LATERIAL CHEMISTRYAND LSEVIER Materials Chemistry and Physics 56(1998)256-261 Factors determining the diameter of silicon carbide whiskers prepared by chemical vapor deposition Ing-Chi Leu", Yang-Ming Lu 1. Min-Hsiung Department of Materials Science and Engineering, National Cheng-Kung University, Tainan, Taiwan Abstract Since the whisker diameter is one of the important parameters for determining the characteristics of whisker-related systems, an understand ing of the factors that affect its size is of great value for whisker preparation. In this study, chemical vapor deposition(CVD)of silicon carbide( Sic)whiskers using a gas mixture of methyltrichlorosilane and hydrogen has been conducted in a hot-wall reactor on graphite plates coated with Ni as a liquid-forming agent, The deposited Sic whiskers are then characterized by scanning electron microscopy(SEM)to determine their nucleation and growth behavior. Experimental results show that the diameter of Sic whiskers is determined by both the vapor liquid-solid (VLS)mechanism and vapor-solid( VS)radial deposition, where the former is affected by the area of the solid-liquid interface from which the crystal precipitates and the latter by the thickening kinetics of vapor-deposited SiC on the lateral face. However, a comparison of the two factors indicates that an appropriate choice of the diameter of liquid droplets for VLS whisker growth is more effective than radial s deposition for obtaining whiskers of desired diame D 1998 Elsevier Science S.A. all rights reserved Keywords: Silicon carbides; Chemical vapor deposition: Whiskers 1. Introduction common method for the preparation of almost any kind of elemental and compound whisker with high quality and high The vapor-liquid-Solid (VLS) mechanism as one of the most commonly accepted mechanisms for whisker growth Because of their unique combinations of physical and during vapor-phase reaction was demonstrated by Wagner chemical properties, whiskers are expected to be useful in a and Ellis [1] in the 1960s. For its academic and practical variety of engineering fields. The application of silicon car- importance, various aspects of the VLS mechanismhave been bide(Sic) and other kinds of strong and lightweight whisk extensively studied ever since. The presence of a liquid layer ers as a second-phase reinforcement into metal and ceramic situated between the vapor and the growing crystal is what matrix composites is one of the most well-known examples distinguishes this mechanism from previous theories [2]. There also exist other kinds of applications which are less That is, with the appropriate addition or the accidental pres- established and still under development. As far as their ence of a liquid-forming agent, a liquid solution containing mechanical and functional applications are concerned, the the crystalline material to be grown and fed from the vapor characteristics of whiskers are of great importance in deter- through the liquid-vapor interface will be formed [3]. As mining the properties of whisker-related devices and articles he constituents for whiskers become supersaturated within For example, the diameter, strength, surface chemistry and the liquid solution, crystal growth proceeds by precipitation morphology of the reinforcing whiskers can definitely influ of crystalline materials from the solid-liquid interface. The ence the strengthening and toughening effects obtainable in liquid solution is consequently lifted upward at the same time reinforced composites [4]. In the case of studying the pho have the precipitation process operating continuously toluminescence properties of ultrathin GaAs whiskers grown With the advance of fundamental is by metal-organic vapor-phase epitaxy on GaAs substrate ndustrial practice, the VLS-based process is now the most Hiruma et al. [5] observed a shift of the luminescence peak energy to higher values with decreasing whisker diameter Corresponding anthor. Accordingly, the control of whisker characteristics is an Currently with the China College of Medical Technology, Tainan lmportant ste for optimizing the performance of whisker related applications. In general, the control of the character- 0254-0584/98/s- see front matter e 1998 Elsevier Science S.A. All rights reserved. PHS0254-0584(98)001898
ELSEVIER Materials Chemistry and Physics 56 (1998) 256--261 MATERIALS CHEMISTRYAND PHYSICS Factors determining the diameter of silicon carbide whiskers prepared by chemical vapor deposition Ing-Chi Leu *, Yang-Ming Lu 1, Min-Hsiung Hon Department of Materials Science and Engineering, National Cheng-Kung University, Tainan, Taiwan Received 26 February I998; received in revised form 20 June 1998; accepted 22 June 1998 Abstract Since the whisker diameter is one of the important parameters for deternffning the characteristics of whisker-related systems, an understanding of the factors that affect its size is of great value for whisker preparation. In this study, chemical vapor deposition (CVD) of silicon carbide (SIC) whiskers using a gas mixture of methyltrichlorosilane and hydrogen has been conducted in a hot-wall reactor on graphite plates coated with Ni as a liquid-forming agent. The deposited SiC whiskers are then characterized by scanning electron microscopy (SEM) to determine their nucleation and growth behavior. Experimental results show that the diameter of SiC whiskers is determined by both the vaporliquid-solid (VLS) mechanism and vapor-solid (VS) radial deposition, where the former is affected by the area of the solid-liquid interface from which the crystal precipitates and the latter by the thickening kinetics of vapor-deposited SiC on the lateral face. However, a comparison of the two factors indicates that an appropriate choice of the diameter of liquid droplets for VLS whisker growth is more effective than radial VS deposition for obtaining whiskers of desired diameters. © 1998 Elsevier Science S.A. All rights reserved. Keywords: Silicon carbides; Chemical vapor deposition; Whiskers 1. Introduction The vapor-liquid-solid (VLS) mechanism as one of the most commonly accepted mechanisms for whisker growth during vapor-phase reaction was demonstrated by Wagner and Ellis [1] in the 1960s. For its academic and practical importance, various aspects of the VLS mechanism have been extensively studied ever since. The presence of a liquid layer situated between the vapor and the growing crystal is what distinguishes this mechanism