Solubility of Ethylene in Liquid, Low-Density Polyethylene at Industrial-Separation Pressures Dennis P. Maloney and John M. Prausnitz. Chemical Engineering Department, University of California, Berkeley, California 94720
High-pressure, gas-liquid chromatography is used to measure Henry's constants and infinite-dilution, partialmolar volumes of ethylene in liquid polyethylene. Combined with previous measurements, these high-pressure data yield Henry's constants from 130 to 300°C to 600 atm. When the experimental data are combined with Flory-Huggins solution theory, solubilities of ethylene can be estimated to 200 atm. At still higher pressures, a previously published free-volume theory is more reliable. The results presented here are useful for the design of separation equipment in the high-pressure polyethylene process.
Introduction Most polyethylene is manufactured by the high-pressure process a t pressures of 2000 atm or more. Since conversion of ethylene is never complete, separation equipment is required to separate polyethylene product from unreacted ethylene. Rational design of such equipment requires information on the equilibrium solubility of ethylene in liquid polyethylene at separator conditions. Since direct experimental data are not available, and since they are difficult to measure, thermodynamic calculations coupled with chromatographic data are useful for estimating these solubilities. Thermodynamic Framework Consider a binary mixture of ethylene (1)and polyethylene ( 2 ) , at temperature T and pressure P,consisting of a liquid phase and a vapor phase. The equation of equilibrium is
@IYiP =
Qi*1~.1Hi
(1)
where, in the vapor phase, y is the mole fraction and @ is the fugacity coefficient and where, in the liquid phase, L(: is the weight fraction, R* is the activity coefficient, and H is the weight-fraction Henry's constant for ethylene in polyethylene. Superscript * indicates that the activity coefficient is normalized according to the unsymmetric convention lim lt
+0
R1*
=1
-
(2)
At pressures well below critical, y 1 1, and $1 can be readily calculated from the thermodynamic properties of pure ethylene given by Bender and von Koch (1955). Fugacity coefficients used here are shown in Figure 1. Henry's constant is independent of composition but depends on temperature and pressure. The effect of pressure is given by (3)
where R is the gas constant and 01" is the partial-molar volume of ethylene in polyethylene a t infinite dilution. Superscript PO denotes zero pressure. In a previous publication (Maloney and Prausnitz, 19751, experimental results for H1(pO)are reported; these were obtained using ordinary gas-liquid chromatography. In this work, Henry's constants are measured using high-pressure gas-liquid chromatography yielding 81". At high pressures, 216
Ind. Eng. Chem., Process Des. Dev., Vol. 1 5 , No. 1 , 1976
the exponential term in eq 3, the Poynting correction, becomes extremely important.
Experimental Section A high-pressure, gas-liquid chromatograph was constructed to measure the solubility of ethylene in liquid polyethylene. The apparatus is shown in Figure 2. A polyethylene-coated support is packed into a stainless steel tube, coiled, and placed in the column oven heated with air. The temperature of the air bath is controlled by a Hallikainen Thermotrol; temperatures and temperature gradients are measured with copper-constantan thermocouples. A thermal conductivity detector is used to measure conoentrations of ethylene and nitrogen in the helium gas stream. After detection, the carrier gas stream is saturated with water and its flowrate is measured by a bubble flowmeter. High-pressure helium gas is generated as follows. The exhaust and throttle valves are closed and the high-pressure bomb is surrounded with liquid nitrogen to lower its temperature to near 80 K. Helium is compressed into the high-pressure bomb until the column pressure gauge reaches 200 atm. The filling valve is then closed and the bomb is allowed to warm to room temperature. The pressure in this constant-volume system thereby increases by a factor of 3 or more. The filling process takes about 4 hr, while the warmup to room temperature requires about 10 hr. Leaks were checked by observing if the system could maintain high pressure for 24 hr. If any significant leaks were detected, the experiment was terminated. If high pressure was maintained, the throttle valve was opened slightly to allow carrier gas to flow through the detector, gas saturator, and bubble flowmeter. Flowrates a t the bubble flowmeter were between 1 and 6 ml/sec. In a typical experiment, ethylene and nitrogen were alternately injected every half hour. Fifteen minutes after each injection, the injection valve was quickly opened and closed t o allow a small amount of carrier gas to purge the injection system of any residual injection gas. Further purging was accomplished by twice filling and exhausting the injection system with helium. The gas to be injected was allowed to enter the injection system where it was compressed to a pressure slightly above column pressure. The filling valve was closed and a quick opening of the injection valve allowed a small sample of nitrogen (or ethylene) to enter the carrier-gas stream. Simultaneous with injection, a voltage signal was sent to the recorder to mark the time of injection. The time from injection to elution of the concentra-
