4 Reaction Mechanisms Part (iii) Free-radical Reactions By D. GRILLER Division of Chemistry National Research Council of Canada Ottawa Ontario Canada KIA OR6 1 Synthesis Giese and his colleagues have made good use of known chemistry to effect carbon- carbon bond formation.’ For example a yield of 87% (gas chromatography) of (1) was obtained when RX was t-butyl iodide and the olefin was acrylonitrile (Scheme 1). Tributyltin hydride was only required in sub-stoicheiometric amounts since it was regenerated in situ by the simple expedient of having sodium borohydride present in the reaction mixture. R.+ \/ I IC=C + R-C-C. /\ I I I I I 1 R-C-C. + Bu3SnH -+ I 1 I 1 R-C-C-H + Bu,Sn. (1) Bu,Sn. + RX -+ Bu,SnX + R-Scheme 1 An elegant use of stannyl radicals was made by Baldwin et al.,’ who devised an addition-elimination sequence for carbon-carbon bond formation (Scheme 2).This method avoids the risk of telomerization inherent in approaches that rely upon simple radical addition to an olefin as a propagation step of 8 chain reaction. R R*+ -Bu,Sn C0,Et Bu,Sn C0,Et I R Rx etc. -Bu,Sn-+ C0,Et Scheme 2 ’ B. Giese J. A. GonzLlez-G6mez and T. Witzel Angew. Chem. Int. Ed. Engl. 1984 23 69. ’ J. E. Baldwin D. R. Kelley and C. B. Ziegler 1. Chem. Soc. Chem. Commun. 1984 133. 69 70 D. Griller Three further communication^^-^ have appeared on the use of thiohydroxamic- 0-esters as synthetic sources of alkyl radicals. One of them' describes a method for the radical decarboxylation of N-protected a-amino-acids after first converting them into their N-hydroxypyridine-2-thioneesters (Scheme 3).This method is a slight variation on previous work6 where tributyltinhydride was used as a hydrogen donor instead of t-butyl thiol. +*SBu' + +RCO,* NS N S-SBu' I 0 I co I R Scheme 3 Silylmethyl radical cyclizations have been used to convert allylic alcohols into 1,3-diols ~tereoselectively,~ (Scheme 4). In general the 5-exo mode of cyclization was dominant and in some instances exclusive. Isolated yields were normally greater than 75YO. I AH :vH R ,BuiSnH Si Si HO OH Me/ \Me Me/\Me Scheme 4 2 Mechanism 1984 saw few striking developments in the area of free-radical reaction mechanisms although many workers continued to develop existing themes.Some showed con- siderable entrepreneurial skill in convincing editors to continue publishing these extensions as urgent communications. D. H. R. Barton and D. Crich J. Chem. SOC,Chem. Commun. 1984 774. D. H. R. Barton D. Crich and W. B. Motherwell J. Chem. SOC.,Chem. Commun. 1984 242. 'D. H. R. Barton Y.HervC P.Potier and J. Thierry J. Chem. SOC.,Chem. Commun. 1984 1298. D. H. R. Barton D. Crich and W. B. Motherwell J. Chem. SOC.,Chem Commun 1983 939. 'H. Nishiyama T. Kitajima M. Matsumoto and K.Itoh J. Org. Chem. 1984,49 2298. Reaction Mechanisms -Part (iii) Free-radical Reactions 71 Two have appeared in defense of the hypothesis that the sigma and pi states of the succinimidyl radical can be differentiated on the basis of their chemical reactions.The arguments for and against are now so intricate and conflicting that it is almost impossible to make an objective judgement on the subject. However several simple facts emerge. First results from one laboratory are not reproduced in another. This is not entirely surprising since much of the debate rests upon the presence or absence of small quantities of P-bromopropionyl isocyanate an extremely moisture-sensitive product which is formed as the result of ring-opening of the succinimidyl radical (Scheme 5). Secondly the reaction mixtures are frequently heterogeneous making a precise description of reaction conditions and concentra- tions essentially impossible. Thirdly chain lengths are seldom determined so that the extent to which initiation and termination reactions contribute to the product distribution is unknown.Fourthly complete product balances are rarely undertaken. As noted in last year's report the principal players in this drama continue to ignore a key paper in the