J. Chem. Soc., Perkin Trans. 2, 1999, 937–945 937 Cascade rearrangement of spiroepoxymethyl radicals into 2-oxocycloalkyl radicals: evaluation of a two-carbon cycloalkanone ring expansion Mohammad Afzal and John C. Walton * University of St. Andrews, School of Chemistry, St. Andrews, Fife, UK KY16 9ST Received (in Cambridge) 13th January 1999, Accepted 2nd March 1999 Series of 2-bromomethyl- and 2-hydroxymethyl-1-oxaspiro[2.n]alkanes were prepared from cycloalkanones by initial Wadsworth–Horner–Emmons methodology to aVord ester-substituted methylenecycloalkanes.The latter were selectively reduced to hydroxymethylmethylenecycloalkanes which were epoxidised with peroxyacetic acid. Homolytic reactions were studied by EPR spectroscopy which enabled transient 3-oxoalk-1-enyl radicals, and their cyclisation products, 2-oxocycloalkyl and 2-oxocycloalkylmethyl radicals, to be characterised. This evidence, together with end product analyses of organotin hydride reductions of the 2-bromomethyl-1-oxaspiro[2.n]alkanes, established that the initial spiroepoxymethyl radicals rearranged by a three-stage cascade of two consecutive b-scissions followed by a cyclisation. Cyclisations of the 3-oxoalk-1-enyl radicals took place mainly in the endo-mode to aVord 2-oxocycloalkyl radicals, except for the 5-oxohept-6-enyl radical for which exo-cyclisation to generate the 2-oxocyclohexylmethyl radical was preferred.Kinetic data for the exo- and endo-cyclisations of the 4-oxohex-5-enyl radical were obtained from tributyltin hydride mediated reactions of 2-bromomethyl-1-oxaspiro[2.3]hexane. Introduction The most useful free radical cascade rearrangements are those which involve carbon–carbon bond formation, and thus multiple cyclisation sequences leading to the assembly of polycyclic compounds have received the most attention.1–3 In contrast, multiple b-scissions usually lead to degradation of the original molecular architecture, as illustrated by the comprehensive disassembly of the 3D-cage of cubylmethyl radicals by a cascade of three b-scissions producing a biscyclobutenyl structure.4,5 A fair range of unimolecular cascades combining these two rearrangement steps, i.e. b-scissions with cyclisations, has been studied.For example, b-scissions of suitably unsaturated epoxymethyl radicals followed by cyclisations led to the production of vinyltetrahydrofurans.6,7 Several of the penultimate vinyltetrahydrofuranylmethyl radicals underwent a second cyclisation to aVord 7-oxabicyclo[2.2.1]heptane derivatives.Intramolecular additions of alkyl radicals to carbonyl bonds, followed by b-scission of the resulting alkoxyl radicals have been exploited as a means of ring expansion by one, and by three or more C-atoms, but not by two C-atoms.8 The rapid b-scission of appropriately unsaturated cyclopropylmethyl radicals, followed by cyclisation, has been utilised for the production of a fledgling range of spirocyclic and polycyclic structures.Noteworthy examples employed functionalised alkynylbicyclo[4.1.0]heptanes, and the corresponding bicycloheptan- 2-ones, as precursors.9–11 To date, however, unimolecular cascades embodying two b-scissions and a cyclisation are rare and select phenomena. Certain b-scissions, notably those of alkoxyl radicals which yield ketones, do not require excessive ring strain as the driving force.12 In principle, therefore, this fragmentation could be combined with several unimolecular radical steps to achieve potentially valuable molecular reorganisation. An intriguing cascade sequence incorporating three of these homolytic steps is outlined in Scheme 1.Simple epoxymethyl radicals related to radical 2 are known to undergo very rapid b-scissions 13,14 of their carbon–oxygen bonds7,15 and hence 1-vinylcycloalkoxyl radicals 3 should form with great ease. Intermediate 3 is expected to undergo a second b-scission, at moderate temperature, to selectively aVord the more stabilised 3-oxoalkenyl radical 4, in preference to scission of the vinyl–carbon bond.Radical 4 has a choice of either ring closure in the exo-mode with production of a 2-oxocycloalkanylmethyl radical 7 or in the endo-mode to yield the 2-oxocycloalkyl radical 5 as a reorganised product of elegant simplicity. The presence of the 3-oxo group in 4 might predispose this radical to endo-cyclisation as a result of its electron-withdrawing character inducing a favourable polar eVect in the transition state and because the endo-radical 5 is stabilised by resonance delocalisation of the unpaired electron onto oxygen.This mode of the cascade is potentially of synthetic value because the overall transformation amounts to a rare two-carbon ring expansion process. A cascade closely related to this was first reported by Kim and Lee who examined the addition of organotin and phenylthiyl radicals to 2-vinylspiroepoxides. They observed mainly one-carbon ring expansion on starting with 1-oxa-2-vinylspiro[ 2.3]hexane derivatives and analogous spiro[2.4]heptane derivatives, presumably as a result of final exo-cyclisation.16 Galatsis and co-workers 17 briefly reported a product study of the organotin hydride induced rearrangement of iodomethylspiroepoxides analogous to 1.Moderate to low yields of the Scheme 1 SY = initial radical source (Y = H or halogen). O X O CH2 O CH2 O O O Y CH2 O O Y n = 1 2 4 • n n • n n • • 1 2 3 4 5 n n SY • 6 7 8 n n SY a b c938 J.Chem. Soc., Perkin Trans. 2, 1999, 937–945 two-carbon ring-expanded products (6, Y = H) were isolated for several medium ring sizes, but for the cyclopentane ring (1, n = 2) the exo-product (8, Y = H) predominated in a low yielding reaction carried out in refluxing benzene. In this paper we report our systematic study of this cascade in which we have probed the regioselectivity and kinetics of the cyclisation steps as a function of ring size and of reaction temperature and have characterised several intermediates by EPR spectroscopy. Results and discussion Synthesis of spiroepoxyalkyl precursors 1 Spiroepoxymethyl bromides (12) and alcohols (11) were chosen as the radical precursors and two flexible synthetic routes were examined (Scheme 2).Cycloalkanones were converted to unsaturated esters 9 using Wadsworth–Horner–Emmons conditions. Selective reductions to the unsaturated alcohols 10 were troublesome due to co-production, during use of LiAlH4, Red-Al or DIBAL-H, of some of the saturated analogue.However, for the unsaturated esters 9b,c with larger ring sizes, selective reduction to 