a N 6). An attempt was made to study the catalyst droplets sharp, indicating that the whisker has nucleated on the surface of the that terminate some of the whiskers. Phosphorous additions to the reaction mixture were found to increase the possibility for finding whiskers terminated with

3. Results and Discussion 3.1. Interrupted experiments 3.1.1. TaC We followed the reaction forming TaC whiskers at the optimum synthesis temperature, 1220C. TaC whiskers were found as soon as the synthesis temperature had been reached, with the same length and diameter as those present after completed carbothermal VLS growth. This indicates that, once a whisker has nucleated, it grows to full length very rapidly. The rate-determining step should then be the formation of supersaturated catalyst droplets from which whiskers can grow. The time dependence of the TaC formation is extracted from the data given in Tables 3 and 4. The fraction of expec￾ted weight loss is thus plotted versus reaction time (see Fig. 4a). These data show that the reaction is complete after 210 min and that TaC is then the only phase that can be identi®ed. The change in chemical composition during synthesis can be followed in a graph showing the observed weight change as a fraction of expected vs. reac￾tion time for the di€erent elements, assuming that the starting mixture will transform to TaC during the reaction (see Fig. 4b). The sodium level rapidly dropped to about 10% of the initial amount, but stayed at this level during the remaining TaC whisker growth period, just until the reaction came to completion after 210 min. According to XRD analysis, this is due to that sodium tantalates (NaTaO3 and Na2Ta4O11) are formed as intermediate phases during synthesis (see Table 4). As the last reduc￾tion step those sodium tantalates react. The main impurity in the synthesis product, found from chemical analysis, is remnants of the Ni catalyst (see Table 3) that, however, is too low to be detected by use of XRD. 3.1.2. TiC Two series of experiments interrupted after di€erent reaction times were performed, one at 1250C and one at 1400C. At 1400C the TiC whisker yield is slightly higher and the residual level of oxygen is lower than at 1250C. Primary chemical analysis data are given in Tables 5 and 6. The reaction time required to form TiC whiskers at 1250C is about the same as for TaC, deduced both from weight loss curves and from chemical analysis data (Fig. 5a±c). Just as for TaC synthesis, a few TiC whiskers had already formed when the synthesis tem￾perature was reached. No sodium titanates were found as intermediates during the reaction, according to the XRD studies, and neither did chemical analysis show any plateau in the Na content (see Fig. 5b and c), which was the case for TaC synthesis. Traces of reduced tita￾nium oxides (e.g. Ti3O5 and Ti2O3) were, however, found in some of the interrupted experiments (see Table 7). 3.2. The catalyst metal Ni has proved to be the best catalyst metal for synthesis of TiC and TaC whiskers.2,3 After completed whisker growth, however, only a minor fraction of the whiskers terminate with a Ni droplet (<5%), and even those droplets are found at di€erent levels of erosion (see Fig. 6). An attempt was made to study the catalyst droplets that terminate some of the whiskers. Phosphorous additions to the reaction mixture were found to increase the possibility for ®nding whiskers terminated with a Fig. 6. (a) TEM micrograph of a TiC whisker terminated by a Ni droplet. The interface between the whisker and the droplet is very sharp, indicating that the whisker has nucleated on the surface of the droplet; (b±d) droplets terminating whiskers in di€erent degrees of erosion. 2612 N. AhleÂn et al. / Journal of the European Ceramic Society 20 (2000) 2607±2618