field." (2)-Br ** + OCN(CO)CH,CH -OCN(CO)CH,CH,Br (2) Scheme 5 The ephemeral capto-dative effect again eluded all attempts to prove its existence. The simple requirement for proof is that an electron-donating and an electron- withdrawing substituent at a radical centre produce a stabilizing effect which is greater than the sum of the effects produced by the individual groups. t-Butoxyl radicals were added to a series of 'capto-dative' olefins and for each case the yield of radical dimers formed was taken as a guide to the efficiency of the initial addition reaction (Scheme 6)." It was concluded that electron-rich olefins were the most efficient scavengers of t-butoxyl radicals although a complete interpre- tation of the data was reserved for a future publication.The experiments were generally carried out using photolysis at 20°C to decompose the peroxide. Yields were low ca. 30% and the basis for their calculation was not given. Moreover there were few experimental details so that it was impossible to ascertain if the starting olefins and/or products were stable under the photolysis conditions. Without this information interpretation in terms of the capto-dative and frontier molecular- orbital approaches hinted at in this work must be treated with caution.H X X \/ / Bu'O.+ C=C + Bu'O-CH,-C. H' \Y 'Y (3) (3) + (3) -(3)-(3) Scheme 6 P. S. Skell and S. Seshadri J. Org. Chem. 1984,49 1650. P. S. Skell J. Am. Chem. SOC.,1984 106 1838. 10 A. G. Davies B. P. Roberts and J. M. Smith J. Chem. SOC.,Perkin Trans. 2 1973 2221. " S. Mignani Z. Janousek R. Merenyi H. G. Viehe J. Riga and J. Verbist Tetrahedron Lett. 1984 25 1571. 72 D. Griller In similar vein relative rate-constants for radical addition to diphenylethylenes were interpreted in terms of a capto-dative interaction.” The results were analyzed in terms of a Hammett plot. Taken as a whole the 11 data points for a range of substitution patterns showed a fair correlation with the sum of substituent constants.However two capto-dative substituted olefins were cited as being unusually reactive. By removing these from the graph it became possible to fit the remaining nine points to two separate correlation lines and claims were made that the quality of this revised fit was proof that the mechanism changed with substituent. However the total spread in relative rate-constants was only a factor of six. There was no description of the experimental work nor were any estimates of error given making all of the conclusions somewhat suspect. A serious attemptt3 was made to measure the combined effect of capto-dative substituents by taking the rotational barriers in ally1 radicals to be a measure of the radical stabilization energies.Some values were CH,=CHCHD (15.7 f l.O) CH,=CHCH(OCH,) (14.3 f 1.4) CH,=CH-eHCN (9.8 f l.l) and CH,=CH-C(OCH3)CN (6.0 f 0.4) (kcal mol-I). Despite the care with which the experiments were carried out it is clear that the combined stabilizing effect of a ‘capto’ and a ‘dative’ group is within experimental error equal to the sum of the effects provided by the groups individually. Hence there was no evidence for a substantial synergistic interaction. Photolysis of the peroxydisulphate dianiont4 was shown to be an effective source of SOT particularly when acetone was present in the solvent mixture to act as a photosensitizer. The radical was an effective oxidizing agent. For example when sulphides were present as substrates dimer radical-cations were observed as products (Scheme 7).S04T+ R2S -SO,,-+ R2S’ R,S? + R,S -R2StSR2 Scheme 7 Zimmt Doubleday and Turro” investigated the magnetic field dependence of CIDNP signals obtained during the photolysis of dibenzyl ketones in micellar media. The photolysis produced pairs of radicals which were effectively trapped within the micelles. The singlet-to-triplet To,energy gap for these pairs was determined from the magnetic field dependence of the polarization. This exhibited a maximum when there was efficient S-T- mixing. Analysis of the data suggested that the energy gaps were of the order of .03 cm-’ and as expected decreased with increasing micellar size. The experimental technique was both simple and effective. Samples were flowed through the field of a small electromagnet where the photolysis was carried out and then onto an n.m.r.spectrometer where the polarization was detected. 