,- i I " Journal J. Am. Soc.. 78 !31 (1 995) Effect of Stacking Faults on the X-ray Diffraction Profiles of I3-SiC Powders Vijay V Pujar· and James D. Cawley" Depanm e nt of Mate rial s Sc ie nce and En gin ee ring, Case Sch oo l of Engin ee ring, Case Western Re erve nive rsi ty, Clevela nd , O hi o 44 106 X-ray diffraction patterns of (3C or the cubic poly- ty pe of SiC) powders often exhibit an addi tional p eak at d = 0.266 nm, high background intensity arou nd the ( Ill ) p ea k, and relative intensities for peaks which diff er from those predicted from the crystal st ru ct ur e. Com puter s imul at ions we re used to s how that aH these feat ur es are du e to stac kin g fault s in the powders and n ot du e to the presence of ot h er polytypes in the powders. Such s imul at ions allow diffr ac - tion pattern s to be gener ated for differe nt types, frequen- cies, and spatial di strib ution of faults. Co mp a ri son of the s imulation results to the XRD d ata indicates that the B-SiC pa rticles cons ist either of heav il y faulted clusters dis trib- uted irregularly between regions that have only occasional faults or twins , or the powders consi t of two types of parti- cle wit h different populations of faults: tho e with a hioh density of faults an d tho se with only twins or occ as ional fault s. Additiona l information is n ecessary to dete rmine which description is correct . However, the simulation res ults can be used to rule o ut ce rt a in fault config uration s. I. Introduction S ILlCD CARBIDE exists in a number of different polyrypic forms. The different polytypes have nearly the arne dens i- ti es, but different crystal sy mme tri es. Although many different polytypes have been reported in the Iiterature,1.2 th e polytypes 3C (zincblende structure), 4H, 6H, and 15R t are th e most com- mon. In addition, 2H (wurtzite structure) ha been frequently c it ed as a minor phase. Th e 3C polytype is often referred to as S-SiC, and all the o th er polytypes are co ll ectively referred to as a-SiC. Aspects of the rela ti ve tability of polytypes and th e condi- ti ons under whi ch a given poly type transform to another are not clear, and this has been a ubject of intere t for several years (e.g., see Refs. 4- 1 2) . The fo rm a ti on of any partic ul ar polytype is influenced by th e presence of spec ifi c impurities and hea t treatment. s It is di f fi cult to synthesize s in gle-polytype SiC pow- der and it is common for commercia ll y ava il a bl e powder to consist of several polytypes. Also, polytypic transforma- ti ons can occur in th ese powders during sintering, of which the 3C --7 6H transforma ti on is th e most common and widely s tu died. 6.'3- '6 Such transformatio ns have a significant influence, often de trimental, on the resulting microstructure and proper- ties, and th erefore are of impo rt ance in th e fa brication of s ili co n T. E. edilO r Manusc ripl o. 193617. Rece ived May 4. 1994: approved Seplembe r 20. 1 99-1. Based in pan on Ihe Ihesis submilled by V. V. Pujar fo r Ihe M.S. degree in Malerials Science and Engineering. Case Wes lern Reserve Universily, Cleveland. OH. 1994. Supported by Ihe NASA·Lewis Research Celller, Cleveland. OH. under Gralll No. NCC·3·139. ·Member. American Ceramic SocielY. ' In Ihe Ramsdell notal ion, ' used for disling ui s hi ng bel ween differel1l po lYlypes. Ihe symbol 11M re fers 10 a polytype wi lh /I number of Si-C layers along Ihe c ·axis of Ihe hexagonal unil ce ll. and M refers 10 lh e cell sy mmclr y. i.e .. eilher cubic (C). hexago nal (H), or rho mbohedral (R). 774 carbide for use in eng in eering applica ti on . 1 J.- 16 Furthermore, th e occurrence of polytypi c transforma ti on and th e nature of th e resulting polytype have been shown to be closely related to th e pre ence of stacking faults and the di ·tributi on of such faults in th e starting mate ri a l. 6-s .13 It is therefore de ira bl e to know the type a nd th e re la ti ve amounts of d ifferent poly type , and also th e type and distribution of stack in g fa ults, present in any given tarting mate ri al and to monitor the progres of th e partic ul ar transformati ons th at occur during process ing. The fir st such studies are th ose of J agodzin ki (ci ted in Ref. 17) and Jagodzin ki and Arnold, 17 who attempted a quanti- tative interpreta ti on of XRD data obta in ed from powders con- taining a mi xture of polytypes. They were una bl e to completely describe th e patterns assuming any combina ti on of polytype and attributed th e re idual mismatch at that time to "unknown disorder effects." They also noted ample-to- sample va ri a ti ons in the number and relative intensities of XRD pe aks in different s in gle crystals of nomina ll y th e same polytype. In th e interven- in g time, several techniques have bee n reported tn the literature for measuring polytype di tributions in SiC spec im ens, a ll of which are essentially variants of a least-squa re s fit of experi- menta ll y ob erved XRD data to those calculated theore ti ca ll y for an arbitrarily chosen set of polytypes. Ruska et ai .l s and Bartram (c it ed in Re f. 19) ca lculated the polytype distributions by employ in g a least-squares fit of intensities of six major peaks in the 30° -45° 28 region assuming th at only th e 3C, 4H, 6H, and 15R polytypes were present in the powders. Since this me th od in vo lves olv in g six to eight (depending on th e number of pea ks used in the 30°-45° 28 range) equa ti ons for the unknowns by multiple regression, the results are hig hl y sens i- tive to moderate fluctuation in the intensities of these peaks. Recentl y, Frevel el al . '9 have shown that the accuracy of such ca lcula ti ons can be improved by cons id ering th e charact eristic nonoverlapping reflections of the noncubi c polytypes to deter- mine the concentra ti on independently (whi ch is essentia ll y an extension of th e method orig in a ll y proposed by Jagodz in ski ). Atte mpts have al so been made to apply th e Rietve ld re fin ement method 20 for calculating polytype distributions 21 Although th ese techniques yie ld re li able res ult fo r powder m ix tures in whi ch each particle is in gle pha e with we ll -defined lattice parameters, such analyses are genera ll y unreliable in the case of SiC polytype mi xtures for at least two reasons. First, the diffrac ti on patterns from s ili con car bide powders conta in peaks th at are often common to two OJ' more polytypes. Second, and more sig ni ficant, is the consequence of ubiquitous stack in g fa ults on th e relative di ffrac ti on intens it ies and th ereby results of polytype distribution obta in ed from these. II. Background (1) Polytype Distribution Calclllations from XRD Data Fi gure I show a calculated XRD pa tt e rn for 3C-SiC together with a pattern obtained from a ty pi ca l S-SiC powde r. * The two 'Ibiden Company. Tokyo. Japa n. The data were obtained from a diffraclomeler (Philips model ADP 3250) wilh a CuKa X·ray generator and Ni filter. u in g a 0.02° step scan wilh I s at eac h slep. https://ntrs.nasa.gov/search.jsp?R=20010061714 2020-05-27T19:16:21+00:00Z