E. B. BRACKETT, T. E. BRACKETT, AND R. L. SASS
2132
results of Grossmeiner and fiIatheson,a who found that the presence of HzOz suppressed fluorescence in ice. An alternative explanation which overcomes this difficulty is that the fluorescence arises from de-excited OH radicals formed by radiation excitation of water which are themselves subsequently excited. Both the radiation emission and the H202production would then be directly related to the rate of formation of OH radicals, but there would be no direct conversion of one to the other. Excitation of OH itself would imply a relatively high concentration of these species, and therefore such a mechanisni would suggest that OH* is formed in an excited state by a reaction involving another (long-lived) species. I n addition to the above features, the band at 4600 A. requires comment. It is not possible to explain this emission in terms of any known excitation levels of water or OH, or other species likely to be present. If, however, we accept the "polaron" theory of WeissJZ2 (22) J. Weiss, Nature, 186, 761 (1960).
Vol. 67
this emission could be associated with some reaction of the hydrated electron, such as recombination with the holes. Using Smith and SymonsZ3and KuperZ4models the emission energy of the hydrated electron is calculatedZ6to correspond to 4700 i 200 A. This accords with our observations of emission a t 4600 8. It does not, however, predict absorption a t other valuesforwhich a more refined treatment is necessary. V. Acknowledgments.-D. N. S. wishes to thank the Anti-Cancer Council of Victoria (Australia) for the award of a Post-doctoral Fellowship during the tenure of which this work was done. We also thank the Gamma Sterilization Pty. Ltd., Dandenong (Victoria), for allowing access to and the use of the 500,000 curie cobalt-60 7-irradiation source in the thermoluminescence studies. (23) M. Smith and M. C. R. Symons, Trans. Faraday Soc., 64, 346 (1958). (24) G. Kuper, Phys. Rea., 122, 1007 (1961). (25) Forthcoming publication.
THE CRYSTAL STRUCTURES OF BARIUM CHLORIDE, BARIUM BROMIDE, AND BARIUM IODIDE BY ELIZABETH B. BRACKETT, THoMAi? E. BRACKETT, AND RONALD L. SASS Department oj Chemistry, W i l l i a m Marsh Rice University, Houston, Texas Received December 26, 196d The crystal structures of the orthorhombic modification of BaClZ,BaBrz, and BaIz have been determined by X-ray diffraction of powder samples. The object of the investigation has been t o obtain accurate atomic parameters necessary for the calculation of lattice energies. These parameters were refined by the method of least squares.
Introduction I n connection with an attempt to calculate the lattice energies of the alkaline earth halides, the authors became interested in the crystal structures of barium chloride, bromide, and iodide. The principle X-ray diffraction investigation of these salts was done by Doll and K1emm.l They showed that all of these salts have an orthorhombic modification stable at room temperature which is similar to that of lead chloride. They also obtained the cell constants of these crystals, but their work was not sufficiently detailed to give the positional parameters of the atoms. Other modifications of these salts exist. I n particular, barium chloride exists in a cubic form studied with electron diffraction by T'ainshtein2 and a monoclinic form as well as the orthorhombic form. Experimental Experimental Preparation and Handling of Samples.-Simple drying of the dihydrate of barium chloride in an oven a t 200" yields a mixture of the monoclinic and orthorhombic forms. The monoclinic form converts to the more stable orthorhombic modification very slowly, the transition not being completed after 1 week. I n an attempt to prepare a pure sample of the orthorhombic form, a sample of the dihydrate was heated under vacuum a t 60' overnight. This procedure yielded the cubic form which is stable in the temperature range 925-960°.8,4 The cubic form did not convert appreciably to the orthorhombic form a t room temperature, but a t 200" the conversion was complete after 2 days. Thus this procedure yielded pure crystalline samples of (1) W. Doll and n'. Klemm, Z. anorg allgem. Chem., 241, 239 (1939). (2) B. K. Vainshtein, Dokl. Akad. N a u b SSSR, 60, 1169 (1948). (3) H. Gemsky, A'eues Jahrb. Mineral., Ceol., Palaeontol., 96, 513 (1913). (4) E. Vortisch, ibad., 38, 185 (1915).
both the cubic and orthorhombic forms. An X-ray diffraction powder record of the cubicoform resulted in a measured value of a. equal to 7.324 & 0.005 A . in agreement with the value of '7.34 A. reported by Vainshtein.2 Anhydrous BaBrz was prepared readily by drying the dihydrate a t 200'. Drying a t temperatures of 120" yielded a product which is either a lower hydrate of the bromide or a different crystalline modification of the anhydrous material. Several attempts to prepare anhydrous BaIz by methods mentioned in the literaturelrj proved unsuccessful. Pure anhydrous BaI2 eventually was prepared by heating the dihydrate under vacuum to 400'. No evidence of other crystalline modifications of anhydrous BaL waa found. A product which is presumably the monohydrate can be prepared by heating the dihydrate in the presence of H I a t 500' or by heating the dihydrate under vacuum a t 150-200°.
