960
The Vapor Pressures of Silicon Tetrachloride and Germanium Tetrachloride below Their 21elting Points
by P. Balk and D. Dong International Business Machines Corporation. Thomas J . Watson Research Center, Yorktown Heights, S e w York (Received October 1 7 , 1963)
I n the course of an investigation of the interaction of chlorine with silicon and germanium, the need arose for accurate data on the temperature dependence of' the saturation vapor pressures of SiC1, and GeCl, in the range from 0.01 to 0.5 mm. Since no published data below 1 mm. were available, it mas decided to measure these quantities. For both substances the pressures a t the melting points are about 0.6 mm.; thus in the pressure range of interest both substances are solids. Considerable undercooling (by more than 20') was observed when trying to freeze the liquids in both cases. This provided a convenient opportunity to extend the range of liquid vapor pressure data, particularly since the published values show considerable scatter between 1 and 10 mm. fpr SiCld2as well as for GeCLa The present note describes the measurements and discusses their results. Experimental A 100-g. sample of semiconductor grade SiC1, (or GeCI4) was transferred in a dry nitrogen atmosphere to a glass apparatus consisting of a purification section, a pressure-measuring section, and a, vacuum station, separated from each other by a set of metal valves (Hooke A 434 E(). After seaIing off the apparatus, the sample was subjected to several cycles of freezing, evacuating, and melting to remove traces of dissolved permanent gases. Sext, the sample was distilled over mercury to remove chlorine. Finally, the liquid was fractionated, retaining a middle fraction of about 20 g. After flushing several times with vapor, about 0.5 g. of sample mas introduced into the evacuated pressure measuring section. Here, the compound was frozen out in a side arm and subsequently sublimed into a cryostated container kept at a temperature where the saturation vapor pressure is approximately 0.01 mm. The residual vapor was pumped off. Mercury was used in the cryostat as a heat transfer medium. The desired temperatures, measured with a calibrated copper-constantan thermocouple, were obtained by adjusting the input to a built-in heater, while cooling continuously with liquid nitrogen. The pressure was monitored continuously with a thermistor gage.4 After the temperature and the vapor pressure The Journal of Physical Chemistry
?;OTES
had reached constant values, the pressure was measured with a 1IcLeod gage. Then the heater current mas increased slightly and the next point was taken. When repeating a run, the sample was removed from the pressure measuring section by pumping and a new charge was introduced. From this point, the same procedure was applied as described above. The melting points or, more accurately, the triple points, were taken from the onset of a plateau in the plot of the thermistor bridge off -balance voltage us. temperature, for a continuous slow temperature rise. During the measurements, the McLeod gage was separated from the cryostat by a U-tube immersed in an ice-salt mixture. The readings were corrected for the mercury vapor pressure. Thermomolecular pressure effects a t the cryostat were estimated following Liang,b using collision diameters calculated according to Jona and Mandel.6 The corrections were always less than 0.2% and mere therefore omitted. The uncertaaintyof the IIcLeod gage readings (the gage had a 3-mm. precision bore capillary) ranged from 2% at the lowest pressures to 0.2% at the melting points. The accuracy of the temperature readings was better than 0.1'. It is thought that the scatter in the experimental data is mainly caused by material condensed in the temperature transition zone a t the top of the cryostat and by sticking of the mercury in the gage. The vapor pressures for the undercooled liquids were obtained using the thermistor gage, after sealing off the McLeod gage from the system. The thermistor gage was first calibrated, using the vapor pressure data for the corresponding solids, a t increasing temperatures. After passing through the melting point, the direction of temperature change was reversed and the liquid vapor pressure data were taken. The error in the pressures obtained with the thermistor gage is about 0.3y0 over the entire range, discounting systematic errors in the calibration. Results and Discussion The data points for independent runs on each solid are indistinguishable and yield straight lines when plotted as log pressure us. 1/T, as shown in Fig. 1. They can thus be represented by an equation of the form
log p (mm.)