from previous theories [2]. That is, with the appropriate addition or the accidental presence of a liquid-fornfing agent, a liquid solution containing the crystalline material to be grown and fed from the vapor through the liquid-vapor interface will be formed [3]. As the constituents for whiskers become supersaturated within the liquid solution, crystal growth proceeds by precipitation of crystalline materials from the solid-liquid interface. The liquid solution is consequently lifted upward at the same time to have the precipitation process operating continuously. With the advance of fundamental issues and the maturing of industrial practice, the VLS-based process is now the most * Corresponding author. L Currently with the China College of Medical Technology, Tainan, Taiwan. common method for the preparation of almost any kind of elemental and compound whisker with high quality and high efficiency. Because of their unique combinations of physical and chemical properties, whiskers are expected to be useful in a variety of engineering fields. The application of silicon carbide (SIC) and other kinds of strong and lightweight whiskers as a second-phase reinforcement into metal and ceramic matrix composites is one of the most well-known examples. There also exist other kinds of applications which are less established and still under development. As far as their mechanical and functional applications are concerned, the characteristics of whiskers are of great importance in determining the properties of whisker-related devices and articles. For example, the diameter, strength, surface chemistry and morphology of the reinforcing whiskers can definitely influence the strengthening and toughening effects obtainable in reinforced composites [4]. In the case of studying the photoluminescence properties of uttrathin GaAs whiskers grown by metal-organic vapor-phase epitaxy on GaAs substrates, Hiruma et al. [5] observed a shift of the luminescence peak energy to higher values with decreasing whisker diameter. Accordingly, the control of whisker characteristics is an important step for optimizing the performance of whiskerrelated applications. In general, the control of the character- 0254-0584/98/$ - see front matter © 1998 Elsevier Science S.A. All rights reserved. PII S0254-0584(98) 00 i89-8
Materials Chemistry and Physics 56(1998)256-262 ics of VLS-grown whiskers, as affected by the growth the present study. a heating rate of about 10.C/min was conditions, can be accomplished by proper selection of the employed to reach the desired temperature and a total pressure chemical vapor deposition(CVD)parameters and liquid- of 100 torr was chosen for each deposition run. After an forming agents. In particular, the liquid solution, in its usual appropriate duration of the vapor-phase reaction, the depos form of a droplet on the top of the growing whisker, may ited Sic whiskers were then examined by seM to determine play a decisive role in the determination of the whisker their nucleation and growth characteristics The whisker diameter, being one of the important whisker characteristics, has been studied to some extent. It was gen- 3. Results and discussion erally reported in the literature that the diameter of whiskers was determined by the size or the diameter of liquid droplets The presence of solidified alloy droplets on whisker tips [3, 6-10] during VLS whisker growth. However, it is com- when they are cooled from the deposition temperature is monly found that the diameter of whiskers is variable during commonly regarded as evidence for the successful operation vapor-phase growth, particularly at the root of the whisker of the VLS mechanism. As discussed in our previous report where growth starts. The conditions at which the proportion- [12] and in the SEM micrograph shown later, the VLS mech- ality between the whisker diameter and that of the anism was responsible for Sic whisker growth in the present droplet prevails and the underlying factors that control the Ni-activated CVD experiment diameter of whiskers should be clearly elucidated among the The deposition of VLS-grown whiskers proceeds by the various process parameters involved precipitation of crystalline materials at the solid-liquid inter- In this study the factors that determine the diameter of face of the supersaturated liquid droplet. The area of the First, an estimation regarding the variation of liquida ated. interface for precipitation will then become the cross section liquid droplet on the solid substrate or crystalline Sic is deter- tionship between liquid droplet and whisker will be given. mined by two major factors, i.e., the volume of the liquid Then the validity of the estimated value and the effect of droplet and the interfacial equilibrium at the VLS triple-phase vapor-phase deposition on the thickening of the whisker will Inction. For two liquids having the same volume but with be experimentally verified. Growth of Sic whiskers was car- different wetting characteristics on a substrate, the liquid that ried out by thermal decomposition of methyltrichlorosilane wets the substrate better will cover the substrate with a larger (MTS)in a hot-wall reactor in the temperature range 1100 area. This phenomenon can be explained in terms of Fig. I to 1300"C. Elemental nickel as a liquid-forming agent for in which the cross sections taken through the center of the roplet formation was chosen to activate whisker growth by spherical liquid droplets(or cap-shaped droplets)are shown the VLs mechanism. The deposited whiskers were examined with respective areas covered being expressed by the diam- by scanning electron microscopy(SEM) and their growth eters of circular areas d, and d2. If liquid A and liquid B are characteristics were determined of the same volume, but the former wets the substrate better than the latter, a smaller liquid/ solid contact angle would be obtained for liquid A(ad2) and meter per minute), respectively, for the various CVD runs in with different volumes(liquid A vs liquid C, d, >d3)
