Journal 1. Am. Cenam. Soc.. 86 [101 1683.90 (2003) Method for Production of a-Alumina Whiskers via Vapor-Liquid-Solid Deposition Victor Valcarcel, Carmen Cerecedo, and Francisco Guitian* lad de Santiago de Compostela. S-15782. Spain 1120}<1100> prism because of the ed in advanced many new B. Derby-contributing editor M 2 ed une :2 2 Manuseript No. 187085. Received May 29, 2002: approved June 2. 2003. 1683
1684 Journal of the American Ceramic Sociery-Valcarcel et al Vol 86. No. 10 work, the best conditions(e. g, temperatures and percentages) for The tensile strength of some of the so-obtained whiskers, one of fiber production have been established in relation to the metal the most important features for their reinforcing applications, was estimated. The measuring procedure was as follows: One hole was The aim of the present work was to attempt ve fiber drilled on a rectangular piece of autoadhesive tape. One fiber yield. To achieve this goal, one simple approach was to increase ribbon) then was pasted onto this hole, using two drops of an d sub epoxy-resin-based glue. Special care had to be taken to keep the more deposited Al-O,. However at >1500 C, the growth of sample parallel to the longitudinal edge of the plastic strip. AL,O, fibers via VLS is restrained. We hypothesized that this Carefully, the plastic with the fiber was placed on the appropriate growth restraint occurs because the liquid silicon (Si()drops test machine. Finally, the plastic strip was cut with a hot blade, needed for the VLS process become unstable at such high and, therefore, the fiber started to hold the force by itself. when the temperatures. Because of the increasing temperature, the equilib- force exceeded the tensile strength of the fiber, it fractured rium between Si(n formation and Si(n) evaporation is altered The fracture was then examined using SEM. Sometimes the causing the drops to disappear eventually and, thus, restraining the breakage resulted from debonding between the fiber and the glue. growth of c-AL,O, whisker in which case the recorded data were rejected. If the fiber was This theory prompted us to use different metals that could be actually broken, the cross section was measured using SEM, which melted (and would. eventually, form drops)at temperatures allowed us to calculate the tensile strength of the fiber. The tensile 1500C. We then started to grow a-Al, 0, fibers by adding strength was estimated dividing the force by the surface area of the certain selected metals to the SiO, powder. The resulting fibers were chemically characterized and their tensile properties esti mated. The c-Al,O, fibers and ribbons were analyzed using duction-coupled plasma(ICP), which showed that they had not llL. Results been contaminated by the metals used during their production described earlier, the present VLS method consisted of small pieces of aluminum over a shallow bed of quartz, inert furnace atmosphere of argon. at temperatures IL. Experimental Procedure between 1300 and 1500C. a-Al, O, crystals were readily Aluminum wires(diameter 4 mm, length 1 cm) of 99.9 wt% produced by this method and grew into a white, cottonlike mass purity were placed onto a powder bed, The powders in the bed of crystals around the aluminum piece(Fig. 1).Detailed SEM were composed of quartz sand, with the eventual addition of analysis revealed that, although these crystals showed varied various percentages of metals (or metal oxides), The metals used morphologies, two classes could be distinguished: fibers and ribbons(Fig. 2). Each fiber had a hexagonal section, drops at were in the form of either powders or small pieces (balls, etc. ) one of its tips. and one basal hexagonal pyramid. Strikingly, the Hm). The Co,O, powder used in our experiments actually con- fibers had aspect ratios of >10 and, sometimes, up to 10. The sisted of a mixture of cobalt ions coexisting under two oxidation -I cm, branch width-10 um thickness.5 um), triangular states(2+ and 3+) tips with eventual drops and branching out at fixed angles(60") A. The programmed temperature cycles consisted of an initial mperature- increase rate of 10 C/min to a plateau temperature to produce large single crystals with elaborate dendritic struc that was maintained for 2-4 h, after which the system was cooled tures, Dendrites usually appear in crystals developed inside liquids, contrary to the gaseous atmosphere in which ribbons to room temperature. at a rate of 10C/min, The plateau temper- develop. The drops at the tips of the fibers clearly showed that of the times, Furnaces with inert atmospheres of argon gas were the fibers had grown via VLS deposition. Although the growth mechanism of the ribbons remains unclear. our available data used, and a 0.2 L/min argon flux was maintained during the indicate that it was vapor-solid(VS) deposition. mperature-increase and the temperature-decrease steps. The furnace tube remained closed and