N. Ahlen et al. Journal of the European Ceramic Society 20(2000)2607-2618 8 EDS-TEM analyses of Ni droplets terminating TaxTiI-C whiskers and Fe-droplets terminating TiCx Ni- whiskers Whisker Droplet Ni T1 Na composition at (at.%) (at%) (at % (at.%) (at.%) TaxTik-C 123456667888 0024 7873 04000 26.5 ccccc TIC NE 6062 TIC NE 78.3 TIC NE 85.8 TIC NE 8.6 TIC NE 001222222333344444 97.2 1.7 TIC NE 82.0 TIC NE 84.6 TIC NE 86.6 .2.0 124 TIC NE 1.1 TIC NE 990 0.1 99.5 84.2 15.6 5.5 Red phosphorus was added to the starting mixtures in order to enhance the possibility of finding droplets. Different spot analysis on the same us content due to Ni-droplet in the product. TEM-EDS point analyses of elements also being present, e.g. O and Na, are lowering Ni droplets terminating whiskers show very little Ti the eutectic temperature and or Ta left in the Ni catalyst after solidification(see Cobalt also works as catalyst metal for whisker Table 8). A SEM-EDS line scan over a Ni droplet ter- growth, but the whisker yield is not as good as when N minating a TiC whisker shows that there is no Ni dis- is used. According to calculations with the computer solved into the structural framework of the carbid program ThermoCalc, cobalt behaves similarly to nickel whisker (see Fig. 7). The experimental results do not and can dissolve both Ti and Ta, but in much lower contradict that Ti and/ or Ta dissolve in the droplet at concentrations (Table 9). The reason why cobalt is not the synthesis temperature as efficient a catalyst as nickel is unclear but may be due Equilibrium calculations with the program Thermo- to the lower solubility of Ta (a factor of 8), especiall Calcusing data from "the Hard Materials Thermo- considering that both Ni and Co can form gaseous dynamic Database gives for hand that both C and oxochlorides and chlorides at the synthesis temperature Ti/Ta are soluble in liquid Ni (see Table 9). The carbide which may account for transport during the synthesis phase thus seems to have precipitated from a super In our studies we have found that iron does not saturated Ni droplet. The eutectic temperatures found function as a catalyst for Ta Ti-C whiskers, which is from the equilibrium calculations are, however, a bit most likely due to poisoning of the catalytic effect by higher than the temperatures that are working for the formation of Fe3C as observed by Wokulski and synthesis. The fact that the catalyst metal forms a melt Wokulska, 2 in a CVd study of TiC whisker synthesis at the synthesis temperature may be due to that other However, iron catalyses growth of TiC,Ni-y whiskers

Ni-droplet in the product. TEM-EDS point analyses of Ni droplets terminating whiskers show very little Ti and/or Ta left in the Ni catalyst after solidi®cation (see Table 8). A SEM-EDS line scan over a Ni droplet ter￾minating a TiC whisker shows that there is no Ni dis￾solved into the structural framework of the carbide whisker (see Fig. 7). The experimental results do not contradict that Ti and/or Ta dissolve in the droplet at the synthesis temperature. Equilibrium calculations with the program Thermo￾Calc9 using data from ``the Hard Materials Thermo￾dynamic Database''10 gives for hand that both C and Ti/Ta are soluble in liquid Ni (see Table 9). The carbide phase thus seems to have precipitated from a super￾saturated Ni droplet. The eutectic temperatures found from the equilibrium calculations are, however, a bit higher than the temperatures that are working for the synthesis. The fact that the catalyst metal forms a melt at the synthesis temperature may be due to that other elements also being present, e.g. O and Na, are lowering the eutectic temperature. Cobalt also works as catalyst metal for whisker growth, but the whisker yield is not as good as when Ni is used. According to calculations with the computer program ThermoCalc, cobalt behaves similarly to nickel and can dissolve both Ti and Ta, but in much lower concentrations (Table 9). The reason why cobalt is not as ecient a catalyst as nickel is unclear but may be due to the lower solubility of Ta (a factor of 8), especially considering that both Ni and Co can form gaseous oxochlorides and chlorides at the synthesis temperature, which may account for transport during the synthesis. In our studies we have found that iron does not function as a catalyst for TaxTi1ÿxC whiskers, which is most likely due to poisoning of the catalytic e€ect by formation of Fe3C as observed by Wokulski and Wokulska11,12 in a CVD study of TiC whisker synthesis. However, iron catalyses growth of TiCyN1ÿy whiskers Table 8 EDS±TEM analyses of Ni droplets terminating TaxTi1ÿxC whiskers and Fe±droplets terminating TiCxN1ÿx whiskersa Whisker composition Droplet no. Ni (at.%) Fe (at.%) Ti (at.%) Ta (at.%) Na (at.%) Cl (at.%) P (at.%) TaxTi1ÿxC 1 99.1 ± 0.1 0 0.5 0.3 0 TaxTi1ÿxC 2 97.5 ± 0.6 0.4 1.5 0 0 TaxTi1ÿxC 3 92.7 ± 0.3 0.4 2.8 1.4 2.4 TaxTi1ÿxC 4 94.3 ± 0.3 0.2 1.0 0 4.2 TaxTi1ÿxC 5 86.1 ± 0.4 0 0.4 0 13.1 TaxTi1ÿxC 6 96.4 ± 1.0 0.5 0.3 0.5 1.3 TaxTi1ÿxC 6 77.9 ± 0.2 0 0.3 0 21.6 TaxTi1ÿxC 6 78.8 ± 0.2 0 0 0 21.0 TiC 7 77.7 ± 1.0 ± ± ± 21.3 TiC 8 73.7 ± 0 ± ± ± 26.3 TiC 8 73.1 ± 0.4 ± ± ± 26.5 TiC 8 73.8 ± 1.0 ± ± ± 25.2 TiC 9 70.6 ± 1.0 ± ± ± 28.4 TiC 9 71.9 ± 1.0 ± ± ± 27.1 TiC 10 69.4 ± 0.6 ± ± ± 30.0 TiC 10 73.9 ± 2.6 ± ± ± 23.5 TiC 11 77.6 ± 1.0 ± ± ± 21.4 TiCxN1ÿx 12 ± 98.0 0.6 ± ± ± 1.4 TiCxN1ÿx 12 ± 97.4 0.2 ± ± ± 2.4 TiCxN1ÿx 12 ± 78.3 0.8 ± ± ± 20.9 TiCxN1ÿx 12 ± 85.8 0 ± ± ± 14.2 TiCxN1ÿx 12 ± 91.1 0.3 ± ± ± 8.6 TiCxN1ÿx 12 ± 97.2 1.1 ± ± ± 1.7 TiCxN1ÿx 13 ± 82.0 1.1 ± ± ± 16.9 TiCxN1ÿx 13 ± 84.6 1.2 ± ± ± 14.2 TiCxN1ÿx 13 ± 86.6 1.0 ± ± ± 12.4 TiCxN1ÿx 13 ± 97.9 2.0 ± ± ± 1.1 TiCxN1ÿx 14 ± 99.0 0.9 ± ± ± 0.1 TiCxN1ÿx 14 ± 81.5 0.6 ± ± ± 17.9 TiCxN1ÿx 14 ± 99.5 0.5 ± ± ± 0 TiCxN1ÿx 14 ± 84.2 0.2 ± ± ± 15.6 TiCxN1ÿx 14 ± 99.5 0.5 ± ± ± 0 TiCxN1ÿx 14 ± 93.8 0.7 ± ± ± 5.5 a Red phosphorus was added to the starting mixtures in order to enhance the possibility of ®nding droplets. Di€erent spot analysis on the same droplet may give a wide spread in the phosphorous content due to that it is present in small discrete Ni± or Fe±phosphide particles. N. AhleÂn et al. / Journal of the European Ceramic Society 20 (2000) 2607±2618 2613

N. Ahlen et al. Journal of the European Ceramic Society 20 (2000)2607-2618 (see Fig. 8), most likely due to that e presenc of 3. 3. Whisker growth nitrogen suppresses the formation of Fe3c at the synthesis temperature. The whisker diameter is gen Most observations indicate that whisker growth starts erally larger with Fe than with Ni catalysts, indicating a from where nickel is in contact with carbon. 2 Both SEM difference in the surface tension between the droplet and and TEM micrographs of interrupted experiments show the solid whisker phase. EDS analysis of Fe droplets that whisker growth occasionally also may start on also gives very low contents of Ti(<I at. %) Copper oxide particles that are in contact with both carbon and does not work as catalyst metal. It has no solubility for nickel, giving very short transport distances(see Fig carbon at the synthesis temperature 9a). Such oxide particles were found in interrupted experiments for all types of whiskers studied. They were detected by giving a distinct EDS peak for oxygen in the SEM. For TaosTio.sC the oxide phase found contained both Ta and Ti, which showed it to be a secondary pre- cipitate. Whiskers also frequently terminate in a carbide particle that is actually a part of the same crystal as the whisker itself(see Fig. 9b). Such a carbide particle can be the result of direct carbothermal reduction of oxide remnants Some whiskers are thick at one end and much thinner at the other (see Fig. 10). Some narrowing whiskers terminated in a small Ni droplet at the thinnest end This narrowing may be explained by that the catalyst droplet has been eroded during whisker growth by Distance reacting with chlorine gas, giving a progressively smaller diameter. and since the surface tension balance with the solid whisker must be retained, this also leads to a pro gressively thinner whisker Fig. 7.