I.-C, Leu et al, /Materials Chemist©' and Physics 56 (1998) 256-26] 257 istics of VLS-grown whiskers, as affected by the growth conditions, can be accomplished by proper selection of the chemical vapor deposition (CVD) parameters and liquidforming agents. In particular, the liquid solution, in its usual form of a droplet on the top of the growing whisker, may play a decisive role in the determination of the whisker characteristics. The whisker diameter, being one of the important whisker characteristics, has been studied to some extent. It was generally reported in the literature that the diameter of whiskers was determined by the size or the diameter of liquid droplets [3,6-10] during VLS whisker growth. However, it is commonly found that the diameter of whiskers is variable during vapor-phase growth, particularly at the root of the whisker where growth starts. The conditions at which the proportionafity between the whisker diameter and that of the liquid droplet prevails and the underlying factors that control the diameter of whiskers should be clearly elucidated among the various process parameters involved. In this study the factors that determine the diameter of VLS-grown SiC whiskers will be thoroughly investigated. First, an estimation regarding the variation of liquid droplet diameter at different growth stages and the dimensional relationship between liquid droplet and whisker will be given. Then the validity of the estimated value and the effect of vapor-phase deposition on the thickening of the whisker will be experimentally verified. Growth of SiC whiskers was carried out by thermal decomposition of methyltrichlorosilane (MTS) in a hot-wall reactor in the temperature range 1100 to 1300°C. Elemental nickel as a liquid-forming agent for droplet formation was chosen to activate whisker growth by the VLS mechanism. The deposited whiskers were examined by scanning electron microscopy (SEM) and their growth characteristics were determined. 2. Experimental Growth of SiC whiskers by CVD was conducted in an externally heated reactor using SiC heating elements. A detailed description of the apparatus and related specifications can be found in a previous report [ 11 ]. High-density isotropic graphite plates with dimensions of 20 ram× 20 mm× 2 mm were used as substrates. Elemental Ni as the liquid-forming agent for whisker growth was applied to the graphite substrates by electroplating with a coating thickness of 2.5 ~xm. The source reactant MTS was fed into the reactor by bubbling the MTS saturator held at 0°C with purified hydrogen. Hydrogen was also used as the main gas to adjust the reactants appropriately to the desired concentration. To evaluate the thickening kinetics of whiskers, CVD of SiC whiskers was performed in the temperature interval 1100 to 1300°C for constructing the Arrhenius plot of the growth rate in the radial direction. The flow rates of H2 and MTS vapor were metered at 1360 and 3.4 sccm (standard cubic centimeter per minute), respectively, for the various CVD runs in the present study. A heating rate of about 10°C/rain was employed to reach the desired temperature and a total pressure of 100 torr was chosen for each deposition run. After an appropriate duration of the vapor-phase reaction, the deposited SiC whiskers were then examined by SEM to determine their nucleation and growth characteristics. 3. Results and discussion The presence of solidified alloy droplets on whisker tips when they are cooled from the deposition temperature is commonly regarded as evidence for the successful operation of the VLS mechanism. As discussed in our previous report [ 12] and in the SEM micrograph shown later, the VLS mechanism was responsible for SiC whisker growth in the present Ni-activated CVD experiment. The deposition of VLS-grown whiskers proceeds by the precipitation of crystalline materials at the solid-liquid interface of the supersaturated liquid droplet. The area of the interface for precipitation will then become the cross section of the accompanying VLS whisker. The contact area of the liquid droplet on the solid substrate or crystalline SiC is determined by two major factors, i.e., the volume of the liquid droplet and the inteffacial equilibrium at the VLS triple-phase junction. For two liquids having the same volume but with different wetting characteristics on a substrate, the liquid that wets the substrate better will cover the substrate with a larger area. This phenomenon can be explained in terms of Fig. 1 in which the cross sections taken through the center of the spherical liquid droplets (or cap-shaped droplets) are shown with respective areas covered being expressed by the diameters of circular areas dl and d> If liquid A and liquid B are of the same volume, but the former wets the substrate better than the latter, a smaller liquid/solid contact angle would be obtained for liquid A ( a d2) and with different volumes (liquid A vs. liquid C, dl> d3)
I.-C. Leu et al. Materials Chemistry and Physics 56(1998)256-261 iation may alter the contact area of a liquid droplet on the substrate, which in turn gives rise to the growth of whiskers with corresponding diameter variation. However, the depend ence of the diameter of liquid droplets on their wetting behav- ior during whisker growth has not been discussed in detail yet. In the following, a simple estimation as well as its exper imental verification for assessing the geometricalrelationship between the diameter of a liquid droplet lying on the subst and that on the top of a whisker will be carried out. In the 6=65 meanwhile, the relationships between the diameter of a liquid droplet(2ra)on a whisker tip and wh e2=30 the root(2rc)and at the steady-state growth stage(2rw)will also be considered. The reported conclusion that the whisker (b) r"r, sin62 Fig. 2. Schematic cross sections taken through the center for (a)a liquid diameter is determined by the diameter of liquid droplets [10] will then be critically evaluated. Fig. 2(a)and(b) larger volume of liquid A will result in more area being cap from now on in this study lying on the substrate and a overed(d,>d, ) Therefore, the wetting behavior of a liquid VLS-grown whisker with a spherical liquid droplet on its tip, droplet on a solid substrate balanced by its volume can infu- respectively. The variation of droplet shape is due to changes ence the contact area of the VLS liquid droplet, which in turn in the contact angle of the liquid droplet at different stages of affects the size of the whisker precipitated. This wetting growth. According to a geometrical estimation using the behavior can be used to explain the reported results that the appropriate parameters shown in Fig. 2, the volumes of the diameter of liquid droplets is generally found to be larger cap and the droplet can be expressed as follows [6, 13, 14] or at least equal to