TABLE I UNITCELLCONSTANTS OF BaClr, BaBrs AND BaL BaClr, A. BaBrr, A. BaIn, A. a b c
7.865& 0.008 4 . 7 3 1 i .004 9.421ic
,008
8.276f0.008 4.956zt ,004 9.919&
,008
8 . 9 2 2 f 0.008 5 , 3 0 4 f .004 10.695ic ,008
All of the salts under investigation are deliquescent. This fact is especially true of the iodide, for which extreme precautions must be taken t o exclude water. All preliminary handling including preparation of flat plate powder samples was done in a drybox. The sample was transferred in a desiccator to the X-ray diffractometer. The diffrartometer was completely covered by a polyethylene tent through which was flushed nitrogen gas dried by silica gel. The sample was removed from the desiccator after having been left under the tent for several hours and inserted into the goniometer. The goniometer mas quickly covered with a radiation shield fitted with a polyethylene windox (5) M. D. Taylor and
L.R. Grant, J . Am. Chern. Soc.. 77, 1507 (1966)
Oct., 1963
-
~ sin2
hlcl 101 002 011 102 200 111 201 112 210 202 103 211 013 212 301 113 203 020 004 302 104 311 121 213 022 312 122 114 220 204 303 221 400 401 222 2 14 313 123 105 410 402 411 304 015 321 115 223 205 412 024 403 322 314 124 215 413 006 031 106 224 305 131 a
2133
CRYSTaL S T R U C T U R E S O F B A R I C > f H A L I D E 8
@obsd
0.03321 ,03634 ,03826 .04287 .06286 ,06482 ,06972 .07159 ,08665 ,09304 09631 .0986Y ,10621 .lo717 .11275 ,11946 ,12520
,14276 ,14546
,15330 .17179 .17302 .17650
.18697
.19895 .20371
.21350 .21925 .23238 .24024 ,24560 ,25136 .25531
_
OBSERVED AND CALCUL.4TED
_
BaCle
6onlcd
{ ::E) ,06987 ,07168 .0 ~ 6 8 1 ,09176 09316 .OY642 .OH869 .lo620 ,10713 ,11324 ,11674 ,11971 ,12251 ,12524 .13298 .13879 142,89 .14329
.
sins U* U
24.42 35.13 35.98 23.13 U
16.71 18.45*a 10.89* 15.42 38.12 42.84 U
14.99 26.98 12.3Y 54.83 55.69 49.26 U
28.70 U
30.41 U U
31.28 U
{ :E;) . U
U
32.55 U
{ :E:} { ::E:) 10.88* ,17327
Fcalcd
FQhsd
33.51* 35.53%
146'72 .15113 ,15371 .i6014i
BaBrz
r
ain!! 0.01631 ,02678 ,03325 ,03639 ,031343 ,04286 ,04513 ,06294
TABLE I1 DIFFRACTION DATAOF BnCIz, BaBrz,
8.27* 7.97% 17.56 10.90*
.18026
U
.la049 .186Y6 .1935Y .19395 .19936 ,20356 ,20489 ,20583 ,20704
U
14.99 U
U
14.67 24.42 U U U
0.03 2.86 -28.25 -34.97 -38.52 -25.71 2.01 17.98 -15.36 9.08 -11.14 41.35 -41.87 - 3.08 14.89 28.06 - 9.29 -61.47 -52.63 47.18 1.29 28.12 0.02 33.59 - 2.17 1.69 28.95 2.84 31.35 33.24 9.46 1.32 36.32 7.23 6.85 6.32 -18.66 8.27 7.99 1.96 4.36 -14.76 2.12 9.39 -11.58 20.11 7.22 6.44 4.72 43.46 4.67 -39.32 -12.12 0.85 -26.14 -17.40 4.05 19.33 30.18 -28.32 5.26 18.24
