Sept., 1958
DECOMPOSITION OF MAGKESIUM HYDROXIDE
1079
as well as of dyes, in microscopic objects. excellent technical assistance. J. K. welcomes this Acknowledgments.-The authors wish to express opportunity to thank Prof. T. Caspersscm for their appreciation to Ing. J. Kudynowsky for his extending the facilities of this Institute to a guest.
APPLICABILITY OF THE KNUDSEN EFFUSION METHOD TO THE STUDY OF DECOMPOSITIOK REACTIONS. THE DECOMPOSITION OF MAGNESIUM HYDROXIDE BY ERICKAYAND N. W. GREGORY Contribution from the Department of Chemistry at the University of 'Washington, Seattle, Washington Received A p r i l 7 , 1968
Expected equilibrium water pressures over Mg(0H)z and MgO and a condensation coefficient of unity are observed in effusion experiments a t 28" only in the initial stages of the decomposition. As the reaction progresses pressures within the cell drop to immeasurably low values. At higher temperatures a constant steady state effusion pressure has been observed between 5 and 55% decomposition when the cell is kept continuously in vacuo. Under these conditions the system shows a pseudo-equilibrium behavior; Pa shows a systematic variation with cell orifice area; however, extrapolation to zero orifice area gives an apparent equilibrium pressure only a ten-thousandth that of the expected value, a condensation coefficient of 10-8, and an apparent A H 58% larger than the thermodynamic value. A possible kinetic explanation is discussed. It is concluded that the effusion method can give very misleading results when applied t o the determination of equilibrium characteristics of decomposition reactions involving substances of very low volatility.
With increasing interest in and need for characterization of chemical systems a t high temperatures, the applicability of the Knudsen effusion method' to the study of decomposition equilibria has become of considerable importance. Because of the experimental simplicity of the method, it appears ideally suited for the study of reactions of the general type, solid (I) = solid (11) gas, commonly encountered in the decomposition of halides, hydroxides, carbonates, etc. Recently a number of authors have pointed out that serious differences between steady-state pressures and true equilibrium pressures may result (even for simple vaporization processes) if the condensation coefficient is very small.2-6 This coefficient may be defined as the fraction of the molecules colliding with the surface which actually condense. Its magnitude may be estimated by observing the dependence of steady-state effusion pressures on cell geometry and their relation to the true equilibrium pressure.6 We have made a study of the effusion method as a means of determining the equilibrium characteristics of the reaction
+
Mg(OH)z(s) = MgO(s)
+ HzO(g)
(1)
This system was chosen primarily because thermodyamic properties of the substances involved are ~ e l l - k n o w n . ~Giauque ~~ and Archibald have carefully determined equilibrium vapor pressures of water in this sytem by a static manometric method in the vicinity of 200". X-Ray diffraction workg (1) M . Knudsen, A n n . P h y s i k , 28, 1002 (1909). ( 2 ) C. I. Whitman, J . Chem. Phys., 2 0 , 161 (1952). (3) S. Dushman, "Vacuum Technique," Chapter 2 , John Wiley and Sons, Inc., New York, N. Y . , 1949. (4) K. Motzfeld, THISJOURNAL, 69, 139 (1955). (5) R. Speiser and H. L. Johnston, Trans. Am. SOC.Metals, 42, 283
(1950).
(6) J. H. Stern and N. W. Gregory, THIs JOURNAL, 61, 1226 (1957). (7) W. F. Giauque and R. C. Archibald, J . A m . Chem. Soc., 69, 561 (1937). (8) W. F. Giauque, ibid., 71, 3192 (1949). (9) R. Fricke, et al., Z . anorg. Chem., 166, 244 (1927); G. F. Huttig and W . Frankestein, ibid., 186, 403 (1929); W. Bussem and F. Koberich. Z . physik. Chem., B l 7 , 310 (1932).