=
-a/T
+b
(1)
~
(1) P. Balk, to be published. ( 2 ) A. Stock, C, Somieski, and R. Wintgen, Ber., 50, 1754 (1917). (3) A. W. Laubengayer and D. L. Tabern, J . Phys. Chem., 30, 1047 (1926). (4) 4 . J. Rosenberg, J . Am Chenz. Soc., 78, 2929 (1956). (5) S. C. Liang, J . Phys. Chem., 57, 910 (1953). (6) F. Jona and G. Mandel, J . Chem. Phys., 38, 346 (1963).
NOTES
96 1
obtained for the entire set. I t should be rioted that the straightness of thc plot of the data for the undercooled liquid is conditioned by the straightness of that for the solid, since the latter data were used i n the form of eq. 1. to calibrate the theinmistor gage for the measuremerits or1 the undercooled liquid. 1Jpon solidification Table I : Values for a, 0, arid A RanRe, K.
"'.OO\ 4+-d.--L-I--I.-L
SiCI, (Solid) SOLID\
'
5.5
5.0
-lo3
IT
(Undercoolcd liq.) (1:nderc:ooled liq.)
(*KY
GeC14 (Solid) (Undercooled liq.)
Figure 1. Vapor pressurm of solid Sic], and GeC1,. IIifTerent symbols indicate different independent runs. Solid lines are Icast-square representations (eq. 1, 'I'able I ) of experiniental points.
-I
Figure 2. Vapor pressures of undercooled liquid SiCl, and G e ( L 1):tshed and solid lines are least-square representations (eq. 1 , Table I ) of experimental points for liquids and solids, respwtively.
Values of the constants a and b, obtained by a leastsquare treatment, are listed in Table I, along with the probable errors (A) in the pressures. Experimental points for the undercooled liquids (Fig. 2 ) can also be reprcscnted by a n expression of the form of eq. I (see Table I). There seems to be a very slight curvature in the plot of the SiC1, data points, and the values for a and b computed for the first half set of points from the melting point down are slightly different from those
A* 0
6
c/u
174. t5-1~1,~. 2202 i 5 182.7-n1.p. 1879 f 2
10,8(i f 0 . 0 3 1 , 3 8 , 9 9 f0.01 0 . 5
193.0--rn.p. 187 1 i 4
8 . 9 5 i0 . 0 2 0 . 3
186.8--111.p. 2437 d= 8 1 0 . 7 7 f 0 12 2 . 0 1!37.8-ni.p. 2036 f 1 8 . 9 5 9 d= 0.004 0 . 3
of the liquid the data points fall on the vapor pressure line of the solid (E'ig. 2 ) , which is indicative of the absence of volatile impurities in thc tetrachlorides. The melting points (triplc points) were found to be 204.3 f 0.1OK. for SIC], (lit.?204.4OK.) and 221.6 f 0 . l 0 K . for GeCL. The melting point for GeCl, seems to indicate that the solid was identical with the metastable 8-phase, for which meltiiig point8values of 221.4OI.,D . I'fnlilar, and 11. RWtair. ibid., 70, 381X (194X). (IO) See ref. i , p. 52. ( I 1) E. A. Gugeeriheim. "Therniod~tiamit:s." 4th PX., NGr1,hHolln nd I'ublishirig Co., Amsterdam, 195!), u p . 148, 148. ( 8 ) 1'.
(9) 11. FI. Sisler, W. .J.
