NOTES
Jan., 1962 greater than po gives linear RET plots in the region
of monolayer coverage. This implies that the second
adsorbed layer is considerably more disordcrcd than the bulk adsorbate, and that as furlher adsorbed layers are built up, thc dccay of adsorption forces originating from the solid is offset by an increasing degree of order in the adsorbate layers. AndersonL7 also found that the range of linearity of BET plots was increased by using a value of PO’ higher than
Po.
The object of plotting adsorption data according to the BET equation is to obtain repeatable values of the constant v,, for the purpose of comparing, not necessarily on an absolute basis, thc surface areas of various adsorbent samples. For this reason, it seems justifiable to adopt a value of p ~ which ‘ will give, in as many cases as possible, straight BET plots from which v,’ may be obtained. It is not suggested that the BET equation is sufficiently well-founded to enable its use as a criterion for the selection of the correct value of a physical constant such as the saturation vapor pressure of krypton. On the contrary, the empirical nature of the BET theory when applied in conjunction with a value of the saturation vapor pressure other than the true one at the adsorption temperature is regarded as justification for a wholly empirical choice of pot; this procedure seems in any case to be necessitatcd by the scatt,er in the recorded vapor pressure data. The choice of the supercooled liquid vapor pressure to represent PO’ appears to have no more theoretical validity, and less practical value, than the choice of a po’ which will give a straight BET plot. The selection of the best value of po’is best made by examination of adsorption data on surfaces giving low c-constants, since these are more sensitive to such choice. On the basis of a number of isotherms on various clay minerals, similar to that in Fig. 1, our preliminary conclusion that pot = 30 cm. at 90.1”K. gives the most linear BET plots, and, when used in conjunction with a molecular area UKr of 18.5 A.,2 consistent agreement is obtained with nitrogen monolayer volumes. However, the c-constants uscd all lie between 30 and 100, and the examination of isotherms with lower cconstants may well yield a better choice of pot. An alternative, avoiding this difficulty altogether, would be to plot krypton adsorption data according to the Harkins-Jura relative equation. The specific surface is determined from the slope of such a plot without reference to the value of p,, by application of an empirical constant derived from measurements on a solid of known area. The BET method also requires such a calibration to obtain the value of UKr; however, UKr is dependent on PO’,whilst the Harkins-Jura constant is not. Rcfcrred to a solid vapor pressure of 21 mm. a t 90.l0K., our valuc of PO’ gives 6 = 0.70, whereas Gaines’ valuelo gives 6 = 0.77. Other values of the extrapolated liquid vapor pressure quoted in Table I give d as low as 0.62. Figure 1 shows that considerable variation in urn‘ is to be expected from the use of these various values of PO’, especially on ~ * J Q
(17) R. R. Anderson, J . Am. Chem. Soo., 68, 688 (1940). (18) W. D. Harkins and G. Jura, J . Chcm. Phys., 11, 431 (1943). (19) W. D. Harkins and G. Jura, J . Am. Chem. Soc., 66, 1306 (1944).
18.5
surfaces giving low c-constants. Since UKr is evaluated by comparison of krypton monolaycr volumes with those obtained with other adsorbates, it is clear that this quantity cannot, he determined rcliably without standardizing the choicc of PO’, and the variability in the rcportcd value of qir,mcntioned by Chines and Cannon,* easily could havc arisen from this cause alone. Similarly, the reasons for the existence of curvature in BET plots cannot be discussed without reference to the value of PO’ uscd in obtaining them, except in cases where c is high enough to render the BET plot relatively insensitive to the choice of PO’. Acknowledgments.-This note is published by permission of the Research Director of English Clays Lovering Pochin & Co. Ltd. The author is indebted to Mr. P. J. Maldenfor invaluable discussions. CUBIC CADMIUM SULFIDE BY 11. AHLBURGAND It. CAINES Research Ih’rision, Rayfhcon Company, Waltham, Massachusetis Receiaed Augusl5, 1961
Cadmium sulfide crystals of large size have been grown only in the hexagonal form (a-CdS) no matter whether the crystals were grown from from the vapor p h a ~ e , or ~ , from ~ the melt under Using X-ray diffraction methods, Rohm and Niclassen* showed that CdS precipitated under certain conditions can have the cubic (p-) or zincblende structure; this result was confirmed by Ulrich and Zachariasen.9 Milligan’O investigated the structure of CdS precipitated from salt solutions a t different temperatures. Stuckertll described conversion from a- to pCdS, while Rittner and Schulmanlz reported essentially the opposite result. Trail1 and BoyleI3 found that a mineral of very small grain size hitherto assumed to be greenockite (cy-CdS) was actually P-CdS. SatoI4 reported growth of both cy- and BCdS from solution on a lead sulfide substrate. Drickamer’b assumed that a discontinuity in the optical absorption edge shift a t 27,500 atm. implies transformation to the cubic phase at this pressure. Experiments on C stal Growth.-Our attempts to repare macroscopic P-ZdS crystals will now be descrifed. Although our objective of growing such crystals was not (1) E. T. Allen, J. L. Crenshaw and H. E. Merwin, A m . J . Sn’., [1] S4.. 