has shown that only the two simple solid phases, Mg(OH)2 and MgO, exist in the system. Furthermore Mg(OH)2occurs in nature in a high state of purity as brucite. Use of this material in initial work avoided much of the difficulty resulting from inability to prepare comparable samples of starting material.7~10-12Kinetic studies have shown rates of decomposition to be slow and a hysteresis loop was observed on comparing hydration and dehydration parts of the cycle.13 Hence it was concluded that this system should provide a good test of the applicability of the effusion method. Experimental Samples of Mg(0H)z were placed in various carefully calibrated effusion cells and the rate of effusion of water vapor measured under molecular flow conditions.l4 Cells were placed in a glass tube immersed in a thermostat or, a t higher temperatures, an electric furnace. Once a run was initiated the sample was maintained under high vacuum and water vapor removed continuously. The effusing water vapor was collected in a liquid air-cooled trap for measured intervals of time. To stop a run a stopcock connecting the collector to the sample was closed but evacuation continued by switching t o an alternate pumping route. All the water vapor removed was collected so as to follow the extent of decomposition of the original sample as measurements progressed. The effusate was transferred from the collecting trap to a small calibrated, thermostated volume in which its pressure was determined manometrically. As little as l o + mole of water could be measured. It has been shown that the ideal gas equation is adequate to determine the number of moles of water vapor as long as its pressure does not (10) D. R. Torgeson and T. C. Sahama, J . Am. Chem. SOC.,'70, 215G (1948). (11) C. H. Shomate and E. H. Huffman, ibid. 66, 1625 (1943). (12) K. Taylor and L. 8. Wells, Bur. Standards J . Res., 21, 133 (1938). (13) R. I. Razouk and R. Sh. Mikhail, THIS JOURNAL, 59, 636 (1955). (14) Considerable care was taken in the design of effusion cells and the trapping system to keep Clausing factors as large as poseible in accordance with well-established principles of effusion work.*,@ Details may be found in the Doctoral Thesis of Eric Kay: "The Applicability of the Knudsen Effusion Technique to the Study of Deoornposition Reactions. The Mg(OH)pMgO-HtO and NaOH-Na20-HzO Systems," University of Washington, 1958.
.
1080
ERICRAYA N D K.W. GREGORY
randomly over the range plotted. The marked dependence of P, on f (the ratio of orifice areas AOto cell cross-section areas A , ; the latter (from 2-4 cm.2) is taken as an approximation to the effective vaporizing and condensing areas of the sample6) suggests a small condensation coefficient. Assuming the equation
4.5
.
h
5.0
E'
2 5.5
v
n, M
I
Vol. 62
6.0
PolP,
=
1
+fla
(2)
6.5
derived from steady-state condition^,^^^ the apparent equilibrium pressure P e and the condensa7.0 tion coefficient CY may be calculated. Alternately, 2.0 2.1 2.2 2.3 2.4 Pe may be estimated by plotting the observed lOOO/T. steady-state pressure as a function of AO(keeping Fig. 1.-Steady-state effusion pressures above ground the cell cross-section constant) and extrapolating samples of brucite; 5-55'34 decomposition, 140-230': 0, to zero orifice area. Both methods give reasonably f = 5.35 x 10-4; a,f = 2.93 x 10-3; e , . j . = 3.91 x 10-3; good extrapolations and essentially the same rea, f = 2.02 x 10-2; - - - -, apparent equilibrium pressure. sult; CY = 1.8 X 10+ and Pe = 2.5 X atmosphere a t 463°K. No variation of CY with tempera10-6 p 1 ture is indicated. The slopes of the lines in Fig. 1 are essentially the same and give an apparent heat of reaction of 31.2 kcal. The temperature dependence of P e (apparent), dotted line Fig. l, may be represented by the equation -6820 log P = -T
Results According to characteristics determined by Giauque and Archibald, the equilibrium vapor pressure of water above Mg(OH)2 and MgO is ca. atmosphere a t room temperature. Initial effusion experiments (at 28") gave pressures of this magnitude (values were higher until absorbed water was removed) for a very limited time only; pressures then fell off to immeasurably low values even though less than 0.1% of the sample had been decomposed. T o bring the pressure back into a measurable range it was necessary to increase the temperature of the sample by ca. 150". At these higher temperatures effusion steady-state pressures were finally established (after ca. 5y0decomposition) which did not change on further continuous decomposition of the sample but which were many orders of magnitude lower than expected equilibrium values. Figure 1 shows the dependence of these steady-state pressures on temperature and on effusion cell geometry. The straight lines shown were obtained by varying the temperature
Actually, at 463°K. P e = 2.62 X lop2 atmosphere and AH = 19.78 kcal. as determined by Giauque and Archibald. Thus P, (apparent), Fig. 1, is lower than the actual equilibrium value by a factor of lo4. This surprising result mill be discussed in more detail after presenting other observations. Because of the very small net rate of decomposition in the effusion cell, it is possible to follow the change in steady-state pressure in the initial stages of the decomposition. Fig. 2 shows the dependence of pressures on time, ie., on degree of decomposition. Results shown are from two independent samples, one large sheets cleaved from native brucite crystals17 and the other a ground sample of brucite of the indicated particle size. The two set of curves correspond to different effusion rates; one cell had an orifice ca. ten times that of the other. Zero time represents the beginning of the first measurement, initiated as soon as molecular flow conditions could be established (a matter of a few minutes). The first measurements gave pressures above the expected equilibrium value, shown by the'dotted line, which we attribute to desorption of water from the sample and/or walls of the cell. Of considerable interest is the brief leveling off (of pressures) at the equilibrium pressure of Giauque and Archibald. The extent of decomposition when the steady-state pressure falls below this equilibrium value is nearly the same in both cases. Grinding the sample increased this period slightly. Figure 3 illustrates the effect of a more drastic change in particle size on the length of time equilibrium pressures were maintained; the rate of decomposition was the same as the larger value in Fig. 2. The behavior of a sample of Baker N.F.