10b,c was achieved cleanly with lithium aluminium ethoxide hydride.18 For 9a however, the best results were obtained by reduction using DIBAL-H and subsequent chromatographic purification of 10a. An alternative strategy avoiding this reduction was examined.Normal Wittig chemistry was used to convert cyclopentanone to the THP-protected allylic alcohol (Scheme 2) by use of the triphenylphosphonium salt of THP-protected 2-bromoethanol. Deprotection with PPTS aVorded the same allylic alcohol 10b but the overall yield was inferior to that of the first route. Epoxidations were carried out eYciently with peroxyacetic acid and the resulting epoxyalcohols 11 were converted to the spiro-bromides 12 by reaction of the corresponding mesylates with lithium bromide.EPR spectroscopic study of the cascade EPR spectra of the transient radicals generated by bromine abstraction in the temperature range 150–200 K were examined by photolysis of a cyclopropane solution containing the 4- membered spiroepoxide (12a) triethylsilane and di-tert-butyl peroxide directly in the microwave cavity of an EPR spectrometer [reactions (1)–(3)]. At higher temperatures a tertt- BuOOBu-t 1 hn æÆ 2t-BuO? (1) t-BuO? 1 Et3SiH æÆ Et3Si? 1 t-BuOH (2) Et3Si? 1 12 æÆ 2 1 Et3SiBr (3) butylbenzene solution was used in which hexamethylditin replaced the silane and peroxide.Scheme 2 i, (EtO)2P(O)CH2CO2Et, NaH, THF, reflux 5 h; ii, LiAlH3- (OEt), Et2O, 17 h; iii, AcO2H, AcOH, DCE, reflux 4 h; iv, MeSO2Cl, Et3N, DCM, then LiBr, Me2CO, reflux 1 h; v, Ph3P1CH2CH2OTHP, BuLi, Et2O, reflux 12 h; vi, EtOH, PPTS. n n 9 O CO2Et O OTHP OH OH O Br n n n n 10 11 12 i ii iii iv v vi a b c n = 1 2 4 In the temperature range 150 to 190 K a nine line spectrum (Fig. 1a) was observed with EPR parameters (Table 1) which enabled this to be reliably attributed to the primary radical 4a. As expected therefore, the b-scissions of both 2a and 3a were very rapid even at 150 K. At higher temperatures (>210 K) the spectrum of 4a was replaced by a new one containing a total of six lines, the inner two being strongly broadened (Fig. 1b). The EPR parameters were identical to those recorded in the literature 19,20 for radical 5a.This spectroscopic observation of two of the intermediates corroborates the cascade sequence and shows that cyclisation in the endo-mode dominates at these low temperatures. The intermediate from exo-cyclisation, i.e. 7a was not detected up to ca. 240 K, above which the spectra became too weak for analysis. Hydroxy- and deuteroxy-substituted spiroepoxymethyl radicals 13aH and 13aD were generated by hydrogen abstraction, with photochemically generated tert-butoxyl radicals, from alcohol 11a and the corresponding deuterium-substituted compound (Scheme 3).In both cases good spectra due to the primary ring-opened radicals 15aH and 15aD were observed (Table 1) but cyclised radicals could not be identified with certainty at temperatures up to 260 K. Thus, endo-cyclisation was slower, as would be expected for a hex-5-enyl type radical containing a substituent at the attacked centre. The spectra at lower temperatures (T < 190 K) showed the presence of an additional radical with a doublet hyperfine splitting (hfs) [a(1H) = 19.9 G, g = 2.001] which disappeared at higher temperatures.The Fig. 1 9.4 GHz EPR spectra of radicals derived from 2-bromomethyl- 1-oxaspiro[2.3]hexane 12a in cyclopropane solution. Upper spectrum shows the 4-oxohex-5-enyl radical 4a at 190 K with one line displayed under higher resolution in the inset. Lower spectrum shows the 2-oxocyclohexyl radical 5a at 235 K; note the broadening on the two inner lines. Table 1 EPR data for radicals generated from 2-bromomethyl-1- oxaspiro[2.n]alkanes a Radical 4a 15aH 15aD 4b 15bH 15bD 4c 5a 7b 5c 17bHd 17bDd T/K 155 150 150 240 235 200 235 210 290 240 280 H(a) 22.3 (2H) 22.8 (2H) 22.8 (2H) 22.4 (2H) 22.2 (2H) 22.6 (2H) 18.0 (1H) 22.6 (2H) 18.0 (1H) 17.4 (1H) 17.4 (1H) H(b) 29.2 (2H) 30.0 (2H) 30.2 (2H) 28.4 (2H) 28.8 (2H) 29.0 (2H) 34.2 (2H) c 30.4 (1H) 21.0 (2H) 17.4 (1H) 17.4 (1H) H(other) 0.55 (2Hg) 0.5 (2Hg) 0.52 (2Hg) 0.70 (2Hg) 0.75 (2H) 1.5 (OH) Tmid/Kb 205 >260 >260 <200 250 270 a Hfs in G (10 G = 1 mT), all g-factors 2.003 ± 0.001 except as noted below.b Temperature at which the cyclised and uncyclised radical are equal in concentration. c Lines broadened due to ring inversion: hfs of the 2 non-equivalent H(b) were ca. 24 and 43 G at 210 K. d g-factor = 2.0031.J. Chem. Soc., Perkin Trans. 2, 1999, 937–945 939 g-factor identifies this as an acyl radical and the doublet hfs suggests an unsaturated species of the general type >C]] CH– C?]] O, similar to MeCH]] CH–C?]] O [g = 2.0005, a(Hb) = 19.5 G] and analogous species observed by Davies and co-workers.21 A likely structure for this additional radical is 18a formed via a 1,5-hydrogen migration in radical 14a followed by b-scission of the second ring and a second hydrogen transfer as outlined in Scheme 3.The EPR spectra obtained on bromine atom abstraction from the next higher spiroepoxymethyl bromide 12b were too weak and broad at T < 200 K for definite observation of the uncyclised radical 4b.Above this temperature a well-marked spectrum of a triplet of doublets was obtained due to radical 7b, the product of exo-cyclisation (Table 1). The magnitude of a(Hb) decreased as the temperature was increased and hence, in the preferred conformation of radical 7b, the SOMO eclipses the C–Hb bond. A spectral study of hydrogen abstraction from spiroepoxymethanol 11b revealed the primary radical 15bH at 240 K accompanied by a second species with a spectrum consisting of a triplet of doublets (Table 1).The EPR parameters of the latter radical are consistent with it being radical 17bH from exo-cyclisation, and this was corroborated by observation of the deuterium isotopomer 17bD with similar EPR parameters. The lower Tmid values for 4b and 15bH indicate that cyclisation occurs significantly faster than for the shorter chain 4a and 15aH.Interestingly, however, the preferred cyclisation mode changes from endo for 4a and 15aH to exo for the next higher analogues. The uncyclised radical 4c was readily detected at temperatures below 250 K on bromine abstraction from the 7- membered ring spiroepoxymethyl bromide 12c. By 290 K this had been completely replaced by a new spectrum (Table 1) which we attribute to the endo-cyclised radical 5c, none of the product of exo-cyclisation could be detected in this temperature range.Only a single