Organometallics 1986, 5, 215-218 described before. The rate of gas evolution corresponded to the rate of acyl decomposition between 50 and 80 O C . During the decomposition of 1 or 2, isomerization of the acyl compounds did not occur. The composition of the reaction mixture was checked by IR with the v(C0) bands of terminal CO ligands as a control of concentrationand the acyl v ( C 0 ) bands as a control of structure. The organic products of decomposition were investigated in experiments performed at 80 OC and 1-h reaction time in sealed 20-mL ampules with 2-mL samples of 0.02 M solutions of 1 and 2 in n-octane containing n-pentane as internal reference for quantitative GC analysis (3 m, n-octanelPorasi1 C, 50 "C, 15 mL/min of Ar, FID). The decompositions yielded 45-48% propene, 1%propane, and 37% n-butyraldehyde (ifstarting from 1) or 45% isobutyraldehyde (if starting from 2). Stability of 1 and 2 at 80 OC under CO Pressure. The 5-mL samples of 0.014 M solutions of 1 or 2 in n-heptane were placed at room temperature under CO in a 20-mL stainless-steelrocking
215
autoclave and pressurized to 80 bar with CO. After 1-h agitation at 80 O C , the IR spectra of the reaction mixtures at room temperature showed within 1% accuracy unchanged concentrations of the acyl complexes as measured at 2105 cm-I. No traces of isomerized acylcobalt tetracarbonyl could be recognized in the characteristic organic v(C0) range of 1 and 2. Reaction of Ph3P with 1 and 2 was performed in the gasometric apparatus described above and started by injecting the acyl stock solutions into solutions containing known amounts of Ph3P. In all kinetic experiments initial rates were calculated from graphical plots below 15% conversion. Registry No. 1, 38722-67-7; 2, 38784-32-6; HCo(CO),, 16842-03-8;Hz, 1333-74-0;CO, 630-08-0; n-PrCHO, 123-72-8; 10210-68-1;n-PrC(0)Co(C0)3(PPh,), i-PrCHO, 7884-2; CO~(CO)~, 34151-25-2;~-P~C(O)CO(CO)~(PP~J, 99495-72-4;PPh3,603-35-0; propene, 115-07-1.
Rhodium( I11)-to-Rhodium( I ) Alkyl Transfer: A Rhodium Macrocycle Which I s Both Nucleophile and Leaving Group James P. Collman,' John I. Brauman, and Alex M. Madonik Department of Chemistty, Stanford University, Stanford, California 94305 Received February 72, 1985
The rhodium(1) macrocyclesRh'(BPDOBF2) (la)and Rh'(PPDOBF2) (lb) are strong nucleophiles toward alkyl halides and may also be alkylated by their rhodium(II1) alkyl halide adducts R'-Rhnl(BPDOBFz)-X (2a and 2b), respectively. In the latter process, la,b act as both nucleophile q d R'-Rh"'(PPDOBFJ-X and leaving group. We have studied the kinetics of alkyl exchange between, e.g., l a and 2b by 'H NMR M-' s-l for methyl exchange (R' = Me, X = I) and (6.4 f 1.0) and the rate constants are (5.8 f 0.5) X X M-' s-l for benzyl exchange (R' = PhCH2, X = Cl). Other alkyl groups do not participate in this reaction. These data are consistent with rhodium(1) attack at carbon; the process may be viewed as a carbon-bridged two-electron transfer. Halide-bridged electron transfer is implied in the much faster exchange ( K > 0.5 M-' 9-l) between la,b and their rhodium(II1) dichloride derivatives 3a and 3b.
