2570
Organometallics 1995, 14, 2570-2574
Investigation of Organolithium Dimerization on Irradiated CdS Powder Michael Lorenz and Timothy Clark* Computer-Chemie-Centrumdes Institutes fur Organische Chemie der Friedrich-Alexander- Universitat Erlangen-Niirnberg, Nagelsbachstrasse 25, D-91052 Erlangen, Germany Received January 18, 1995@
Irradiation of CdS pawder suspended in ethereal solutions of organolithium compounds, RLi, yields elemental Cd and the photooxidation products R-R. The mechanism is shown to be a radical one, the products R-R at least partly resulting from intraaggregate dimerization of radicals R' if the parent RLi is aggregated. Free solution-phase radicals have not been detected. The configuration of a benzyl anion adsorbate has been investigated using AM1 semiempirical molecular orbital calculations on a ZnS model surface. These suggest that ql- and y7-bound species compete with each other. A comparison of the reactivity of PhCH2Li and two alkyl-substituted, sterically hindered derivatives shows that the substituted benzylic species, which are +bound, react faster than PhCH2Li. We therefore propose that PhCHZLi prefers a q7 configuration on the active CdS surface. Considerable interest has been focused on the use of semiconductors in photovoltaic cells and in photoelectrosynthesis of potential fuels.' Their application in synthetic organic chemistry is, however, less common, often because of high demands on reaction conditions and photocatalysts. Only few examples of photoassisted syntheses on semiconductors are known that compete successfully with classical synthetic methods (e.g. the photo-Kolbe reaction2 and olefin oxidation3) or lead to formerly unknown organic molecules (synthesis of bidihydrofuryl~~).However, the synthetic use of such reactions is often less important than their significance as probes into the mechanisms of heterogeneous electron transfer and subsequent surface reactions or into microscopic characteristics of photosemic~nductors.~ Our studies on the photooxidation of simple organolithium compounds of CdS powder should be viewed thus. Competitive reactions of sterically different organolithium adsorbates and supporting semiempirical calculations provide information about important features of this type of heterogeneous reaction and on adsorption equilibria and mass transfer. With the exception of certain dilithio compounds6that can be oxidized directly to their radical anions by irradiation, many organolithiums are photooxidized only Abstract published in Advance ACS Abstracts, April 1, 1995. (1)(a)Fujishima, A.; Honda, IC Nature 1972,238,37.(b) Memming, R. Top. Curr. Chem. 1988,143,79and references therein. (c) Fox, M. A. Acc. Chem. Res. 1983,16,314 and references therein. (2)Kraeutler, B.; Bard, A. J. J . Am. Chem. SOC.1977,99,7729; 1978, 100,5985. (3)Kanno, T.;Oguchi, T.; Sakuragi, H.; Tokumaru, K. Tetrahedron Lett. 1980,21,467. (4)Zeug, N.;Buecheler, J.; Kisch, H. J.Am. Chem. SOC.1985,107, 1459. (5)(a) Fox, M. A.; Chen, C.-C. J . Am. Chem. SOC.1981,103,6757. (b) Hetterich, W.; Kisch, H. Chem. Ber. 1988,121,15. (c) Henglein, A. Top. Curr. Chem. 1988,143,113. (6)(a) Wilhelm, I.; Courtneidge, J. L.; Clark, T.; Davies, A. G. J . Chem. SOC.,Chem. Commun. 1984,810. (b) Lorenz, M.; Clark, T.; Schleyer, P. v. R.; Neubauer, K.; Grampp, G. J. Chem. SOC.,Chem. Commun. 1992,719. @
