Impurity Phases in Hot-Pressed Si,N, L. K. V. LOU,* T. E. MITCHELL,* and A. H. HEUER* zyxw Department of Metallurgy and Materials Science, Case Western Reserve University, Cleveland, Ohio 44106 Impurity phases in commercial hot-pressed Si,,N4 were investi- gated using transmission electron microscopy. In addition to the dominant ,B-Si:,N, phase, small amounts of Si2N,0, Sic, and WC were found. Significantly, a continuous grain-boundary phase was observed in the -25 high-angle boundaries examined. This film is zyxwvutsrq = 10 zyxwvutsrq b thick between ,B-Si:,N, grains and -30 b thick between Si2N,0 and P-Si:,N, grains. I. Introduction ICROSTRUCTURAL characterizations of many hot-pressed M Si,N, ceramics have been reported'-5 and much attention has been given to the grain-boundary regions. The presence, struc- ture, composition, and distribution of a glassy grain-boundary phase, which reportedly assists both densification and the zyxwvutsr a+@ transformation during hot-pressing but also is assumed to give rise to poor creep properties, has been of prime concern. Transmission electron microscopy (TEM) studies by Nuttall and Thompsod indicated the presence of amorphous areas at triple points. Other workers4 made the same observation but also noted that these amorphous areas were susceptible to electron radiation damage; they were able to correlate the amount of amorphous phase with the quantity of MgO added. Furthermore, in studying the shape of thep-Si,,N, grains, they pointed out that many ,B crystals exhib- ited a hexagonal prism morphology, which was attributed to precip- itation from a liquid. Mosher et zyxwvutsrq ~ 1 . ~ used a high-temperature internal friction technique to measure the viscosity of the grain- boundary phase and suggested that it was 100 to 1000 A thick. On the other hand, workers using TEM4*5 reported that this grain- boundary film was not continuous. A simple calculation, assuming 5 vol% of glass in a material with an average grain size of 0.3 p m , yields a grain-boundary thickness of -50 A; if the glass concen- trates at triple points, the zyxwvutsr coatinuous-grain-boundary film would b : substantially less than 50 A. Clarke and Thomas' reported 10-A films at a few grain boundaries and pointed out the possibility that complete grain-boundary wetting may occur only at higher tempera- tures. Alpha Si:,N, starting powders often contain very high impurity concentrations (in particular W, Ca, C, Al, Fe, and Ks) in addition to 0.02 to 0.12 mol fraction SiO?; further contamination may be introduced during pro~essing.~ Most impurities are thought to be incorporated in the grain-boundary phase: but second-phase parti- cles such as Sic's5 and Si2N20'*3 have been detected by X-ray diffraction (XRD). The main aim of the present work was to identify the presence and distribution of the impurity phases at grain-boundary regions and as discrete particles. 11. Experimental Techniques Billets used were HS-I 30* and NC-132* hot-pressed Si,N,.+ Samples for analysis were prepared from the central portion of these billets to avoid the hot-pressed surfaces. The TEM specimens were then prepared by mechanically polishing 0.5-mm thin sections down to 20 p m and ion-thinning at 6 kV. Specimens for XRD analysis (CuKa radiation) were ground with a 30-pm diamond wheel to produce flat surfaces and analyzed using a 0.2" slit on the Received September 12, 1977; revised copy received December 8, 1977. Supported by the Air Force Office of Scientific Research under Grant No. Member, the American Ceramic Society. *Norton Co., Worcester, Mass. tObtained from zyxwvutsrqpon NASA Lewis Research Center. Berea, Ohio, and Air Force Materi- A!OSR75-2789. als Laboratory, Dayton, Ohio. detector. The TEM studies were performed on an electron micro- scopet with a 45" double-tilt stage at 125 kV. To use conventional TEM efficiently in studying hot-pressed Sit3N4, it is necessary to identify positively individual grains. In a complex body such as commercial hot-pressed Si:,N,, where the possibility of a large number of phases must be considered, phase identification on the basis of a single diffraction pattern is in- adequate, as there is always the possibility of a similar diffraction zone of another phase; it is thus always necessary to obtain a number of diffraction patterns. The technique used in the present investiga- tion involved tilting individual grains through large angles into various diffracting zones in a calibrated double-tilt stage; results were then plotted onto stereograms, which were compared with standard projections of suspected phases. In this way, various phases could be unambiguously identified. For studying grain-boundary regions in TEM at high magnifica- tions, both axial lattice fringe and conventional bright field/dark field techniques were applied. For the former technique, an image was formed by using a 300-pm condenser aperture in the illumina- tion system and either no aperture or a 100-pm o!jective apertureo, with the objective lens underfocused by - 1000 A to optimize 6-A fringe contrast. (More-closely-spaced fringes were observed in very thin areas but with less underfocus.) Objective astigmatism was corrected by focusing on the "salt and pepper" contrast at the edge of the specimen. For the grain-boundary study, it was neces- sary to have adjacent grains oriented such that symmetrical rows of spots appeared in the diffraction pattern; under this condition both grains gave lattice fringes. The essence of the conventional bright fielddark field technique is to manipulate the specimen so that the grain boundary is exactly parallel to the beam direction (edge-on grain boundary) while adja- cent grains are strongly diffracting, i.e. rows of low-index spots are present in the diffraction pattern. Under such conditions, any weakly diffracting grain-boundary film (e.g. Figs. 5 to 8) will appear bright between two dark grains in bright field. If two dif- fracted spots from adjacent grains are close to each other, a dark- field image may be obtained which will show the reverse contrast. The tilt angle of the grain boundary is less critical in this case, as the contrast arises mainly from diffraction. It is interesting to note that grain-boundary films were sought in hot-pressed A1203containing 0.1 %MgO and in a low-angle boundary between two@-ShN, grains (see below). In neither case could a grain-boundary film be de- tected, even after scrupulous and prolonged tilting efforts. 111. Results The samples of HS-130 and NC-132 gave essentially similar results and will be discussed together. (I) X-Ray Dgfractwn AllP-S&N, peaks were present, except for the (0001) reflection, which is very weak. In addition, there were small but distinct peaks which corresponded to all possible SizNzO reflections; in particular, the following four peaks, which occurred in two groups, were very prominent and could be used to identify SizNzO unambiguously: diiu=4.68 A, 1/Ii=80 anddzu,=4.44 A, I/Ii=lOO do,= 2.42 A, [/I, =80 and doZI =2.38 A, 111, =80 The amount of Si2Nz0, although not determinedquantitatively by X ray, appeared to be constant from sample to sample. $Siemens 102, Siemens Corp., Iselin, N.J 392