-
-
-
-
-
-
-
,05710
,22294 ,23238 .24052 ,241136 ,24565
U
23.57 18.42 U
18.83 33.00* 30.96%
{ :E} ,25386 .25526
U
21.42
,06034 ,06495 ,07856
,08307 .Ob724
.OS913 ,09668 .lo221 .lo843 ,11325 .12096
1
.08724 ,08907 .09608) .09664) .lo225 ,10537 .lo833 .11152 ,11327 ,12096 ,12645
,13163
.13894 ,14485
,13755 ,13883 .14487
,15589 ,15665
,18115 .18388
,19348
,18093 .18388
.18719 ,19344
.19909
.21008
-
,22593 .22886
U
34.30 46.60% 23.44% 31.32 45.15 50.00 10.70* 26.43* 30.14 33.31 74.06* 4'3,49* 47.63 U
31.31 U
30.52 7.19 U
U
30.07 17.71* 11.75" 9.46* U
0.80* .16300 ,16907
,18602 .18719
11.53 17.85 24.51 29.21 16.58
18.05* 10.77*
.15971 .16300
U
18.82% 4.43* 19,42* 22,79* 15.51*
,12960
-
-
Fobsd
dca Ied
0.01472 ,02416 ,03024 ,03284 ,03471 ,03892 ,04076 ,05704
,06315 ,06495 ,07854
AND
BaIz
t20212 ,20991 .21739 .21744 ,22384 ,22612 ,22815 .22909 .23252
16.44" U **a
24.46 17.65 18.93* 19.19* 10.84 34.16* 10.07* 33.70* 23.27* U
23.73 U U U
26.99 13.44 U U
I",&lCd
0 0262 ,0281 ,029b
-
ocaird
Fobsd
Faalod
0.01267 ,0207b ,02633 0282;i ,01986 ,03380 ,07506 ,04938
U
0.71 10.20 -13.40 -1'0 a 6 -10 57 - 5 44 -11.863 . i ( j 62
0492
i
9.36 9 20 10 40 U 11
,0507 ,0540 ,0660 0678 ,0723 ,0752 ,0766 0842 ,0832 ,0877 ,0903 ,0935 ,0977
-
-
sin2
oobbd
0.19 11,65 -18.97 -22.89 -24.72 -16.42 - 6.02 36.53 -39.11 -19.67 -30.33 50.78 -54.44 -11.76 29.04 29 63 -25.11 -66.25 -57.64 55.54 5.99 29.65 0.04 36.68 8.78 4.86 15.16 - 3.57 16.76 19.67 -13.39 4.70 28.55 19.36 15.47 12.45 4.33 24.42 14.58 0.71 -15.21 2.66 8.26 -22.76 -24.33 13.71 20.79 -21.34 11.55 45.38 13.38 -43.79 -30.25 5.74 -26.39 2.07 9.21 9.61 28.91 -14.39 4.07 7.27
---
BaIz
-
sin*
,05910
[ %j\ 36.64% 3.81* 37.703 11.50*
sin2
oohsd
0.02415 ,03024 ,03287 ,03472 .03890
7
,05432 .0,5619 ,0678Y ,07177' \ .07230] ,07836 ,07662 ,084.52 ,08314 ,08797 .01061 ,09352 09719) :09775] . 10530 .I0910
1
{
I
,1133
,11438 ,113001
,1195
1 .11946j
,1247
.12466 ,13516)
,1350
,1385
.14059 .14024 ,14579 ,15033 .15103 ,15691 ,15850
,1433 ,1509 .1570
7
,1720
\
.1762 ,1809
0.18.
36.52* U U
4 79* 10.58* 5.01% 19.52% 23.04% 13.57% 33.03% 31.59 21.07% 23.50* 3.83* 38.73* 23.79* 0.38* 28.90* 21.06 U
40.25 42.51 U
.1611 .1673
47.18 64.49% 24,08* 53.04 57.84 65.30 20.90* 48.30% 39.58 43.57 118.21 74.37 73.30 25.32 35.72
,16766 ,166220 .17249 ,171461 ,17513 .18089
'.