(a) SEM microgel f a TiC whisker terminated by a Ni droplet; (b) linescan of the whisker in(a)showing that no Ni is incor- porated in its structural framework and that very little Ti remains dissolved in the Ni droplet after solidification. The Ti level is the same Table 9 the solubility of the whisker constituents in some System Eutectic temperature(C) at the eut = Ni-Ta-C Ni6.7 at.%o Ta-78 at.%C Ni-TI-C Ni9.0 at.%o T1-5.6 at.%oC Co-Ta-c 1311 Co-0.8 at.% Ta-ll 8 at. %C 200nm Co-TIC Co-3.9at.%T}-15.5at.%C Fe-Ti-C 1151 Fe-0.2 at. %T-17.5 at. %C sker terminated by a Fe droplet. Those The eutectic temperature given is the lowest temperature were generally have larger diameters than those grown with Ni as catalyst liquid is in contact with carbide phase, e.g. TiC or TaC

(see Fig. 8), most likely due to that the presence of nitrogen suppresses the formation of Fe3C at the synthesis temperature. The whisker diameter is gen￾erally larger with Fe than with Ni catalysts, indicating a di€erence in the surface tension between the droplet and the solid whisker phase. EDS analysis of Fe droplets also gives very low contents of Ti (<1 at.%). Copper does not work as catalyst metal. It has no solubility for carbon at the synthesis temperature. 3.3. Whisker growth Most observations indicate that whisker growth starts from where nickel is in contact with carbon.2 Both SEM and TEM micrographs of interrupted experiments show that whisker growth occasionally also may start on oxide particles that are in contact with both carbon and nickel, giving very short transport distances (see Fig. 9a). Such oxide particles were found in interrupted experiments for all types of whiskers studied. They were detected by giving a distinct EDS peak for oxygen in the SEM. For Ta0:5Ti0:5C the oxide phase found contained both Ta and Ti, which showed it to be a secondary pre￾cipitate. Whiskers also frequently terminate in a carbide particle that is actually a part of the same crystal as the whisker itself (see Fig. 9b). Such a carbide particle can be the result of direct carbothermal reduction of oxide remnants. Some whiskers are thick at one end and much thinner at the other (see Fig. 10). Some narrowing whiskers terminated in a small Ni droplet at the thinnest end. This narrowing may be explained by that the catalyst droplet has been eroded during whisker growth by reacting with chlorine gas, giving a progressively smaller diameter, and since the surface tension balance with the solid whisker must be retained, this also leads to a pro￾gressively thinner whisker. Fig. 7. (a) SEM micrograph of a TiC whisker terminated by a Ni droplet; (b) linescan of the whisker in (a) showing that no Ni is incor￾porated in its structural framework and that very little Ti remains dissolved in the Ni droplet after solidi®cation. The Ti level is the same as the background level. Table 9 Composition of the liquid phase at the eutectic temperature to indicate the solubility of the whisker constituents in some catalyst metals testeda System Eutectic temperature (C) Composition of the liquid phase at the eutectic temperature Ni±Ta±C 1347 Ni±6.7 at.% Ta±7.8 at.%C Ni±Ti±C 1257 Ni±9.0 at.% Ti±5.6 at.%C Co±Ta±C 1311 Co±0.8 at.% Ta±11.8 at.%C Co±Ti±C 1284 Co±3.9 at.% Ti±15.5 at.%C Fe±Ti±C 1151 Fe±0.2 at.% T±17.5 at.%C a The eutectic temperature given is the lowest temperature were liquid is in contact with carbide phase, e.g. TiC or TaC. Fig. 8. TiCyN1ÿy whisker terminated by a Fe droplet. Those whiskers generally have larger diameters than those grown with Ni as catalyst metal. 2614 N. AhleÂn et al. / Journal of the European Ceramic Society 20 (2000) 2607±2618