Table I. Polyethylene Properties ~~
c z w -
-
0
-
u . -
Designation
MNQ
CH,/ 1000 Melt &VI MZI C i'n- DenM N Mwa ~ atoms dexb sityc
Supplier
~~
PE1408.5 16600 5.0 3.3 25.5 27 0.924 Gulf a MN,Mw, and M Z are number-, weight-, z-average molecular weights. b ASTM D1238-65T, g/10 min. C ASTM D1505-63T, glee. 0
IO00
500
PRESSURE,
I500
ATM
Table 11. Chromatographic Columns
Figure 1. Fugacity coefficient for pure ethylene. LUMN
PRESSURE
DETECTOR
Column
Polymer mass, g
I I1
1.015 1.685
Antioxidant Chromosorb Percent mass, g mass, g Polymer 0.0014
5.025
0.0026
5.217
16.8 24.4
In this work, t , is estimated by measuring nitrogen's retention time (tNz)and solving eq 4 for t , in terms of t ~ ~
FLOW-
ER
CARRIER GAS I HELIUM)
He
N2
Henry's constant for nitrogen at zero pressure has been previously estimated by Maloney and Prausnitz (1975)
ZH4
Figure 2. High-pressure gas-liquid chromatographic apparatus, tion maximum of the injected species was recorded as the retention time. Materials and Column Preparation. All gases were used as supplied; they had advertised purities of a t least 99.5 mol %. The polyethylene is a branched, low-density sample typical of those polyethylenes made by the highpressure process. Its properties are summarized in Table I. Columns were prepared by dissolving weighed amounts of polyethylene and Irganox 1010 antioxidant and thermal stabilizer into m-xylene at 130OC. Preheated Chromosorb W (acid-washed, DMCS treated, 60/80 mesh) was stirred in. Slow solvent evaporation precipitated the polymer onto the Chromosorb. The packing was dried in a vacuum oven, then packed into a 15-ft length of %2-in. i.d. stainless steel tubing. Two columns were prepared and their loadings are reported in Table 11.
Data Reduction The gas-liquid chromatography experiment uses an insoluble carrier gas, helium, flowing through a column packed with a polymer-coated, inert support. A solute sample is injected into the carrier-gas stream and the time from injection to the detection of its concentration peak maximum is recorded. This retention time (ti) can be separated into two parts: the time the solute spends in the gas phase (t,) and the time the solute spends in the liquid phase (ti t,). Assuming that fugacities of solutes in the liquid phase are proportional to their weight fractions, and that partitioning of solute molecules between liquid and gas is only determined by the thermodynamic equilibrium between the bulk fluids, the chromatographic data yield Henry's constant (4) where mz is the polymer mass, di is the solute's fugacity coefficient in the helium carrier gas, P is column pressure, Mi is the solute's molecular weight, and ri is the carrier gas molar flowrate.
In H ~ J (atm) ~ ( ~= ~7.49 )
+ 666 T
where T is in kelvins. Nitrogen's infinite-dilution, partialmolar volume was estimated by fitting data of nitrogen's solubility in polyethylene (Lundberg and Mooney, 1969) with eq 7
Fugacity coefficients for nitrogen were calculated from the tabulations of Din (1961). Each isotherm was well fit with a straight line when In (+N$/WNJ was plotted against pressure. Thus a constant ON^" could be calculated from the slope of each line. These values were then fit to an equation linear in temperature D N (cc/g-mol) ~ ~ = 16.7
+ 0.1T
(8)
where T is in kelvins. Since Q N ~ *is assumed to be unity in the derivation of eq 7 from equations similar to eq 1 and 3, the calculated D N are ~ slightly ~ lower than the true values. However, they are only needed as correction terms; therefore, D N ~ " need not be known precisely. Fugacity coefficients needed in eq 4 and 5 were calculated from the modified Redlich-Kwong equation as given by Chueh and Prausnitz (1967). These fugacity coefficients are for nitrogen and ethylene infinitely dilute in helium. Binary parameters, kij, are 0.16 for helium-nitrogen, as listed by Chueh and Prausnitz (1967), and 0.40 for heliumethylene, as reported by Hiza and Duncan (1970). Knowing ambient pressure, Pa,and ambient temperature, T,, the molar flowrate of the carrier gas is easily calculated with the ideal-gas law from measurements of volumetric flowrates, V,, with the bubble flowmeter
The presence of water vapor in the gas flowing through the bubble flowmeter is accounted for by P H ~ Othe , vapor pressure of water a t T,. Since no flow controllers were used, the column pressure and the carrier-gas flowrate, hence retention times, Ind. Eng. Chern., Process Des. Dev., Vol. 15, No. 1, 1976
217
.