weak spectrum was obtained during hydrogen abstraction from the 7-membered ring alcohol 11c and this consisted of a pentet of doublets [a(4H) = 19.8, a(1H) = 1.8 G at 240 K]. Hydrogen abstraction from the deuteroxy analogue 11cD gave rise to a spectrum consisting of only a pentet [a(4H) = 19.8 G at 290 K] which demonstrated that the small doublet hfs was due to an OH group attached to the radical centre. The observed hfs are very similar to those of the 1-hydroxycycloheptyl radical [a(4H) = 19.0, a(1H) = 1.0 at 200 K] 22 and hence our spectrum is almost certainly that of the 1-hydroxy-3-oxocyclononyl radical 20c.It is probable that this is formed by hydrogen abstraction from the product Scheme 3 O OR CH2 O O O OR OR OR 11 t-BuO • n • n • • n 13 15 16 17 1 2 4 aH bH cH aD bD cD R = H: R = D: n = n • O R O OH O OH O • n R = H • n • O HO • O • HO n n n 14 18 3-hydroxycyclononanone 19c as shown in Scheme 4, and this is indirect evidence that endo-cyclisation predominates in this system.The EPR spectra provide good evidence in support of the 3-stage cascade and imply that the final cyclisation is the slow rate-controlling step in each case and is predominantly endo, at temperatures below 298 K, except for the 5-membered ring spiroepoxide 1b (5-oxohept-6-enyl radical 4b). The mid point of each cyclisation under EPR conditions is denoted by the temperature (Tmid) at which the concentrations of the uncyclised and cyclised radicals are equal, and these are listed in Table 1.These Tmid values are related to the rate constants for cyclisation of the 3-oxoalk-1-enyl radicals 23 and hence these decrease in the following order for the temperature range 200 to 270 K: 4b(exo) > 4a(endo) > 15bH(exo) > 4c(endo). The fastest rate is therefore exo-cyclisation of the 5-oxohept-6-enyl radical, which exceeds that of endo-cyclisation of the 4-oxohex-5-enyl radical.The slower rates for longer chain radicals, and for the hydroxy substituted radicals, are in accord with expectation. Reaction of spiroepoxymethyl bromides with organotin hydrides The photo-initiated reaction of each spiro-bromide 12 with triphenyltin hydride and/or with tributyltin hydride in benzene furnished a mixture of four main products: a cycloalkanone 22, a 2-methylcycloalkanone 26, an alk-1-en-3-one 23 and an alkan-3-one 29 (Scheme 5). The proportions of these products depended on temperature and organotin hydride concentration. The 1-vinylcycloalkanols 21 were not detected under any reaction conditions or for any of the ring sizes, except for traces of 21c identified in a low temperature (22 8C) reaction of the 7-membered ring spiroepoxy bromide 12c.It can be concluded that b-scission of all the alkoxyl radicals 3a–c must be fast in comparison with hydrogen abstraction from the organotin hydrides. The alkan-3-ones were only produced in substantial quantities when a large excess of organotin hydride was used.It is probable that they were formed from the initially produced alk-1-en-3-ones 23 by addition of a stannyl radical to the carbonyl group to generate resonance stabilised allyl type radicals 27 which then abstracted hydrogen to aVord stannyl ethers 28. The latter would hydrolyse and tautomerise to ketones 29, the overall reaction being reduction of the double bond (Scheme 5). Production of open chain ketones 23 and 29 was reduced to a very low level when gradual addition of the organotin hydride was employed. The product yields under various reaction conditions are listed in Table 2 which shows that for the 4-membered ring substrate 12a endo-cyclisation yielding cyclohexanone 22a predominated over exo-cyclisation yielding 2-methylcyclopentanone 26a at 80 8C.The predominance of 22a at lower temperatures is in gratifying agreement with the EPR results which showed radical 5a (the precursor of 22a) to be the main cyclised intermediate at low temperatures.Reductions were also carried out at 120 and 180 8C in tert-butylbenzene and hexadecane as solvent, respectively, and the [22a]/[26a] ratio decreased from 3.2 at 80 8C to ca. 0.9 at both higher temper- Scheme 4 O OR O OR O OR R = H, D 15c • 16c t-BuO • 19c 20c •940 J. Chem. Soc., Perkin Trans. 2, 1999, 937–945 atures. Clearly, the relative amounts of the products are governed by a fine balance between stereoelectronic eVects and thermodynamic factors.Rearrangement of radical 7a by intramolecular addition to the neighbouring carbonyl group via bicycloalkanoxyl radical 24a to 3-oxocyclohexyl radical 25a could be envisaged. Hydrogen abstraction by radical 25a would then produce additional cyclohexanone. If this ring expansion took place it would be expected to increase in importance at higher temperatures. Literature precedent 8 indicates that b-scission of the interring bond of bicycloalkanoxyl radicals of type 24 only occurs rapidly if the ring-enlarged radical 25 is stabilised by an adjacent electron-delocalising substituent (Ph, CO2R).In a recent paper Chatgilialoglu et al.24 reported the direct generation of radical 30 which is closely related to 7. From a careful study of the five- and six-membered ring products they established that the rate constant for rearrangement of 30 to the corresponding six-membered ring radical was given by log- (kr/s21) = 10.51 2 (39.3 kJ mol21)/2.3RT and that the reverse ring contraction was at least an order of magnitude slower. It follows that at the lower end of our temperature range this ring expansion will be negligible (kr/s21 = 4.2 × 103 at 298 K) but Scheme 5 O OH O n • n 2 O 12 R3Sn • 3 R3SnH 21 n O • 4 R3SnH n O n 23 OSnR3 endo 5 R3SnH n 22 O exo • R3Sn O • • 7 n • n • R3SnH OSnR3 24 25 27 R3SnH O n n H2O n 28 29 n = 1 2 4 a b c O R3SnH n 26 k5 k6 kH kH kH O kr • 30 Table 2 Product yields from reduction of 2-bromomethyl-1- oxaspiro[2.n]alkanes 12 with organotin hydrides a,b Substrate 12a 12a 12b 12b 12c 12c 12c Tin hydride Bu3SnH Ph3SnHc Bu3SnHc Bu3SnHd Bu3SnH Ph3SnH Ph3SnH Temp/8C 80 80 82 150 75 20 75 22 45 58 25 63 8 [20] [36] 26 14 22 50 13 44 23 28 17 53 13 [11] 29 14 — trace trace 13 a Photolytic initiation in benzene solution with 1 equiv.of organotin hydride. b Yields in mol% as determined by GC, except those in parenthesis which are isolated yields. c Yields determined by 1H NMR.d Tin hydride added in small portions. that at the upper end of the range it will make a significant contribution. By way of contrast, the 5-membered ring substrate 12b yielded 2-methylcyclohexanone 26b as the major product derived from