airtight during the periods of c-axis growth. In contrast, the ribbons showed preferential growth constant temperature. -packed A series of assays was designed to investigate and monitor fiber basal plane. Thus. the fiber-growth direction was normal to the ield. Only one variable was changed at a time. Thus, the crucibles ribbon-growth direction This finding explains how fibers can use used were always of the same size and shape; the same amounts of ribbons as crystallization surfaces, with each ribbon resulting in a SiO, powder and aluminum wires were always used. and all the set of fibers and, thus, the use of other (expensive) materials other considered variables(temperature ramps, gas flux, etc. )were avoided. At>1500 C, the production of such crystals decreased held constant. The only difference between the two assays was ignificantly. Nevertheless, some additives can alter the situation either the addition of different percentages of powdered metals( romoting fiber development at higher temperatures. their oxides) or the plateau temperature used. The fibers were The first case study on the effect of metals addition was haracterized using optical microscopy. X-ray diffractometry performed using nickel powder. Several assays were conducted to (XRD), transmission electron microscopy(TEM), and scanning show the effect of adding different amounts of powdered nickel, at electron microscopy(SEM) with energy-dispersive X-ray spec- different temperatures. For a comparison of the results, diagran oscopy analysis(EDS were designed based on the ratio between the weight of the Al,O ICP analysis revealed that the purity of the fibers obtained using an additive and that of the fibers with no as of the fib metals added. Thus, improvement resulting from the additives SEM and EDS showed that the actual purity of the fibers could be read at a glanc have been higher. Three possibilities for this discrepancy were Figure 3 shows the effect of adding 20% powdered nickel (< the fibers: (ii) impurities may have been inside the VLS drops: and improvement was observed at <1500C. but the effect became (iii)impurities may have been present because of solid solutions of considerable at 1550"C:20 times more c-Al,O, tibers were uch elements into the c-Al, O. To distinguish between"clean obtained protocol was developed purities inside the fibers, a cleaning purities and im Figure 4(A) shows the effect of adding various percentages of nickel at 1550.C. Clearly, percentages >5% did not cause further Theoretically, impurities should be removable ng. improvement. We also tested the effects of adding NiO, cobalt, and because they are not truly"inside "the fibers. This was Co,O,(Figs. 4(B)(D)). ucted in a 35% HF-65% HNO, solution EDS microanalysis clearly showed that nickel and cobalt might immersion in HCl ppeared at the drops. whereas only aluminum was localized at th
October 2003 Method for Production of a-Alumina Whiskers via Vapor-Liquid-Solid Deposition (A) (C) Fig. 1. (A) Scheme indicating the position of aluminum wires inside crucibles, placed over a SiO, bed inside a tube furnace(B) Development of corundum fibers around aluminum wires.(C) After adequate treatment, aluminum wires are completely surrounded by a white cottonlike mass of a-Al,O, fibers, fibers(Fig. 5). ICP analysis revealed that the fibers were composed polycrystalline fibers exhibited a great range of properties, bu of a-Al,O, with a purity of.5%. Nevertheless, ICP also heir tensile strength was( Fig. 7) showed that only -0 I% of the contaminating elements remained after leaching. As explained earlier. leaching was conducted in a IV. Discussion 35% HF-65% HNO, solution, followed by overnight immersion in HCL. The first leaching treatment was adequate to remove A feasible explanation for the increased production of fibers silicon from the system, and the second treatment removed aluminum. In principle, the vast majority of the other elements following arguments. First, we consider the growth of a-Al,O, present at the drops should also have been removed, Becau fibers when only aluminum and SiO, powder were used; second Al2O3 is resistant to acids, the fibers and ribbons remained we study the effect of adding certain metals to the system. unchanged after treatment. This cleaning protocol allowed u As explained earlier, when pieces of aluminum were placed analyze the impurities that were"inside"the crystals over quartz at 1300%-1500oC, a-Al,O, crystals were produced and Some of the measured single-crystal Al,O, fibers exhibited a grew into a white cottonlike mass of crystals around the aluminum tensile strength slightly >6 GPa(Fig. 6). The maximum measured piece. The drops at the tip of the fibers clearly showed that the tensile strength of the Al,O, ribbons was --260 MPa. The fibers had grown via VLS deposition