0276-7333/95/2314-2570$09.00/0
in the presence of a suitable electron a ~ c e p t o r .One ~ might expect photosemiconductors to be well-suited for this purpose because of reduced back electron transfer and their chemical inertness in dark reactions. Fox and Owen were the first to take up this idea. Using a doped single-crystal Ti02 electrode, they succeeded in photooxidizing fluorenyllithium and (tetraphenylcyclopentadienylllithium in a photoelectrochemical ce1L8
Results and Discussion Mechanism. The photooxidation of organolithiums can be achieved with commercially available hexagonal CdS powder and leads to dimeric products and metallic Cd (eq 1).The formation of metallic cadmium is a well2RLi
+ CdS hv R-R + Cd + Li2S
(1)
R = PhCH,, n-Bu, Ph known photocorrosion process in the absence of oxygen.g Overall, yields are moderate (Table l),although the reaction is hampered by a continuous blackening of the sulfide powder. After 6-12 h of irradiation the reaction stops and unconsumed RLi remains. The 3-9% conversion obtained, however, is adequate for our purposes and so no attempt to increase the yield was made. Care was taken to find reaction conditions that preclude all possible reactions except that described by eq 1. First, photostability of BuLi and PhLi without CdS was ensured by using filters, and second, the reaction mixture was cooled.1° The photoinduced electron trans(7)(a)Winkler, H. J. S.; Winkler, H. J. Org. Chem. 1967,32,1695. (b) Fox, M. A,; Ranade, A. C.; Madany, I. J . Organomet. Chem. 1982, 239,269. (8) Fox, M. A.; Owen, R. C. J . Am. Chem. SOC.1980,102,6559. (9)See, for instance: Meissner, D.; Memming, R.; Kastening, B. J . Phys. Chem. 1988,92,3476.Henglein, A. Top. Curr. Chem. 1988,143, 113. (10)(a) Photodecomposition of organolithiums: van Tammelen, E. E.; Braumann, J. I.; Ellis, L. E. J . Am. Chem. SOC.1865,87,4964. Glaze, W. H.; Brewer, T. L. J. Am. Chem. SOC.1980,91,6559. (b) PhCHzLi slowly decomposes in the dark a t room temperature but not at -30 "C.
0 1995 American Chemical Society
Organometallics, Vol. 14, No. 5, 1995 2571
Organolithium Dimerization on CdS Powder
Table 1. Consumption and Reaction Conditions for Different RLi Species RLi PhCHzLi
conditionsb
9
6 h, ,I> 340 n m 6 h, ,I=. 340 n m 10 h, 1 > 410 n m
5 3
+ - R-R + Li 2Li + CdS - Li,S + Cd
R'ad/solv RLi,,sol,
consumptionu (%)
n-BuLi PhLi
Scheme 2
(7)
(8)
fer from RLi to CdS profits from an enormous electromotive force, and the ratio of the yields in Table 1 is due partly to a lower RLi oxidation potential on going from PhCHzLi (-1.43 V) and BuLi (-1.41 V) to PhLi (-0.34 V).ll In the cases of R = Bu, Ph we propose the mechanism given in Scheme 1. With light of wavelengths longer
reaction 7 proceeds on the surface, it would be facilitated by a reduced carbon-lithium binding energy of adsorbed RLi. To distinguish between the two possibilities (eqs 4/5 versus eqs 7/8), we performed a competitive reaction using PhCHzLi and BuLi in a molar ratio of 1.0:2.5. In this case, nearly equal amounts of benzyl and butyl radicals were formed, as can be deduced from the quantitative analysis of the three reaction products bibenzyl, 1-phenylpentane, and n-octane, which occurred in a molar ratio of 1.2:1.0:1.3(Scheme 3).13 If the
Scheme 1
Scheme 3
The fraction of RLi that reacts to give R-R. 150 W highpressure mercury lamp, temperature -30 "C, EtzO.