341 (1912). . , (2) A. KremheUer, A. K. Levine and C. Gashurov, J . EZecfrochem. SOc., 107, 12 (1980). (3) D. R. Hamilton, Brit. J. Appl. Phya., e, 104 (1958). (4) L. C. Gmne, D. C. Reynolds, 9. J. Czyzak and W. M. Baker, 3. cham. P l y & , 29, 1375 (1958). (5) A. richer, 2.Naturfweeh., lSa, 105 (1968). ( 0 ) A . Addamiano and M. Aven, J . A p p l . Ph?& 81, 38 (1960). (7) J. Medaalf and R. Fahris, J . Electrochem. Soc., 106, 719 (19.58). (8) Bohm and Niclasson, 2.anotg. u. allgem. C k m . , 182, 7 (1928). (9) F. Ulrich find W. Zaohariaeen, 2. Kn’el.. 61, 280 (1925). (10) 0. Milligan, J . Phya. Chem., 88, 797 (1934). (11) L. Stuckert. Die Glashuette, 80 011 (1935). (12) E. S. Rittner and J. R. Sahulman, J . Phys. Chem., 47, 537 (1943). (13) R. J. TmiU and R. W. Boyle, Am. Mineralogist. 40, 555 (1965). (14) R. Sato, Nature, 184, 2005 (1959). (16) A. h Edwards, T. E. Slykhouse and H. G. Drickamer, J . Phys. and Chem. Soli&, 11, 140 (1959); A. L. Edwards and H. 0. Drickamer, Phys. Rev., -1, 1148 (1961).
W.
186
NOTES
achieved, we believe the information collected may be of interest. Since it is known that a-CdS is the stable form a t temperatures above 800°, growth from the melt was ruled out. Growth from the vapor phase and many chemical transport methods furthermore cannot be used a t low enough temperatures. Cadmium sulfide exhibits very low solubilitv a t normal temperature and pressure in al1"non-reacting sblvents. I n mole/l.le pure water at 25" its solubility is 1.5 X Preparation of CdS from solution leads to high supersaturation which gives rise to spontaneous nucleation and hence minute crystallites. We tried to avoid excessive supersaturation in order to favor the growth of fewer nuclei into larger crystallites. Some of Allen and Crenshaw's' experiments using salt solutions therefore were repeated and extended. Following Milligan,lo we varied the temperature (0 to loo"), pressure (10 to 800 mm. of H n S ) ,Cd++ concentration ( I to 120 mg./ m!.), and p? (HzSO4 or HNOa, pH 2.5 to 0.5) of the nutrient solution. I n some experiments we used alkaline NazS solutions. I n others we added a wetting agent, sodium lauryl sulfate. Sulfur compounds such as thiourea (NH&CS, thioacetamide CHLXNHz, and sodium thiosulfate Na2S203were employed as sources that would release sulfur ions slowly. I n all these experiments, the particle size of the resultant precipitate did not exceed 1 or 2 ,u. I n agreement with Milligan, the best cubic X-ray pattern and the coarsest grains were obtained when CdS was precipitated by bubbling HzS through a boiling solution of CdS04 (10 g. Cd++ per liter) in dilute HeSOr (20 ml. per liter). Electron microscope examination17 showed owder prepared this way consisted of spherical particyes with an average diameter of 0.8 to 1 p ; no crystal faces were evident. On the other hand, when a Cd++ solution at 0" mas mixed with water which had been saturated with HzS near O " , the resulting fine powder consisted of roughly spherical fragments 0.15 p in diameter, but exhibiting cube corners and edges. This material thus showed better, although minute, crystals but its X-ray pattern indicated the presence of some hexagonal CdS. Hydrothermal growth n",s attempted at seed temperatures between 120 and 250 , well below the 350" where hydrothermal treatment reported by Kremheller, et a!.,2 had yielded a-CdS. Seeds of cubic ZnS, HgS,'* or mica were used. Precipitated 0-CdS powder served as nutrient. Practically no growth was observed, yet the temperature was high enough to turn the nutrient CdS into the hexagonal modification. This result ruled out the hydrothermal method. Other growth methods which proved to be unsuitable included an aqueous circulation method, Allen and Crenshaw's double tube method, dissolving CdS in molten sulfur, and ion diffusion in water. In no case was significant cubic growth observed. Bchneider'sl9 method of growth also appeared useless because of the temperature necessary. Transition Temperature .-According to Ulrich and Zachariasen,B 8-CdS is transformed into the hexagonal form between 700 and 800". If the transformation rate is high enough, one observes conversion of the thermodynamically unstable form into the one which is stable a t that temperature, but never the reverse. (See ref. 1, 3, 5 on ZnS.) Samples of both 01-20 and @-materialwere heated in vacuo from 200 to 700" in steps of approximately 100" for five days at each temperature. These runs gave evidence of the conversion of the p- to a-material, the a-modification being completely stable. Conversion occurred a t temperatures between 400 and 450°, the rate increasing with temperature. Several differential thermal analysis runs supported these findin s in general. The smallness of the effect, however, limitei interpretation of the curves to general aspects Solvents facilitated conversion. The X-ray patterns of samples of each polymorph, which had been sealed individually in tubes containing distilled water and heated (16) S.F. Ravitz, J . Phys. Chem., 40, 6 1 (1936). (17) Kindly done by Professor P. Dorain, Brandeis Cniversity, Waltham, Mass. (18) Kindly supplied by Professor C. Frondell, Mineralogy Department, Harvard University, Cambridge, Mass. (19) R. Schneider, J . pvakt. Chem., [21 8,38 (1873). (20) Luminescent grade CdS from the General Electric Co., Lamp Division, Cleveland, Ohio.
