Ind. Eng. Chem. Res. 1988,27,1925-1929 Martel, C. R.; Angelo, L. C. Technical Report AFAPL-TR-72-103, 1973, pp 11-15; Air Force Aero Propulsion Laboratory. Selvig, W. A.; Gibson, F. H. Chemistry of Coal Utilization; Wiley: New York, 1945; Vol. I, pp 132-44. Stein, T. R.; Cabal, A. V.; Callen, R. B.; Dabkowski, M. J.; Heck, R. H.; Simpson, C. A.; Shih, S. S. "Upgrading of Coal Liquids for Use as Power Generation Fuels". EPRI AF-873, 1978.
1925
Technical Data Book-Petroleum Refining, 3rd ed.; American Petroleum Institute: Washington, DC, 1976; Vol. 111, Chapter 14, p 3. Received for review September 25, 1987 Revised manuscript received May 27, 1988 Accepted June 28, 1988
Addition Reaction of Allyl Radical and Butadiene Daisuke Nohara* and Tomoya Sakai Department of Chemical Reaction Engineering, Faculty of Pharmaceutical Sciences, Nagoya City University, Mizuho-ku, Nagoya 467, J a p a n
Pyrolysis of diallyl oxalate in the presence of butadiene was carried out at 430-510 "C in order to study the kinetics and mechanism of the addition reaction of the allyl radical and butadiene. It is revealed that there are four categories of addition between the two reactants. Addition products of the allyl radical and butadiene were cyclopentadiene, 1-pentene, 3-methyl-1,5-hexadiene, and l,g-heptadiene, while products produced by the Diels-Alder reaction of butadiene with other olefins were 1,3-cyclohexadiene,benzene, and 4-vinylcyclohexene. The second-order rate constant, activation energy, and A factor characterizing the product formation from an allyl radical and butadiene were obtained. I t was revealed that the rate constant for cycloaddition of the allyl radical to olefins is 104-105 times as large as that for cycloaddition of butadiene to olefins. Mechanistic discussions were put forward. Two model reactions proposed by us for the polycyclization process as the early stage of coke formation are as follows: (1) polycyclization by gradual cycloaddition of adhesive fragment hydrocarbon such as butadiene or the allyl radical to the unsaturated bond of olefins or polycyclics; (2) further cycloaddition taking place between congenial polycyclics, sometimes involving participation of "fragment hydrocarbon". Actually, the polycyclization may proceed under entwinement of the two reactions described above, accompanied or followed by dealkylation and/or dehydrogenation at higher severity. Although, in the latter model reaction, mono- or polycyclics with some kinds of branched chain may be considered to be more polycyclizable, the great difficulty of the precise product analysis retards the progress of this field of investigation. On the other hand, investigations concerned with the former model reaction, which stand on the substantiated theoretical basis, clarified the mechanism and kinetic features of the primary cycloaddition stage positively (Sakai et al., 1970). A series of experiments on thermal cycloaddition of the allyl radical to unsaturated hydrocarbons, one of the basic researches related to polycyclization, have been studied. Among them, the reactions between the allyl radical and ethylene, propylene, acetylene, or methylacetylene have been reported. (See, for example, Nohara and Sakai (1980a,b).) In each reaction, as Bryce and Ruzicka (1960) commented, one or more C, cyclic compounds were obtained as the main addition product of the allyl radical to unsaturated hydrocarbons. That is, cyclopentene, methylcyclopentenes, cyclopentadiene, and methylcyclopentadienes corresponded to the individual substrate hydrocarbon described above. I t is considered that the formation of these C5 cyclic products is attributed to the diene character of the allyl radical, based on the analogy of the mechanism of c6 cyclic products formation by the Diels-Alder reaction of butadiene and olefins as illustrated as follows: 0888-5885/88/2627-1925$01.50/0
