Thermodynamics of the System NaF-AlF 3 : Part V. Solid Solution in Cryolite
- PDF / 549,453 Bytes
- 4 Pages / 612 x 792 pts (letter) Page_size
- 99 Downloads / 188 Views
the previous paper in this ser-ies! a theoretical analysis of the shape of the cryolite liquidus curve was used to demonstrate that the solid material in equilibrium with the liquid at high temperatures was not stoichiometric. The composition indicated was 2.824 NaF ·AIF a. The aim of the present work was to provide some direct physical evidence for the existence of nonstoichiometry, and to try to establish the phase boundaries. The existence of solid solutions of CaF 2 in cryolite was demonstrated" by means of the variation of electrical resistivity; substitution of Ca 2+ for Na+ is accompanied by the formation of cation vacancies and the conductivity is increased by a factor of up to 50. Since a structure deficient in NaF also almost certainly has cation vacancies," it may be anticipated that nonstoichiometry would be associated with changes of resistivity. The same technique was therefore used to look for it. EXPERIMENTAL Mixtures were made up by weight from hand picked Greenland cryolite, reagent grade NaF (Baker Chemical Co.) and sublimed AIF a. About 15 g of each mixture was melted in a platinum crucible and poured out onto a cold aluminum tray. Freezing took about 15 s or less. The cold material was crushed by hand in a mortar. Resistivity specimens were made by compressing the powder into alumina tubes 5 mm ill x ~3 cm long with two stainless steel pistons squeezed in a vise. Silver powder (precipitated, Fisher Scientific Co.) was similarly compressed onto the ends of the specimen to act as electrodes. The specimen was then mounted between two stainless steel leads in a Vycor tube (12 mm ill) flushed with argon, together with a Pt/pt-lO pct Rh thermocouple. Resistance measurements were made with a Wheatstone bridge fed from a Hewlett-Packard 3311A oscillator. The detector was a Keithley 840 lock-in ampliE. W. DEWING is with the Kingston Research Center of the Aluminum Co. of Canada, Kingston, Ontario, Canada, K7L 4Z4. Manuscript submitted March 15, 1978. METALLURGICAL TRANSACTIONS B
fier. Having verified that plots of resistance vs 1/{f if is frequency) were effectively linear in the range above 0.5 kHz, measurements were normally made only at 1 and 5 kHz and were extrapolated to infinite frequency. At the end of a run the pellet of cryolite (which had invariably sintered) was recovered, if necessary by breaking the alumina tube, and its diameter and length (usually 5 to 7 mrn) were determined. Measured resistances were then converted to resistivities. The mean density found for the pellets was 2.45, with a standard deviation of 0.04. This is 83 ± 2 pet of theoretical. It was found that readings tended to drift towards higher resistance with time. XRD analysis indicated that AlFa was being lost, and some etching of the inside of the Vycor tube occurred. The readings reported here were therefore obtained rapidly (about 2 h for a complete run) to mitigate the effects of composition changes. There was sometimes hysteresis between heating and cooling, which is responsible for some of the scatter observed, but
Data Loading...