l686 Journal of the American Ceramic Sociery-Valcarcel et al Vol. 86. No: 10 100pm 100pm Fig. 2.(A)Fibers have lengths well above hundreds of micrometers, with cross-sections <10 um (B)Ribbons have very particular features, as described n the lexl. (C) Drops at the tip of fibers demonstrate that they grow via VLS deposition. (D) Schemes of a fiber. a ribbon, and the initiation of a fiber on The VLs process can be described as follows: When the aluminum evaporates, it reacts with the quartz powder in the crucible, and new gaseous species, such as aluminum and silicon oxides (i.e. AlO(g), Al,O(g), or Sio(g)) are generated Such gaseous species can react to produce Al,O,(allowing epitaxial growth to continue)and Si(l). This Si(/) forms drops in which gaseous species can dissolve, increasing their probability of reacting to produce Al,O The excess Si(l) diffuses back He-Si0, *Ni into the atmosphere. because of the Si( equilibrium vapor 9 si(h) is critical for drop formation. These drops at the fiber tips slowly dissolve gaseous phases rich in aluminum and silicon oxides, thus increasing their probability of reacting. In this way, a-Al,O, that can be deposited onto an adequate crystallization surface is produced. The fiber tip itself always acts as a crystallization surface, allowing epitaxial growth Ie continue with a constant cross section (fibers as long as -2 cm have been attained). Si(/) drops are the main factor responsible for the unidirectional growth of the c-Al,O, whiskers. 130013501400145015001550 The use of an argon-gas atmosphere with an initial oxygen Temp(C) content of about 2 ppm (Po.=10" atm) restrains aluminum oxidation and, therefore, its"passivation. "In addition, Al(g)reacts Fig 3. Effect of the addition of a 20%E of powder with oxygen, causing an even lower Po (<63m)to he SiO, at various temperatures. Below 1500 C, a sli provement can In summary, two phenomena are needed for VLS deposi be observed, but, above this temperature. the effect becomes considerable Deposition can progress as long as gaseous species are produced about 20 times more corundum fibers are obtained
October 2003 Method for Production of a-Alumina Whiskers via Vapor-Liquid-Solid Deposition (A)20 (B)2 NiO Percentage of additives Percentage of additives ① Co CoO Percentage of additives Percentage of additives Fig. 4.(AHD) Effect of the addition of various percentages of nickel, NiO, cobalt, and Co, O4 at 1550'C and drops exist. In principle, the drops are stable, whereas Si() eventually producing nickel. We considered the following reac- evaporation is balanced by Si() production tions, although other possibilities could oc If the temperature is increased to >1500C, two effects are expected. First, more gaseous species are produced, increasing 2A(g)+3NO→→AO3+3Ni(g) their probability of reacting to produce a-Al,O, fibers. Second, it seems reasonable that Si(l) at the drops evaporates more rapidly. eventually causing restraint of the VLS mechanism. To take AG=-133 kcal/mol advantage of the first effect, it seems to be necessary to stabilize those drops AlO(g)+2Nio 1550 Al,O,+ 2Ni(s, 1) In principle, then, a good strategy seemed to be to add elements that melt at temperatures >1500 and <2050.C ( the a-AL,O, △G=-162keal/mol(1) melting point). Among those elements, we decided to discard those with high toxicity and to choose the most economical of the The effects of adding cobalt or Co, O4 were similar to those for aining elements with the appropriate melting/boiling adding nickel and NiO, Co, O4 is a nonstoichiometric"commer Two candidates were studied in depth, although different possibil cial"formula for an"actual"mixture of Co, O, and CoO. Cobalt ities were tested. Results are presented here only for nickel and oxide powders are produced in this way, because it is difficult to separate them. But, for calculations, our programs cannot deal with The first element under study was nickel. As shown in Fig. 3 Co, O4. Then we use CoO as an example for the calculations, although Co, O, could be used as well increasing the weight of nickel to >5% did not result in an The existence of Co() can be explained using the followin improved yield of a-Al,O, fibers. This finding must be attributed reaction as an example to the fact that the availability of Ni(g) depends on the Ni(D) equilibrium vapor pressure, which, in turn, depends on the tem 2Al(g)+3Co0-Al,O,+3Co(n) perature. Once the nickel equilibrium vapor pressure was attained further additions of nickel did not result in more Ni(g).In accordance with the same principle, no relevant differences were △G=-238kcal/mol(2) found between the use of powdered nickel or small pieces of nickel Checking the adequate phase diagrams shows that the solid- of various shapes olution range of nickel and cobalt into AlO, is very small. Thus, When NiO was used, it reacted with the gaseous species(mainly although metals were present at the drops, the fibers grew without Al,O(g)and Al(g)) in the furnace, and the NiO was reduced, significant