< + r ; " - f y " - y y- ~0" * The reason that the concerted mechanism is excluded in the latter scheme is because the concurrent production of acyclic products was observed in most cases. In the present study, butadiene was employed as another substrate not only to extend the kind of substrate species but also to directly compare the diene capability of the allyl radical with that of butadiene from the kinetic performance of the C5 cyclic product formation and that of the c6 cyclic product formation. It was revealed that four different parallel reaction pathways prevailed in the present reaction: (a) biallyl formation by recombination of the allyl radical; (b) product formation by addition of the allyl radical to butadiene; (c) product formation by the DielsAlder reaction of butadiene and other olefins; (d) 4vinylcyclohexene formation by the Diels-Alder reaction of butadiene itself. The C5cyclic product formation was classified in (b) and the formation of c6 cyclic product except 4-vinylcyclohexene in (c). Experimental Section The apparatus used was of an ordinary atmospheric flow system, and the reactor was a quartz, annular cylinder heated by an electric furnace. The details of the apparatus and procedures were previously described (Sakai et al., 1976). Butadiene was diluted with deoxygenated nitrogen to about 1.6 vol % concentration. As the source material of the allyl radical, diallyl oxalate (DAO) was employed. J h vapor concentration was set around 2.5 vol % of the total inlet gas. Five reaction temperatures were designed be0 1988 American Chemical Society
1926 Ind. Eng. Chem. Res., Vol. 27, No. 10, 1988 Table I. Typical Experimental Results" temp, "C 510 residence time, s 7.36 27.0 DAO conversn, % products, 10-lo mol cm-3 ethylene 113 propylene 353 1-butene 75.8 1-pentene 52.6 cyclopentadiene 139 biallyl 2200 44.3 benzene 1,3-cyclohexadiene 75.3 1,6-heptadiene 40.0 3-methyl-1,5-hexadiene 28.0 octadiene 22.1 octatriene 19.5 4-vinylcyclohexene 12.2
510 3.51 15.2
490 7.24 15.4
490 3.32 7.95
470 7.14 6.03
470 3.53 3.34
450 7.86 2.88
450 3.76 1.41
430 7.66 1.20
430 3.61 0.735
54.5 235 35.3 27.0 58.1 1120 13.5 26.5 19.3 14.5 9.90 9.07 6.53
67.2 292 35.0 36.6 74.2 1030 12.9 24.2 18.9 14.9 10.9 10.2 7.80
28.2 145 14.3 15.9 33.6 542 4.89 10.7 10.3 7.49 6.29 5.80 6.25
34.3 135 11.7 15.5 40.5 441 4.62 9.70 9.83 8.03 5.94 6.32 7.75
15.2 64.6 6.40 10.2 20.9 276 1.85 4.32 4.63 3.54 3.09 3.25 4.49
13.8 44.5 4.29 6.76 18.2 129 1.42 3.27 3.44 3.27 2.23 2.57 4.15
7.72 30.5 3.28 3.83 9.69 69.9 0.894 1.76 1.97 2.02 1.45 1.72 3.63
8.77 27.5 2.92 3.45 9.99 58.1 0.908 1.51 1.61 1.66 1.08 1.50 3.54
3.31 11.3 1.94 1.37 3.67 16.0 0.564 0.615 0.565 0.719 0.426 0.612 2.56
"DAO concentrations, 1.0 X lo* mol cmT3;butadiene concentration, 0.62
X
lo4 mol ~ m - ~ .
x 801
125
130
1.35
140
A
490'C
1
145
( l / T ) / 10'3K-'
Figure 1. Arrhenius plots for the DAO decomposition. Residence time / s
tween 430 and 510 "C. Because each experimental run caused its own characteristic temperature profile in the reactor tube, the equivalent reactor volume a t each designated temperature was calculated by the trial and error method of Hougen and Watson (1943). Residence times thus obtained were between 2 and 15 s. Reactants and products were analyzed by FID and TCD gas chromatographies and GC-MS spectroscopy. Hydrogenation of product gas in the presence of Pd black at room temperature or 100 OC was another useful tool for the identification of products obtained in the present experiment.
Results and Discussion DAO is pyrolyzed to produce allyl radicals and carbon dioxide. The extent of its decomposition was measured by the amount of COz formation. The thermal decomposition of DAO fitted the fint-order reaction kinetics, and the temperature dependence of its rate constant, k d , is shown in Figure 1, from which the following equation was obtained:
kd =
lo1'.'
exp(-186000/RT)
(in sS1)
These kinetic parameters were almost consistent with those obtained previously in the reaction of the allyl radical with acetylene (Nohara and Sakai, 1980a,b). The independent kinetic feature of DAO decomposition in the presence of either acetylene or butadiene implies the versatility of the use of DAO as an allyl source. Products. The main products in the present reaction, in order of decreasing amount, were biallyl, propylene,
Figure 2. Product distributions: ( 0 )biallyl; (0) cyclopentadiene; (0) 1-pentene; ( 0 )3-methyl-1,5-hexadiene;(@) 1,6-heptadiene; (e) 1,3-cyclohexadiene; ( 0 )benzene; (+) 4-vinylcyclohexene.