44.70% 42.40* 22.41* 62.00% 23.34* 59.43* 56.06% 22.21 46.51
-6li.28 .-%4 7.5
-50.22 65..?iZ -tj6.94 '-18.70 45.22 38.62 43.79 118 10 72 41 69. 59 20.24 31.68 0.18 36.0ti - 8.09 12.22 2 93 6.44 3.04 11.88 19.81 9.14 22.22 35.34 20 62 22.98 3.74 42.17 25.89 0.42 -31.46 18.42 4.94 41.93 -37.46 1.63 36.62 -34.60 18.37 58.11 21.88 -54.84 -51.73 -18.04 -41 48
-
-
-
-
See text for explasation.
and containing dry MgC104. In this way the sample was kept dry for several hours. X-Ray Examination.--All diffraction data were recorded on a Norelco automa tic recording powder diffractometer equipped with a geiger counter detector. Cu K a radiation (Aal = 1.5405 A., A, = 1.5418 A.) was used throughout. The diffractometer was calibrated against the known spectrum of a-quartz (a = 4.913 A., c = 5.405 i.). The diffraction patterns for each salt were taken by continuouu scanning a t a rate of 0.5' in 28/min. from 10 to 70". I n all cases the initial portions of the diffraction patterns were re-examined in order to detect any decomposition of the salt by reaction wii h water. No evidence of such decomposition was found. The observed peaks were indexed satisfactorily on the basis of the orthorhombic cell constants given by Doll and Klemm.1 The best values of the cell constants were then chosen according to a least squares fit of the calculated values of sin2 e to the obuerved values. The resulting values of
a, b , and c together with estimated standard deviations are listed in Table I. Final observed and calculated values of sin2 e are presented in Table 11. Assuming that the cell parameters listed for these salts by Doll and Klemm are in k X units, our parameters are consistently larger by an average value of 0.477, in BaClz, 0.24y0 in BaBrz, and 0.65% in BaIz. The observed systematic absences
OM; k
+I
IzlcO; h
=
=
l(inod 2 )
l(mod 2 )
are consistent with the probable space group C2h'6-Pnma reported by Doll and Klemm or the noncentric space group C2,@Pnaal. Density measurements agree with the assignment of four molecular units per unit cell. Sets of relative intensities were obtained for each of the three
2134
E. B.
BRSCKETT,
T. E. BRACKETT, AND R. L. SASS
Vol. 67
in a table of 64 entries of equally spaced values of sin B/X from 0 to 1; linear interpolation was used. The final values of the discrepancy factor, R = [ZJIFol- ~ F , 1 ~ l / [ Z , F 0were ’ ] , 0.09, 0.09, and 0.07 for BaC12, BaBra, and Ba12, respectively. The final observed and calculated values of the structure factors are given in Table 11. The observed values of IF, for unresolved reflections were obtained from the observed intensity of the unresolved peak as follows. The intensities of the various component reflections of the unresolved peak were calculated from the least squares final parameters. The observed total intensity was then divided into parts proportioned to the calculated values of the individual reflection intensities. The resulting values were then reduced in the usual way t o “observed” IFI values. These values appear in Table I1 followed by an asterisk. Observations not distinguishable from background are presented as “u” in Table 11. A double asterisk indicates reflections for which an accurate measurement of intensity could not be made. The final values of the atomic parameters obtained from the least squares calculation appear in Table 111.
TABLE I11 ATOMICPARAMETERS OF BaC12, BaBrz, A N D BaI: BaC12
BaBrp
BaIa
0.2514 0.2500 0.1209 0.87
0.2447 0.2500 0.1149 3.37
0.2366 0.2500 0.1213 4.33
0.1504 0.2500 0.4130 1.14
0.1422 0.2300 0.4272 0.70
0.1393 0.2500 0.4265 0.94
0 0290 0.2300 0.8392 0.98
0.0284 0 2500 0 8401 0.92
0.0290 0 2500 0 8387 0.96
Ba X
Y
Fig. 1.-Stereographic projection of the nearest neighbor environment of the barium ion in orthorhombic BaC12. salts by taking the height of the peak times its half width as an estimate of the peak area. Unresolved peaks with more than one maximum were treated as sums of gaussian curves. For each salt, data were recorded for a t least two different powder samples. The resulting intensities were examined for differences due to impurities, orientation, or other effects. In no case did individual intensities differ by more than the amount expected from statistical deviations. In order to determine the various atomic parameters, the structures were assumed to be of the PbClz type with atoms occupying the fourfold special positions 4c of space group Dzh“-Pnma. The PbCl2 atomic parameters given by Braekken6 were checked by collecting the single crystal diffraction data for PbClz and synthesizing the P(uw) Patterson projection. All atoms could be clearly placed and the resulting structure differed only slightly from that of Braekken. A refinement of the PbCL structure is now in progress and will be published later. These parameters, modified somewhat t o more closely fit the data of each salt, then were used as trial parameters for the barium and halogen ions. A refinement of these parameters was carried out by the (AF)2. method of least squares minimizing the function hkl