the diameter of whiskers for various kinds of whiskers prepared by the addition of liquid- Veap3sin 6, /(2+cos 01)(1-cos 01)2 forming agents with varying wetting characteristics. Under the condition where liquid droplets of the same volume are considered, the contact area of the liquid droplet can now be Varoplet 3T7G-3ad(2+cos B2)(1-cos 62) determined by the wetting behavior of the liquid-solid pair concerned. Unfortunately, the wetting characteristics of liq- where Vean is the volume of the liquid cap on graphite, Vd tip, re the radius of the circular contact area of the liquid cap tion of processing parameters. on the graphite substrate as well as the radius of the whisker According to the evolution of whisker morphology, the root, ra the radius of the liquid droplet on the Sic whisker nucleation and growth process of whiskers can be divided tip, B, the contact angle of the liquid cap on the graphite into three stages. These are the nucleation stage, the transient substrate, o the contact angie of the liquid droplet on the Sic growth stage and the steady-state growth stage, which are whisker and B2 is the complementary angle to as shown characterized by an incubation period, growth of whiskers with variable diameter and growth of whiskers with constant Fig 3 shows SEM micrographs for the droplet at different diameter, respectively [15]. As described above, variation in stages of whisker growth. The measured magnitudes of 8, the wetting characteristics as exhibited by contact-angle var- and c are about 65 and 150, respectively. An angle B2 of 30 25k3;B 当Hm日2日42 ig. 3. SEM micrographs of the droplet at different growth stages for(a) a liquid cap on the substrate and (b)a whisker at the steady-state growth stage
258 L-C. Leu et al. ~Materials Chemistry and Physics 56 (t998) 256-26t substrate NN~ r~ sinO~ Ol = 65° ,,,o =150' 02=30 ° (a) (b) r, =r~.sin02 Fig. 2. Schematic cross sections taken through the center for (a) a Iiquid cap on the substrate and (b) a liquid droplet on the whisker tip. larger volume of liquid A will result in more area being covered (dl > d3). Therefore, the wetting behavior of a liquid droplet on a solid substrate balanced by its volume can influence the contact area of the VLS liquid droplet, which in turn affects the size of the whisker precipitated. This wetting behavior can be used to explain the reported results that the diameter of liquid droplets is generally found to be larger [6,t3,14] or at least equal to the diameter of whiskers for various kinds of whiskers prepared by the addition of liquidforming agents with varying wetting characteristics. Under the condition where liquid droplets of the same volume are considered, the contact area of the liquid droplet can now be determined by the wetting behavior of the liquid-solid pair concerned. Unfortunately, the wetting characteristics of liquid droplets on solid substrates are highly sensitive to the growth ambient and are thus easily influenced by the fluctuation of processing parameters. According to the evolution of whisker morphology, the nucleation and growth process of whiskers can be divided into three stages. These are the nucleation stage, the transient growth stage and the steady-state growth stage, which are characterized by an incubation period, growth of whiskers with variable diameter and growth of whiskers with constant diameter, respectively [ 15 ]. As described above, variation in the wetting characteristics as exhibited by contact-angle vatiation may alter the contact area of a liquid droplet on the substrate, which in turn gives rise to the growth of whiskers with corresponding diameter variation. However, the dependence of the diameter of liquid droplets on their wetting behavior during whisker growth has not been discussed in detail yet. In the following, a simple estimation as well as its experimental verification for assessing the geometrical relationship between the diameter of a liquid droplet lying on the substrate and that on the top of a whisker will be carried out. In the meanwhile, the relationships between the diameter of a liquid droplet (2ra) on a whisker tip and whisker diameters both at the root (2ro) and at the steady-state growth stage (2rw) will also be considered. The reported conclusion that the whisker diameter is determined by the diameter of liquid droplets [ 10] will then be critically evaluated. Fig. 2(a) and (b) depicts a spherical-cap-shaped liquid droplet (called a liquid cap from now on in this study) lying on the substrate and a VLS-grown whisker with a spherical liquid droplet on its tip, respectively. The variation of droplet shape is due to changes in the contact angle of the liquid droplet at different stages of growth. According to a geometrical estimation using the appropriate parameters shown in Fig. 2, the volumes of the cap and the droplet can be expressed as follows: 1[ ro ,~3 V~p = g r~ si---~l ) (2+cos 0,)(1-cos 0,) z (1) 4 1 3 Vdrop,e, = grrr2- ~rra (2+cos 02)(1--cos 02) 2 (2) where Vcap is the volume of the liquid cap on graphite, Vdropl~t the volume of the spherical liquid droplet on the SiC whisker tip, re the radius of the circular contact area of the liquid cap on the graphite substrate as well as the radius of the whisker root, ra the radius of the liquid droplet on the SiC whisker tip, O, the contact angle of the liquid cap on the graphite substrate, q~ the contact angle of the liquid droplet on the SiC whisker and 02 is the complementary angle to q~ as shown in Fig. 2. Fig. 3 shows SEM micrographs for the droplet at different stages of whisker growth. The measured magnitudes of 0t and q~ are about 65 and 150 °, respectively. An angle 02 of 30 ° Fig. 3. SEM micrographs of the droplet at different growth stages for (a) a liquid cap on the substrate and (b) a whisker at the steady-state growth stage