40001
I
I
I
1
Table 111. Partial Molar Volumes at Infinite Dilution. Comparison of Lyckman Correlation with Experiment Infinite-dilution, partial-molar volume (cclg-mol)
I 4 c
c
4 z
138°C
300°C
45 79
54 110
58
74
65
95
Y)
z
Ethylene, experimental Ethylene, Lyckman correlation Nitrogen, eq 8 Nitrogen, Lyckman correlation
0 Y)
E
z
I Y
OBTAINED INDEPENDENTLY FROM EXTENSIVE LOW-PRESSURE CHROMATOGRAPHIC MEASUREMENTS
IO00
0
200
400
600
800
PRESSURE, ATM
Figure 3. Effect of pressure on Henry's constant. Experimental data from high-pressure,gas-liquid chromatography. changed during the course of an experiment. This change is due mainly to the depletion of helium from the 500-ml high-pressure bomb. However, these changes were gradual. Column pressure, carrier-gas molar flowrate, and ethylene and nitrogen retention times were plotted vs. time so that interpolations could be made for a hypothetical simultaneous injection of nitrogen and ethylene. The small retention-time difference between ethylene and nitrogen, on the order of 1 or 2 min, precluded a simultaneous injection into the column. Column pressures were taken at the mean time between ethylene's injection and elution, while the carrier gas molar flowrate was taken a t the mean time between nitrogen's and ethylene's elutions. Results Equations 4, 5, 6, 8, and 9 were used to reduce retentiontime data a t 138OC. Figure 3 shows the results of these measurements with Henry's constant plotted on a logarithmic scale. The data define a straight line which passes through the zero-pressure Henry's constant previously calculated by Maloney and Prausnitz (1975) from extensive low-pressure, gas-liquid chromatographic data. This implies (see eq 3) that the infinite-dilution, partial-molar volume of ethylene in polyethylene, D i m , is not a function of pressure. Since the elution peaks did not show any tailing and the results from columns I and I1 are superimposable, there is no reason to suspect extraneous adsorption phenomena. The results are also independent of injection size and carrier gas flowrate in the ranges investigated. The scatter in the data is primarily due to random errors in measuring retention times. From the slope of the straight line in Figure 3, D1" (138OC) = 45.5 cc/g-mol. Data a t 3OOOC gave ul" (3OOOC) = 54.2 cc/g-mol. These values are compared in Table I11 to calculations made with the correlation of Lyckman et al. (1965). Comparisons for nitrogen are also shown. Solubility parameters (square root of cohesive energy density) needed in the Lyckman correlation are from Maloney and Prausnitz (1974). The experimental values are always lower than those calculated from the correlation. For nitrogen these discrepancies could be due to neglecting Q N ~ *in eq 7. For ethylene, the differences are outside the range of estimated experimental errors of about 10 cc/g-mol at 138OC and 20 cc/g-mol a t 300OC. Only data for gases in low molecular-weight solvents, such as water, heptane, benzene, and carbon tetrachloride, were available to Lyckman et al. (1965). For similar-sized solvents and solutes Lyckman's correlation works well. However, the density of polyethylene is larger than that of 218
Ind. Eng. Chem., Process Des. Dev., Vol. 15, No. 1, 1976
Table IV. Maximum Errors Expected in Ethylene's Henry's Constants Calculated from Eq 3,10,and 11 Maximum expected errors, % Conditions
138°C
300°C
Zero pressure 600 atm
6 15
13 33
its low molecular-weight homologs; therefore, it has a lower free volume. Modern solution theory (e.g., Patterson, 1969) shows that free-volume differences cannot be ignored in calculating mixture properties. Since free-volume differences are extreme for ethylene and polyethylene, it is not surprising that Lyckman's correlation gives only a rough estimate, larger than that which would be obtained if freevolume effects had been taken into account. For interpolations and slight extrapolations, the infinitedilution, partial-molar volume of ethylene is D1"
= 23.35
+ 0.0538T
(10)
where 81" is in cc/g-mol and T is in kelvins. The zero-pressure Henry's constant for ethylene in polyethylene (Maloney and Prausnitz, 1975), in the range 124 to 3OO0C, is given by the empirical equation In
148300
= 8.173 - T2