PhCH,Li
+ BuLi Chv~
S
PhCH,'
PhCH,CH,Ph
CdS
+ 2eWcb+ 2Li'
-
Cd
+ LizS
(5)
than 340 and 410 nm, respectively, the initial process must be the creation of electron-hole pairs in the semiconductor followed by the oxidation of RLi by the positive holes. The primary oxidation products, RLi'+, have been investigated previously. Ab initio calculations suggest that the n-propyllithium radical cation is a quite stable intermediate in the gas phase, Li+ dissociation being endothermic by 23.3 kcal/mol.lZ However, taking into account that Li+ is much better stabilized by solvation than by complexation with R', we cannot regard R L P as a stable species in solution. It dissociates immediately, and, as a consequence, radicals R' are involved in the subsequent reactions. This is shown by the fact that traces of the coupling products 1 and 2 were found for R = Bu. Their formation can only be explained if one assumes a radical H abstraction from Et20 by R' as a side reaction (eq 6). n-Butane was not observed directly. No byproducts were found when the reaction was performed in nhexane. Bu'
+ EtOCH2CH3
-
Bu-H + EtOCHCH3
k04
(6)
Bu
A0J0- 1
2
The final products R-R result from a simple dimerization of R' (eq 4). However, the R' concentration in solution is low and an excess of RLi is present throughout the reaction, and so an alternative route initiated by radical attack on RLi is conceivable (Scheme 2). If (11)The values for PhLi and BuLi are not very reliable because of electrode-fouling effects, but the tendency should be described correctly. See: Jaun, B.; Schwarz, J.; Breslow, R. J.Am. Chem. Soc. 1980,102, 5741. (12)Clark, T. J. Chem. Soc., Chem. Commun. 1986,1774.
+
+ Bu'
+
PhCHzBu Bu-BU
dimeric products R-R (R-R) resulted from radical attack on RLi (eq 7), one would have expected that butyl and benzyl radicals would preferably attack excess BuLi with n-octane and 1-phenylpentane as the main products, unlike the experimental results. Hence, it follows that the reactions shown in Scheme 2 do not contribute significantly to the overall reaction. Surprisingly, the product ratio does not fit the statistical distribution of 1:2:1. This is approximately expected for a simple solution-phase recombination of the butyl and benzyl radicals, which are known t o self-terminate by diffision contr01.l~ A possible explanation is that the products R-R (R-R) are formed, a t least partly, by radical recombination on the CdS surface. Assuming that BuLi exists predominantly as a tetramer, as in its pure Et20 solution,15 it is possible that a portion of the butyl radicals formed does not escape from the adsorbed aggregate and is quenched on the surface by intraaggregate dimerization with a second butyl radical. As a consequence, the mixed dimer 1-phenylpentane is not the main product. In contrast, mixtures of PhCHzLi and m-alkyl-substituted benzyllithiums, which are both monomers in Etz0,16do not show this aggregation effect and give high yields of the corresponding mixed dimers, as expected. These reactions will be discussed in detail below. We tested for free solution-phase radicals R' by adding an excess of tetramethylethylene (TME), which is capable of quenching radicals1' but is unreactive toward organolithiums. The reaction mixture RLVI?ME/ CdS behaves as if no olefin were present when exposed to light. GC analysis gives no indication of olefin addition products. Although radical addition is slower than self-termination by 105- lo6, radical addition to TME should be able to compete if a significant amount (13)Using a 1:l mixture of PhCHzLi and BuLi, the products based on the butyl radical are slightly reduced. (14) Self-termination rates: 2kt(benzyl) = 4.0 x los M-ls-l; 2kt(npropyl) = 3.4 x lo9 M-1 s-l. Values from: Ingold, K. U. In Free Radicals; Kochi, J. K., Ed.; Wiley: New York, 1973; Vol. I, p 37. (15)Seebach, D.; Hiissig, R.; Gabriel, J. Helu. Chim. Acta 1983,66, 308. (16) West, P.; Waack, R. J.Am. Chem. SOC.1967,89, 4395. (17) Rate constant k for CH$ addition to TME: k = 0.9 x lo4; see: Tedder, J. M.; Walton, J. C. Adu. Phys. Org. Chem. 1978,16, 51.