cyclopentadiene, ethylene, 1-butene, 1,3-cyciohexadiene (not 1,4-cyclohexadiene),1-pentene, benzene, 3-methyl1,5-hexadiene, 1,6-heptadiene, octadiene, octatriene, and 4-vinylcyclohexene. The minor products were found to be methane, methylcyclopentene, an unidentified C7 hydrocarbon ( m l e 94), methylcyclohexene, toluene, ethane, and smaller amounts of unidentified C7's. The expected product, vinylcyclopentene, could not be explicitly identified, even in the minor C7 products. However, the production of a small amount of vinylcyclopentene was justified by the detection of ethylcyclopentane after the hydrogenation of the product gas. No C7 cyclic product such as cycloheptene, cycloheptadiene, or cycloheptane was produced. Typical experimental results are listed in Table I. The distributions of the main products are shown in Figure 2, which illustrates moles of product per 100 mol of DAO converted. Ethylene and 1-butene were excluded from Figure 2 because these were not concerned with the reaction of the allyl radical with butadiene, at least a t the initial stage. Octadiene and octatriene, which seemed to be produced through addition of butadiene to butadiene, were also excluded because the discussion on their formation was not suitable for the aim of this work. In Figure 2, extrapolation of the curve to the axis of ordinates implies that all these products appear to be the initial ones. The rate of formation of these products could be measured from the initial slope of each formation curve. Examples of the
Ind. Eng. Chem. Res., Vol. 27, No. 10, 1988 1927 3
'5 E Yo 2 \ '=\
-c 1 m 2.
v
0
"00
"
5 10 15 Residence time / s
Figure 3. Example of measurement of rate of product formation: ( 0 )biallyl; (0) cyclopentadiene; (e)1,3-cyclohexadiene.
-20
-.> C
-24
\
\
\
1 25
1.30
1.35
1.40
I I 45
( 1 1 ~ 1/ IO.~K-'
Figure 4. Temperature dependency of rate of each product formation: (@) biallyl; (0) cyclopentadiene; (0)1-pentene;).( 3methyl-l,5-hexadiene; (0) 1,6-heptadiene; (e) 1,3-cyclohexadiene; ( 0 )benzene; (-# 4-vinylcyclohexene.
formation curves are shown in Figure 3 for biallyl, cyclopentadiene, and 1,3-cyclohexadiene. The temperature dependences of the measured rates for main products formation are illustrated in Figure 4. Examination of the above results enabled us to classify the product formation routes into the four groups described earlier. In the first place, (a) a large amount of biallyl is produced by recombination of the allyl radical generated from DAO pyrolysis. This is confirmed from the fact that the rate of biallyl formation was as large as that obtained in another substrate case. (See, for example, Nohara and Sakai, (1980a,b).) Next, (d) 4-vinylcyclohexene should be formed from butadiene itself. In the reaction of butadiene, i.e., in the absence of DAO, the product was almost all 4-vinylcyclohexene, and no compound between C5 and C7 was detected. The rate of 4vinylcyclohexene formation without addition of DAO was nearly equal to that with addition of DAO. Products other than biallyl and 4-vinylcyclohexene should be produced either through (b) addition of the allyl radical to butadiene or (c) the Diels-Alder reaction between butadiene and other olefins. Of these products, c6