N. Ahlen et al. Journal of the European Ceramic Society 20 (2000)2607-2618 33. Growth directions and with smooth surfaces. Also in the Tao.s TiosC e The growth directions of the whiskers were identified by sample(1250 C), only the [100] growth direction was ectron diffraction In TaC whisker samples(synthesised found, but the surface shape of these whiskers was quite at 1250C)only the growth direction [100 was found. different. If they are studied at fairly low resolution, or The Tac whiskers are generally well shaped-straight if they are not properly oriented, they may appear to have smooth surfaces, but if the crystals are investi- gated carefully(aligned perpendicular to the whisker direction) it is clear that the surfaces are wavy. This appearance can be compatible with a screw-like three- dimensional shape of the whisker that was also found for TiC (1275C)(see Fig. Ila and b). The screw-like shape can be more or less pronounced than in this example Studying the appearance of the TiC whiskers as a function of temperature, we found that at low tempera tures(around 1150-1200oC) the whiskers were fairly straight. At about 1250C the screw-like shape were most prominent(see Fig. 12). Samples at moderately high temperatures (1275C) had screw-like shape similar to the TaosTio.sC(1250C)sample(see Fig 13) At 1300C, the crystals are again straighter, but still not as perfectly shaped as the whiskers grown at lower temperatures (see Fig. 13). At even higher tempera tures (1400oC) the shape of the whiskers is again smooth The majority (roughly three quarters) of the TiC whiskers obtained at 1250C were screw-shaped. The growth direction of these"screws"is a crystallographic direction which would be either [100]or [lll]. On many occasions there was a sharp bend in the crystals, where 500nm the growth changed to another direction, either crystal lographically equivalent or the other one possible. For instance, the whisker in Fig. 12b changes growth direc- tion from [100] to [010]. In addition to these screwed whiskers in the TiC (1250C) sample, slightly thinner, straight whiskers could also be found [00 TaC and Ta.Til-C whiskers cannot be prepared in a high yield above about 1250C. At higher temperatures TaxTil-C particles are instead formed by a direct car 100 bothermal reduction reaction between the oxide and 111 minor fraction(w from t Fig 9. (a)Whisker that may have started to gro an oxide particle; (b) TiC particle terminati 200nm particle may be the result of a direct carbothermal Fig. 10. Narrowing Taos TiosC whisker. At least two mechanisms for the formation of this type of whiskers are possible(see text)

3.3.1. Growth directions The growth directions of the whiskers were identi®ed by electron di€raction. In TaC whisker samples (synthesised at 1250C) only the growth direction [100] was found. The TaC whiskers are generally well shaped Ð straight and with smooth surfaces. Also in the Ta0:5Ti0:5C sample (1250C), only the [100] growth direction was found, but the surface shape of these whiskers was quite di€erent. If they are studied at fairly low resolution, or if they are not properly oriented, they may appear to have smooth surfaces, but if the crystals are investi￾gated carefully (aligned perpendicular to the whisker direction) it is clear that the surfaces are wavy. This appearance can be compatible with a screw-like three￾dimensional shape of the whisker that was also found for TiC (1275C) (see Fig. 11a and b). The screw-like shape can be more or less pronounced than in this example. Studying the appearance of the TiC whiskers as a function of temperature, we found that at low tempera￾tures (around 1150±1200C) the whiskers were fairly straight. At about 1250C the screw-like shape were most prominent (see Fig. 12). Samples at moderately high temperatures (1275C) had screw-like shapes similar to the Ta0:5Ti0:5C (1250C) sample (see Fig. 13). At 1300C, the crystals are again straighter, but still not as perfectly shaped as the whiskers grown at lower temperatures (see Fig. 13). At even higher tempera￾tures (1400C) the shape of the whiskers is again smooth. The majority (roughly