Lorenz and Clark
2572 Organometallics, Vol. 14, No. 5, 1995
of R'desorbs from the surface. The absence of a reaction between R' and TME, therefore, indicates that free solution-phase radicals do not play an important role.18 For R = PhCH2, Scheme 1 may be extended by a further mechanism, which takes into account that both CdS and PhCH2Li absorb in the blue region. Initially, PhCH2Li is excited to its SI state, a potent reductant, which transfers an electron to CdS (Scheme 4).19 A
Scheme 4 PhCH2Li hv PhCH2Li* 2PhCH2Li*
+ CdS - BPhCH,' + Cd + Li2S
(9) (10)
remaining radical dimerizes as described. If we keep in mind that the higher radical yield for PhCH2Li in comparison with that for BuLi presumably does not ensue from a very different electromotive force (Table 11, the operation of an alternative mechanism for PhCHzLi according to Scheme 4 may provide an explanation for the different reactivity. So far, a decision in favor of one of the routes discussed for PhCH2Li (Scheme 1 versus Scheme 4) is not possible. Semiempirical Adsorbate Model. Little is known about the steric arrangement of anionic organic adsorbates. Benzylic species prefer a 7 7 configuration on Pt(111)20 and on a Rus cluster.21 Here, semiempirical calculations may shed some light on the configuration of the benzyl aniodCdS adsorbate. Problems arise in the search for a model surface which fits reality best. PM3,22the only semiempirical method with Cd parameters, was not suitable, because it shows no tendency to bind the benzyl anion in a r7 way t o a model surface; therefore, we chose AM123and contented ourselves with a stable surface of the isomorphous ZnS, the (1010)layer of hexagonal wurzite. ZnxSxclusters have already been investigated using ab initio MO theory in order to find a suitable surface model for ZnCl2 chemi~orption.~~ We preferred a cluster with a suitable charge distribution in the vicinity of the site of adsorption. Therefore, all metal ions were 3-fold coordinated with sulfur ions. This made it necessary to consider a second layer and saturate with H2S molecules (Figure 1). Unfortunately, AM1 optimization strongly distorted the chosen cluster; therefore, we used experimental values for the lattice parameters25instead of optimized data. According to the rigid lattice the lattice parameters were (18)In addition, Kisch and Kiinneth recently suggested that the photodimerization of dihydrofurans also proceeds on the semiconductor surface. See: Kisch, H.; Kiinneth, R. In Photochemistry and Photophysics; Rabek, J. F., Ed.; CRC Press: Boca Raton, FL, 1991; Vol. IV, p 131. (19) Photoinduced electron transfer from excited adsorbates to CdS has already been described by: Watanabe, T.; Takizawa, T.; Honda, K. J. Phys. Chem. 1977,81, 1845. (20)Avery, N. J. Chem. Soc., Chem. Commun. 1988,153. (21) Bullock, L. M.; Field, J. F.; Haines, R. J.; Minshall, E.; Smit, D. N. J. Organomet. Chem. 1986,310, C47. (22) Stewart, J. J. P. J. Comput. Chem. 1989,10, 209, 221. (23) Dewar, M. J. S.; Zoebisch, E. G.; Healy, E. F.; Stewart, J. J. P. J.Am. Chem. SOC.1986,107, 3902. (24) Lindblad, M.; Pakkanen, T. A. J.Comput. Chem. 1988,9,581. (25) Gmelins Handbuch der Anorganischen Chemie; Verlag Chemie: Weinheim, Germany, 1959; Cd Suppl., p 595. Ibid., Zn Suppl., p 916. (26) Somorjai, G. A. In Bonding Energetics in Organometallic Compounds; American Chemical Society: Washington, DC, 1990; p 218.
(b)
(C)
Figure 1. The (1010) surface model molecule Zn,Sn.6HzS: (a) from above; (b) from the side with two layers ( n = 12);(c) from the side with three layers ( n = 15). Metal ions are black, and the sites of adsorption are marked. For wurzite the lattice constant a = 3.84 A, and for a-CdS, a = 4.14
A.