cyclic compounds such as 1,3-cyclohexadieneand benzene should be formed through (c), while C5 cyclic compounds should be formed through (b). As for 1,3-cyclohexadiene and benzene, such a large amount has never been obtained in the reactions of the allyl radical and unsaturated hydrocarbons other than butadiene. (See, for example, Sakai and Nohara (1975)J Besides, the temperature dependence of their rates of formation is similar to the dependence of c6 cyclic product formation in the thermal reaction of butadiene and 1-butene (Sakai et al., 1970). Moreover, they must be formed through addition of butadiene to the C=C double bond of DAO, which exists in the highest concentration in the system. They would not be formed as the secondary products through addition of butadiene to biallyl produced in situ, because the temperature dependence of the rate of biallyl formation, 224 kJ mol-', was larger than 219 kJ mol-' for the 1,3-cyclohexadiene formation and not much smaller than 249 k J mol-l for the benzene formation. Nor are they through devinylation of vinylcyclohexene because the amount of 1,3-cyclohexadiene and benzene obtained in the present experiment is far beyond the amount evaluated from the authorized second-order rate constant of 4-vinylcyclohexene formation. At the present stage, however, the reasonable route for these C6 cyclic products formation from butadiene and DAO remains undecided. Under the temperature conditions, no C8 acyclic product was formed from the thermal reaction of butadiene (Sakai et al., 1970). Owing to the presence of DAO in this experiment, C8 acyclic products must be formed by another radical chain mechanism than addition of butadiene itself. Next, the products classified into (b), addition products of the allyl radical to butadiene, are cyclopentadiene, 1pentene, and C7 compounds such as 3-methyl-1,5-hexadiene and 1,6-heptadiene. It has been confirmed that an allyl radical adds, for example, to ethylene to form mainly cyclopentene and 1-pentene (Sakai and Nohara, 1975), to propylene to form methylcyclopentenes and hexene (Sakai and Nohara, 1976), to acetylene to yield mostly cyclopentadiene (Nohara and Sakai, 19801, and to methylacetylene to produce methylcyclopentadienes (Nohara and Sakai, 1981). The scheme for the products formation, for example, in the ethylene case, is believed to be as follows: 0
0
L
-
Q
/
TThe expected amounts of vinylcyclopentenes, if they existed, were little. A considerably large amount of cyclopentadiene and 1-pentene produced in the present experiment implies that elimination of the C2fragment from the intermediate species such as vinylcyclopentyl or heptenyl radicals may take place. Detail discussions on this point will be developed in the next section. The contribution of addition of the allyl radical to DAO or to biallyl for these products formation should be common to the case of any substrate hydrocarbon. Then, it was evaluated to be less than 5% of the total amount of these products formation. It is reasonable to conclude that these two products, cyclopentadiene and 1-pentene, as well as C7 products in the present system are undoubtedly formed through addition of the allyl radical to butadiene. Kinetic Studies. The second-order rate constant for each product formation was tentatively evaluated on the basis of the following simplified schemes deduced from the above discussion. The schemes proposed are ones for
1928 Ind. Eng. Chem. Res., Vol. 27, No. 10, 1988 Table 11. Kinetic Parameters reaction
product
c=c-e. + c=c-c=c
c=c-c. + c=c
log A , cm3 mol-' s-'
7.54
46.8
10.8
&v-
7.22
64.8
11.6
A
7.02
87.4
12.9
Pe-J-.\
6.87
61.0
11.0
0 0
6.52
48.0
9.77
6.38
69.8
11.1
[y
c=c-c=c + c=c-c=c
E,
kJ mol-'
0
-
c=c-c. + c=c
log k m * c l
cm3 mol-' s-'
7.64
104
14.6
2.97
105
10.1
"
classification of the type of product formation route and do not account for carbon or hydrogen balance: (a) allyl radical + allyl radical biallyl (b) allyl radical + butadiene cyclopentadiene, 1-pentene, 3-methyl-l,5-hexadiene, 1,6-heptadiene (c) butadiene
+ DAO
(d) butadiene
- -
1,3-cyclohexadiene, benzene
+ butadiene
4-vinylcyclohexene
The characteristic rate constant for each product formation may be equated as follows: d[biallyl]/dt = k(,)[allyl radicalI2 d[cyclopentadiene]/dt, d[l-pentene]/dt = ko)[allyl radical] [butadiene] ?L
I
*.