three quarters) of the TiC whiskers obtained at 1250C were screw-shaped. The growth direction of these ``screws'' is a crystallographic direction which would be either [100] or [111]. On many occasions there was a sharp bend in the crystals, where the growth changed to another direction, either crystal￾lographically equivalent or the other one possible. For instance, the whisker in Fig. 12b changes growth direc￾tion from [100] to [010]. In addition to these screwed whiskers in the TiC (1250C) sample, slightly thinner, straight whiskers could also be found. TaC and TaxTi1ÿxC whiskers cannot be prepared in a high yield above about 1250C. At higher temperatures TaxTi1ÿxC particles are instead formed by a direct car￾bothermal reduction reaction between the oxide and carbon. A minor fraction (<5%) of the TaxTi1ÿxC whiskers (1250C) show a secondary growth of carbide phase out Fig. 9. (a) Whisker that may have started to grow from the surface of an oxide particle; (b) TiC particle terminating TiC whiskers (1300C) the whiskers and the particle constitute the same crystal. The carbide particle may be the result of a direct carbothermal reduction of oxide. Fig. 10. Narrowing Ta0:5Ti0:5C whisker. At least two mechanisms for the formation of this type of whiskers are possible (see text). N. AhleÂn et al. / Journal of the European Ceramic Society 20 (2000) 2607±2618 2615

N. Ahlen et al. Journal of the European Ceramic Society 20 (2000)2607-2618 from the whiskers. The whiskers themselves have the ondary growth is usually flat(the slab thickness is les same [100 growth direction as the others, and the than the width of the whisker). viewing crystals like secondary growth can be described as termi- those in Fig. 14 perpendicular to the whisker direction nated slabs with edges(see Fig. 14) The sec- and to the viewing direction of this figure would show a whisker"with perfect edges. It is obvious that this growth has taken place after the whisker growth was finished. This secondary growth is always more or less pure TiC and may have formed by precipitation directly from the vapour ph TiCl3(g)+2C0(g)- TiC(s)+Co(g)+3Cl(g)(8) The CO2(g) thus produced can then react with carb CO2(g)+C(s)→2COg) 500nm [100 [100 110 [0101 Fig. Il.(a) Tic(1275.C)whisker with a screw-like shap c Fig 12.(a)Overview of TiC whiskers(1250C) showing the frequent (1250.C)have similar morphology(b) TiC whiskers(12 ft screw-like appearance; (b) close up of one screw-like whisker changing one grown along [100] and the right one along [lll e growth direction. Growth started along [100 and continued along 010]

from the whiskers. The whiskers themselves have the same [100] growth direction as the others, and the secondary growth can be described as termi￾nated slabs with edges (see Fig. 14) The sec￾ondary growth is usually ¯at (the slab thickness is less than the width of the whisker). Viewing crystals like those in Fig. 14 perpendicular to the whisker direction and to the viewing direction of this ®gure would show a ``whisker'' with perfect edges. It is obvious that this growth has taken place after the whisker growth was ®nished. This secondary growth is always more or less pure TiC and may have formed by precipitation directly from the vapour phase. TiCl3… †‡ g 2CO g…†! TiC s… †‡ CO2… †‡ g 3Cl g…† …8† The CO2(g) thus produced can then react with carbon to form CO(g) CO2… †‡ g C s…†! 2CO g…† …9† Fig. 11. (a) TiC (1275C) whisker with a screw-like shape. Ta0:5Ti0:5C (1250C) have similar morphology. (b) TiC whiskers (1275C), the left one grown along [100] and the right one along [111] (note that the right one is not perpendicular to the beam). Fig. 12. (a) Overview of TiC whiskers (1250C) showing the frequent screw-like appearance; (b) close up of one screw-like whisker changing growth direction. Growth started along [100] and continued along [010]. 2616 N. AhleÂn et al. / Journal of the European Ceramic Society 20 (2000) 2607±2618