; I
(b)
165'
I 1 1
4 I
160' I I
(C)
Figure 2. Details of the AM1-optimized 7' and r7 configurations of PhCH2-/Zn&24H2S: (a) vl, a = 4.14 A (for a = 3.84 A, the benzyl anion structure is nearly identical); (b) r7,a = 4.14 A; (c) r7,a = 3.84 A. also held constant during the optimization of the adsorbed benzyl anion. The difference between the calculated adsorption energies of the q1 and v7 configuration^^^ (Figure 2) of the benzyl anion/Zn,Sn.6H2S complex is a function of the number of layers and of the lattice constant a (Table (27)q7 means that PhCH2- lies flat over the surface; strictly configuration. speaking, the calculated '3'" structure has a q1 (u):q'(n)
Organolithium Dimerization on CdS Powder
Organometallics, Vol. 14, No. 5, 1995 2573
Table 2. AM1-Calculated Difference of q1 and q7 Adsorption Energies of the Model Adsorbate Complex Benzyl Anion/Zn&.6HaS as a Function of Lattice Constant a and Number of Layers lattice paramsn
product distribution which indicates that about twice as many radicals are formed from 4 than from benzyllithium (eq 11).Furthermore, when the two mixtures
E.d(n7) - E.d(nl) (kcaUb
-
~
a = 3.84A (wurzite) n = 12 (two layers) a = 4.14A (a-CdS) n = 12 (two layers) a = 4.14A (a-CdS) n = 15 (three layers)
+
11.1
PhCH,Li 4-Li 1.O:l.O PhCH,' 4' 1.0:1.8
3.0
+
0.4
See Figure 1. All calculated adsorbate complexes are (at least local) minima on the energy surface. For positive values, q1 is more strongly bound t h a n q7.
Table 3. AM1 Adsorption Energies of Alkyl-SubstitutedBenzyl Anion Adsorbate Complexes Referred to the Unsubstituted 1 1 Species
0 0.3 -0.2
-
+
+
0.4 11.2 no minimum found
For negative values, the anion is bound more strongly to the model surface than the unsubstituted benzyl anion.
2). The v7 configuration is more favorable because of reduced strain if the distance between the two binding metal ions is increased. In this case, the angle between the benzylic CH2 group and the plane of the aromatic ring is also increased, as can be seen in Figure 2. The artificial procedure of extending the ZnnSn*6H2Smolecule to the lattice constant of a-CdS therefore reduces the difference between q1 and q7 adsorption energies considerably. The further addition of a third layer renders the two configurations nearly equal in energy. Finally, one should consider that the rigid lattice model will underestimate the stability of the q7 complex, because this configuration is likely t o be more suscep. tible to the flexibility of the surface than ~ l Therefore, a preponderance of the y7-benzylanion adsorbate seems to be probable on the surface in question.28 Whether these results apply to the reactive surface as well may be deduced from a competitive experiment using PhCH2Li and an alkyl-substituted, sterically hindered derivative. Steric Effects. Bulky substituents force a benzyl anion to adopt a q1 configuration and offer the opportunity to compare the reactivity of a +bound species with that of the unsubstituted benzyl anion. Alkyl groups in the meta position are very suitable to investigate steric effects, because they have negligible influence on the electronic structure of the benzyl anion. In order to estimate the adsorption energies of the two meta-substituted derivatives m-(X,X')-PhCHZ- (3,X = X = Me; 4, X = Me, X' = t-Bu), we performed semiempirical calculations and compared the results t o those for the unsubstituted benzyl anion adsorbate complex (Table 3). Interestingly, for 3 we also found a q7 adsorbate complex that is a local minimum and lies minimum, whereas 4 only about 11 kcal above the exists in a r1 configuration as expected. Irradiating a 1:l mixture of PhCHzLi and 4-Li, which is definitely a +bound species, with CdS gives a (28) Here we emphasize that we intended to investigate whether a q7 adsorbate is energetically feasible on a certain surface. We are not
aware which surface plays the active part in the photoreaction.