(15) H. S. Frank, THISJOURNAL, 83, 970 (1929). (16) I. R. McHaffie and 5. Lehner, J , Chem. Soc., 137, 1569 (1925); THISJOURNAL, 31, 719 (1929).
(17) Brucite used was part of a piece of Texas Brucite obtained from Ward's Natural Science Estahishment, Inc., 3000 Ridge Rd. E., Rochester 9, New York.
10-8
I
0
I
200
400 600 Time (min.).
I
4
800
1000
Fig. a.-Steady-state effusion pressures in the initial stages of the decomposition (28'); A curves, 1 g. of large sheets of brucite; B curves, 1 g. (-20 60 mesh) of ground brucite: 8 , 0, effusion (decomposition) rate 4.36 X 10-3 P moles/second; 0,0 , effusion (decomposition) rate 4.63 X loT4P moles/second; - - - -, equilibrium pressure according to Giauque and Archibald.
+
exceed ca. 0.7 that of the saturation value.lbJ8 The quantity collected was related to the pressure in the effusion cell by the usual Knudsen equation.
.
+ 9.118
Sept., 1958
DECOMPOSITION OF ~ ~ A G N E S I UHPDROXIDU; M
Mg(OH)2 (c,heinicslly precipitated) and of rehydrated MgO (originally produced by decomposition of brucite) is shown and may be compared with those in Fig. 2. Even the rehydrated mnterial, which is probably of colloidal dimensions, only maintains the equilibrum pressure until ea. O.25y0 decomposed. The condensation coefficient for the equilibrium portion of the decomposition must be very near unity. This is demonstrated by the independence of P, (observed) on orifice area. I n Fig. 4 a comparison of steady-state pressures from the various samples of Mg(0H)z throughout the major part of their decomposition is shown. I n identical cells, P, values for the ground brucite particles are somewhat larger than pressures above the large sheets. On the other hand, the chemically precipitated Mg(OH)z, which would be expected to have a still larger total surface area, gives the smallest steady-state pressure of the three shown. However, considering the magnitude of these pressures relative to true equilibrium values, these differences are small. Changing the quantity of a given kind of sample in the effusion cell (for example from 0.6 to 1.2 g. in a cell with largest f factor) did not cause a discernible change in the steady-state pressure. By the time the constant low steady-state pressures were reached (at the higher temperatures) less than 5% of the brucite samples and 9% of the N. F. samdes were decomDosed. Figure 5 shows a
1081 1
14
4 3
L GOO
O
1200 1800 2400 Min.
Fig. 3.-Duration of equilibrium pressures for finely divided magnesium hydroxide: 0 , 1g. of rehydrated MgO; 0 , 1 g. of .Mg(OH), N. F. owder; - - - -, equilibrium pressure (Giauque and ArchiEald). 5.0
2 5.5 44
6.0 ba
6.5 7.0
t
I 2.0
2.2
2.1 1000/T.