Introduction Low-valent transition-metal complexes exhibit nucleophilic behavior in many of their oxidative-addition reactions with alkyl halides, leading to the formation of metal-carbon bonds via an SN2-like pr0cess.l The ability of metal centers to function as leaving groups is less widely recognized, despite scattered reports.2 These include several examples of essentially degenerate nucleophilic reactions in which the same type of metal center serves as both nucleophile and leaving group2a,b,c,e (eq 1). The M:
+ M R -2M R -'y,';
I
M;
(1)
latter reactions result in the transfer of an alkyl group from one metal center to another and may be viewed as carbon-bridged inner-sphere two-electron-transfer processes? (1) Collman, J. P.; Brauman, J. I; Madonik, A. M. Organometallics, following paper in this issue. (2)(a) Johnson, R. W.; Pearson, R. G. J. Chem. SOC.,Chem. Commun. 1970,986-987. (b) Dodd, D.; Johnson, M. D. J. Chem. Soc., Chem. Commun. 1971,1371-1372. (c) Mestroni, G.;Cocevar, C.; Costa, G. Gazz. Chim. Ztal. 1973,103,273-285. (d) Schrauzer, G.N.; Stadlbauer, E. A. Bioinorg. Chem. 1974,3, 353-366. Stadlbauer, E. A.; Holland, R. J.; La", F. P.; Schrauzer, G. N. Bioinorg. Chem. 1974,4,67-77.(e) Dodd, D.; Johnson, M. D.; Lockman, B. L. J. Am. Chem. SOC. 1977, 99,
3664-3673. (3)Rsviews of inner-sphere electron transfer: (a) Taube, H. 'Electron Transfer Reactions of Complex Ions in Solution"; Academic Press: New York, 1970. (b) Linck, R. G. S u m B o g . Chem. 1976,7,89-147. (c) Sutin, N. "Inorganic Biochemistry"; Eichhorn, G. L., Ed.; Elsevier: New York, 1974;Chapter 7.
0276-7333/86/2305-0215$01.50/0
When the two centers are distinguished by a minor ligand modification, these exchange reactions are nearly thermoneutral, and their study should throw light on the kinetic factors which control two-electron processes that make and break metal-carbon bonds.4 We encountered a novel example of the process represented by eq 1in the course of our studies on the oxidative addition of alkyl halides to a Rh(1) ma~rocyc1e.l~~ We (4)Several groups have undertaken the analysis of "intrinsic" kinetic barriers to nucleophilic reactions in which the Marcus t h e o e of electron transfer is applied to reactivity data for methyl transfer in the gas phase'bd or in solution.' Endicott's group has made an extensive analpis of the analogous S H mechanism ~ implicated in methyl and halide transfer between Co(I1) and Co(II1) macrocycles.'-h (See also ref 4i.) (a) Marcus, R. A. Annu. Reu. Phys. Chem. 1964,15,155-196.(b) Shaik, S.S.; Pross, A. J. Am. Chem. SOC.1982,104,2708-2719. (c) Wolfe, S.;Mitchell, D. J.; Schlegel, H. B. J. Am. Chem. SOC.1981,103,7694-7696.(d) Pellerite, M. J.; Brauman, J. I. J. Am. Chem. SOC. 1983,105,2672-2680.Pellerite, M. J.; Brauman, J. I. J. Am. Chem. SOC.1980,102,5993-5999.(e) Albery, W. J.; Kreevoy, M. M. Adu. Phys. Org. Chem. 1978, 16,87-157. (f'j Endicott, J. F.; Wong, C.-L.; Ciskowski, J. M.; Balakrishnan, K. P. J.Am. Chem. SOC.1980,102,2100-2103.(9) Endicott, J. F.; Balakrishnan, K. P.; Wong, C.-L. J.Am. Chem. SOC.1980,102,5519-5526. (h) Durham, B.; Endicott, J. F.; Wong, C.-L.; Rillema, D. P. J. Am. Chem. SOC. 1979, 101,847-857. (i) Johnson, M. D. Acc. Chem. Res. 1983,16,343-349. (5)Preliminary report: Collman, J. P.; MacLaury, M. R. J.Am. Chem. SOC.1974,96,3019-3020.We have since adopted a modified nomenclature for the macrocyclic ligands: (PPDOBF,) = [difluoro[N,N'-bis(3pentanon-2-ylidene)-1,3-diaminopropane)dioximato] borate], and (BPDOBF,) = [difluoro[N~-bis(3-butanon-2-ylidene)-1,3-diaminopropane)dioximato]borate]. Other abbreviations: PPN+ = bis(tripheny1phosphoranediyl)nitrogen(l+) ion.
0 1986 American Chemical Society
216 Organometallics, Vol. 5, No. 2, 1986
Collman et al.
report here a kinetic study of this reaction, using two similar macrocyclic ligands which can be differentiated by ‘HNMR. The process is quite sensitive to steric bulk in the alkyl group to be transferred; nonetheless, in some cases alkyl exchange proceeds a t a rate competitive with alkylation of Rh(1) by the corresponding halide (eq 2 and 3). F’, I F
F,,
o’B’o
O’B‘O
U
U
b R=Et
h
._ la R z M e
R=Me R=Et
(3)
?a 2b_ .lo .. A much faster Rh(1)-Rh(II1) exchange process, possibly involving a chloride-bridged transition state, was also observed (eq 4). l.b.