+
PhCH2Li 3-Li 1.O:l.O PhCH,' 3' 1.0:2.0 3-Li 4-Li 1.o:l.o
~
X=X'=H X = X' = Me (3) X = Me, X' = t-Bu (4)
Scheme 6
-
dimeric products (11)
-
dimeric products (12)
- 3'+ 4' 1.0:0.8
dimeric products (13)
PhCHzLV3-Li and 3-LV4-Li are irradiated under the same conditions, radical ratios of 1:2.0 (eq 12) and 1:0.8 (eq 13) (Scheme 5) are found. These results clearly indicate that meta alkyl substitution increases the reactivity of benzyllithiums toward CdS and that methyl and tert-butyl groups behave nearly equivalently. It seems reasonable (also with the AM1 results in mind) to accept a rJ, configuration for both 3 and 4. Probably, the small difference in the radical yields of eq 13 can be explained by a somewhat lower adsorption energy of 4 because of the stronger repulsive effect of the tertbutyl The results of eqs 11and 12 are less easy t o explain. Assuming that both the benzyl anion and its m-alkylsubstituted derivatives are $-bound, it is difficult to see why the latter should react Bo much better. Alternatively, the benzyl anion differs from 3 and 4 in that it adopts a r7 configuratino on the surface. We are then forced t o assume that a +bound benzylic species is more reactive than a y7-boundone. As a possible clue to the problem, it should be noted that the metal ions of the surface have a strong polarizing effect on the electron density of the anion. A v7-benzylicanion is both u- and n-bound to metal ions and resembles a naked anion with a more diffuse electron density, whereas a q1 species is strongly influenced by one a-bound counterion, leading to a charge concentration on the benzylic carbon atom, which therefore should be oxidized more easily. However, AM1 calculations do not show significant differences in the charge distribution of the rl-and v7-bound benzyl anions. Possibly the rigid 1010 ZnS surface is not the best choice to test for this effect. Calculations on the benzyl anion adsorption on different ZnS surfaces are under way and will be discussed in the future.
Conclusion The photoredox reaction between organolithium compounds RLi and CdS powder is somewhat unusual within the increasing class of chemical conversions with irradiated semiconductors in that in lacks a reducible solution-phase species. The semiconductor acts both as (29)This effect is not reflected by the AM1 results in Table 3, because the model surface is flat, and even meta tert-butyl groups of a q 1 adsorbate do not interfere with the surface. However, this is surely not valid for rough surfaces or surfaces with impurities, which prevail in reality.
2574 Organometallics, Vol. 14,No. 5, 1995
a sensitizer (at least in the reaction with BuLi and PhLi) and as a reactant, decomposing to elemental Cd in the course of the reaction. This complicates investigations on mass transfer or adsorption equilibrium because the surfaces are gradually decomposed to, or poisoned with, Cd metal. Nevertheless, we can conclude that radicals R'dimerize preferably on the semiconductor surface and that intraaggregate dimerization of R' dominates if the parent compound RLi is aggregated. The interesting result that alkyl substitution in the meta position of PhCHzLi increases the reactivity can be explained by assuming two different adsorption configurations. We believe that bulky groups in the meta position force the benzyl anion to switch from a r7 to a r1 configuration and that the latter is more reactive because of a higher oxidation potential.
Experimental Section and Calculational Methods All irradiations were carried out in a n acetone bath with a Schlenk tube and the coolingjacket of a high-pressure mercury lamp (Hanau TQ 150 W) placed at a distance of 10 mm. The reaction mixture in the Schlenk tube was stirred with a magnetic stirrer. An Ultra Cryomat (Lauda K 120 W) was connected to the cooling jacket, both cooling the lamp and tempering the acetone bath (between room temperature and -70 "C). An immersion lamp apparatus was less suitable because Cd metal deposited on the glass in the region of the highest radiation density. All products were detected by GC analysis (HP 5890; SE54, 25 m, 0.23 mm) and identified by comparison with an authentic sample and/or GC-MS (Finnigan MAT 90). Quantitative evaluations were carried out by adding a suitable GC standard to the reaction mixture. CdS (99.999%, Strem Chemicals, hexagonal by powder diffraction) was degassed, dried in a drying apparatus for 5 h at 0.05Torr and 100 "C, and kept under argon. Its suspension in Et20 was activated in an ultrasonic unit for 15 min immediately before it was used. Benzyllithium and its alkylsubstituted derivatives 3-Li and 4-Li were obtained as orange, crystalline products by the method of Boche et aL30 The ethereal solutions should be stored at -78 "C and used fresh, because small amounts of the corresponding bibenzylic species gradually emerged after several days. For determinations of the anion concentration and tests for decomposition products, small amounts of the benzyllithium solutions were quenched with trimethylchlorosilane ((TMS)Cl)and washed with water and a GC standard was added for quantitative GC analysis. PhLi was obtained by the method of Seebach and B a ~ e rand ,~~ n-BuLi was purchased as a 1.6 N solution from Aldrich. Et20 was refluxed over Na-K alloy under argon before it was used, and all reactions were performed under an argon atmosphere. Photostability without CdS. A 20 mL portion of an ethereal RLi solution (R = PhCHz, Bu; 0.05 m) was irradiated at -30" in an acetone bath (A 2 340 nm). For PhLi a solution of NaNOz (1 m ) in MeOWHzO (1:l)was used (A L 410 nm) instead of acetone. After 6 h of irradiation the solutions were (30) Zarges, W.; Marsch, M.; Harms,K.; Boche, G. Chem. Ber. 1989, 122,2303. (31)Bauer, W.;Seebach, D. Helu. Chim.Acta 1984,67,1972.