predominant exo-cyclisation (Table 2). This was also in good agreement with the EPR spectroscopic finding of radical 7b as the major cyclised intermediate at lower temperatures, and with the major product reported by Galatsis 17 from the analogous spiro-iodide.Reduction of 12b at 150 8C indicated that the [22b]/[26b] ratio increased from 0.50 at 80 8C to 0.85 at the higher temperature. This trend is as expected from the greater thermodynamic stability of the endoradical 5b. Treatment of the 7-membered ring spiroepoxymethyl bromide 12c with organotin hydrides aVorded the products of both endo- 22c and exo-cyclisation 26c in proportions which varied with temperature (Table 2).Of these two, only cyclononanone could be isolated from larger scale reactions employing triphenyltin hydride. The chromatograms disclosed the presence of trace amounts of 1-vinylcycloheptanol 21c in this case. This is understandable because the low ring strain in the 7-membered ring alkoxyl radical 3c is expected to diminish its b-scission rate in comparison with radicals 3a,b and hence a small proportion is trapped as alcohol 21c by hydrogen abstraction from the organotin hydride.Tin hydride reactions of 12c at higher temperatures (>100 8C) showed a lot of by-products including heptane which can be attributed to degradative conversion of the substrate. Kinetics of cyclisation of the 3-oxohex-5-enyl radical 4a The EPR spectroscopic evidence and the product analyses demonstrated beyond any reasonable doubt that the slowest, rate-determining step of the cascade is the final cyclisation. A kinetic appraisal of the endo- and exo-cyclisations of the 3-oxohex-5-enyl radical 4a was made by measuring the proportions of the products, at a series of temperatures, from reactions of 12a with a fourfold excess of tributyltin hydride (Table 3). Under these conditions, the unsaturated ketone 23a was wholly replaced by hexan-3-one 29a probably arising as shown in Scheme 5.Analysis of the reaction scheme shown in Scheme 5, assuming negligible ring contraction of radical 25a, leads to relationships (4) and (5). Assuming that the conversion of 23a [29a]f/[26a]f = (kH[Bu3SnH] 1 kr)/k5 (4) [29a]f/[22a]f = kH[Bu3SnH]/{k6 1 k5kr/(kH[Bu3SnH] 1 kr)} (5) to 29a was quantitative, that the rate constants for hydrogen abstraction from tributyltin hydride (kH) by radicals 7a and 5a, 25a are equal to the literature values for primary and secondary alkyl radicals,25,26 and that kr was the same as that reported for radical 30, the Arrhenius data in eqns.(6) and (7) for log(k6/s21) = 8.7 ± 1.5 2 (20.0 ± 4.0 kJ mol21/2.3RT) (6) Table 3 Kinetic data for 3-oxohex-5-enyl radical cyclisation a T/K [29a]/[22a] [29a]/[26a] k6/105 s21 k5/105 s21 278 11.2 11.7 0.88 0.76 323 10.1 9.43 2.4 2.7 393 9.09 7.04 7.9 10.4 458 4.93 5.38 29.0 27.9 aReactions with [12a] = 0.176 mol dm23, [Bu3SnH] = 0.685 mol dm23 with photolytic initiation in benzene, tert-butylbenzene and hexadecane as solvents.Rate constants obtained from eqns. (1) and (2) as described in the text.J. Chem. Soc., Perkin Trans. 2, 1999, 937–945 941 log(k5/s21) = 8.8 ± 1.5 2 (21.0 ± 4.0 kJ mol21/2.3RT) (7) endo- (k6) and exo-cyclisation (k5) were derived from the product ratios. Neglect of the ring expansion reaction channel (kr) simplifies the kinetic equations but has only a minor eVect on the cyclisation parameters, and changes the activation energies by £1 kJ mol21 and the pre-exponential factors by £0.2 log units. The rate constant for exo-cyclisation of radical 4a (Table 3) is very similar to that for exo-cyclisation of the hex-5-enyl radical (2.5 × 105 s21 at 25 8C).25,26 However, the rate constant for endocyclisation of the 3-oxohex-5-enyl radical is very much greater than that of endo-cyclisation of hex-5-enyl (4 × 103 s21 at 25 8C).The major eVect of 3-oxo-substitution on the cyclisation is to substantially favour the endo-mode and this is logically accounted for in that the 2-oxo substituent is expected to thermodynamically stabilise the product 2-oxocyclohexyl radical.Conclusions Spectroscopic examination of the intermediate radicals, and end product analysis, paint a consistent picture in which spiroepoxymethyl radicals with a range of ring sizes undergo a rapid one-pot cascade of two b-scissions and one cyclisation. This cascade entails changes in hybridisation at five of the original atoms and yet the molecules traverse this complex reaction coordinate with apparent ease. The carbonyl substituent in the intermediate oxoalkenyl radical 4 enhances endo-mode cyclisation and this predominates, except for the 5-oxohept-6-enyl radical 4b at low temperatures. The final cyclisation is not suYciently regioselective for the cascade to be of general use as a 2-carbon ring-expansion method.However, if an alkyl substituent were incorporated at C-2 of the 1-oxaspiro[2.n]alkyl moiety this should hinder exo-cyclisation and hence augment the endo-cyclisation product. Spiroepoxides of type 1 are easily made by several alternative methods17,27 and therefore this provides some scope for incorporating the cascade into selected syntheses.Experimental 1H NMR spectra were recorded at 200 or 300 MHz and 13C NMR spectra at 75 MHz, in CDCl3 solution with tetramethylsilane (dH = dC = 0) as reference. Coupling constants are expressed in Hz. Ether refers to diethyl ether. Light petroleum refers to the fraction boiling in the range 40–60 8C. EI mass spectra were obtained with 70 eV electron impact ionisation and CI spectra were obtained with isobutane as target gas on a VG autospec spectrometer. GC-MS analyses were run on a Finnigan Incos 50 quadrupole instrument coupled to a Hewlett Packard HP 5890 chromatograph fitted with a 25 m HP 17 capillary column (50% phenyl methyl silicone).For the calculation of yields from GC data, the detector response was calibrated with known amounts of authentic materials (or close analogues) and n-dodecane, or n-heptane was added as a standard. EPR spectra were obtained with Bruker ER 200D and Bruker EMX 10/12 spectrometers operating at 9.1 GHz with 100 kHz modulation.Samples of the substrate (ca. 40 mg) in di-tert-butyl peroxide (500 ml) or in tert-butylbenzene (0.5 cm3) (occasionally cyclopropane) were degassed by bubbling nitrogen for 20 min and photolysed in the resonant cavity by light from a 500 W super pressure mercury arc lamp. EPR spectra were simulated with programs provided by Heinzer 28 and Whitwood.29 The deuteriated alcohols 11D were obtained from