Experimental Section The preparation of the Rh complexes has been described, as has the purification of solvents.’ Inert-atmosphere techniques were required for all experimentsbecause of the oxygen sensitivity of the Rh(1) complexes. NMFt spectra were recorded on a Varian XL-100 instrument equipped with a Nicolet Technology data system6 Inorganic salts and benzyl chloride were purified according to the recommendations of Perrin, Armarego, and Perrin? PPN+Cl- (Alfa) was recrystallized from ethanol and dried under vacuum. TLC analyses used silica gel plates supplied by Analtech, Inc., eluting with 30% MeCN/CH2C12. Kinetics Stock solutions of the Rh(1) reagents in CD3CN (10 or 20 mM) were prepared in the glovebox, where weighing errors resulted in uncertainties of f5-10% in their concentrations. Rh(II1) compounds were weighed precisely before transfer to the glovebox. The Rh(1) solution (0.5 mL, 0.005 or 0.01 mmol of Rh(1)) was measured with a volumetric pipette or syringe, stirred with the Rh(II1) (1.0 equiv) compound to dissolve it, and the mixture filtered into an NMR sample tube, which was capped with a rubber septum. Reacting samples were transferred to the probe of the spectrometer (at 28.5 “C), and sequential spectra were acquired automatically using the kinetic routine provided by Nicolet. At the minimum concentration used, 10 mM in each of the reacting species, reasonable signal-to-noise ratio was achieved in a 16-transient spectrum which took slightly under 1 min to acquire. Reactions having rate constants significantly greater than 0.1 M-’ s-l could not be studied by this technique. Kinetic data were obtained by separately integrating the distinct ethyl group triplets of the Rh(1) and Rh(II1) species bearing the (PPDOBF2)ligand. The ratios of the integrated signals were used as input for the kinetic analysis.* Attempts to use an internal standard (cyclo(6) NMR data are reported relative to internal Me&; they were calibrated relative to the residual solvent peak at 6 1.93 for CD!CN; (7) Perrin, D.O.; Armarego, W. L. F.; Perrin, D. R. “Purification of Laboratory Compounds”; (Pergamon Press: New York, Oxford, 1966.
0
,
I
I
1
5
10
15
20
TIME
(minutes)
Figure 1. Least-squares plots of NMR kinetic data for the alkyl exchange reaction (at 28.5 “C in CD&N solution).8 (a) Methyl exchange between lb and 2a (R’= Me, X = I), each at 20 mM. (b) Benzyl exchange between la and 2b (R’ = PhCH,, X = Cl), each at 10 mM.
hexane) for comparison of integrals gave erratic results, probably owing to the large differences in peak widths and relaxation times. We believe the peak ratios are reliable measures of [product]/ [reactant], since rate constants were unchanged by variations in the pulse width and time delays used to acquire the spectra.
Results The facility of alkyl exchange is extremely sensitive to the structure of the alkyl group R’. Thus, reaction is fastest for R’ = benzyl; methyl transfer also proceeds a t a moderate rate a t room temperature. Contrary to our preliminary r e p ~ r t we , ~ could find no evidence for the exchange of bulkier R’ groups such as n-butyl or isopropyl, even over periods of several hours a t 70 “C (in CD3CN, with Rh(1) and Rh(II1) complexes each a t 20 mM). A 1-phenylethyl complex and an acetyl complex were also unreactive toward the Rh(1) reagent. (8) The exchange reaction of eq 3 can be treated according to the kinetic scheme used for isotope exchange,” provided that the equilibrium constant is equal to one (kz = k-&: In (1 - F) = RLt, where R = the constant rate of exchange, k2fRh’(I)]o[Rh(III]o,L = (l/[Rh’(I)lo + I/ [Rh(III)l0,and F = the ”fraction of exchange”; ([Rh’(III)]t([Rh’(I)]o)+ [Rh(III)]o)/([Rh’(I)]~[Rh(III)]~) = L[Rh’(III)], (Rh and Rh’ represent the two distinct macrocycles.) Since the NMR kinetic data were in the form of a ratio (Q = [profluct]/[reacfa“t] = [Rh’(mj]([ph’(I)]), they were converted to concentration data wing the known initial concentrations: [Rh’(III)] = [Rh’(I)],Q/(Q + 1). The initial concentrations of both Rh and Rh’ are required to compute the input for the lease-squares analysis. If one was not known with sufficient accuracy, it could be calculated from the other in conjunction with the observed equilibrium ratio Qhfinity= [Rh(III)IO/[Rh’(I)lo In practice, the value of Qxhw ww sometimesvaried slightly in order to obtain the best possible least-squares fit to the data. is plotted against time, and the slope is In Figure 1,-(log(l - F))/(R’L) equal to the rate constant k (R’ = [Rh’(I)]o[Rh(III)]o). (a) McKay, H. A. C. “Principles of Radiochemistry”; Butterworths: London, 1971; p 296. (b) Wilkins, R. G. ‘The Study of Kinetics and Transition Metal Complexes” Allyn and Bacon: Boston, 1975; p 152 ff.