12 F. Lahousse R. Mertnyi J. R. Demurs H. Allaime A. Borghese and H. G. Viehe Tetrahedron Lett. 1984 25 3823. 13 H. Korth P. Lommes and R. Sustmann J. Am. Chem. Soc.. 1984 106. 663. 14 M. J. Davies B. C. Gilbert and R. 0.C. Norman J. Chem. SOC.,Perkin Trans. 2 1984 503. 15 M. B. Zimmt C. Doubleday jun. and N. J. Turro J. Am. Chem. Soc. 1984 106 3363. Reaction Mechanisms -Part ( iii) Free-radical Reactions 73 3 Structure Wendt and Hunziker16 obtained the gas-phase U.V. spectra of a variety of alkyl radicals including ethyl i-propyl s-butyl and t-butyl. The radicals were generated by mercury photosensitization and were detected by modulation spectroscopy.The pattern of bands agreed with ab initio predictions for the 3s 3p and 3d Rydberg transitions. The spectra became increasingly red-shifted with successive methyl substitution at the radical centre. This observation was consistent with the trend in ionization potentials for the radicals. The experimentally difficult technique of time-resolved microwave dielectric absorption was used to measure the dipole moment for the benzylperoxy radical.” The dipole moment was found to be 2.4 f0.2 D. Since the benzyl radical is thought to have a dipole moment of 0.2 D in the opposite sense it was concluded that simple alkylperoxyl radicals should have dipole moments of ca. 2.6 f 0.2 D. This value falls between those for alkyl chlorides and cyanides and indicates that peroxyls should be fairly hydrophilic; a property that may be of importance for oxidations carried out in vesicles or micelles.The e.s.r. spectrum of the bromoalkyl radical (Scheme 8) was interpreted in terms of a rapid migration of the bromine atom between the two central carbons at 100 K.’* At 77 K this motion was thought to be effectively frozen. The rapid migration of the bromine atom was consistent with conclusions drawn from earlier mechanistic studies which were carried out in solution at room temperature. However a complete analysis of the spectrum depended upon the interpretation of hyperfine splittings which were only discernible after resolution-enhancement techniques were applied and they were therefore interpreted with some caution.Br Br Me,C-&Me Me,L-CMe Scheme 8 The need for care in applying the radiolysis method for the generation of radicals was emphasized in a corrigendum” withdrawing the assignment of a spectrum to the p-benzoquinone radical cation. The signal detected was actually due to an impurity in the sample. An analysis of p-hyperfine uH@splittings in the 3-methyl-3-phenylbut-1 -yl radical has been reported,20 which is more important from the point of view of the underlying principle involved than for the specific example in question. It was shown that the simple equation (1) is frequently used in error. This equation relates uH@ to two aH8 = A + B cos‘ 8 (1) constants (A and B) and the dihedral angle 0 between the C-H bond and the orbital containing the unpaired electron.Errors in applying the equation have arisen because the observed value of aH@ is an average over quantum states and corresponds 16 H. R. Wendt and H. E. Hunziker J. Chem. Phys. 1984 81 717. 17 R. W. Fessenden A. Hitachi and V. Nagarajan J. Phys. Chem. 1984,88 107. 18 S. P. Maj M. C. R. Symons and P. M. R. Trousson J. Chem. SOC Chem. Commun. 1984 561. 19 H. Chandra and M. C. R. Symons J. Chern. Soc. Chem. Commun. 1983 1044. 20 K. U. Ingold D. C. Nonhebel and T. A. Wildman J. Phys. Chern. 1984 88 1975. 