the protio-analogues by shaking with D2O; one drop of D2O was included in the EPR tube to ensure complete exchange was maintained during photolysis in the resonant cavity. ·-(Ethoxycarbonyl)methylenecyclobutane 9a To a suspension of sodium hydride (6.28 g, 0.157 mol) in dry THF (300 cm3) under nitrogen, was added dropwise over a 45 min period a solution of triethyl phosphonoacetate (32.03 g, 0.143 mol) in dry THF (15 cm3).The rate of addition was regulated such that the temperature of the stirred mixture did not rise above 30 8C. After this addition was complete, the reaction mixture was stirred until hydrogen liberation had stopped. Cyclobutanone (10.0 g, 0.143 mol), diluted with THF (15 cm3) was added to the flask while maintaining the temperature of the reaction mixture around 30 8C. Immediately after the addition of the ketone, a sample of the solution was analysed by TLC and the progress of the reaction was subsequently monitored every 30 min.After 2 h the reaction was arrested by pouring the contents of the flask slowly into a slurry of ice and water. Once the ice had melted the aqueous layer was separated and extracted with ether (3 × 50 cm3). The ethereal extracts were combined, dried over anhydrous sodium sulfate and the solvents were removed under reduced pressure. The product 9a was obtained as a colourless liquid by distillation using a Vigreux column (16.48 g, 82%), bp 48–50 8C/1.0 mmHg; dH (300 MHz, CDCl3) 1.25 (3H, t, J = 7.4, CH3), 2.08 (2H, q, J = 8.0), 2.83 (2H, t, J = 8.0), 3.12 (2H, t, J = 8.0), 4.13 (2H, q, J = 7.4), 5.57 (1H, m, CH); dC 12.5 (CH2), 15.7 (CH3), 31.2 (CH2), 37.1 (CH2), 57.2 (CH2), 59.6 (CH), 80.1 (C), 172.5 (CO); m/z 140 (M1, 46%), 112 (80), 95 (100), 84 (23), 67 (86), 55 (31), 41 (74) (Found: M1, 140.0843.C8H12O2 requires 140.0837). ·-(Hydroxymethyl)methylenecyclobutane 10a DIBAL-H (45.58 g, 0.321 mol) was added to a-(ethoxycarbonyl) methylenecyclobutane 9a (15.0 g, 0.107 mol) in dry THF (120 cm3) under nitrogen, in small aliquots at hourly intervals, and the progress of the reaction was monitored by TLC.After the last addition of the reducing agent, the reaction mixture was stirred for a another 60 min. Finally, the excess reagent was decomposed by the slow addition of methanol– THF (40 cm3/80 cm3 resp.) to the reaction mixture. This hydrolysis was fully completed by the dropwise addition of water (25 cm3), with cooling.The precipitated aluminium salts were filtered oV and washed with methanol (3 × 50 cm3). After drying the methanolic solution over anhydrous sodium sulfate, it was concentrated at the rotary evaporator and the remaining solvents were removed from it by distillation at atmospheric pressure. The residue was distilled using a Vigreux column to aVord alcohol 10a as a colourless liquid (5.4 g, 36%), bp 39– 48 8C/0.9 mmHg.The product was purified by column chromatography on silica eluting with EtOAc–petroleum ether (2 : 8, v/v resp.); dH (300 MHz) 1.99 (2H, q, J = 7.8, CH2), 2.10 (1H, br s, OH), 2.70 (4H, t, J = 7.8, 2 × CH2), 4.01 (2H, d, CH2OH), 5.32 (1H, m, CH); dC 17.2 (CH2), 29.3 (CH2), 31.1 (CH2), 59.3 (CH2), 119.3 (CH), 145.1 (C); m/z 98 (M1, 10%), 83 (24), 79 (43), 70 (100), 69 (45), 67 (20), 55 (35), 53 (24), 41 (72) (Found: M1, 98.0734. C6H10O requires 98.0732). 2-Hydroxymethyl-1-oxaspiro[2.3]hexane 11a To anhydrous sodium carbonate (0.98 g, 0.01 mol) in anhydrous dichloroethane (100 cm3) was added the alcohol 10a (1.20 g, 0.012 mol).Peroxyacetic acid (1.07 g, 0.014 mol) was introduced dropwise over a period of 15 min. The resulting mixture was stirred for 1.5 h after which period it was refluxed. TLC analysis performed on the reaction mixture after 5.5 h of heating indicated the complete absence of the starting material. After cooling, the crude material was concentrated under reduced pressure followed by the removal of residual acetic acid under high vacuum to aVord 11a as a pale yellow viscous liquid (1.18 g, 86%); dH (300 MHz) 1.85 (2H, m, CH2), 2.32 (2H, m, CH2), 2.50 (2H, m, CH2), 2.86 (1H, br s, OH), 3.07 (1H, dd, J = 6.1, 3.7, CH), 3.53 (1H, dd, J = 12.2, 6.1, CH2), 3.82 (1H, dd, J = 12.2, 3.7, CH2); dC 13.2 (CH2), 28.8 (CH2), 31.2 (CH2), 61.0 (CH), 62.1 (CH2), 63.9 (C) (Found M1 1 1, 115.0764.C6H11O2 requires 115.0759).942 J.Chem. Soc., Perkin Trans. 2, 1999, 937–945 2-Bromomethyl-1-oxaspiro[2.3]hexane 12a 2-Hydroxymethyl-1-oxaspiro[2.3]hexane (1.01 g, 8.86 mmol), was added to triethylamine (1.09 g, 11.0 mmol) in dry dichloromethane (30 cm3) under nitrogen, cooled by an ice-salt bath. Methanesulfonyl chloride (0.93 g, 8.14 mmol) was added slowly to the reaction mixture which was stirred for a further 30 min. Following the addition of water (25 cm3), the dichloromethane layer was separated and washed successively with 2 M hydrochloric acid (35 cm3), 5% brine (20 cm3) and saturated sodium bicarbonate solution (35 cm3).The mesylate was obtained from the dried solution (anhydrous Na2SO4) by evaporation of the solvent. To the mesylate in dry acetone (25 cm3) was added dried lithium bromide (1.81 g, 20.80 mmol) and the solution was refluxed until all the lithium mesylate had precipitated out of the solution as a white solid. The latter was filtered oV and the acetone was evaporated at room temperature.The residue was treated with water (15 cm3) and this aqueous mixture was extracted with pentane (3 × 15 cm3). The pentane was evaporated, residual solvent being removed under high vacuum. The 2-bromomethyl-1-oxaspirohexane was purified by high vacuum distillation, at room temperature, using a specially designed micro-distillation apparatus which yielded a colourless liquid (1.09, 76%); dH (300 MHz) 1.90 (2H, m, CH2), 2.32 (2H, m, CH2), 2.44 (1H, t, CH2), 2.55 (1H, t, CH2), 3.05 (1H, dd, J = 10.0, 7.5), 3.17 (1H, dd, J = 7.5, 5.4), 3.45 (1H, dd, J = 10.0, 5.4); dC 12.9 (CH2), 28.0 (CH2), 30.7 (CH2), 31.0 (CH2), 59.2 (CH), 65.7 (C); m/z 176 (M1 1 1, 11%), 137 (5), 113 (12), 112 (17), 111 (10), 99 (32), 97 (100) (Found M1 1 1, 176.9921. C6H10O79Br requires 176.9915).