Organometullics, Vol. 5, No. 2, 1986 217
Rhodium(IIl)-to-Rhodium(I)Alkyl Transfer
Table I. Kinetic Data for Rh(1)-Rh(II1) Alkyl Exchange' Rh(I),d mM
R
Rh(III), mM
14b 206 22b 20b 29* 18b 196 15" 12" 11" 16" 11" 9" 15b 12" 2 1"
Me Me Me Me Me Me Me BZ BZ BZ Me Me Me Me Me Me
20" 20" 20" 20" 20" 20" 20" 1Ob lob 1Ob
20b lob
20b 20" 206 206
Na+Y-, mM
Qd
I200 I200 I200 C104 200
1.40 1.02 0.89 1.00 0.68 1.12 1.04 1.49 1.15 1.11 0.80 1.09 0.45 1.35 0.58 1.07
I I I I I I
I
c1 c1 c1 BF4 BF4 I I I I
103 k, M-1
5-1
5.6 f 0.11 5.7 f 0.10 6.1 f 0.10 5.9 f 0.14 4.4 0.04 6.7 f 0.07 6.0 f 0.05 66 f 5.1 66 f 5.5 60 f 3.5 49 f 2.1 34 f 2.5 5.0 f 0.13 6.0 f 0.17 2.2 f 0.02 9.0 f 0.16
*
R' 0.997 0.995 0.997 0.994 0.999 0.999 0.999 0.944 0.942 0.964 0.991 0.990 0.993 0.988 0.999 0.996
"Ligand = (BPDOBFJ. bLigand = (PPDOBF2). cReactions a t 28.5 OC in CD3CN. dCalculated from the initial concentration of the Rh(II1) complex using Qid8. eProduct/reactant ratio a t t = infinity, after adjustment to give the best least-squares fit. 'Correlation coefficient.
The Rh(II1)-alkyl derivatives of the two macrocyclic ligands (PPDOBF,) and (BPDOBF,) could be readily distinguished by TLC analysis. However, the progress of these exchange reactions was most conveniently followed by 'H NMR. The equilibrium constants are equal to 1.0 within the uncertainties of the experiments (f5% in the product/reactant ratio). They exhibit straightforward second-order behavior, with a rate constant of (5.8 f 0.5) X M-' s-l for methyl exchange (eq 3, R' = Me, X = I) and (6.4 f 1.0) X M-' s-' for benzyl exchange (eq 3, R' = Bz, X = Cl). Representative kinetic data are plotted in Figure 1. (Zero time corresponds to the first spectrum acquired after transfer of the reacting sample to the spectrometer probe, so the plots do not intercept the origin a t t = 0). The benzyl exchange reaction was almost too fast to study by this method, hence the greater uncertainty in its rate constant. Kinetic data are listed in Table I for all runs. The dependence of the Rh(1)Rh(II1) exchange rates on the structure of the alkyl group implies Rh(1) attack at the carbon bound to Rh(II1) in the crucial step leading to group transfer (see below). Methyl exchange was also studied by using the cationic methyl donor 4, and it was found to be 8-10 times faster than with the neutral methyl iodide adduct (2a or 2b, R' = Me, X = I). The neutral and cationic methyl complexes exhibit distinct lH NMR spectra which do not broaden or merge in the presence of one another. By this NMR criterion, the neutral complex does not dissociate appreciably at concentrations as low as 5 mM in CD,CN. Addition of sodium iodide to solutions of the neutral complex does cause slight shifts in its NMR spectrum, and the presence of NaI may retard Rh(1)-Rh(II1) exchange somewhat (sodium perchlorate appears to have the opposite effect). Our data on these points are inconclusive, but given the implied low dissociation constant for coordinated iodide, it seems unlikely that formation of a five-coordinate intermediate precedes methyl transfer in the reactions of the neutral complex. The chloride exchange reaction of eq 4 was found to be much more rapid than alkyl exchange. The presence of all four species at equilibrium was confirmed by quenching the reaction with benzyl chloride and identifying the two dichloride derivatives and the two benzyl chloride adducts by TLC. The equilibrium was completely established by the time an NMR sample could be transferred from the glovebox to the spectrometer (CD,CN solution, Rh(1) and Rh(II1) species each at 10 mM). This result indicates that the rate constant is greater than 0.5 M-' s-l. However, no
significant broadening of the Rh(1) or Rh(II1) ligand resonances was observed, which requires that the rate constant be less than 100 M-' s-'.*~ Addition of a fivefold excess of PPN+Cl- did not appreciably retard this reaction.