Volume 68, n'umber 4
A p r i l , 1904
NOTES
962
are 9.31 f 0.01, 11.14 f 0.04, and 1.83 f 0.02 kcal. mole -I, respectively. The data permit calculation of the entropies of the solids. Starting from Kelley and King's value for the standard entropy of SiC1, vapor (79.1 f 1.0 e.u.),13 one obtains the entropy a t the melting point and 760 mm. by subtracting 7.75 e.u. (from Kelley's specific heat equation),14 and the entropy of the vapor a t the saturated vapor pressure a t the melting point by subsequently adding a pressure correction term (to 0.617 mm.) of 14.14 e.u. By subtraction of the entropy of sublimation: 10350/204.3 = 50.65 e.u. (present data), a value of 34.8 f 1.3 e.u. is found for the entropy of solid SiC1, a t the melting point. This value is close to that obtained in ref. 13 from a recalculation of Latimer's datal2 (35.05 e.u.), and it is still compatible with Latimer's original value (33.85 e.u.).12 The entropy of GeC14 vapor a t the melting point and saturated vapor pressure (0.590 mm.) is obtained similarly by subtracting 6.53 e.u.I6 from the standard entropy of the vapor (83.0 f 0.8 e.u.)16 and next adding a pressure term of 14.22 e.u. A value of 40.4 f 1.2 e.u. for solid GeC14 follows after subtraction of 11,140/221.6 = 50.30 e.u. for the entropy of sublimation (present data). No value for solid GeC14has been listed in the literature. Vapor pressure equations for the liquid tetrachlorides also follow from Kelley's free energy equations." They are of the form logp
=
-A/T
+ B + Clog T
(2)
where C is the difference between the specific heats of vapor and liquid, divided by the gas constant R, and the constants A and B were apparently chosen to fit the experimental data above 10 mm. The present data for undercooled Sic&are about 5% higher than the values obtained from Kelley's expression. Inspection shows that a satisfactory fit to the present data and the higher pressure data can be obtained by a slight adjustment in the values of A and B. For undercooled GeCl,, the data reported in this note are 20% higher. It is not unlikely that this discrepancy is caused by the larger scatter in the experimental points from which the constants of eq. 2 were derived, and by the rather high estimate of the value for the difference in the specific heats. (12) W.M. Latimer, J. Am. Chem. SOC.,44, 90 (1922). (13) K. K . Kelley and E. G. King, Bureau of Mines Bulletin 592, U. S. Govt. Printing Office, Washington, D. C., 1961, p. 83. (14) K. K. Kelley. Bureau of Mines Bulletin 584, U. S. Govt. Printing Office, Washington, D. C., 1960, p. 163. (15) See ref. 14, p. 72. (16) See ref. 13, p. 42. (17) K. K. Kelley, Bureau of Mines Bulletin 383, U. S. Govt. Printing Office, Washington, D. C., 1935, pp. 47, 91.
The Journal of Physical Chemistry
Electrolytic Separation and Purification
of Oxygen from a Gas Mixture by Stanley H. Langer" and Robert G. Haldeman Central Research Division, American Cyanamid Company, Stamford, Connecticut (Received September 9, 1963)
Oxygen is usually prepared by fractional distillation of air or electrolysis of water. The latter has the disadvantages of overvoltage, resulting in greater than theoretical power requirements, and the need to separate and dispose of simultaneously evolved hydrogen. An alternate convenient electrolytic technique would be separation of oxygen from air or other gaseous mixtures using porous catalytic oxygen electrodes working in opposite directions with an electrolyte transport and barrier phase between them. Oxygen is selectively removed, transported, and regenerated according to the following equations in basic aqueous solution.
O2
+ 2H20 + 4e +40H-
(at the cathode) (1)
40H- transported through the 40H- +O2
electrolyte to the anode
(2)
+ 2H20 + 4e (at the anode)
(3)
Thus, oxygen is separated from other less electroactive materials which are relatively insoluble in the electrolyte barrier phase. The applied external voltage must be sufficient to overcome the characteristic overvoltage loss for reactions 1 and 3 as well as concentration and ohmic polarization. This may still be considerably less than the 1.5 v. or more generally necessary for the electrolysis of water ; furthermore, no hydrogen is evolved. I n acid solution, reaction may be represented as
+ 4H+ + 4e 2Hz0 (cathode) 2H20 -+ O2 + 4Hf + 4e (anode)
O2
---t
(4) (5)
Using the schemes above and molded catalystpolyethylene or catalyst-polytetrafluoroethylene1v2gas permeable electrodes, we have obtained substantially pure oxygen from air (purity > 97%, mass spectrometrically) with acid, base, and ion-exchange membrane
* Chemical Engineering Dept., University of Wisconsin, Madison 6, Wisc. (1) (a) G. V. Elmore and H. A. Tanner, J.Electrochem. SOC.,108, 669 (1961); (b) W. T. Grubb, "Proceedings of the 16th Annual Power Sources Conference," PSC Publications Committee, Red Bank, N. J., pp. 31-34. (2) S. H. Langer and R. G. Haldeman, Science, 142, 225 (1963).