For each salt only those reflections which were completely resolved were used in the refinement. These intensities were reduced in the usual way to a set of relative FI values. The number of reflections included in the refinement of BaCL, BaBr2, and BaIz were 27, 25, and 20, respectively. The least squares calculation was carried out on the Rice computer using a program written for that machine. The diagonal approximation of the normal equation matrix was employed. The weighting scheme used was
Individual isotropic temperature factors n-ere refined for all atoms. Generally, seven cycles of refinement were first run varying only the positional parameters. -4fter partial convergence had been reached, the temperature parameters were also allowed to vary. The refinement was terminated when the indicated shifts in the positional parameters were less than 5/100,000 of the corresponding unit cell constant. The atomic form factors used were obtained from Tables 3.3.1-4 and 3.3.1B of the International Tables for X-Ray crystallography, Vol. 111.’ The form factors were modified by the real part of the dispersion correction found in Table 3.3.2B of the International Tables for X-Ray Crystallography, Vol. 111. The form factors were stored (6) H. Braekken, 2. Krist., 83,222 (1932). (7) “International Tables for X-Ray Crystallography,” Vol. 111, The Kynoch Press, Birmineham. 1962.
z
B(A.2) XI X
Y z
B( A . 2 ) X
Y 2
B( A. 1 2 )
Results and Discussion The nearest neighbor environment of the barium ion is essentially the same in each of these three halides. There are nine halogen ions surrounding each barium ion, three on the same crystallographic mirror plane and three each on the two equivalent mirror planes above and below. Figure 1is a stereographic projection of this arrangement for BaC12. The projection axis was chosen perpendicular to the mirror plane containing the barium ion. One can see that the configuration is a slight angular distortion from Dshsymmetry. The various Ba-X and X-X nearest neighbor distances are given in Table IT’. The quoted errors are estimated standard deviations calculated from the least squares normal equation matrix. These values are perhaps a little low in the case of those distances involving the barium ion. Because the barium x parameter is rather close to one-fourth the unit cell constant, many reflections are rather insensitive to this parameter, thus making the barium position underdetermined. This fact is also evident in the abnormally large values obtained for the barium ion thermal vibration parameters since this parameter interacts with the positional parameter in our calculation. Comparing the niiie Ba-C1 distances reported in Table IV with the value of 3.16 A. given by Paulings as the sum of the ionic radii, six are quite ti-ithin experimental error as reported. The in-plane Ba-C1 distance (8) L. Pauling, “The Nature of the Chemical Bond,” Cornell University Press, Ithaca, N. Y . >1960, D. 514
Oct., 1963
REACTION BETWEEN OXYGENAND EVAPORATED FII.MSOF SODIUM
of 2.86 8.is probably not significantly shorter than this value but the two equivalent out-of-plane separations of 3.58 8.are significantly larger. I n the case of the cubic form of BaC12, the barium ion coordination sphere is eightfold with each Ba-C1 distance equivalent and equal to 3.17 8. in length. Although the average Ba-C1 distance in the orthorhombic form is somewhat larger than that in the cubic modification, the coordination number is higher and the structure is more closely packed. This fact is evident by comparing the molecular volume of 98.22 A.3/molecule in the cubic form to 87.64 8.3/molecule in the orthorhombic form. The environment around the chloride ions would require an average of four and one-half nearest barium ions to satisfy the stoichiometry of the compound. There are two crystallographically different chloride ions. That which is labeled in Table I11 as XI is surrounded by four barium ions in a distorted tetrahedral configuration a,t distances comparable to the combined ionic radii. The chloride ion labeled Xz is surrounded by five near ba,rium ions, two of which are at distances significantly aonger than the sum of the ionic radii, namely 3.58 A. Both types of chloride ions have seven nearest chloride ion neighbors at distances of less than 4.00 8. The distances observed range from 3.53 to 3.87 8.as presented in Table IV. These values are comparable to the accepted value of the sum of the ionic radii, namely 3.62 A. The packing in both BaBrz and Barz is quite similar
2135
to that of BaCL The Ba-X and X-X distances observed in these salts are also presented in Table IV. TABLE IV IXTERATOMIC DISTANCES IN BaClz, BaBrz, AXD Barl No. of equiv. Interaction distances
BaC12,
8.
