Method of Strengthening Silicon Nitride Ceramics

This invention relates to the field of ceramics and particularly to the field of silicon nitride (Si.sub.3 N.sub.4) ceramics.

During the last decade, Si.sub.3 N.sub.4 ceramics have received considerable attention primarily directed toward heat engine applications. Their intrinsic properties such as low thermal expansion and moderate elastic modulus minimize stresses developed during severe thermal transients better than other candidate structural ceramics. In practice, the chemical formula Si.sub.3 N.sub.4 represents a family of ceramic materials in which Si.sub.3 N.sub.4 constitutes the major phase.

Si.sub.3 N.sub.4 ceramics are prepared by mixing Si.sub.3 N.sub.4 powder with additive powders, and densifying the mixture by either hot-pressing or pressureless sintering. Hot pressing is usually carried out in graphite dies at temperatures between 1650.degree. C. and 1750.degree. C. for several hours under high pressures. Pressureless sintering is accomplished by forming the mixed powders into shapes before densification. These shaped, powdered parts are then sintered by heating them at the sintering temperature.

Si.sub.3 N.sub.4 powder oxidizes when heated to high temperatures in air, and such oxidation is undesireable because it changes the composition of the powder, i.e., increases the SiO.sub.2 content. Therefore, special precautions are taken during the processing and sintering of Si.sub.3 N.sub.4 ceramic to prevent it from oxidizing. Such precautions include providing a protective atmosphere such as a nitrogen atmosphere, or packing the ceramic in a protective Si.sub.3 N.sub.4 powder as described in U.S. Pat. No. 4,179,486 to Frederick F. Lange (one of the present inventors).

Additives, called densification aids, are required for densification, and these densification aids are mixed with Si.sub.3 N.sub.4 powder. Current commercial, hot-pressed materials are fabricated using MgO as a densification aid. Other known densification aids include ZrO.sub.2, Al.sub.2 O.sub.3, BeO, and rare earth oxides such as Y.sub.2 O.sub.3, CeO.sub.2, Ce.sub.2 O.sub.3, Nd.sub.2 O.sub.3, and La.sub.2 O.sub.3.

The densification aids react with Si.sub.3 N.sub.4 and SiO.sub.2 at high temperatures to form a liquid, which upon cooling results in a glassy phase located between the primary phase of crystalline Si.sub.3 N.sub.4 grains so that the material is a dense, polyphase ceramic. Unfortunately, this glassy phase becomes viscous at high temperatures and causes the mechanical properties to degrade at high temperature. The temperature at which degradation begins depends on the composition of the glassy phase. The extent of degradation depends on the volume fraction of the glassy phase. Within a given system, compositions containing a smaller volume fraction of glassy phase have significantly higher strength and creep resistance at high temperature. However, compositions containing little or no glassy phase are difficult or impossible to densify.

It is an object of the invention to provide an improved Si.sub.3 N.sub.4 ceramic.

It is an object of the invention to provide a method of increasing the higher temperature strength of Si.sub.3 N.sub.4 ceramics.

It is an objective of the invention to provide a method of fabricating a Si.sub.3 N.sub.4 ceramic utilizing a densification aid to obtain rapid densification and still obtain a high strength Si.sub.3 N.sub.4 ceramic.

It is an object of the invention to provide a method of reducing the amount of glassy phase in Si.sub.3 N.sub.4 ceramics.

It is an objective of the invention to provide a method of increasing the oxidation resistance of Si.sub.3 N.sub.4 ceramics.

According to the invention, silicon dioxide (SiO.sub.2) is provided on the surface of a Si.sub.3 N.sub.4 ceramic which has a glassy secondary phase located between grains and/or at grain triple junctions of the primary phase of crystalline Si.sub.3 N.sub.4. The SiO.sub.2 can be provided by immersing the Si.sub.3 N.sub.4 alloy in powdered SiO.sub.2, or by heating the Si.sub.3 N.sub.4 alloy in air to oxidize the Si.sub.3 N.sub.4 and form a thin scale of SiO.sub.2.