2.3
Fig. 4-Steady-state pressures in various effusion cells for various samples of Mg(0H)t: - - - -, data shown in Fig. 1; 0 , 1 g. of large brucit,e sheets; 0 , 1 g. of N. F. powder; A, f = 3.91 X 10-3; 13, f = 2.93 X 10-3; C, = 5.35 x 10-4.
s sures suggests a slow structural change. Once t h e i p p e r value was reached, it Femained unchanged throughout decomposition of the major part of the sample. In one experiment of the sample was decomposed in a continuous series of measurements a t various (randomly chosen) temperatures with reproducible variation of P, along the lines shown in Fig. 1. I n another the sample was further dehydrated by heating briefly to high temperatures and then cooled again to ea. 200'; essentially the same P, value was observed with the sample about 80% decomposed. I n the intermediate stages of the decomposition the build up of de-
I-
A
-
@
-
2
-
9
/F
-
a, 5 -
(18) R. J. Peavler and A. W. Searcy, J . Am. Chem. Soc.. 78, 2076
r
% decomposition
70 d:;;position
155%
s"
.&
system. l8 If evacuation of the sample were discontinued, water pressure within the cell increased slowly. On renewing effusion experiments, after a short interval, the first pressure measurement was high (approaching true equilibrium) but subsequent values dropped quickly to the same constant steady-state pressure observed before closing off the system. Water take-up by the dehydrated material is very slow at room temperature. An excess of water vapor (at its saturation (liquid) vapor 1956)
I
1
I
-1
pressure) was allowed to stand three weeks over a one-gram sample of Mg(0H)z 90% decomposed; at the end of this period the composition still corresponded to 70y0 MgO. Renewal of effusion experiments with this material showed the kind of behavior indicated for rehydrated MgO on Fig. 3. Surface cooling does not appear important in the effusion experiments. Net rates of vaporization and hence energy demands are very small. No apparent change in the steady-state pressure in the equilibrium region was observed on changing the orifice area by a factor of ten. A change would surely be observed if surface cooling by evapora-
1082
Riuc KAYAND N. W. GREGORY
Vol. 62
tion were an important effect. This same change in orifice area (and corresponding rate of decomposition) has a marked effect on the steady-state pressure in later stages of the decomposition as shown in Fig. 1. This comparison offers evidence that the fall-off in the latter case is associated with the low condensation coefficient rather than n manifestation of surface cooling.lg It should be pointed out that the estimate made of the condensation coefficient from equation 2 involves considerable uncertainty. The effect of a and A , cannot be separated. The small value we have attributed to a! might instead be characteristic of A,; however this would require the effective vaporizing area to be exceedingly small relative to the total sample area; one would expect a more pronounced effect on grinding the sample than is observed. Some justification for assuming A, to be the cell cross-sectional area has been found in a study of the vaporization of iodine.6 While the two solid phase system of present concern is more complex, it has been demonstrated that the variation of P, (in the 5-55% decomposition range) with the cell-cross sectional area A,, A. held constant, is in reasonably good agreement with equation 2 . Discussion Although a definite explanation of the effusioil pressures in terms of the complex structural and diffusion effects involved on the reacting surfaces is not possible at this time, we suggest that the following features are of major importance. In the beginning stages of the decomposition only a small fraction of the magnesium hydroxide at the surface appears to be in sites most favorable for decomposition. In this part of the process the decomposition activation energy essentially corresponds to the thermodynamic heat of decomposition. Magnesium oxide formed appears to be in close contact with, perhaps essentially an integral part of, the magnesium hydroxide lattice since the recapture of water molecules with subsequent ref ormation of the hydroxide occurs readily. Here the condensation coefficient is very near unity. This is the first instance of which me are aware in which unit condensation coefficient has been clearly demonstrated for a decomposition reaction. After the magnesium hydroxide in these favorable sites has been decomposed, both the rate of decomposition and the condensation coefficient decrease markedly, as concluded from P, and its relationship t o f . If only the rate of decomposition were to change, no dependence of the new steadystate pressure on orifice area should be observed unless the vaporizing area continously changes. If A, is changing continuously, it is difficult to see how constant steady-state pressures would be observed over such a large range of per cent. decomposition. In addition P, was found relatively insensitive to the quantity of sample in the cell. If the rate of decomposition remains constant and the condensation coefficient alone decreases, steadystate pressures should increase rather than decrease.