Discussion The normal second-order kinetics of the exchange reactions and the higher rate of benzyl vs. methyl transfer are consistent with nucleophilic attack at carbon. The Rh(1) macrocycle can be assigned a "leaving group ability" on a par with that of chloride, since benzyl chloride alkylates Rh(1) a t a rate similar to that for the benzyl-Rh(111) complex.' No alkyl group more bulky than benzyl or methyl participates in this reaction. This result.strongly suggests that attack at carbon is crucial to alkyl transfer. This reactivity sequence (benzyl > methyl >> n-butyl, isopropyl) is entirely consistent with our studies on the nucleophilic behavior or Rh(1) complex (1)toward alkyl halides.' The faster rate of methyl transfer from the cationic methyl-Rh(II1) complex 4 compared to the neutral methyl iodide adduct (eq 2, R' = Me, X = I) is also consistent with Rh(1) attack a t an electrophilic center. If Rh(1)-Rh(II1) exchange were initiated by Rh(1) attack on the coordinated halide (or on rhodium itself at the vacant axial coordination site of a cationic alkyl-Rh(II1) intermediate), there would be no reason for dramatic reactivity differences among the various alkyl groups. Furthermore, such a mode of attack would generate an anionic alkylRh(1) species, an energetically unfavorable intermediate (eq 5). (5)
La Where Rh(1)-Rh(II1) exchange actually does take place via Rh(1) attack on coordinated halide (eq 4),the process is much faster than alkyl-mediated exchange. The presumed mechanism involves formal transfer of C1+ to Rh(I), in which case the new Rh(1) complex is associated with an energetically reasonable chloride anion. Macrocyclic alkyl-Co(II1) complexes participate in exchange reactions with both Co(1) and Co(I1) macrocy~ l e s . ~ ~ , ~The , ~ , finding ~ g , ~ of second-order kinetics and lb
(9) (a) van den Bergen, A,; West, B. 0. J. Chem. SOC.,Chem. Commun. 1971,52-53. van den Bergen, A.; West, B. 0. J . Oganomet. Chem. 1974, 64, 125-134. (b) Chrzastowski, J. Z.; Cooksey, C. J.; Johnson, M. D.; Lockman, B. L.; Steggles, P. N. J . Am. Chem. SOC.1975, 97, 932-934.
218
Organometallics 1986, 5, 218-222
inversion of configuration at carbon supports SN2 and SH2 mechanisms for the reactions involving C O ( I )and ~ ~ Co~ (II)2””pgb, respectively. In either case, alkyl transfer is much more facile than in the rhodium system. Given the structural similarities among these macrocyclic complexes, this rate difference can best be accounted for by the higher metal-carbon bond strength in the case of rhodium (c.f. ref 4g). Furthermore, methyl transfer to Rh(1) complex l b from methyl-Co(II1) complex 5 proceeds quantitatively on mixing.’O Thus, the cobalt macrocycle appears to be r
1
+ B
5-
b R=Et a better leaving group than the rhodium macrocycle, presumably owing to the difference in metal-carbon bond strengths. However, the “intrinsic” kinetic barriers for SN2-like processes involving Co(I/III) and Rh(I/III) macrocycles cannot yet be ascertained precisely, as no activation parameters or overall free energy changes (for transfers between the two different metals) have been determined. (10)Collman, J. P.; Finke, R. G.; Sobatka, P. A., unpublished observations. Prepartion of cobalt complexes: Finke, R. G.; Smith, B. L.; McKenna, W. A.; Christian, P. A. Znorg. Chem. 1981,20, 687-693.