BaBra.
BaIz, 8.
8.
B a-X 1 Insameplane 1 1 2 Outofplane 2 2
2.86f0.08 3 . 1 5 f .08 3 . 1 8 f .08 3 . 1 7 f .08 3 , 2 5 2 ~ .08 3 . 5 8 & .OS
3.21f0.04 3 . 3 2 f .04 3 . 2 6 f .04 3 . 1 9 f .04 3 . 3 8 & .04 3 . 7 3 f .04
3.38f0.04 3 . 6 3 f .04 3 . 5 5 % .04 3 . 5 5 f .04 3 . 5 8 5 .04 4 . 1 0 f .04
Xl-XI utofplane
2
3 . 7 3 & .10
3.71f
.04
3.96f
.04
2
3.87f
4.05f
.04
4.38f
.04
2 2 1
3 . 6 4 5 .10 3.53-1: .10 3 . 7 8 & .10
3 . 8 9 f .04 3 . 7 8 f .04 3 . 9 4 & .04
4.16f 4.08f 4.29f
.04 .04 .04
x2-x2
Outofplane
.10
xrx2 Outofplane In plane
Acknowledgments.-This work was supported by a grant from the National Aeronautics and Space Administration. The Rice Computer is supported by grant No. AT-(40-1)-1825 from the Atomic Energy Commission.
THE REACTION BETWEEN OXYGEN AND EVAPORATED FILNIS OF SODIUM BY J. R. ANDERSON AND N. J. CLARK Chemistry Department, University of Melbourne, Parkville N.2, Victoria, Australia Received January 7 , 1968 The kinetics of the reaction between oxygen and evaporated films of sodium have been studied a t 90, 195, and 273°K. in the pressure range 10-3 t o 10-1mm. Provided an adequate standard of purity was achieved for the reactants and provided the oxygen pressure above the film wm sufficiently high, a protective oxide layer was produced and a t 90 and 195°K. the results obeyed a logarithmic inverse rate law. At 273'K. no single rate law gave a satisfactory fit over the entire uptake range but a cubic rate law was found t o hold after an oxide layer of sufficient thickness had been formed. Data, including those from an electron diffraction examination, indicated that under these conditions the oxide layer corresponded to sodium superoxide, NaOt. The surface potential, measured in the initial stages of oxidation by the diode method, was positive. The slow addition of oxygen to the system which inaintained an oxygen pressure above the film not greater than about 10-3 mm. resulted in an augmented total oxygen uptake. Impure reactants resulted in extensive and irreproducible oxidation with no protective behavior and the oxide formed corresponded approximately to sodium monoxide, XasO. The mechanism of oxidation is discussed.
Introduction Although mainy kine tic studies have been made on the rate of oxidation of a wide variety of metals, little attention has been paid to the reaction between oxygen and sodium. Esome quantitative data for this system have been presented by Cathcart, Hall, and Smith,l but very high oxygen pressures were used, so that little information was obtained about the initial stages of the reaction and, fnrthermore, the structure of the oxide product was not determined. For the present study, this system was chosen because, by the absence of any cation other than Nai- in the reaction product, a less ambiguous interpretation of the mechanism was expected than may be possible for those metals in which the oxidation product contains variable valency cations. (1) J. V. Cathcart, L. L. Hall, and G . P. Smith, Acta Met., 5, 245 (1957).
Experimental A simple constant volume apparatus, Fig. I , was used for the determination of the kinetics. Normal high-vacuum techniques with Apiezon-greased taps were used. With the reaction vessel and protecting trap baked to 620°K. and then cooled, the pressure prior to the deposition of a film was -lo-? nim. The reaction vessel (RV) was made from a 250-ml. round-bottom flask. The filament ( F ) and the Pyrex glass thimble (C) a t the base of the vessel formed an electrolytic cell for the preparation of sodium using a sodium nitrite melt maintained a t 750-780°K. A detailed description of this technique is given by Strong2 and the modification3 for the use of an all-Pyrex apparatus was adopted. The electrolysis current was determined almost entirely by the filament temperature, and, for 0.1-mm. diameter tungsten wire operated at 2750'K. (1.8 amp.), an electrolysis current of about 50 ma. was achieved. Faraday's laws are obeyed4 so that a (2) J. Strong, "Modern Physical Laboratory Practice," Blackie and Sons Ltd., London, 1948. (3) E. W. Pike, Rev. Sei. Instr., 4, 687 (1933).