The Si.sub.3 N.sub.4 ceramic and the SiO.sub.2 are then heated to a temperature which is sufficient to cause diffusion of cations from the silicates in the glassy phase into the SiO.sub.2 scale or SiO.sub.2 powder on the surface of the ceramic. This cation diffusion takes place because the SiO.sub.2 on the surface forms a diffusion or reaction couple with the silicate glassy phase inside the ceramic. This outward cation diffusion and the inward diffusion of oxygen has been discovered to result in a change of both the chemistry and phase content of the bulk material. Any surface pits formed during the heating are removed by grinding.

High strength Si.sub.3 N.sub.4 ceramics can be fabricated using sufficient quantities of densification aids to promote rapid densification during sintering despite the fact that the densification aids are known to decrease the high temperature strength of the ceramic. Then the high temperature strength of the ceramic can be improved by the oxidation-induced compositional changes procedures of the invention. This is possible because the compositional changes decrease the volume fraction of glassy phase.

FIG. 1 is a curve showing the distribution of Mg across the thickness (both scale and subscale) of an oxidized sample of an Si.sub.3 N.sub.4 -SiO.sub.2 -MgO ceramic;

FIG. 2 is a curve showing the oxidation of a Si.sub.3 N.sub.4 -SiO.sub.2 -MgO ceramic from its surface below the scale to its center plotted as a function of Si.sub.2 N.sub.2 O/Si.sub.3 N.sub.4 ratio vs distance below the surface; and

FIG. 3 is a log-log plot of the strain rate vs compressive stress at 1400.degree. C. for unoxidized and oxidized samples of Si.sub.3 N.sub.4 ceramics.

Silicon nitride based ceramics are inherently unstable in oxidizing atmosphere, and since they are polyphase materials, their oxidation behavior is quite complex. Current commercial ceramics have a primary phase of Si.sub.3 N.sub.4 grains and a glassy secondary phase used to densify the alloy. In the case of ceramics which use MgO as a densification aid (generally referred to a composition in the Si-Mg-O-N system), the glassy phase has a eutectic composition (in mole fraction) of approximately 0.1 Si.sub.3 N.sub.4, 0.3 Si.sub.2 N.sub.2 O, and 0.6 Mg.sub.2 SiO.sub.4. The glassy phase also contains small amounts of cation impurities such as Ca, Fe, Al, Mn, Na, and K. The glassy secondary phase is principally a continuous intergranular phase.

The oxidation of Si.sub.3 N.sub.4 ceramics was investigated in work leading to the present invention. A variety of ceramics throughout the Si.sub.3 N.sub.4 -SiO.sub.2 -MgO system were investigated. These included two series containing 0.833 and 0.755 mole fraction of Si.sub.3 N.sub.4. The materials were prepared by hot-pressing composite powders of Si.sub.3 N.sub.4, MgO, and SiO.sub.2 as known in the art. Specimens for oxidation were cut from each material and the constituent phases were identified by x-ray diffraction analysis. Specimens were then oxidized in air at 1400.degree. C. in an open-ended furnace for 24 and 288 hours.

After oxidation, the specimens were cut in half. One half was used to determine the compositional gradient from the exterior to the interior. The procedure was to successively grind the surface and determine the surface phase with x-ray diffraction until the center of the specimen was reached. The other half was used for a non-dispersive x-ray microanalysis determination of the spatial distribution of elements from the oxidized surface to the interior. Samples having the same compsitions, but unoxidized, were sectioned and analyzed to provide calibrations for the x-ray microanalyses.

When the oxidized specimens were sectioned and polished, three distinct regions could be discerned in all the materials: a friable, white oxide scale, a sub-scale reaction zone beneath the scale and a central portion. For each specimen, it was found that the scale was greatly enriched in Mg and Ca relative to the unoxidized samples. There was no detectable enrichment or depletion of silicon in the material beneath the scale. In each of the specimens there was, however, a magnesium depletion in the subscale material, the extent of which was a function of the MgO/SiO.sub.2 ratio of the starting composition. In the unoxidized samples, the Mg/Si ratio was independent of position, indicating that during oxidation, Mg had diffused to the scale from as far as the center of the specimens.

FIG. 1 shows the distribution of Mg in the scale 2 and in the sub-scale 4 based upon the Mg/Si ratios measured by x-ray microanalysis of a specimen having 0.755 mole fraction of Si.sub.3 N.sub.4 which was oxidized for 288 hours at 1400.degree. C. The specimen was 3360 microns thick and had developed a scale 180 microns thick on both exposed sides. The Mg/Si ratio in the sub-scale is shown in lower curve 8. THe ratio prior to oxidation is shown by dashed line 10. Since Mg is depleted in the immediate region below the scale, it appears that Mg is drawn into the scale faster than it diffuses from the interior into the scale.