L.C. Leu et al. Materials Chemistry and Physics 56(1998)256-261 thus results. Substituting the above values into Eqs.(1)and (2), volumes of the cap and the droplet of about 1. 136r. and 4.135ra are obtained, respectively. Fig 3(a)was obtained erupting the vapor-phase reaction at the end of the nucleation stage. The constituents of SiC whiskers should be about to be saturated in the liquid droplet in order to be able to precipitate at the solid-liquid interface. It is reasonable to argue that the volume of the liquid droplet (or liquid cap) ge, since no significant volume increase due to all Fig. 4. Schematic cross sections taken through the center of wh would be possible, As a consequence, the assumption of constant volume for both the liquid cap on the graphite sub rate and liquid droplet on the whisker tip could be employe to estimate their dimensional relationships. The following impurities were used as liquid-forming agents for SiC expression can be deduced whisker growth, respectively. The reported values of the ratio of the droplet diameter(2ra) to the whisker diameter(2rw) fell within the ranges 1.7-2. 1 and 3-3.5, respectively, as After appropriate algebraic operat compared to about 1. 5-2 for whiskers obtained from Ni n. a rato containing droplets in the present study [15]. The difference 1. 538: 1 is obtained. The value of rw. as shown in Fig. 2(b)is in the values accounts for the fact that droplet size as well as equal to the product of ra and sin Bz. The dimensional rela- tionship between re, ra and rw is therefore represented the ratio of droplet diameter to whisker diameter cannot nec essarily be the dominant factor in determining the diameter with experimentally measured values for the whisker indi- of the whiskers grown. The conclusion can generally be cated by an arrow in Fig 3(b), where the values of the anism. another observation to explain the effect of wetting diameter(2re, 2ra and 2rw )measured at different sections of whisker are 13. 1,8.8 and 4.5 mm, respectively. A measured characteristics on whisker diameter is the presence of knotted dimensional relationship of 1.49: 1: 0.51 can be obtained. In whiskers or the periodic variation of whisker diameter along its axis with essentially the same volume of liquid droplet view of the changes in wetting characteristics of liquid drop- located on the whisker tip. It was found that changes in wet- lets during whisker growth, especially at the transient growth stage exhibiting a consequently variable whisker diameter, ting characteristics caused by perturbations in supersaturation the reported conclusion 10] that the diameter of whiskers is of the liquid solution might give rise to the growth of a determined by the diameter of liquid droplets can only be whisker with variable diameter along its axis [14, 17] valid in the steady-state growth stage with the characteristic In addition to the abovementioned factor in determinin of an essentially constant contact angle the diameter of whiskers grown by the vls mechanism, the According to the above results and discussion, a general effect of CVd of Sic on the lateral surface of whiskers should conclusion can be drawn that applies to all whiskers grown also be discussed. That is, the thickening process by two- by the VLS mechanism. The exact factor for determining th diameter of a VLS-grown whisker is the contact area of the during CVD should not be overlooked. The JANAF Ther solid-liquid interface from which the crystal precipitates. The mochemical Tables [18] indicate that the Gibbs free energy wetting behavior, characterized by the value of the contact change for the decomposition of MTS becomes negative at angle, may vary during whisker growth. It is especially true temperatures higher than 800C. However, due to the exis that at the initial stage of whisker growth a tapered cross tence of a nucleation barrier for the formation of Sic crystal section grown from a liquid droplet with increasing contact lites, nothing could be found even at temperatures above angle will be found prior to the transition to the steady-state 1150"C with the low reactant concentration employed in the growth stage. Even with the appropriate choice of the kind deposition runs without the addition of liquid-forming agent of liquid-forming agent, the volume of liquid droplets is ne [12]. Under conditions when the nucleation barriers are once not necessarily the decisive factor for determining the reduced by the assistance of other kinetic factors, the vapor- whisker diameter. As a consequence, a liquid droplet with a solid(vs)deposition of Sic could occur continuously and large volume located on a whisker having a small diameter contribute to the increase in the radial dimension of whiskers. might be possible due to its large contact angle. On the other The growth kinetics concerning the increase in the radial nd,a small contact angle for a liquid cap on a whisker dimension during Ni-activated whisker preparation have crystal would lead to the growth of a whisker with the same been found and will be discussed in detail elsewhere [15] diameter for only a small volume of liquid-forming agent An explanation underlying the nucleation and growth mech- added. The results are depicted in Fig 4 and can be verified anism for whisker thickening will be employed to clarify the as compared with those reported by Bootsma et al. [14] and apparently contradictory arguments mentioned above. Wag DeJong and Mc Cauley [6], where iron and iron-containing ner and Ellis [2] proposed a mechanism in which the pres-
L-C. Leu et al. / Materials Chemistry and Physics 56 (1998) 256-262 259 thus results. Substituting the above values into Eqs. ( 1 ) and (2), volumes of the cap and the droplet of about 1.136r; ~ and 4.135rd 3 are obtained, respectively. Fig. 3(a) was obtained by interrupting the vapor-phase reaction at the end of the nucleation stage. The constituents of SiC whiskers should be about to be saturated in the liquid droplet in order to be able to precipitate at the solid-liquid interface. It is reasonable to argue that the volume of the liquid droplet (or liquid cap) would be essentially the same after passing the nucleation stage, since no significant volume increase due to alloying would be possible. As a consequence, the assumption of constant volume for both the liquid cap on the graphite substrate and liquid droplet on the whisker tip could be employed to estimate their dimensional relationships. The following expression can be deduced: V~ ~r = 1.136re 3 = 4.1351h 3 = Vdropie t After appropriate algebraic operation, a ratio rc:rd = 1.538:1 is obtained. The value of rw as shown in Fig. 2 (b) is equal to the product of ru and sin0,. The dimensional relationship between to, rd and r,,, is therefore represented as ro:rd:rw = 1.538:1:0.5. The result is in reasonable agreement with experimentally measured values for the whisker indicated by an arrow in Fig. 3(b), where the values of the diameter ( 2r¢, 2r~ and 2r,,.) measured at different sections of a whisker are 13.1, 8.8 and 4.5 re_m, respectively. A