Lorenz and Clark quenched with (TMSICl and washed with water. GC analysis shows no consumption of the corresponding RLi. Stability against CdS in the Dark. The organolithium solutions in question were mixed with an excess of CdS and stirred for 6 h at -30 "C in the dark. Again, no reaction was detectable. Photodimerization of RLi. Into 30 mL of a cooled, ethereal suspension of 200 mg of CdS (1.4 mmol) in a 100 mL Schlenk tube was syringed 1.4 mmol of RLi (as a solution in Et20 or hexane) under argon. After 6 h of irradiation (10 h for PhLi) at -30 "C the black suspension was cooled to -78 "C and an excess of (TMS)C1was added. The cooling bath was removed, and the mixture w2s stirred for l/2 h. After washing with water and separating the organic layer, a fixed amount of a GC standard was added and the products were submitted to GC analysis. Yields: 0.120 mmol of bibenzyl for R = PhCH2, 0.073 mmol of n-octane for R = Bu, and 0.034 mmol of biphenyl for R = Ph. In the case of R = Bu, traces of 1 and 2 were detected by GC-MS. Cadmium metal was detected by filtering off the solids after quenching and washing with water. Treatment with dilute NaHS04 solution led t o gas development, and the black suspension became yellow. The solution was filtered off and treated with dilute NazS solution to give a yellow CdS suspension. Irradiation with Tetramethylethylene ("ME). To a mixture of 1.4 mmol of CdS, 0.7 mmol of PhCHzLi, and 30 mL of Et20 was added 0.5 mL (4.2 mmol) of TME. Half of the suspension was quenched before and the other half after 6 h of irradiation at -30 "C. GC analysis showed that both samples contained equal amounts of TME. Bibenzyl was the only reaction product detected. CompetitiveReaction between PhCHLi and BuLi. A suspension of 0.4 mmol of PhCHzLi, 1.0 mmol of BuLi, and 1.4 mmol of CdS in 30 mL of Et20 was irradiated under the conditions described. The product mixture consisted of 0.12 mmol of bibenzyl, 0.10 mmol of 1-phenylpentane, and 0.13 mmol of octane. Competitive Reaction between Benzylic Species. As an example, reaction 11 is described: a mixture of 0.9 mmol of PhCHZLi, 0.9 mmol of 4-Li, 1.4mmol of CdS, and 30 mL of Et20 yielded, when irradiated for 6 h at -30 "C, 0.012 mmol of bibenzyl, 0.042 mmol of 1-(3-tert-butyl-5-methylphenyl)-2phenylethane, and 0.039 mmol of 1,2-bis(3-tert-butyl-5-methylpheny1)ethane. Calculations. Semiempirical calculations used the AM1 Hamiltonian with the VAMP 4.3 program32on a Convex C220. Zn parameters were those of Dewar and mer^.^^ Adsorption energies Eadwere obtained from the calculated heat of formation of the absorbate complex minus the sum of the heats of formation of the described model surface and of the corresponding free gas-phase anion. As solvation effects were omitted, we confined ourselves to present relative values of E a d Only.
Acknowledgment. We are indebted to Prof. H. Kisch and R. Kiinneth for helpful discussions and to D. Schaefer for carrying out all GC-MS analyses. OM950038Y (32)Based on AMPAC (QCPE 539); vectorized and adopted to Convex C220, written by T. Clark. (33) Dewar, M. J. S.; Merz, K. M. Orgammetallics 1988,7, 522.