·-(Ethoxycarbonyl)methylenecyclopentane 9b Sodium (2.76 g, 0.12 mol) was added, in small amounts, to absolute ethanol (80 cm3). Once the metal had dissolved, triethyl phosphonoacetate (21.50 g, 0.096 mol) was added to the cooled solution of the ethoxide.The resulting mixture was stirred at 0 8C for 2 h and cyclopentanone (10.08 g, 0.12 mol) was added to it at such a rate so as to maintain the temperature of the reaction mixture below 10 8C. The ice-bath was removed and the contents of the flask were stirred for an additional 14 h. Dilution with brine (200 cm3) was followed by extraction of the product with hexane (6 × 50 cm3).The combined extracts were dried over anhydrous magnesium sulfate and concentrated at the rotary evaporator. Distillation, using a Vigreux column, gave the unsaturated ester 9b as a colourless liquid (11.03 g, 75%); (bp 38 8C/0.7 mmHg); dH (300 MHz) 1.28 (3H, t, CH3), 1.72 (4H, m, 2 × CH2), 2.44 (2H, t, CH2), 2.78 (2H, t, CH2), 4.14 (2H, q, CH2), 5.80 (1H, m, CH); dC 14.4 (CH3), 26.6 (CH2), 27.7 (CH2), 32.5 (CH2), 36.6 (CH2), 59.4 (CH2), 111.8 (CH), 127.8 (C), 168.8 (CO), GC-MS tR = 11.17 min. m/z 154 (M1, 41%), 125 (11), 108 (36), 81 (100), 80 (85), 79 (99), 67 (27), 53 (21), 41 (20), 29 (24) (Found M1 1 1, 155.1069.C9H15O2 requires 155.1072). ·-(Hydroxymethyl)methylenecyclopentane 10b Preweighed lithium aluminium hydride (5.29 g, 0.135 mol) was quickly added, under nitrogen gas, to stirred ether under nitrogen in a three-necked flask equipped with a dropping funnel. A separate flask was charged with dry ether (250 cm3) under an atmosphere of nitrogen gas and a-(ethoxycarbonyl)methylenecyclopentane (10.0 g, 0.065 mol) was added to it.Ethanol (4.99 g, 0.108 mol), diluted with dry ether (10 cm3), was placed in the dropping funnel containing the hydride suspension. The ethanol was added dropwise to the contents of the flask. After allowing the resulting lithium aluminium ethoxide hydride to stir for 30 min, aliquots (10.0 cm3, 2.70 mmol) of it were added, hourly, to the flask containing the ester.The progress of the reaction was monitored by TLC which, after 17 h, showed the reaction to be complete. The reagent (170.0 cm3, 0.046 mol) had been consumed in achieving a complete reaction which was terminated by decomposing the excess reducing agent initially with wet diethyl ether (2 × 50 cm3) and subsequently with water (50 cm3) and with periodic cooling of the reaction vessel. The precipitated aluminium salts were filtered, washed with water (3 × 50 cm3) and then with ether (3 × 50 cm 3).The aqueous layer was separated and extracted with ether (3 × 50 cm3). These ethereal extracts were combined with the organic layer obtained previously and the resulting solution was dried over anhydrous sodium sulfate. a-(Hydroxymethyl)methylenecyclopentane was obtained as a colourless, viscous liquid after evaporation of the solvent (7.87 g, 95%); dH (300 MHz) 1.65 (4H, m, 2 × CH2), 1.90 (1H, br s, OH), 2.28 (4H, dt, J = 2.3, 1.2, 2 × CH2), 4.12 (2H, dq, J = 7.0, 1.2, CH2), 5.49 (1H, tt, J = 7.0, 2.3, CH); dC 26.0 (CH2), 26.3 (CH2), 28.6 (CH2 ), 33.7 (CH2), 60.9 (CH2), 119.1 (CH), 147.6 (C); m/z 112 (M1, 32%), 94 (48), 83 (58), 79 (100), 67 (61), 55 (46), 41 (52) (Found M1, 112.0883.C7H12O requires 112.0888). Reduction of 9b with sodium bis(1-methoxyethoxy)- aluminium hydride (Red-Al) yielded a 50 : 50 mixture of 10b and the saturated analogue (75%). Use of diisobutylaluminium hydride (DIBAL-H) yielded an 85 : 15 mixture of 10b and the saturated analogue (73%). 2-(Hydroxyethyl)triphenylphosphonium bromide 2-Bromoethanol (18.80 g, 0.15 mol) and triphenylphosphine (26.60 g, 0.10 mol) in dry ether (30 cm3) were refluxed for 6 h. The precipitated phosphonium salt was filtered and the crude solid was purified by dissolving it in chloroform and then reprecipitating it by dropwise addition of ether. Filtration aVorded the phosphonium salt as a white solid that was dried under high vacuum (22.90 g, 60%); dH 3.70 (2H, dt, CH2), 3.97 (2H, dt, CH2), 4.70 (1H, br s, OH), 7.53–7.80 (15H, m, 3 × Ph).THP protection of the phosphonium salt 2-(Hydroxyethyl)triphenylphosphonium bromide (4.40 g, 0.01 mol) was added to dry dichloromethane (85 cm3), dihydropyran (1.26 g, 0.015 mol), and the catalyst pyridinium toluene-psulfonate (0.31 g, 1.22 mmol). This mixture was stirred at room temperature for 4 h and then diluted with dry ether (90 cm3). The catalyst was decomposed by use of half-saturated brine (15 cm3) and the organic layer was evaporated to aVord a viscous oily liquid which, when washed with dry ether (5 × 15 cm3), and dried overnight under high vacuum, gave the THP-protected phosphonium salt as a white solid (4.70 g, 99%); dH 1.20–1.40 (4H, m, 2 × CH2), 3.32 (2H, dt, CH2), 3.42 (2H, t, CH2), 3.80 (2H, t, CH2), 4.02 (2H, dt, CH2), 4.22 (1H, m, CH), 7.54–7.81 (15H, m, 3 × Ph).·-(Tetrahydropyranyloxymethyl)methylenecyclopentane n-Butyllithium (12.5 cm3) was added to the THP-protected phosphonium salt of 2-bromoethanol (7.10 g, 0.015 mol) in dry ether (50 cm3) under nitrogen over a 10 min period.The resulting suspension was stirred for 4 h. Cyclopentanone (1.26 g, 0.015 mol) was added and, after stirring for 1 h, refluxing was continued for a further 12 h. The precipitated triphenylphosphine oxide was removed by filtration. The residue was washed with dry ether (3 × 50 cm3) and these washings were combined with the filtrate from the earlier stage.The resulting solution was treated with portions of water (25 cm3) until the latter was neutral. After removal of the solvent, the product was obtained as a colourless liquid by distillation at 95–105 8C/1.3 mmHg using a Kugelrohr (1.10 g, 37%); dH 0.89 (2H, m, CH2), 1.28 (2H, m, CH2), 1.69 (4H, m, 2 × CH2), 1.94 (2H, q, CH2), 2.30 (4H, m, 2 × CH2), 2.55 (2H, m, CH2), 2.78 (2H, m, CH2), 3.64 (1H, t, CH), 5.38 (1H, m, CH).J. Chem. Soc., Perkin Trans. 2, 1999, 937–945 943 Deprotection of ·-(tetrahydropyranyloxymethyl)methylenecyclopentane a-(Tetrahydropyranyloxymethyl)methylenecyclopentane (0.40 g, 2.04 mmol) was added to ethanol (20 cm3) and pyridinium toluene-p-sulfonate (0.05 g, 0.21 mmol). After 3 h the catalyst was decomposed with half-saturated brine (10 cm3) and the product was extracted with dry ether (3 × 20 cm3). These ethereal extracts were combined and dried over anhydrous sodium sulfate. After removing the solvent, the residue was distilled using a Vigreux column and a-(hydroxymethyl)methylenecyclopentane 10b was collected as a colourless, viscous liquid (0.12 g, 52%) (bp 80–85 8C/1.7 mmHg) (spectrum as above). 