measured dimensional relationship of 1.49:1:0.51 can be obtained. In view of the changes in wetting characteristics of liquid droplets during whisker growth, especially at the transient growth stage exhibiting a consequently variable whisker diameter, the reported conclusion [ 10] that the diameter of whiskers is determined by the diameter of liquid droplets can only be valid in the steady-state growth stage with the characteristic of an essentially constant contact angle. According to the above results and discussion, a general conclusion can be drawn that applies to all whiskers grown by the VLS mechanism. The exact factor for determining the diameter of a VLS-grown whisker is the contact area of the solid-liquid interface from which the crystal precipitates. The wetting behavior, characterized by the value of the contact angle, may vary during whisker growth. It is especially true that at the initial stage of whisker growth a tapered cross section grown from a liquid droplet with increasing contact angle will be found prior to the transition to the steady-state growth stage. Even with the appropriate choice of the kind of liquid-forming agent, the volume of liquid droplets is now not necessarily the decisive factor for determining the whisker diameter. As a consequence, a liquid droplet with a large volume located on a whisker having a small diameter might be possible due to its large contact angle. On the other hand, a small contact angle for a liquid cap on a whisker crystal would lead to the growth of a whisker with the same diameter for only a small volume of liquid-forming agent added. The results are depicted in Fig. 4 and can be verified as compared with those reported by Bootsma et al. [ 14] and DeJong and McCauley [6], where iron and iron-containing d,, Fig. 4. Schematic cross sections taken through the center of whiskers with the same diameter but gTown from liquid droplets having different volumes. (Volume of droplet A is smaller than that of droplet B.) impurities were used as liquid-forming agents for SiC whisker growth, respectively. The reported values of the ratio of the droplet diameter (2re) tO the whisker diameter (2r,~) fell within the ranges 1.7-2.1 and 3-3.5, respectively, as compared to about 1.5-2 for whiskers obtained from Nicontaining droplets in the present study [ 15]. The difference in the values accounts for the fact that droplet size as well as the ratio of droplet diameter to whisker diameter cannot necessarily be the dominant factor in determining the diameter of the whiskers grown. The conclusion can generally be applied to various whisker crystals grown by the VLS mechanism. Another observation to explain the effect of wetting characteristics on whisker diameter is the presence of knotted whiskers or the periodic variation of whisker diameter along its axis with essentially the same volume of liquid droplet located on the whisker tip. It was found that changes in wetting characteristics caused by perturbations in supersaturation of the liquid solution might give rise to the growth of a whisker with variable diameter along its axis [ 14,17]. In addition to the abovementioned factor in determining the diameter of whiskers grown by the VLS mechanism, the effect of CVD of SiC on the lateral surface of whiskers should also be discussed. That is, the thickening process by twodimensional nucleation and growth in the radial direction during CVD should not be overlooked. The JANAF Thermochemical Tables [18] indicate that the Gibbs free energy change for the decomposition of MTS becomes negative at temperatures higher than 800°C. However, due to the existence of a nucleation barrier for the formation of SiC crystallites, nothing could be found even at temperatures above 1150°C with the low reactant concentration employed in the deposition runs without the addition of liquid-forming agent [ 12 ]. Under conditions when the nucleation barriers are once reduced by the assistance of other "kinetic factors, the vaporsolid (VS) deposition of SiC could occur continuously and contribute to the increase in the radial dimension of whiskers. The growth kinetics concerning the increase in the radial dimension during Ni-activated whisker preparation have been found and will be discussed in detail elsewhere [ 15]. An explanation underlying the nucleation and growth mechanism for whisker thickening will be employed to clarify the apparently contradictory arguments mentioned above. Wagner and Ellis [2] proposed a mechanism in which the pres-
L.C. Let et al. /Materials Chemistry and Physics 56(1998)256-261 ence of crystalline defects on the substrates to be deposited 1250 1190 11361087 could effectively reduce the activation barrier required for nucleation and growth of crystalline deposits. Moreover, discussed by Kitamura et al. [19], the activation energy for the incorporation of growth units into the potential growth sites demonstrates an increasing order of kink, step, re-entrant 3 comer and surface. Fig. 5(a) shows a transmission electron Activation Energy 180KJ/mole pic image in which contrast bands due to stacking a 100p [20,21] are clearly observed. The presence of such packed stacking faults with random widths is believed to form microtwins in the order of a few lattice parameters thick [20]. These microtwins are expected to develop microfacets on the whisker surface. The points of deflection in microfaceted surface profile were found to coin- cide with planar defects intersecting the whisker surface 16 nd have been shown in lattice images by Nutt [16] and Wang et al. [22]. A schematic illustration of the zig-zag surface as a result of the intersection of defect planes with the whisker surface is depicted in Fig. 5(b). Closely distributed Reciprocal Temp. (x10-4 steps, kinks and re-entrant comers are thus believed to prevall Fig. 6. Arrhenius plot for the thickening kinetics of CVD SiC whiskers for on the surface of the whiskers these low -barrier sites for an MTS concentration of 0. 25% and pressure of 100 torr. deposition will effectively help the process of nucleation and growth of Sic on the lateral surfaces to proceed at the rather which is somewhat smaller than the values of about 200-400 ow reactant concentration employed in the present study as kJ/mol reported in the literature for CVD SiC coatings under the whiskers are nucleated from the liquid droplet by the VLs similar processing conditions [23 ] This discrepancy may be mechanism. The temperature dependence of the average attributed to the abovementioned kinetic assistance of the radial growth rate as measured for whiskers at the steady- microfaceted surface state growth stage is plotted in Fig. 6. An activation energy In summary, as whiskers are grown by the VLS mecha of about 180 kJ/mol is obtained from this Arrhenius plot, nism, the determining factors for whisker diameter should be Fig. 5.(a) TEM micrograph and (b) a schematic drawing of microfacets caused by stacking faults intersecting the whisker surface