The greater efficiency of chloride-bridged vs. methylbridged metal-metal exchange reactions has also been noted for Co(I1)-Co(II1) e ~ c h a n g e . ~ ~Endicott * g * ~ et al.4f attribute this acceleration to the greater electron affinity of the chlorine radical compared to the methyl radical, leading to more favorable bonding in the transition state. The millionfold rate difference observed by these authors can be compared to the result for Rh(1)-Rh(II1) exchange, where the factor is no more than lo5. This smaller acceleration probably results from the requirements of the two-electron transfer. Furthermore, the observed lability’ of coordinated halides in the Rh(II1) complexes may reduce their efficiency as bridging ligands.
Summary The scope of the Rh(II1)-Rh(1) transfer reaction (eq 3) has been defined, and kinetic measurements confirm its SN2-like nature. It is considerably slower than related Co(1)-Co(II1) exchange reactions, probably owing to the difference in carbon-metal bond energies. Alkyl transfer from Co(II1) to RhIII) is quantitative. Further comparisons within a given triad, combined with the measurement of activation parameters for “degenerate” exchange reactions, should serve to characterize the “intrinsic” barriers to the formation and cleavage of metal-carbon bonds via two-electron processes.
Acknowledgment. This work was supported by the NSF Grant CHE78-09443. Alex M. Madonik was the recipient of an NSF graduate fellowship. Registry No. la,41707-60-2;lb, 53335-25-4;2a (R’= Me, X = I), 99355-04-1; 2b (R’= Bz, X = Cl), 99355-15-4; 3a,99355-03-0; 5, 57255-98-8. 4b,99355-24-5;
Reactions of a Rhodium( I ) Macrocycle with Organic Dihalides: Oxidative- Addition and ,6-Elimination Pathways James P. Collman,’ John I . Brauman, and Alex M. Madonik Depat?ment of Chemistry, Stanford University, Stanford, California 94305 Received March 18, 1985
The reduction of organic dihalides by low-valent metal species may occur via several mechanisms. We have examined the reactions of a macrocyclic rhodium(1) supernucleophile, 1, with a variety of vicinal and 1,3-dihalide substrates. Unlike typical one electron reductants, 1 does not induce elimination of 1,3-dihalopropanes to give olefins or cyclopropanes; rather, normal alkyl-Rh(II1) oxidative-addition products are isolated. (When the two halides are identical, the expected statistical mixture of ”mono” and ”bis” adducts is obtained, although reaction in the presence of undissolved 1 can skew the result in favor of the “bis”adducts.) In contrast, the reaction of 1 with vicinal dibromides invariably gives olefiis and the Rh(II1) dibromide, the only partial exception being 1,2-dibromoethane, from which a “bis” adduct is formed in ca. 50% yield. Reduction rates are comparable to those found for the oxidative addition of monofunctional whose reduction to cyclohexene alkyl bromides to 1, except in the case of trans-l,2-dibromocyclohexane, occurs on mixing with 1 (the cis isomer is reduced at least lo5 times more slowly, although it still reacts about 10 times faster than bromocyclohexane). The olefinic products probably result from decomposition where a concerted of Rh(III)-@-bromoa&ylintermediates, except in the case of trans-1,2-dibromocyclohexane, elimination process is proposed. Introduction The reduction of organic dihalides by metals is a wellknown route to olefins1 and cycloalkanes.1a~2Competing
one- and two-electron processes frequently limit the stereospecificity of olefin formation by this route, especially from simple vicinal dihaloalkanes. In contrast, iodidepromoted elimination is highly stereo~pecific.~More re-
(1) (a) House, H.0. “Modern Synthetic Reactions”, 2nd ed.; W. A. Benjamin: Menlo Park, CA, 1972;p 220 ff. (b) Singleton, D. M.; Kochi, J. K. J. Am. Chem. SOC.1967, 89, 6547-6555 and references therein.
(2) Kochi, J. K.; Singleton, D. M. J. Org. Chem. 1968, 33, 1027-1034 and references therein.
0276-7333/86/2305-0218$01.50/00 1986 American Chemical Society