Compositional changes within the sub-scale material were discovered after measuring the Si.sub.2 N.sub.2 O/Si.sub.3 N.sub.4 ratio from the surface of the sub-scale material (just under the scale) to the center of the specimen as shown by curve 12 in FIG. 2. Dashed line 14 shows the Si.sub.2 N.sub.2 O/Si.sub.3 N.sub.4 ratio of the sample before oxidation. The decrease in Mg and the increase in Si.sub.2 N.sub.2 O shown in FIGS. 1 and 2 indicate that the glassy phase in the specimen has been reduced by the oxidation treatment, particularly near its surface.

The diffusion of Mg and cation impurities to the surface during oxidation is apparently caused by the formation of a reaction couple between the SiO.sub.2 scale on the oxidized surface and the Mg containing glassy phase in the bulk of the material. This diffusion produces a compositional gradient and, thus, must produce a depletion gradient of the glassy phase. Since the composition at the scale/sub-scale interface has shifted furthest from the eutectic (or glass) composition, rules for interpreting phase diagrams indicate that the volume fraction of the glassy phase will be smallest at this interface. In terms of an equivalent phase diagram, the composition of the glassy phase is nearly identical to the eutectic composition and the composition in the subscale material has shifted away from the eutectic composition as a result of the compositional changes produced by diffusion.

The SiO.sub.2 on the surface which forms one side of the diffusion couple and draws out the detrimental glassy phase can be provided simply by heating the specimen in air or oxygen as described above. This creates an SiO.sub.2 scale on the oxidized surface as a result of the reaction of oxygen with Si.sub.3 N.sub.4. In a second embodiment of the invention, the SiO.sub.2 is provided by surrounding the Si.sub.3 N.sub.4 ceramic with SiO.sub.2 powder during a high temperature heat treatment. Cations (other than Si) which form the glassy phase will diffuse to the SiO.sub.2 powder on the surface in an attempt to reach equilibrium.

The post-fabrication treatment methods of the invention will improve the high temperature stability of dense Si.sub.3 N.sub.4 ceramics. Oxidation resistance is improved due to the decreased flux of anions (oxygen, nitrogen) and cations (additive and impurities) as the volume fraction of the glassy (or viscous) phase, which is the path for fast diffusion, is decreased by the post-fabrication treatment.

It has also been discovered that oxidation as described above increases the high temperature creep resistance and strength of Si.sub.3 N.sub.4 ceramics. This is an important discovery because it makes Si.sub.3 N.sub.4 ceramics more useful in structural applications. The strength increase is quite substantial, as shown by the following examples:

Specimens (approximately 0.3.times.0.3.times.0.9 cm) were diamond cut from a ceramic of 0.75 mole fraction Si.sub.3 N.sub.4, balance MgO and SiO.sub.2, with a molar ratio of MgO/SiO.sub.2 of 0.09. Some of the specimens were oxidized in an air furnace at 1400.degree. C. for 100 hours. This created an oxide surface scale which was less than 50 .mu.m thick.

Compressive creep testing was performed at 1400.degree. C. in air on the unoxidized and on the oxidized specimens. Log-log plots of this data are shown in FIG. 3, dashed curve I-a being for the unoxidized specimen and solid curve I-b being for the oxidized specimen. As shown, the oxidation treatment greatly improved the compressive creep resistance.

Specimens were prepared as described for Example I except that the molar ratio of MgO/SiO.sub.2 was 2.28. The compressive creep resistance is shown in FIG. 3, dashed curve II-a being for the unoxidized specimen and solid curve II-b being for the oxidized specimen. The creep strength of the oxidized specimen was considerably higher than the unoxidized specimen.

Specimens were prepared as described for Example I except that the mole fraction of Si.sub.3 N.sub.4 was 0.83 and the molar ratio of MgO/SiO.sub.2 was 2.40. The compressive creep resistance is shown in FIG. 3, dashed curve III-a being for the unoxidized specimen and solid curve III-b being for the oxidized specimen.