260 1-C, Leu et aL /Materials Chemistry and Physics 56 (t998) 256-261 ence of crystalline defects on the substrates to be deposited could effectively reduce the activation barrier required for nucleation and growth of crystalline deposits. Moreover, as discussed by Kitamura et at. [ 19], the activation energy for the incorporation of growth units into the potential growth sites demonstrates an increasing order of kink, step, re-entrant corner and surface. Fig. 5 (a) shows a transmission electron microscopic image in which contrast bands due to stacking faults [20,21] are clearly observed. The presence of such densely packed stacking faults with random widths is believed to form microtwins in the order of a few lattice parameters thick [20]. These microtwins are expected to develop microfacets on the whisker surface. The points of deflection in microfaceted surface profile were found to coincide with planar defects intersecting the whisker surface [ 16] and have been shown in lattice images by Nutt [16] and Wang et al. [22]. A schematic illustration of the zig-zag surface as a result of the intersection of defect planes with the whisker surface is depicted in Fig. 5 (b). Closely distributed steps, kinks and re-entrant comers are thus believed to prevail on the surface of the whiskers. These low-barrier sites for deposition will effectively help the process of nucleation and growth of SiC on the lateral surfaces to proceed at the rather low reactant concentration employed in the present study as the whiskers are nucleated from the liquid droplet by the VLS mechanism. The temperature dependence of the average radial growth rate as measured for whiskers at the steadystate growth stage is plotted in Fig. 6. An activation energy of about 180 kJ/mol is obtained from this Arrhenius plot, E e¢ e- (-9 N "o e¢ 1316 looo I I I 10 1250 1190 1136 1087 1 , I , I , I 7.6 8,0 8.4 8.8 9.2 Reciprocal Temp. (xl 0 -4) Fig. 6. Arrhenius piot for the thickening kinetics of CVD SiC whiskers for an MTS concentration of 0.25% and pressure of 100 torr. which is somewhat smaller than the values of about 200-400 kJ/mol reported in the literature for CVD SiC coatings under similar processing conditions [23 ]. This discrepancy may be attributed to the abovementioned kinetic assistance of the microfaceted surface. In summary, as whiskers are gown by the VLS mechanism, the determining factors for whisker diameter should be (a) (b) Fig. 5. (a) TEM micrograph and (b) a schematic drawing of microfacets caused by stacking faults intersecting the whisker surface. =
I.-C. Leu et al. Materials Chemistry and Physics 56(1998)256-261 the size of the solid-liquid interface where crystallization growth rate of Sic whiskers leads to an apparent activation ccurs as well as the thickening rate in the radial direction of energy of about 180 kJ/mol. However, a comparison of the the whisker during vapor-phase deposition. The size of the two factors using the parameters employed in the present solid-liquid interface is determined by the volume of the study indicates that an appropriate choice of the volume of liquid droplet and the equilit liquid droplets during VlS whisker growth is more effective tensions among the phases involved. Proper selection of the than radial vs deposition for obtaining whiskers of desired type and volume of liquid droplets for a specific whisker- diameters substrate system together with a careful choice of processing parameters can facilitate the tailoring of wetting characteris ics of liquid droplets. The controllability of the diameter of Acknowledgements whiskers grown by the VLS-based process will in turn be chanced. Nevertheless the effect of the substrate on the The financial support of this study by the National Science interfacial energy balance can be negligible as the liquid Council of the Republic of China under grant contractnumber droplets are lifted from the substrate [24]. A constant whisker NSC84-2216-E006-012 is greatly appre diameter is obtained at the steady-state growth stage provided nat the interfacial energy balance has not been disturbed during processing. The above reasoning can simplify further References discussion and a certain dimensional relationship between [1]RS. Wagner, w.C. Ellis, Appl Let.4(1964)89 the liquid droplets and the whiskers would exist for a specific [2]RS Wagner, W.C. Ellis, Trans E233(1965)1053 growth stage, i.e, the steady-state growth stage, as was com- [3] J.V. Milewski, F D. Gac, J.J. Petrovic, S.R. Skaggs, J. Mater. Sci. 20 monly reported by others. In the present study, the difference (1985)1160 in the effectiveness of the two factors in determining the [4] P F. Becher, T N. Tiegs, P.A. Angelini, in K.S. Mazdiyasni(Ed whisker diameter is significant. The diameter of Sic whiskers Fiber reinforced Ceramic Composites- Materials, Processing and Technology, ch Il, Noyes Publications, Park Ridge, NJ, USA, 1990 grown by the addition of a 2.5 um thick Ni coating as the [5] K. Hiruma, M. Yaeawa, K. Haraguchi, K. Ogawa, J. Appl. Phys. 74 liquid-forming agent is about 2-3 um due to the disintegrated metal particles being about twice as large, whereas an average 6]R De Jong, R.A. McCauley, J Am Ceram Soc. 70(1987)C radial growth rate of Sic whiskers of about 0. 