where H I ( ~ Ois) in atmospheres and T is in kelvins. Equations 3, 10, and 11 allow calculation of ethylene's Henry's constant in liquid polyethylene from 130 to 3OOOC for pressures to 600 atm. Table IV lists the maximum errors to be expected when calculating Henry's constants from eq 3, 10, and 11. The zero-pressure errors have been previously reported by Maloney and Prausnitz (1975), and the errors at 600 atm were calculated with an error analysis of the terms appearing in eq 4. The most important error source is the uncertainty in Henry's constant for nitrogen which affects the estimate oft, through eq 5. Ethylene Solubility i n Liquid Polyethylene To calculate the solubility of ethylene in polyethylene with eq 1, an estimate of Q l * must be made. The FloryHuggins equation for a solute (1) in a polymer (2) can be written
Equation 12 relates the solute's fugacity in a liquid solution, f l , to its weight-fraction Henry's constant, Hi, through its volume fraction, 91, the polymer's volume frac-
tion, a2, and an interaction parameter tion is defined by
a.
x. The volume frac-
(UiWilMi)
1 -
(vlUi/Mi)
+ (~2~2/M2)
200,
(13)
~
l
=*In a1 w1
- In
- x[1 - @z2]
(14)
I
I
1
A
002
004 0.06 008 WEIGHT F R A C T I O N E T H Y L E N E
0
0 IO
Figure 4. Recommended values for solubility of ethylene in polyethylene. Calculations based on Henry’s constants and FloryHuggins theory with x = 1.0.
The quantity [ ( V ~ / M I ) / ( V ~ / M is ~estimated )] from the group-contribution method of Bondi (1964) as the ratio of van der Waals volumes of an olefinic CH? ZrouD to a Daraffinic CH2 group. This estimate gives [ ( ~ - 1 ~ M 1 ) / ( ~ 2 =/ ~ 2 ) ] 1.167. Quantitative estimates of x are made as follows. The pure-liquid fugacity is related to Henry’s constant and x through eq 15.
The concept of a hypothetical pure liquid ethylene is used to calculate x from eq 15. Estimating f 1 ° from the reference fugacities of Lyckman et al. (1965) and Henry’s constant from eq 11 gives x(130°C, 0 atm) = 1.63. As temperature rises, x decreases; x(3OO0C, 0 atm) = 1.46. However, the concept of pure “liquid” ethylene becomes untenable a t temperatures far above the solute’s critical temperature. An estimate of x at higher pressure is made by utilizing the data of Swelheim et al. (1965). These data indicate that a t 13OOC and 1374 atm, ethylene has a weight fraction of 0.70 in the polymer-rich phase. If the assumption is made that the coexisting phase is pure ethylene, then x can be calculated from eq 1, 3, 10, 11, and 14 with pure ethylene’s fugacity calculated from the equations of Benzler and von Koch (1955) or from Figure 1. This calculation gives x(13OoC, 1374 atm) = 0.53. Current polymer solution theories (Patterson, 1969; Flory, 1970) suggest that, in general, x may vary with temperature, pressure, and composition. However, for our purposes here we use x = 1.0 over a restricted range of conditions. Uncertainties in the thermodynamic analysis do not allow a more accurate estimate. Using x = 1.0 with eq 1, 3, 10, 11, and 14, the solubility of ethylenein liquid polyethylene is calculated from 150 to 300°C to 200 atm. Figure 4 shows these calculations. Ethylene’s fugacity in the vapor phase is calculated from the equations of Benzler and von Koch (1955) with the assumption that the ethylene-rich phase is pure ethylene. This is a reasonable assumption for pressures less than 200 atm. The curves presented in Figure 4 represent the best estimates which can be made a t this time for the equilibrium weight fraction of ethylene in liquid polyethylene a t pressures below 200 atm. At higher pressures, errors in measured Henry’s constants and use of an incorrect x can result in errors of more than 50%. This is illustrated in Figure 5 which shows the results of three calculations a t 130OC. The dashed lines show results using eq 1, 3, 10, 11, and 14 for the two extreme values of x. The two values yield similar solubilities at pressures below 100 atm, but begin to diverge greatly above 200 atm. The curve for x = 1.63 predicts total miscibility at pressures near 250 atm, in disagreement with experimental data of Ehrlich (1965), Swelheim et al. (1965), and Steiner and Horl6 (1972). This high value of x,therefore, cannot be used at high pressure. It may appear
I
150’C.