74 D. Griller to an average over cos’ 8 i.e. (cos28) which is not equal to cos’ (8). It is therefore not possible directly to apply equation (1) to obtain physically meaningful values of the average value of 8 from the hyperfine splittings.In fact by assuming the correctness of equation (l) data for the P-hyperfine splittings were used to derive the potential function for the rotation of the radical in question.” Studies of radical cations derived from conjugated systems have continued,21-26 and perhaps the most interesting of these26 deals with that derived from pentamethyl- cyclopentadiene. This radical cation was generated by photolysis of trifluoroacetic acid solutions of its parent hydrocarbon. The hyperfine coupling constants were consistent with values calculated by substituting Huckel coefficients into the McCon- nell equation. The hydrogen on C-1 which lies in the nodal plane of the orbital containing the unpaired electron was correctly predicted to have a very small hyperfine splitting (aH = 0.16 mT).The e.s.r. spectrum of the gallane radical anion has been obtained by photolysis of tetrahydrofuran solutions containing tetra-n-butylammoniumtetrahydrogallate and di-t-butyl peroxide (Scheme 9).27The isotropic gallium hyperfine splittings showed that the radical was pyramidal and that the orbital containing the unpaired electron had ca. 16% s-character. The temperature-dependence of the splittings was extremely small indicating that there was a substantial barrier to the umbrella inversion of the radical. Bu‘OOBu‘ -% 2Bu‘O-Bu‘O. + H,Ga-__* Bu‘OH + H,GaS Scheme 9 The opticai absorption spectra of diarylphosphonyl radicals were detected on flash photolysis of acylphosphine oxides28 and had absorption maxima at 330 nm.Since dialkylphosphonyl radicals absorb below 310 nm the use of aryl groups clearly served to red-shift the spectra to an experimentally accessible range of wavelengths. However a more careful reading of the published literature on these radicals would have prompted the authors to check their assignments by olefin benzene or alkyl bromide quenching of the transient absorption spectra. 4 Kinetics Work has continued on persistent peduoroalkyl radical^.'^ These were generated by radiolysis at 77 K of perfluoroalkane fractions and had general formula C,F2,+1. On warming to 300 K the concentration of radicals dropped to 10-70% of the value initially detected in the frozen solution.However the residual radicals showed no ” J. L. Courtneidge A. G. Davies T. Clark and D. Wilhelm J. Chem SOC.,Perkin Trans. 2 1984 1197. 22 Q. 3. Broxterman H. Hogeveen and R. F. Kingma Tetrahedron Letf. 1984. 25 2043. 23 J. L. Courtneidge A. G. Davies E. Lusztyk and J. Lusztyk J. Chem. Soc. Perkin Trans. 2 1984 155. 24 J. L. Courtneidge and A. G. Davies J. Chem. Soc. Chem. Commun. 1984 136. 2s J. L. Courtneidge and A. G. Davies BulL SOC.Chim. Belg. 1984 93 329. 26 J. L. Courtneidge A. G. Davies and S. N. Yazdi J. Chem. Soc. Chem. Commun. 1984 570. 27 J. C. Brand and B. P. Roberts J. Chem. SOC Chem. Commun.,1984 109. T. Sumiyoshi A. Henne P. Lechtken and W. Schnabel 2.Naru$orsch. Teil A 1984,39 434. 29 S. R. Allayarov S. V.Demidov D.P. Kiryukhin A. I. Mikhailov and I. M. Barkalov DOH.Akad. Nauk SSSR 1984 214,91. Reaction Mechanisms -Part (iii) Free-radical Reactions 75 signs of decay in the absence of oxygen but were immediately destroyed when oxygen was admitted. The persistence of these radicals was ascribed to steric protection of the radical centres. The diphenylketyl radical was formed in its excited state by photolysis of phenyl- benzoin using an intense laser pulse.30 The fluorescence of the excited state was quenched by a variety of electron acceptors at rates close to the diff usion-controlled limit. The quenching reactions for the excited state were several orders of magnitude faster than the corresponding processes for the ground-state radical. The excited-state reactions were therefore thought to be electron-transfer processes which were ther- modynamically viable by virtue of the additional energy of the excited transient.The reaction products were not investigated. Absolute rate constants for the reactions of the cyclopropyl radical have been measured by a flash photolysis technique which makes use of photolysis of bis(cyc1o- propylformyl) peroxide as the radical prec~rsor.