To determine if the method applied to Si.sub.3 N.sub.4 ceramics having a Y.sub.2 O.sub.3 rather than MgO glass forming densification aid, specimens were cut from a ceramic of 0.85 mole fraction Si.sub.3 N.sub.4, balance Y.sub.2 O.sub.3 and SiO.sub.2, with a molar ratio of Y.sub.2 O.sub.3 /SiO.sub.2 of 0.5. Some of the specimens were oxidized in an air furnace at 1600.degree. C. for 120 hours. The higher oxidation temperature was used because of the higher melting temperature of the Y.sub.2 O.sub.3 -containing glass as compared to the MgO-containing glass. The oxidized ceramic (curve IV-b) had a higher compressive creep strength than the unoxidized ceramic (curve IV-a) as shown in FIG. 3.

Standard flexural strength tests were run on specimens in the Si.sub.3 N.sub.4 -SiO.sub.2 -MgO system having the composition shown in Table I. The specimens were oxidized together in air at 1400.degree. C. for 300 hours. After oxidation, each specimen was separately surface ground until surface pits (formed to various depths on the materials of different composition) were no longer observed by visual inspection with a binocular microscope. Previous work by one of the coinventors (F. F. Lange, "Reaction of Iron with Si.sub.3 N.sub.4 Materials to Produce Surface Pitting", Journal of The American Ceramic Society, Vol. 61, No. 5-6, May-June, 1978) has shown that large pits form during the oxidation of many (but no all) Si.sub.3 N.sub.4 ceramic compositions. These pits are responsible for the degradation in flexural strength previously reported for commercial Si.sub.3 N.sub.4 ceramics fabricated with MgO and subjected to long term oxidation.

The flexural strength at 1400.degree. C. is shown in Table I for an unoxidized specimen and for a specimen oxidized in an air furnace at 1400.degree. C. for 300 hours. The oxidized specimens had a significant increase in flexural strength.

Flexural test specimens were diamond cut from a large billet of Si.sub.3 N.sub.4 obtained from the Norton Company and commercially identified as NC-132 hot-pressed Si.sub.3 N.sub.4. Specimens were subjected to oxidation in air at 1400.degree. C. for 548 hours and then surface ground to remove pits as described for Examples V-XIX. The average flexural strength of the oxidized specimens at a test temperature of 1400.degree. C. was 54,500 psi. This compares to an average flexural strength (at 1400.degree. C.) of only 40,900 psi for similarly prepared specimens which did not have the oxidation treatment.

To determine if a higher temperature oxidation treatment could

be used to reduce the treatment time, flexural test specimens from the NC-132 billet were prepared as described for Example XX except that the oxidation temperature was increased from 1400.degree. C. to 1500.degree. C. The average flexural strength (at 1400.degree. C.) of specimens oxidized at 1500.degree. C. for 206 hours was 69,900 psi. For specimens oxidized at 1500.degree. C. for 306 hours it was 82,000 psi. These strengths compare to an average flexural strength (at 1400.degree. C.) of only 40,900 psi for similarly prepared specimens which did not have the oxidation treatments. While the above treatments increased the high temperature strength of the specimens shown, these treatments neither increased nor decreased these same specimens' room temperature strength (i.e., their room temperature strength was within the normal scatter of untreated specimens).

The advantage obtained by the oxidation treatment as shown in the above examples can be utilized during the compounding and fabrication of the ceramic. In normal practice of preparing Si.sub.3 N.sub.4 ceramics, the fabricator adds a densification aid in order to form a liquid phase during sintering and densifying. During cooling from fabrication some or all of the liquid solidifies as a glass. This glassy phase reduces the high temperature strength and oxidation resistance of the ceramic. Therefore, the fabricator must strike a compromise between the amount of densification aid added to improve sintering and the strength of the ceramic. The present invention greatly reduces this problem because it provides a means for drawing out some of the glassy phase after the ceramic is densified to improve its high temperature strength and oxidation resistance.

According to the invention, a sufficient quantity of densification aid is added to the Si.sub.3 N.sub.4 powder to cause rapid densification due to the formation of a liquid, glassy phase during sintering. The mixture of powder is then sintered to form a dense Si.sub.3 N.sub.4 ceramic. The ceramic is cooled and the glassy phase is drawn out by the oxidation treatment previously described.

The amount of grinding required after the oxidation treatment depends upon the depth of the pits formed during the oxidation treatment.