2 um/h is [7] G F. Hurley, J.J. Petrovic, Advanced Composites Conf. Proc Materials Park, OH, USA, 1985, p. 20 obtained under the conditions mentioned above except for a J P.D. Shalek, W.J. Parkinson, Mater. Res, Soc. Symp Proc., vol. 168, deposition temperature of 1300C instead. Consequently, the Mater Res. Soc., Pittsburgh, PA, USA, 1990, p. 255 appropriate choice of the size of activating metal particles fo [9] M. Yazawa, M. Koguchi, A. Muto, M. Ozawa, K. Hiruma, Appl whisker growth, if no further breakup of the liquid droplet Phys.Let.61(1992)2051 takes place during deposition, is more effective than the thick [10] J. Westwater, D. Pal Gosain, S. Usui, Jpn J. Appl. Phys. 36(1997) ening deposition in the radial direction in controlling the [11] DJ.Cheng, WJ. Shyy, D HKuo, MHHon,JElectrochem. Soc diameter of Sic whiskers (1987)3145 [13] G.A. Bootsma, H.J. Gassen, J. Crystal Growth 10(1971)223 [14] G.A. Bootsma, W.F. Knippenberg, G. Verspui, J Crystal Growth 11 4. Conclusions M H. Hon, Y M. Lu, to be published he factors that control the diameter of sic whiskers are 16]SR Nutt, J Am Ceram Soc. 71(1988)149. summarized as the result of a combined operation of the VlS 17] E.I. Givargizov, J. Crystal Growth 20(1973)217. mechanism and VS radial deposition. The former is deter [18] JANAF Thermochemical Tables, 2nd edn., NatL. Stand. Ref, Data Ser 37.1971 mined by the area of the solid-liquid interface and the latter [19] M Kitamura, S Hosoya, I Sunagawa, J, Crystal Growth 47(1979 by the thickening kinetics of vapor-deposited SiC on the face. The solid-liquid contact area is determined by [20] L.I. Van Torne, J Appl. Phys. 37(1966)1849 he volume of the liquid droplets and the equilibrium condi [21] J.J. Comer, Mater Res. Bull. 4(1969)279. [22] L. Wang, H Wada, L.F. Allard, J Mater, Res. 7(1992)148. tion at the vapor-liquid-solid triple-phase junction. An [231 J Schlichting, Powder Metallurgy Intemational 12(1980)141 Arrhenius plot of the temperature dependence of radial 24]Z Wokulski, J, Crystal Growth 82(1987)427
L-C. Leu et al./ Materials Chemistry and Physics 56 (1998) 256-261 261 the size of the solid-liquid interface where crystallization occurs as well as the thickening rate in the radial direction of the whisker during vapor-phase deposition. The size of the solid-liquid interface is determined by the volume of the liquid droplet and the equilibrium condition of interfacial tensions among the phases involved. Proper selection of the type and volume of liquid droplets for a specific whiskersubstrate system together with a careful choice of processing parameters can facilitate the tailoring of wetting characteristics of liquid droplets. The controllability of the diameter of whiskers grown by the VLS-based process will in turn be enhanced. Nevertheless, the effect of the substrate on the interfacial energy balance can be negligible as the liquid droplets are lifted from the substrate [ 24 ]. A constant whisker diameter is obtained at the steady-state growth stage provided that the interfacial energy balance has not been disturbed during processing. The above reasoning can simplify further discussion and a certain dimensional relationship between the liquid droplets and the whiskers would exist for a specific growth stage, i.e., the steady-state growth stage, as was commonly reported by others. In the present study, the difference in the effectiveness of the two factors in determining the whisker diameter is significant. The diameter of SiC whiskers grown by the addition of a 2.5 b~m thick Ni coating as the liquid-forming agent is about 2-3 bun due to the disintegrated metal particles being about twice as large, whereas an average radial growth rate of SiC whiskers of about 0.2 /xm/h is obtained under the conditions mentioned above except for a deposition temperature of 1300°C instead. Consequently, the appropriate choice of the size of activating metal particles for whisker growth, if no further breakup of the liquid droplet takes place during deposition, is more effective than the thickening deposition in the radial direction in controlling the diameter of SiC whiskers. 4. Conclusions The factors that control the diameter of SiC whiskers are summarized as the result of a combined operation of the VLS mechanism and VS radial deposition. The former is determined by the area of the solid-liquid interface and the latter by the thickening kinetics of vapor-deposited SiC on the lateral face. The solid-liquid contact area is determined by the volume of the liquid droplets and the equilibrium condition at the vapor-liquid-solid triple-phase junction. An Arrhenius plot of the temperature dependence of radial growth rate of SiC whiskers leads to an apparent activation energy of about 180 kJ/mol. However, a comparison of the two factors using the parameters employed in the present study indicates that an appropriate choice of the volume of liquid droplets during VLS whisker growth is more effective than radial VS deposition for obtaining whiskers of desired diameters. Acknowledgements The financial support of this study by the National Science Council of the Republic of China under grant contract number NSC84-2216-E006-012 is greatly appreciated. References [ 1 ] R.S. Wagner, W.C. Ellis, Appt. Phys. Lett. 4 (I964) 89. [2l R.S. Wagner, W.C. Ellis, Trans. AIME 233 (1965) 1053. [3] J.V. Milewski, F.D. Gac, J.J. Petrovic, S.R. Skaggs, J. Mater. Sci. 20 (1985) i160. [4] P.F. Becher, T.N. Tiegs, P.A. Angelini, in K.S. Mazdiyasni (Ed.), Fiber Reinforced Ceramic Composites -- Materials, Processing and Technology, ch. i 1, Noyes Publications, Park Ridge, NJ, USA, 1990. [5] K. Hiruma, M. Yaeawa, K. Haraguchi, K. Ogawa, J. Appi. Phys. 74 (1993) 3162. [6] R. De Jong, R.A. McCauley, J. Am. Ceram. Soc. 70 (1987) C-338. [7] G.F. Hurley, J.J. Petrovic, Advanced Composites Conf. Proc., ASM, Materials Park, OH, USA, 1985, p. 207. [8] P.D. Shalek, W.J. Parkinson, Mater. Res. Soc. Symp. Proc., vot. 168, Mater. Res. Soc., Pittsburgh, PA, USA, 1990, p. 255. [9] M. Yazawa, M. Koguchi, A. Muto, M. Ozawa, K. Hiruma, Appl. Phys. Lett. 61 (1992) 2051. [10] J. Westwater, D. Pal Gosain, S. Usui, Jpn. J. Appl. Phys. 36 (1997) 62O4. [ 11 ] D.J. Cheng, W.J. Shyy, D.H. Kuo, M.H. Hon, J. Electrochem. Soc. i34 (1987) 3145. [ 12] I.C. Leu, Y.M. Lu, M.H. Hon, J. Crystal Growth I67 (1996) 607. [ 13] G.A. Bootsma, HJ. Gassen, J. Crystal Growth 10 ( 1971 ) 223. [ 14] G.A. Bootsma, W.F. Knippenberg, O. Verspui, J. Crystal Growth I 1 (I971) 297. [ 15] I.C. Leu, M.H. Hon, Y.M. Lu, to be published. [ 16] S.R. Nutt, J. Am. Ceram. Soc. 7i (1988) 149. [ 17] E.I. Givargizov, J. Crystal Growth 20 (1973) 217. [18] JANAF Thermochemical Tables, 2nd edn., NatI. Stand. Ref. Data Set. (US Natl. Bur. Stand.) No. 37, 1971. [ 19] M. Kitamura, S. Hosoya, I. Sunagawa, J. Crystal Growth 47 (1979) 93. [20] L.I. Van Tome, J. Appl. Phys. 37 (i966) I849. [21] JJ. Comer, Mater. Res. Bull. 4 (1969) 279. ~22] L. Wang, H. Wada, LF. Allard, J. Mater. Res. 7 (1992) 148. [23] J. Schlichting, Powder Metallurgy International 12 (1980) 141. [24] Z. Wokulski, J. Crystal Growth 82 (1987) 427