~' The rate constant for the reaction with P-methylstyrene was 2 x lo6 M-'s-' at 298 K. This substrate was used as a probe of the lifetime of the cyclopropyl radical when other substrates were added as quenchers. Rate constants for a variety of C-H abstractions by cyclopropyl were ca. lo6 M-' s-l and were fairly insensitive to the strength of the bond being broken.They varied by about an order of magnitude when the latter changed by 20 kcal mol-'. Laser flash photolysis was also used to measure rate constants for the reactions of tributyl-stannyl and -germyl radicals with carbonyl compounds and organic halides.32 The y-scission of a series of P-peroxyalkyl radicals has been investigated (Scheme Rate constants for scission were ca. lo6s-' at 298 K. Variations reflected the conformational requirement for the carbon-oxygen bond to be aligned with the axis of the p-orbital containing the unpaired electron. It was also that this scission should contribute to the thermal decomposition of di-t-butylperoxide at room temperature since the t-butoxyl radical formed should abstract hydrogen from the parent peroxide to propagate a chain reaction.The normal rate of peroxide thermoly- sis at room temperature should be sufficient to initiate this chain decomposition and lead to degradation of this peroxide within a few weeks. In fact di-t-butyl peroxide has a long shelf-life. It was concluded that this stability was due to-impurities of hydroperoxide dissolved oxygen and the oxirane itself all of which serve as chain-breaking inhibitors. Quantum mechanical tunnelling in hydrogen-transfer reactions has been treated by a combination of theory and e~periment.~~-~* The reaction of methyl radicals *C-C-0-OR 4 C+ + *OR '0' Scheme 10 30 H. Baumann C. Merckel H.-J. Timpe A. Graness J. Kleinschmidt I. R. Gould and N. J. Turro Chem.Phys. Lett. 1984 103 497. 31 L. J. Johnston J. C. Scaiano and K. U. Ingold J. Am. Chem. SOC.,1984 106,4877. 32 K. U. Ingold J. Lusztyk and J. C. Scaiano J. Am. Chem. SOC.,1984 106 343. 33 A. J. Bloodworth J. L. Courtneidge and A. G. Davies J. Chem. SOC.,Perkin Trans. 2 1984 523. 34 S. A. Davis B. C. Gilbert D. Griller and A. S. Nazran J. Org. Chem. 1984 49 3415. 35 W. Siebrand T. A. Wildman and M. Z. Zgierski 1.Am. Chem. SOC.,1984 106 4083. 36 W. Siebrand T. A. Wildman and M. Z. Zgierski J. Am. Chem. SOC.,1984 106 4089. 37 T. Doba K. U. Ingold and W. Siebrand Chem. Phys. Lett. 1984 103 339. 38 T. Doba K. U. Ingold W. Siebrand and T. A. Wildman J. Phys. Chern. 1984 88 3165. 76 D. Griller in methanol glass was taken as a model system.38 To a first approximation the radical decay was exponentially related to It was shown that this rate law could be reproduced by considering that the methyl radicals occupied a distribution of sites each of which had a characteristic rate constant for decay which depended exponentially upon r the distance between the methyl radical and methanol.The 'most probable* site was found to be that in which the methyl radical and methyl group of the alcohol were separated by a distance ro which was approximately equal to the sum of their Van der Waals radii. The experimental data were reproduced with a very narrow distribution of sites varying by only a few percent from that of the most probable. This approach is certainly the most promising and intuitively reasonable treatment of tunnelling in chemical reactions which has yet emerged.5 Thermochemistry 1984 was a bumper year for papers on thermochemistry. Lossing and Homes reported heats of formation of 18 radicals as measured from appearance energies.39 These included vinyl (AH = 64 f2 kcal mol-') and phenyl (AH = 75.8 * 2 kcal mol-') which lead respectively to carbon-hydrogen bond dissociation energies of 104 f 2 and 108 * 2 kcal mol-I. Two determination^^^*^' of the heats of formation of ethyl (28 kcal mol-I) and one of t-b~tyl~~ (9.2 kcal mol-') were in excellent agreement with the results of an earlier but much criticised solution and led to bond strengths of 100 and 94 kcal mol-' respectively. Stratt and Desjardins& have addressed the frequent criticism that thermodynamic data obtained in solution are often contaminated by solvent effects.They calculated the magnitude of the interaction between the dipole induced on vibration of the methyl radical and solvent molecules. The solvation free-energies due to the 'vibra- tional polarizability' were only 4.0 and 33.0 cal mol-' for methane and dichloromethane respectively. However the difference in the I3C hyperfine splitting as measured in the two solvents was predicted to be 0.028 mT which is about 1% of the total hyperfine splitting. This prediction and the dependence of I3C on the dipole moment of the solvent can clearly be tested by experiment. The relationship between bond dissociation energy and radical stability was reviewed in some detail.45 It was argued that the definition of radical stabilization energy E = BDE(S-X)-BDE(R-X) (where S refers to a standard moiety and R to the radical in question) depended on the nature of X and therefore was not a measure of the intrinsic stability of the radical.However the data cited for a variety of X gave fairly consistent values of E, which were within the normal range of experimental errors. In a few instances the fit for X = halogen was poor but showed no well defined trend suggesting that the experimental determinations were suspect in those cases rather than the E definition itself. The data do not support 39 J. Holmes and F. P. Lossing Int. J. Muss Spectrom. Ion Processes 1984 58 113. 40 J.-R. Coa and M.H. Back Znt. J. Chem. Kinet.1984 16 961. 41 P. D. Pacey and J. H. Wimalasena J. phys. Chem. 1984,88 5657. 42 T. S. A. Islam and S. W. Benson Znt. J. Chem. Kinet. 1984 16 995. 43 A. L. Castelhano and D. Griller J. Am. Chem. SOC.,1982 104 3655. 44 R.M. Stratt and S. G. Desjardins J. Am. Chem. SOC.,1984 106 256. 45 A. M.de P. Nicholas and D. R. Arnold Can. J. Chem. 1984 62 1850. Reaction Mechanisms -Part ( iii) Free-radical Reactions the argument attractive as it may be. Variations of the kind expected by the authors ought only to be expected when perturbation caused by an electronegative substituent shows a strong dependence on the structure of R. In an imaginative approach Nonhebel and Walton showed that a linear correlation existed between the C-H bond dissociation energy of a compound and the barrier to internal rotation (as measured by e.s.r.) of the corresponding radical.& The results suggested that literature value of the C-H bond strength in acetone (98 kcal mol-') was too high by ca.8 kcal mol-' and that BDE(RSCH,-H) = 93 kcal mol-'. This concept was extended in a study of heptatrienyl and other polyenyl radicals47 and an empirical equation was developed which related radical stabilization energies as estimate from e.s.r. rotational barriers to aH-the average of the anti-and syn-hydrogen hyperfine splittings of the terminal methyl groups. While relationships of this kind are unlikely to replace thermodynamic measurements of stabilization energies they are certainly very effective methods for screening the accuracy of the available literature data.Finally Chen and Mendenhal148 measured the heat of combustion of di-t-butyl hyponitrite by bomb calorimetry and determined its heat of vaporization by an effusion technique. These results led to the gas-phase heat of formation of -41.3 2.9 kcal mol-' and to estimates of group equivalents for oxygen-nitrogen combina- tions. This work emphasises the need for fundamental thermochemical data and the importance of simple classical techniques for their determination. 46 D. G. Nonhebel and J. C. Walton J. Chem. Soc. Chem. Commun. 1984 731. 47 I. G. Green and J. C. Walton 1. Chem. Soc. Perkin Trans. 2 1984 1253. 48 H.-T. E. Chen and E. D. Mendenhall 1. Am. Chem. Soc. 1984 106 6375.