Vol, 66
for several days a t 220°, showed transformation of p- to a-CdS. At 130°, this transformation was not noticeable In another experiment, we checked the work of Stuckertll on the room temperature conversion of a-CdS to pCdS in strong acid solution (50 ml. H&04 per liter). Samples of both a- and 8-material were each immersed for one week in small stoppered tubes of this solution. No evidence of conversion in either direction could be observed from X-ray patterns of the resulting dried powders. If heated between 150 and 200", conversion of the p-CdS to a-CdS was observed. I n order to check the results of Rittner and Schulman12 cubic and hexagonal powder was immersed in a highly concentrated solution of ammonium hydrogen sulfide NHaHS, and left at room temperature for three days. Transformation from @- to a-CdS occurred but not the reverse, confirming Rittner and Schulman's statements. At 5", no transformation from 0- to a-CdS could be detected after 5 days. In agreement with Rittner and Schulman, we find the hexagonal modification of CdS is stable above 20". If the cubic modification is stable a t atmospheric pressure, it can only be below 20". Identification of the exact transition temperature is limited by the infinitesimal reaction rate at low temperature.
Acknowledgments.-We wish t o thank Dr. 0. J. Guentert, Mr. W. R. Bekebrede and Mr. R. Hawkes for X-ray measurements and interpretation, Dr. J. Van Hook and Mr. C. Snider for taking differential thermal analysis curves, and Mr. R. C. Ellis, Jr., and Dr. D. M. Warschauer for many discussions and suggestions. THE DIELECTRIC CONSTANT SND LOSS OF IRON PENTACARBONYL AT MICROWAVE FREQUENCIES W. D. HORROCKS, JR., AND E. N. DICARLO Department of Chemistry, Princeton University, Princeton, N . J . Received August 8, 1961
The structure of iron pentacarbonyl has been the subject of considerablecontroversy. Some time ago Bergmann and Engell and Graffunder and Heymann2 found dipole moment values of 0.62 and 0.81 D for Fe(CO), in benzene solution. The atom polarization was taken as zero in the above work, so these values can be considered only upper limits for a permanent dipole moment. More recently Weiss3measured the dipole moment of Fe(C0)s in benzene solution a t radiofrequencies and found a value of p = 0.63 =k 0.06 D, assuming zero atom polarization. In order to make the static dipole moment vanish it was necessary to assume an atom polarization equal to 20% of the electronic polarization, PE. For Ni(CO)44 an atom polarization of 5y0 of the P E yields a zero dipole moment. Evans and Lister6 interpret their electron diffraction data in terms of a trigonal bipyramidal (D3h) structure which would have no dipole moment. On the other hand, O'Dwyer6 analyzed his infrared data in terms of a tetragonal pyramidal (C,,) configuration. More recently (1) E. Bergmann and L. Engel, Z. physilo. Chem., ISB, 232 (1931). (2) W. Graffunder and E. Heymann, ibid.. 15B,377 (1932). (3) E. Weiss, 2. anorg. u. allgem. Chem., 287, 223 (1956). (4) L. E. Sutton, R. G. New and J. B. Bentley. J . Chem. floc., 652
(1933). ( 5 ) R. V. G. Evans and NI. L. Lister, Trona. Faraday SOC.,86, 681 (1939). (6) & F.I. O'Dwyer, J . M o l . Spectroscopy, 2, 144 (1958).