To a great degree an improvement in the properties of aluminum nitride-base ceramics is related to an increase in purity of the raw material, particularly to a reduction in the quantity of oxygen, the impurity most strongly influencing the thermal conductivity and mechanical and other properties of the material. In addition, methods of determination of the oxygen content in nitrides, as in other aluminum compounds, are very inadequate. For these purposes the method of reduction extraction of oxygen by carbon in the form of carbon monoxide, which is distinguished by high sensitivity and rapidity, is normally recommended [1]. However, for aluminum compounds this method has poor reproducibility and accuracy and also low reliability of the results. This is the result of difficulty in reduction of aluminum oxide by carbon, formation of A10 gaseous suboxide not recorded by this method, intense dissociation of aluminum nitride during the analysis, and the fact that the particles of the unreacting substance are carried off by the liberated nitrogen from the reaction zone. In order to eliminate these processes in analysis of oxygen in the nitride and other aluminum compounds, the authors of [1] recommend that platinum or tin capsules with the weighed portion be dropped into a graphite crucible with a platinum bath heated to 2700 K. It is clear that, for largescale analyses, the necessity of which is the result of the wider use of aluminum nitride ceramics, such recommendations are inapplicable. Reference has been made to the possibility of determination of nitrogen in refractory nitrides [2] and in mixture of aluminum and vanadium nitrides [3], and of oxygen in mixtures of aluminum and calcium oxides [4] with use of the TN314, TC-136, and RO-316 instruments (Leco, USA) but the conditions of the analyses are not described. In order to establish the possibility of determination of the oxygen content in aluminum nitride and other aluminum compounds in this work, the temperature and time relationships of extraction of oxygen from aluminum nitride and oxide and mixtures of them under different conditions, including in the presence of a molten nickel---copper mixture and with sealing of the opening of the graphite crucible with a partially fused metal stopper, were studied. The method of the experiment consisted of measurement of the temperature and time relationships of the degree of extraction of oxygen and nitrogen (share of extracted gas in relation to its total content in %) from aluminum oxide and nitride and also of proportioned mixtures of them in pulsed heating of the substances being analyzed in a graphite crucible in a flow of helium. The design of the instrument for pulsed reduction extraction with chromatographic separation of the gases and recording with a detector based on thermal conductivity, the dimensions of the graphite crucibles, and the basic stages of preparation of the samples and conducting the analysis, are similar to those described in [5, 6]. The investigations were made in the 500-4300 K range and one heating pulse for a 5--30 mg sample lasted 3--30 sec. The LKhM-80 chromatograph (model No. 2) was operated with a column thermostat temperature of 333 K, a thermal conductivity detector thermostat temperature of 383 K, a thermal conductivity detector current of 120 mA, and helium consumptions through the measuring column of 1.5" 10 -4 and through the comparison column of 3" 10 -5 m3/min. With the use of the metal bath 1--4 mg of carbonyl nickel powder was added to the graphite crucible together with the sample before analysis. The crucible was covered with well-matched copper plugs (diameter 3 mm, weight about 0.35 g). In turning on of heating a portion of the copper plug adjoining the water-cooled electrode of the pulse furnace remains infused while the opposite portion fuses earlier than interaction of the plug with the graphite starts. The copper--nickel alloy wets the graphite well, and with the help of capillary forces closes the lack of tightness close to the plug, envelops the sample, and covers the inner surface of the crucible, decreasing the gas permeability of its walls. In this way carrying away of the dispersed particles of the sample by the liberated gas is prevented. Powder substances with known oxygen (standard specimens of No. 471, R21 iron powder, tungsten and erbium oxides) and nitrogen (S41b and $44b standard specimens, and also titanium nitride powders, the quantity of nitrogen in which was determined by the Kjeldahl method) were used for calibration of the instrument. The nitrogen content in the aluminum nitride powders was determined by the same method.
Read full abstract