2-Hydroxymethyl-1-oxaspiro[4.2]heptane 11b a-(Hydroxymethyl)methylenecyclopentane 10b (0.21 g, 0.019 mol) in dichloromethane (35 cm3) was stirred and m-chloroperoxybenzoic acid (4.31 g, 0.025 mol) dissolved in dichloromethane (50 cm3) was added to it slowly over 15 min. The reaction was allowed to proceed for a further 30 min and the excess peroxy acid was decomposed by addition of sodium sulfite solution (10%) until a starch–iodide paper gave a negative result.The organic layer was separated and washed with 5% sodium bicarbonate solution to extract the by-product m-chlorobenzoic acid. This was followed by washing it with water (50 cm3) and finally with saturated sodium chloride solution (50 cm3). The separated dichloromethane layer was dried over anhydrous sodium sulfate and concentrated under reduced pressure.High-vacuum removal of the solvent gave 11b as a ca. 2 : 1 mixture with m-chlorobenzoic acid (1.57 g). The latter was diYcult to remove without decomposing 11b and hence the preferred method employed peroxyacetic acid. Alcohol 10b (6.50 g, 0.058 mol) was introduced into a suspension of anhydrous sodium carbonate (13.36 g, 0.126 mol) in anhydrous dichloroethane (300 cm3) under nitrogen. Peroxyacetic acid (12.61 g, 0.063 mol), diluted with dichloroethane (10 cm3), was added dropwise into the reaction mixture over a period of 1 h.After stirring the contents of the reaction flask for a further 1 h, it was heated to reflux for 1 h, when TLC showed reaction had gone to completion, and a gelatinous white precipitate had appeared. The mixture was allowed to cool to room temperature and the white solid was removed by filtration and washed with dichloroethane (3 × 50 cm3). The filtrate was initially concentrated at the rotary evaporator and then the residual solvent was removed from it under high vacuum.The product 2-hydroxymethyl-1-oxaspiro[4.2]heptane was obtained as a colourless, oily liquid (6.75 g, 91%); dH 1.62 (4H, m, 2 × CH2) 1.82 (4H, m, 2 × CH2), 1.92 (2H, m, CH2), 2.91 (1H, br s, OH), 3.21 (1H, dd, CH), 3.60 (1H, dd, CH2), 3.83 (1H, dd, CH); dC 25.4 (CH2), 25.8 (CH2), 29.8 (CH2), 34.4 (CH2), 62.2 (CH), 63.0 (CH2), 70.1 (C); m/z 128 (M1, 4%), 111 (20), 97 (39), 85 (100), 83 (26), 67 (98), 57(22), 55 (40), 43 (24), 41 (56) (Found M1, 128.0840.C7H12O2 requires 128.0837). 2-Bromomethyl-1-oxaspiro[2.4]heptane 12b To the epoxy alcohol 11b (6.50 g, 0.05 mol) and triethylamine (6.36 g, 0.06 mol) in dry dichloromethane (175 cm3) under nitrogen at 25 8C, was added methanesulfonyl chloride (5.38 g, 0.047 mol), drop by drop and the solution was stirred for 30 min. The reaction was terminated by addition of water (145 cm3) and the organic layer was separated and washed successively with 2 M hydrochloric acid (200 cm3), 5% brine (100 cm3) and saturated sodium bicarbonate (200 cm3).The resulting solution was dried over anhydrous sodium sulfate and concentrated at room temperature. Lithium bromide (10.44 g, 0.120 mol), dried under high vacuum with gentle periodic heating for 1 h, was quickly added to continuously stirred acetone (125 cm3) under nitrogen and the mesylate, diluted with dry acetone (10 cm3), was gradually added to the reaction mixture over 15 min.The contents of the reaction vessel were stirred until no further precipitate appeared. The white solid was filtered oV and the filtrate was concentrated using a rotary evaporator operated at room temperature. The residue was treated with water (90 cm3) and the aqueous mixture was extracted with pentane (3 × 50 cm3) which was completely evaporated leaving a very pale yellow oily material. The latter was purified by distillation using a Kugelrohr (30 8C/0.10 mmHg) over several hours to yield a pale yellow oily product 12b (2.75 g, 31%); dH 1.70 (4H, q, 2 × CH2), 1.95 (4H, q, 2 × CH2), 3.26 (1H, dd, CH), 3.33 (1H, dd, CH2), 3.58 (1H, dd, CH2); dC 25.0 (2 × CH2), 28.5 (CH2), 31.1 (CH2), 33.4 (CH2), 60.0 (CH), 71.6 (C); m/z 192, 190 (M1 1%), 97 (24), 83 (39), 67 (30), 55 (100), 53 (25), 41 (89), 39 (63), 29 (37) 28 (27), 27 (99) (Found M1 1 1, 191.0081.C7H12O79Br requires 191.0073). ·-(Ethoxycarbonyl)methylenecycloheptane 9c This was prepared from cycloheptanone by the same method as for 9a as a colourless liquid (91%) (bp 42–46 8C/0.02 mmHg); dH 1.25 (3H, t, J = 7.1, CH3), 1.54 (4H, m, 2 × CH2), 1.68 (4H, m, 2 × CH2), 2.38 (2H, dd, J = 1.2, 6.0, CH2), 2.86 (2H, dd, J = 1.2, 6.0, CH2), 4.14 (2H, q, J = 7.1, CH2), 5.67 (1H, m, J = 1.2, CH); dC 14.4 (CH3), 26.6 (CH2), 28.0 (CH2), 29.0 (CH2), 29.8 (CH), 32.1 (CH2), 39.0 (CH2), 59.3 (CH2), 115.6 (CH), 166.7 (CO) (Found M1 1 1, 183.1392.C11H19O2 requires 183.1385).·-(Hydroxymethyl)methylenecycloheptane 10c The same methodology as for 10a was applied to selectively reduce the unsaturated ester 9c yielding 10c as a viscous, colourless liquid (82%); dH 1.56 (8H, m, 4 × CH2), 2.26 (4H, m, 2 × CH2), 4.14 (2H, d, CH2), 5.40 (1H, t, CH); dC 28.0 (CH2), 29.5 (CH2), 29.6 (CH2), 30.4 (CH2), 30.6 (CH2), 38.4 (CH2), 59.8 (CH2), 124.5 (CH), 146.3 (C); m/z 140 (M1, 14%), 122 (86), 107 (31), 96 (60), 83 (54), 81 (100), 70 (64), 67 (92), 57 (56), 55 (82), 41 (92) (Found M1, 140.1206.C9H16O requires 140.1201). 2-Hydroxymethyl-1-oxaspiro[2.6]nonane 11c This preparation was accomplished in the same manner as described for 11a to give the spiroepoxy alcohol 11c as a colourless oily liquid (87%); dH 1.55 (8H, m), 1.70 (4H, m), 2.70 (1H, br s, OH), 2.97 (1H, dd, CH), 3.65 (1H, dd), 3.83 (1H, dd); dC 25.0 (CH2), 25.3 (CH2), 29.4 (CH2), 29.5 (CH2), 32.1 (CH2), 38.1 (CH2), 61.9 (CH2), 65.4 (C), 65.5 (CH); m/z 157 (M1 1 1, 8%), 139 (20), 121 (26), 113 (100), 95 (62), 81 (37), 67 (46), 55 (46), 43 (38), 41 (46) (Found M1 1 1, 157.1231.C9H16O2 requires 157.1229). 2-Bromomethyl-1-oxaspiro[2.6]nonane 12c This spiroepoxymethyl bromide was prepared by the same method as that described for 12a which gave 12c as a colourless liquid (87%); dH 1.59 (8H, m), 1.74 (4H, m), 3.06 (1H, dd, J = 6.0, 7.6), 3.26 (1H, dd, J = 7.6, 10.4), 3.52 (1H, dd, J = 6.0, 10.4); dC 24.3 (CH2), 24.7 (CH2), 28.9 (CH2), 29.0 (CH2), 29.9 (CH2), 30.9 (CH2), 37.0 (CH2), 63.1 (CH), 66.6 (C); m/z 220, 218 (M1, 6%), 125 (100), 121 (33), 107 (17), 95 (37), 81 (74), 67 (29), 55 (41), 41 (26) (Found M1, 218.0300.C9H15O79Br requires 218.0306). Reaction of 2-bromomethyl-1-oxaspiro[2.3]hexane 12a with tributyltin hydride Tributyltin hydride (128.0 mg, 0.44 mmol) was weighed into an NMR tube and dissolved in benzene (500.0 ml). A single portion of the spiro-bromide 12a (20.0 mg, 0.11 mmol) was added to the diluted organotin reagent via a microsyringe.After introducing heptane (10.0 ml), the contents of the tube were944 J. Chem. Soc., Perkin Trans. 