It is somewhat surprising that pressures in Fig. 5 , showing the slow approach to the new steadystate constant pressure, are rising rather than falling. This is not a thermal equilibration problem as the sample has been a t constant temperature for nearly 100 hours before the pressure plateau is reached. Rather it appears indicative of formation of a new kind of “pseudo-equilibrium” surf ace, with different properties from that involved in the initial phases of the decomposition. This “state” of the system appears then to remain unchanged throughout the major part of the decomposition, a t least as long as water pressures and temperatures are maintained in the range shown on Figs. 1 and 4. Temperatures can be varied a t random in this range and reproducible steadystate pressures observed. On one occasion a cell was cooled briefly and removed from the vacuum system and a new lid (with different orifice area) substituted; on renewing effusion experiments the expected steady-state pressures at the new f were quickly re-established. For these reasons, as well as the correlation of P, with f already discussed, the system in this state has been said to show a pseudo-equilibrium behavior. The apparent decomposition energy, perhaps more nearly an activation energy’, is larger than the thermodynamic heat by 11.4 kcal. It is suggested by the work of othersz0Sz1 that the large cleavage surfaces, 0001, of brucite crystals do not participate in the initial stages of the decomposition. We have observed that even after nearly complete decomposition, the external form of the brucite crystal is preserved, though it crumbles when subjected to mechanical force and is shown by X-ray powder patterns to have the MgO structure. Electron micrographs of replicas of the 0001 surface (taken with a Siemen’s Elmiskop I by Dr. J. H. Luft of the Department of Anatomy, School of Medicine, University of Washington) a t various stages of decomposition up to ca. 5% failed to show development of any detail on this smooth surface with resolution down to ten Bngstroms. Hence decomposition appears to occur largely a t the edges, corresponding to the probable mode of growth of the original crystal. The rate of decomposition does not appear to be controlled by diffusion of water vapor out of changing long pores or crevices between 0001 layers, however; if this were the case, one would expect the pressure t o fall continuously as decomposing sites receded into the crystal, particularly in measurements with large sheets of brucite. Similarly, a large effect on the steady-state pressure would be expected on grinding the crystals. The low value of the condensation Coefficient and rate of vaporization is not surprising in view of these structural characteristics and, as stated earlier, was generally anticipated from previously reported kinetic data. However, it is most surprising that extrapolation of effusion steady-state pressures to zero orifice area leads to an apparent equilibrium pressure too low, but which correlates the observed pressures from all cells satis-
(19) R. Littlewood and E. Rideal, Trans. Faraday Soc., 62, 1598 (1956).
(20) D.R. Garrido, Ion, 11, 206,453 (1952). (21) 9. J. Gregg and R. I. Razouk, J . Chem. Soc., SI, 36 (1949).
Sept., 1958
CRITICAL MICELLE CONCENTRATION OF ETHERALCOHOL SULFATES
factorily. This extrapolation must in fact be inaccurate; the steady-state eff usioii pressure must rise to the true equilibrium value in the limit of zero orifice (orifices materially smaller than those used in this work are not experimentally practical). For the true equilibrium pressures to be maintained in eff usioii cells, the decomposing crystal surfaces must be able to re-form sites kinetically favorable for decomposition. Since MgO and Mg(OH)z both have negligible vapor pressures a t the temperatures of the effusion work, reorientation cannot occur by a vapor phase mechanism and does not appear to occur readily by solid state diffusion. Similar difficulties may well he encountered in other deconipositioii reactions, particularly when the solid phases involved have very low vapor pressures. It is apparent, that effusion results must
1083
be interpreted with caution in such cases and that the method is not well-suited for determination of equilibrium properties. Without the availability of the work of Giauque and Archibald in the present system, the brief leveling off a t the true equilibrium pressure might well have been ignored. It also has been demoiistrated that even though an apparently normal extrapolation of steady-state pressures to zero orifice can be made, it may not lead to the correct result. It may be noted, however, that the effusion method has been applied successfully in a study of the decomposition of lithium hydroxide. 22 This work was done with financial support of the Office of Ordnance Research, U. S. Army. (22) N. W. Gregory and R. H. Mohr, J . A m . Chew&.Sac., 77, 2142 (1955).