d[ 1,3-cyclohexadiene]/dt, d[benzene]/dt = k(,)[butadiene][DAO] d[4-vinylcyclohexene]/dt = k(d)[butadieneI2 The individual rate of product formation in the above equations can be experimentally measured from the initial slope of the formation curve, as exemplified in Figure 3. Here, reaction a is adopted as the reference reaction; i.e., the concentration of the allyl radical in the reaction system can be approximately estimated from [allyl radical] =
(
1 d[biallyl]
)
l''
dt in which kfa)is available from Golden et al. (1969) and Rossi et al. (1979). In practice, the numerical value of kb) was settled to be 5 X 10l2cm3 mol-' s-l, being regarded virtually as independent of temperature. The temperature dependence of k values evaluated by the above method is shown in Figure 5. Numerical values of both activation energy and A factor obtained for the 4-vinylcyclohexene formation are in good agreement with those reported by Sakai et al. (1970) and are considerably smaller than those for the formation of 1,3-cyclohexadieneor benzene. The latter values coincide with those of the Diels-Alder reaction of butadiene with 1-butene and isobutene (Sakai et al., 1970). Now the discussion will be focused on reaction route b, the addition of the allyl radical to butadiene. Kinetic parameters obtained for the formation of the individual addition product of an allyl radical to butadiene are indicated in Table I1 together with those for the formation of cyclopentene and 1-pentene in the ethylene case (Sakai
'+,
5 125
130
, \,e 140
135 ( ii T ) I
'.01
K-'
Figure 5. Temperature dependency of rate constant for each product formation: (0)cyclopentadiene; (0)1-pentene; ( 0 ) 3methyl-1,5-hexadiene; (@) 1,6-heptadiene; (e) 1,3-cyclohexadiene; (6) benzene; (+) 4-vinylcyclohexene.
et al., 1976) and of cyclopentadiene in the acetylene case (Nohara and Sakai, 1980a,b). Each activation energy or A factor for the formation of 1-pentene, 3-methyl-1,5hexadiene, or 1,6-heptadiene is similar to that for the 1pentene formation in the ethylene case. These activation energies correspond to the hydrogen abstraction step by the alkenyl radical as disputed by Sakai et al. (1976). The activation energy, 46.8 kJ mol-l, obtained for the cyclopentadiene formation is quite different from 104 k J mol-' obtained for the cyclopentadiene formation in the acetylene case. This value is nearly equal to 48.0 kJ mol-' obtained for the cyclopentene formation in the ethylene case. This fact indicates that the mechanism for the cyclopentadiene formation in the present case is quite different from that in the acetylene case, in which cyclization of the 1,Cpentadienyl radical and consecutive hydrogen atom elimination were postulated. Also this fact implies that the present mechanism is analogous to that of cyclopentene formation in the ethylene case, in which cy-
Ind. Eng. Chem. Res., Vol. 27, No. 10, 1988 1929 hara and Sakai, 1981) should correspond to the cyclization step of a 2-methyl-l,4-pentadienyl radical. The large difference, as large as 4-5, in log A between the cases of acetylenes and olefins originates at the cyclization step as discussed previously (Nohara and Sakai, 1980a,b). After all, the cyclization of a 1,4-pentadienyl-type radical should be accompanied by a large A factor and high activation energy compared to that of a 4-pentenyl-type radical.
600 542 P
T
200 -
I
4
h209)1 cb
CZ 52
I
'
I
1I
Figure 6. Hypothetical energy diagram pictured if an H-elimination step were rate limiting for the cyclopentadiene formation in the butadiene case: (*) value observed in this work; (**) value observed in the previous work; (w) value which should be evaluated if an H. elimination step were rate limiting in this work; (0)value which should be observed if an H. elimination step were rate limiting in this work.
clization of the 4-pentenyl radical and consecutive hydrogen atom elimination were postulated (Sakai et al., 1976). Accordingly, the following scheme should be adopted:
The discrepancy arises that neither vinylcyclopentenenor vinylcyclopentadiene was explicitly detected in the present experiment. The unique solution to this problem is to postulate that the vinylcyclopentyl radical, which inclines to the character of a 4-pentenyl-type radical, cyclizes again followed by the reverse Diels-Alder reaction to yield the cyclopentenyl radical and ethylene as indicated below:
This reverse Diels-Alder reaction is analogous to the decomposition of norbornadiene surveyed by Walsh and Wells (1975). It was already described by Nohara and Sakai (1980a,b) that the cyclopentenyl radical releases a hydrogen atom to yield cyclopentadiene as follows: J- ( *
-Q
+
He
On the basis of these schemes, it is revealed that the activation energy, 104 kJ mol-l, obtained for the formation of cyclopentadiene in the acetylene case should not correspond to the hydrogen atom elimination step but to the cyclization step of a 1,Cpentadienyl radical. Otherwise, the value of 273 kJ mol-' should be actually measured for the formation of cyclopentadiene in the present butadiene case. The reason is explained in terms of heat content in Figure 6, which is pictured on the hypothesis that an Helimination step from a cyclopentenyl radical is rate limiting. It shows 104 + 52(ethylene) + 227(acetylene) llO(butadiene) = 273 kJ mol-I. The estimated heat content for a cyclopentenyl radical, 209 kJ mol-', might not be a true value, but no difficulty is involved in its comparative use. From the above discussion, the rate-limiting step for the formation of cyclopentadiene in the acetylene case is the cyclization step of a lP-pentadienyl radical in view of energy barrier. From a similar reason, the activation energy, 105 kJ mol-', obtained for the formation of methylcyclopentadiene in the methylacetylene case (No-
Conclusions It was explained that in the pyrolysis of DAO in the presence of butadiene the rate of the C5 cyclic product formation represented that of cycloaddition of the allyl radical to butadiene, while the rate of the c6 cyclic product formation represented that of cycloaddition of butadiene to olefin. The concentration of allyl radicals in the system was estimated to be (1-8) X mol cm-3 and that of butadiene was ca. 6 X lo-' mol ~ m - ~Accordingly, . the concentration term in the rate equation for the cycloaddition of the allyl radical to butadiene was around 10-17-10-'s and that of the Diels-Alder reaction of butadiene and olefin was on the order of Thus, the rate constant for cycloaddition of the allyl radical to olefin is 104-105times as large as that of butadiene to olefin because the rate of total C5 cyclic product formation exceeded the rate of total c6 cyclic product formation in the present observation. The molar ratio of the cyclic to the acyclic addition products of allyl radicals to butadiene was between 1and 2. Although the ratio in the addition of butadiene to olefins could not be determined in this experiment, it was reported (Sakai et al., 1970) that vinylcyclohexene was the sole product in the thermal reaction of butadiene up to 600 "C. In terms of selectivity for cycloaddition to olefins, butadiene is superior to the allyl radical, although the selectivity for cycloaddition of the allyl radical to acetylene is as high as that of butadiene (Nohara and Sakai, 1980a,b). 4-Pentenyl- and 1,Cpentadienyl-typeradicals cyclize to yield C5 cyclic product, and the cyclization of the latter radical proceeds fast, overcoming a considerably larger energy barrier than that of the fo mer. Registry No. DAO, 615-99-6; allyl radical, 1981-80-2; butadiene, 106-99-0; cyclopentene, 142-29-0; ethylene, 74-85-1; propylene, 115-07-1; 1-butene, 106-98-9; 1-pentene, 109-67-1; cyclopentadiene, 542-92-7; benzene, 71-43-2; 1,3-cyclohexadiene, 59257-4; biallyl, 592-42-7; 1,6-heptadiene, 3070-53-9; 3-methyl-1,5hexadiene, 1541-33-9;octadiene, 63597-41-1;octatriene, 30702-87-5; 4-vinylcyclohexene, 100-40-3.
Literature Cited Bryce, W. A.; Ruzicka, D. J. Can. J. Chem. 1960,38,835-844. Golden, D. M.; Gac, N. A,; Benson, S. W. J. Am. Chem. SOC. 1969, 91,2136-2137. Hougen, D.A.;Watson, K. M. Chemical Process Principles; Wilay: New York, 1943; p 884. Nohara, D.; Sakai, T. Ind. Eng. Chem. Fundam. 1980a,19,340-344. Nohara, D.; Sakai, T. J . Jpn. Pet. Inst. 1980b,23, 133-138. Nohara, D.;Sakai, T. J . Jpn. Pet. Inst. 1981,24,122-127. Rossi, M.;King, K. D.; Golden, D. M. J . Am. Chem. SOC.1979,101, 1223-1230. Sakai, T.; Nohara, D. Bull. Jpn. Pet. Inst. 1975,17, 212-217. Sakai, T.; Nohara, D. Seventh Symposium (Japan)on Petroleum Chemistry; The Japan Petroleum Institute: Sapporo, Japan, 1976; pp 2-5. Sakai, T.; Nohara, D.; Kunugi, T. Am. Chem. SOC.Symp. Ser. 1976, 32, 152-177. Sakai, T.;Soma, K.; Sasaki, Y.; Tominaga, H.; Kunugi, T. Adu. Chem. Ser. 1970,97,68-91. Walsh, R.; Wells, J. M. J. Chem. Thermodyn. 1975,7(2), 149-154. Received for review October 2, 1987 Accepted May 26, 1988