No abstract

It is shown that an α→β transformation can be induced in reaction‐bonded silicon carbide by high‐temperature annealing in high nitrogen pressures. The transformation is the reverse of that commonly observed during high‐temperature processing of SiC compacts and is accompanied by the formation of substantial amorphous grain‐boundary films, suggesting a possible liquid‐phase transformation mechanism.

916LE FLUOSILICATE DE MAGNESIUM types de domaines ordonn6s, de grande extension, 6quirepr6sent6s dans le cristal. Un type de domaine contient, pour chaque vari&6 d'anions, deux orientations A pour une orientation B, l'autre deux orientations B pour une orientation A. I1 est remarquable qu'une analyse structurale classique des compos6s NiSiF 6. 6H20 et ZnSiF 6. 6H20 (Ray, Zalkin & Templeton, 1973) ait conduit ~t des proportions identiques pour les deux orientations d'octa6dres (deux orientations A pour une orientation B), dans le cas d'une r6partition d6sordonn6e de ces octa6dres (R3). La structure pr6sent6e (P3) pour MgSiF 6. 6H20 peut ~tre regard6e comme une structure antiphas6e p6riodique, b. p6riode enti~re [2M = 3d(101)], fond6e sur la structure ordonn6e monoclinique de basse temp&ature (P21/c).Cette mani~re d'envisager la structure du fluosilicate de magn6sium devait nous permettre de traiter le cas du fluosilicate de fer dans toutes ses particularit~s (taches de surstructure d'extension variable h indices non strictement entiers). Cette &ude en voie d'ach~vement fera l'object d'une prochaine publication.Nous remercions R. Saint-James pour la pr6paration d'6chantillons, A. Hardy pour sa collaboration h la mesure des intensit+s diffract6es et P. P&io avec qui nous avons eu des fructueuses discussions. Acta Cryst. (1979 AbstractLattice imaging with tilted illumination at the 2.5 A level has been used for the direct determination of stacking sequences in silicon carbide polytypes by observation of electron micrographs. The method employed is first demonstrated by observations of the common short-period polytypes 3C, 4H, 15R and 6H, and is then used to reveal the stacking sequences in the polytypes 9R and 129R, neither of which appear to have been reported previously. Stacking sequences in 0567-7394/79/060916-08501.00 disordered and faulted materials have also been characterised.

High resolution transmission electron microscopy techniques, principally involving direct lattice imaging, have been used as part of a study of the crystallography and phase transformation mechanics of silicon carbide polytypes. In particular, the 3C (cubic) and 6H (hexagonal) polytypes have been examined together with partially transformed structural mixtures. Although direct observation of two-dimensional atomic structures was not possible at an operating voltage of 100 kV, considerable microstructural information has been obtained by careful choice of the experimental conditions. In particular, tilted beam observations of the 0.25 nm lattice fringes have been made in the 3C polytype for two different { 11 1 } plane arrays in order to study the dimensions and coherency of finely-twinned regions together with (0006) and {10T2} lattice images in the 6H polytype which allow the detailed stacking operations to be resolved. Lower resolution lattice images formed with axial illumination have also been used to study the nature of the 3C-+6H transformation and results are presented showing that the transformation interface may originate with fine twinning of the 3C structure followed by growth of the resultant 6H regions. Observations have been made of the detailed stepped structure of this interface together with the stacking fault distribution in the resultant 6H material. The images obtained suggest that tilted beam techniques will be useful for studying the one-dimensional disorder in other S i c polytypes and related ceramic materials. Some of the possible advantages of future observations with high voltage, high resolution instruments are considered. '

are shown in Fig. 7(B).* The preferential consolidation from the wall inward is quite marked, in agreement with the observations of the germanium-doped layer. This correlation between the micrographs and the numerical model provides further verification that viscous sintering is the mechanism responsible for consolidation. An important prediction of the model is the dependence of the thickness of the flocculent layer which can be consolidated on the value of the capillary number. Sample numerical results indicate that decreasing the capillary number by 25%, but holding all other inputs constant, makes it possible to consolidate a layer twice as thick. V. ConclusionViscous sintering was experimentally verified as the mechanism responsible for the consolidation of the particulate layers in the MCVD process. The rate of consolidation is primarily a function of the capillary number. The consolidation rate was experimentally and theoretically shown to be extremely sensitive to physical factors, including gas thermal conductivity and composition of dopants, the latter significantly affecting the viscosity of the material being sintered. Diffusion of gas in the glass was faster than viscous *This computation used a threefold viscosity increase toward the gas-floc interface, corresponding to a decrease in Ge concentration of 8 wt% measured on sample 1 (Fig. I ) by J. W. Fleming, using X-ray fluorescence. The viscosity variation was derived from this information using data obtained from C. R. Kurkjian. sintering, implying that trapped closed pores in general do not create difficulties. However, bubble formation can occur in MCVD as a result of incomplete consolidation or excessive deposition temperatures . Acknowledgments: The writers thank G. M. Homsy for participation in all phases of this work, from the initial modeling to final documentation, M. L. Green for assistance in the use of the transmission electron microscope, and J. W. Fleming forthe X-ray data.