with vi, wi, and Mi being the molar volume, weight fraction, and molecular weight of component i. Since the molar volume of polyethylene is much larger than that of ethylene, (v1/u2) can be neglected in eq 12. Since f l = Q1*w1H1, eq 1 2 gives In
I
1 - 006
I3OoC
0
0.1
0.2
0.3
0.4
WEIGHT F R A C T I O N E T H Y L E N E
Figure 5. Comparison of three methods of calculation. Solubility of ethylene in polyethylene from free-volume theory and from Henry’s constants with two different values of Flory-Huggins parameter.
\
I00
150
200 TEMPERATURE
2 50
3000
300
‘C
Figure 6. Effect of temperature on critical pressure calculated from free-volume theory for monodisperse polyethylene.
strange that in Figure 5 the curve with the highest x shows the largest ethylene solubility. The higher solubility for x = 1.63 is due to the lower estimate of f l o , the fugacity of pure hypothetical liquid ethylene at system temperature and pressure. Since HI(“) is determined from eq 11, eq 15, with x = 0.53, gives f1° (zero pressure) = 260 atm, while x = 1.63 gives f1° (zero pressure) = 88 atm. As expected, highest ethylene solubility is predicted for the curve with the lowest f lo. The solid curve in Figure 5 is calculated from a previously published free-volume theory (Bonner et al., 1974) which attempts to take into account dissimilarities in pure-component free volumes as well as dissimilarities in potential energies. Parameters needed in this theory are evaluated from pure-component pressure-volume-temperature data Ind. Eng. Chem., Process Des. Dev., Vol. 15, No. 1, 1976
219
1000
I
The temperature inversion predicted by free-volume theory has not been previously reported in the literature. I t is remarkable that a simple, one-fluid, free-volume theory predicts such an inversion although no low-density data are incorporated into its parameters.
800
2 W
600 v)
w @=. a
400
200
0
0.2
04
0.6
WEIGHT F R A C T I O N ETHYLENE
Figure 7. Effect of temperature on solubility of ethylene in polyethylene at higher pressures. Calculations from free-volume theory using Ehrlich's critical data.
and from critical-pressure measurements of Ehrlich (1965) for a high-molecular-weight polyethylene. These data, shown in Figure 6, are reduced with the assumption that Ehrlich's polyethylene sample had an infinite molecular weight. Little error is introduced by this assumption since critical pressures are not strong functions of molecular weight for high-molecular-weight polymers. This is shown in Figure 6 where critical pressures, predicted by the freevolume theory, are given for several monodisperse-polymer molecular weights. Above the curve for M = a, the freevolume theory predicts total miscibility at all ethylenepolyethylene compositions. The vast majority of experimental cloud-point data do not contradict this conclusion, although a few cloud-point pressures of Steiner and Horlb (1972) may be higher. Accurate reading of their figures is difficult, but certainly no cloud-point pressures exceed the by more critical pressures shown in Figure 6 for M = than 10%. Since the binary adjustable parameter is fixed from high-pressure data and since theoretical analysis suggests that the free-volume theory is more applicable to high-density solutions, the free-volume theory should be more accurate a t higher pressures. Figure 5 shows that the free-volume theory underestimates solubilities at pressures below 150 atm. This is undoubtedly due to the free-volume theory's poor representation of the equation of state for lowpressure gases. However, a t pressures above 200 atm, the free-volume theory falls between the dashed curves calculated from the two extreme x estimates. Above 200 atm, the free-volume theory provides the best method now available for calculating ethylene solubility in liquid polyethylene. The success of the free-volume theory is illustrated in Figure 7 which shows the solubility of ethylene as a function of pressure for four isotherms. At high pressure, say 800 atm, the free-volume theory