THE CRITICAL MICEL1.E CONCENTRATION OF ETHER ALCOHOL SULFATES, R (OC~H~),OSO~Nal BY J. E(. WEIL,R. G. BISTLINE,JR., AND -4.J. STIRTON Eastern Rqional Research Laboratory, Philadelphia 18, Pennsylziar~ia Received April 21, 1958
The critical micelle concentration (c.ni.c.) of sodium hexadecyl and octadecyl ether alcohol sulfates ( R(OC2H&OSOsNaJ 1,2,3,4) was measured by surface tension, dye titration and conductance methods. In moles per liter X 10,000 the values were 2.1, 1.2, 0.7, 0.8 in the hexadecyl series and 1.9, 0.8, 0.5, 0.4 in the octadecyl series, respectively. Although solubility as measured by the Krafft point increased with the number of ethenoxy groups, the c.m.c. decreased; apparently as a result of the combined effects of increased hydrophilicity and greater chain length.
i
=
Sulfated ether alcohols are surface active agents of increasing importance for detergent and allied uses. In a previous report3 we have shown that sodium salts of sulfated ethenoxylated tallow alcohols with low ethenoxy content are considerably more soluble than the corresponding alcohol sulfates and have similar surface active properties. It seems desirable therefore to study the effect of ethenoxylation on fundamental solution properties such as critical micelle concentration (c.m.c.) for this class of compounds. The commercial method for preparing ether alcohols, i.e., the reaction of ethylene oxide with an appropriate fatty alcohol, results in n mixture, R(OC2H4)%OH, where n is known only as an average value. This study is concerned with chemical individuals where n becomes i , or a specific integer, prepared through the Williamson synt,hesis from known purified starting materials. Experimental Ether Alcohol Sulfates.-Ether alcohols of known composition were prepared from the alkyl bromide and a glycol by a method described to us by W r i g l e ~ . ~ The pure ether alcohols were sulfated in a manner similar to that described elsewhere.6 A 15% excess of chlorosulfonic acid was added (1) Presented a t the Meeting in Miniature of the Central Pennsyl\%nia Section, American Chemical Society, University Park, Pa., March 15, 1958. (2) Eastern Utilization Research and Development Division, Agricultural Research Service, U. S. Department of Agriculture. (3) R. G. Bistline, Jr., A. J. Stirton, J. IC. Weil and E. W. Maurer, J . Am. Oil Chemists' Soc., 34, 516 (1957). (4) A. N. Wrigley, P h . D . Dissertation, Temple University, 1958. (5) J. K. Weil, A. J. Stirton and E. W. Maurer, J . Bm. 0 2 1 Chemists' soc., sa. 148 (1955).
dropwise to a stirred solution of 0.1 mole of ether alcohol in 150 ml. of chloroform cooled in an ice-bath at 10-20". When the addition was completed the reaction mixture was allowed to warm to room temperature and stirring was continued for an hour. After chilling again to lo", 100 ml. of absolute methanol was added and the solution was neutralized with 18 W sodium hydroxide. The product was crystallized and filtered from the reaction mixture, redissolved in hot absolute ethanol, insoluble inorganic salt was filtered from the hot solution and the product was recrystallized from the filtrate a t room temperature. In some instances a second recrystallization from absolute alcohol was required. Carbon, hydrogen, sulfur6 and sodium analyses of the first three members of each series were found to be within 0.3% of the theoretical value. The compound containing four ethenoxy groups was somewhat more difficult to purify. Surface Tension.-The duNouy tensiometer was wed to measure surface tension. Since little qhange was observed with temperature variation over a limited range, measurement was made a t room temperature, 25 & 1'. Appropriate corrections7 were applied to the readings to obtain surface tension in dynes per centimeter. Figure 1 is a Burface tenRion versus log C plot for the ether alcohol sulfates derived from hexadecanol and Fig. 2 is a similar plot for the ether alcohol sulfates derived from octadecanol. The solutions were allowed to age one hour prior to measurement and measurements were repeated until four identical values were obtained. Since Nutting and co-workers8 have shown that most of the change of surface tension with time, for sodium alkyl sulfates, occurs within 2 to 60 minutes, the values reported may be considered to be very nearly true equilibrium values. Dye Titration.-Pinacyanole chloride was used as described by Corrin, Klevens and H n r k i n ~ . ~Five ml. of a (6) Microanalysis for C, H and S performed by Miss Laverne Scroggins. (7) W. D. Harkins and H. F. Jordan, J . A m . Chem. Soc., 62, 1751 (1930). (8) G. C. Nutting, F. A. Long and W. D. Harkins, J . Am. Chem. S o c . , 62, 1496 (1940).