predicts that solubility of ethylene increases with rising temperature. This is confirmed by the data of Ehrlich (1965) shown in Figure 6; for example, a t 1500 atm and 15OoC, certain compositions form two phases, but as temperature increases to 2OO0C, the one-phase region is attained where ethylene and polyethylene are miscible in all proportions. As shown in Figure 7, the free-volume theory predicts a temperature inversion near 500 atm, and below this pressure solubility of ethylene in polyethylene decreases with rising temperature. This trend is experimentally verified by the zero-pressure Henry's constants summarized by eq 11 and it is also shown in Figure 4. 220
Ind. Eng. Chem., Process Des. Dev., Vol. 15, No. 1, 1976
Discussion The ethylene-polyethylene phase equilibrium diagram can be divided into three distinct pressure regions. Above 1000 atm, the cloud-point data of Ehrlich (1965), Swelheim et al. (1965) (with analysis by Koningsveld et al., 1966),and Steiner and Horlb (1972) provide estimates of phase stability, Le., existence of one or two fluid phases. Information of this type is needed to ensure operation of reactors in the one-phase region. Kinetic modeling and reactor control are easier in the one-phase region and precipitation of a highly viscous polyethylene-rich phase causes poor heat transfer from the exothermic reaction, possibly resulting in detonation. At low pressures, from atmospheric to 200 atm, the curves presented in Figure 4 provide accurate estimates of the solubility of ethylene in polyethylene. These estimates are useful for design of the final ethylene-polyethylene separation stage. At intermediate pressures, the free-volume theory provides the best method now available for calculation of equilibrium phase compositions. The results shown in Figure 7 are calculated for a high-molecular-weight polyethylene, so that the vapor phase contains only ethylene; however, the free-volume theory does not make this assumption. The free-volume theory can be used to calculate not only the solubility of ethylene in the polymer-rich phase but also the amount of polymer dissolved in the ethylene-rich phase. The polyethylene molecular-weight-distribution in each phase can also be calculated. A computer program which performs these calculations has been prepared. It enables the calculation of equilibrium phases, including molecular-weight distributions, given temperature, pressure, overall masses of ethylene and polyethylene, and overall polymer molecular-weight distribution. This Fortran IV computer program is available from the authors on request. Acknowledgment For financial support, the authors are grateful to the donors of the Petroleum Research Fund, administered by the American Chemical Society, Gulf Oil Chemicals Corporation, Union Carbide Corporation, and the National Science Foundation. Literature Cited Benzler. H., von Koch, A,, Chem. hg. Tech., 27, 71 (1955). Bondi, A,, J. Phys. Chem.. 68, 441 (1964). Bonner, D. C., Maloney, D. P.. Prausnitz, J. M., lnd. Eng. Chem., Process Des. A%.,13, 91 (1974): Erratum, 13, 198 (1974). Chueh, P. L.. Prausnitz, J. M., lnd. Eng. Chem., Fundam., 6 , 492 (1967). Din, F., "Thermodynamic Functions of Gases", Vol. 3, Butterworths. London, 1961. Ehrlich, P., J. Polym. Sci., A3, 131 (1965). Flory, P. J.. Discuss. Faraday SOC., 49, 7 (1970). Hiza, M. J., Duncan, A. G., AlChE J., 16, 733 (1970). Koningsveld, R., Diepen. G. A. M.. Chermin, H. A. G., Rec. Trav. Chim., 85, 504 (1966). Lundberg, J. L.. Mooney, E. J., J. Polym. S c i A-2, 7, 947 (1969). Lyckman, E. W., Eckert, C. A,, Prausnitz. J. M., Chem. Eng. Sci., 20, 685 (1965). Maloney, D. P., Prausnitz, J. M.. J. Appl. Polym. Sci., 18, 2703 (1974). Maloney, D. P.. Prausnitz. J. M., AlChE J., in press, 1975. Patterson, D.. Macromolecules, 2, 672 (1969). Steiner. R., Horb, K., Chem. lng. Tech., 44, 1010 (1972). Swelheim, T., de Swaan Arons. J., Diepen, G. A. M., Rec. Trav. Chim., 84, 261 (1965).
Received for review June 26, 1975 Accepted September 2,1975