2, 1999, 937–945 thoroughly mixed and photolysed with light from a 250 W Hg lamp in a continuously stirred water-bath maintained at 5 8C for 4 h. The experiment was repeated at 50 8C, 120 8C and 185 8C. In view of the high volatility of benzene at higher temperatures, tert-butylbenzene and hexadecane were used as solvents.The reaction times were appropriately adjusted. GC-MS peak no. 138, tR 2.52 min, heptane standard; peak no. 172, tR 3.15 min, hexan-3-one 29a, peak no. 215, tR 3.92 min, 2-methylcyclopentanone 26a; peak no. 275, tR 5.03 min, cyclohexanone 22a. Product identities were confirmed by matching tR and MS data with that of authentic materials. Reaction of 2-bromomethyl-1-oxaspiro[2.4]heptane 12b with tributyltin hydride 2-Bromomethyl-1-oxaspiro[2.4]heptane 12b (50.0 mg, 0.26 mmol) was placed into an NMR tube.After dilution with benzene (500 ml), an aliquot of tributyltin hydride (10.51 mg, 36.13 mmol) was introduced into the tube which was photolysed for 8 h at 30 8C with the addition of the above mentioned amount of the organotin reagent every 30 min until a total of 16 such portions had been added to the reaction mixture. This experiment was repeated at 80 8C, 150 8C and 180 8C. tert-Butylbenzene was used as a solvent at higher temperatures and the duration of the reactions was appropriately adjusted.Products at 30 8C were identified by GC-MS which indicated the presence of a minor amount of unreacted starting material: peak no. 246, tR 4.51 min, heptan-3-one 29b, (7%), peak no. 310, tR 5.62 min, hept-1-en-3-one 23b, (12%), m/z 112 (M1, 14%), 71 (36), 69 (50), 68 (21), 43 (100), 41 (86), 39 (39), 27 (25), peak no. 340, tR 6.22 min, 2-methylcyclohexanone 26b, (34%), m/z 112 (M1, 18%), 84 (11), 69 (24), 68 (48), 56 (37), 55 (59), 42 (40), 41 (100), 39 (55), 29 (20), 28 (29), 27 (69); peak no. 417, tR 7.62 min, cycloheptanone 22b (34%). In the reaction at 180 8C, the only compounds obtained were: the open-chain enone 23b, the a-methylated cycloalkanone 26b and the cycloalkanone 22b. Reaction of 2-bromomethyl-1-oxaspiro[2.6]nonane 12c with tributyltin hydride 2-Bromomethyl-1-oxaspiro[2.6]nonane 12c (44.09 mg, 0.21 mmol) diluted with benzene (500 ml) was placed in an NMR tube. Tributyltin hydride (66.06 mg, 0.23 mmol) in benzene (200 ml) was divided up into eight equal portions which were added to the photolysing solution at 75 8C every 30 min.The reaction was terminated after 4 h. The experiment was repeated at 150 8C using tert-butylbenzene as a solvent. GC-MS peak no. 532, tR 9.80 min, nonan-3-one 29c; peak no. 585, tR 10.80 min, 2-methylcyclooctanone 26c, m/z 111 (96%), 97 (76), 83 (35), 67 (33), 55 (80), 41 (100), 27 (78); peak no. 559, tR 10.90 min, non-1-en-3-one 23c, m/z 111 (17%), 97 (15), 83 (57), 70 (38), 55 (75), 41 (62), 27 (62); peak no. 591, tR 11.1 min, 1-vinylcyclononanol 21c m/z 140 (2%), 138 (7), 109 (17), 94 (19), 81 (22), 79 (30), 70 (33), 67 (96), 56 (38), 55 (64), 43 (39), 41 (100), 39 (78); peak no. 744, tR 12.10 min, cyclononanone 22c. In addition several unidentified products were observed. The reactions at 150 8C and 180 8C showed little of the above products but indicated the presence of cycloheptane and a range of unidentified materials.Triphenyltin hydride initiated reactions A mixture of bromide 12a (31.0 mg, 0.18 mmol), perdeuteriobenzene (600 ml) and triphenyltin hydride (9.26 mg, 26.38 mmol) was heated to 80 8C and aliquots (9.26 mg) of the organotin reagent were added every 30 min to the reacting solution for 4 h. The contents of the tube were cooled and dichloromethane (9.26 mg, 5 ml) was added as a standard. A 1H NMR spectrum of the resulting solution showed the composition of the reaction mixture to be hex-1-en-3-one 23a (17%), cyclohexanone 22a (58%) 2-methylcyclopentanone 26a (22%).Reduction of bromide 12b with Ph3SnH was carried out in a similar way. Signals from the 1H NMR spectrum could not be analysed with certainty owing to severe overlapping of the peaks. GC-MS analysis showed the same four products as in the tributyltin hydride reductions. A small scale reduction of 12c with Ph3SnH using a similar method at 20 8C showed the same products as in the tributyltin hydride reaction. The reaction was scaled up using 12c (0.43 g, 1.96 mmol), benzene (100 cm3) and the organotin reagent (1.38 g, 3.92 mmol).This mixture was photolysed at 75 8C and progress was monitored by TLC which showed complete consumption of starting material after 10 h of photolysis. The solvent was completely evaporated, following the removal of solid residues by filtration. The residual solution was separated by column chromatography using silica gel (mesh size 40–63 mm).The column was eluted with petroleum ether–EtOAc (initially 40 : 60) of increasing polarity and this yielded cyclononanone 22c (0.10 g, 36%). Similarly, cyclononanone (20%) and 3- oxonon-1-ene 23c (11%) were isolated from a reaction carried out at 20 8C. Acknowledgements We thank GlaxoWellcome and the EPSRC (grant GR/L49185) for financial support of this research, Dr Adrian Whitwood, York University, for help with spectral simulations and a referee for suggesting the structure and mode of formation of radical 18.References 1 D. P. Curran and H. Liu, J. Am. Chem. Soc., 1992, 114, 5863; D. P. Curran, S.-B. Ko and H. Josien, Angew. Chem., Int. Ed. Engl., 1995, 34, 2683. 2 G. Pattenden, A. Smithies, D. Taploczay and D. S. Waller, J. Chem. Soc., Perkin Trans. 1, 1996, 21; L. Chen, G. B. Gill, G. Pattenden and H. Simonian, J. Chem. Soc., Perkin Trans. 1, 1996, 31; A. Batsov, L. Chen, G. B. Gill and G. Pattenden, J. Chem. Soc., Perkin Trans. 1, 1996, 45. 3 K. K. Wang, Chem. Rev., 1996, 96, 207. 4 E. W. Della, N. J. Head, P. Mallon and J. C. Walton, J. Am. Chem. Soc., 1992, 114, 10730. 5 P. E. Eaton and Y. C. Yip, J. Am. Chem. Soc., 1991, 113, 7692; S.-Y. Choi, P. E. Eaton, M. Newcomb and Y. C. Yip, J. Am. Chem. Soc., 1992, 114, 6326. 6 A. Johns and J. A. Murphy, Tetrahedron Lett., 1988, 29, 837. 7 R. C. Gash, F. MacCorquodale and J. C. Walton, Tetrahedron, 1989, 45, 5531. 8 P. Dowd and W. Zhang, Chem. Rev., 1993, 93, 2091, and references cited therein. 9 J. D. Harling and W. B. Motherwell, J. Chem. Soc., Chem. Commun., 1988, 1380. 10 R. A. Batey, J. D. Harling and W. B. Motherwell, Tetrahedron, 1992, 48, 8031. 11 R. A. Batey and W. B. Motherwell, Tetrahedron Lett., 1991, 32, 6211. 12 C. Walling and A. Padwa, J. Am. Chem. Soc., 1963, 85, 1579. 13 A. G. Davies and B. Muggleton, J. Chem. Soc., Perkin Trans. 2, 1976, 502. 14 D. Laurie, D. C. Nonhebel, C. J. Suckling and J. C. Walton, Tetrahedron, 1993, 49, 5869. 15 M. Cook, O. Hares, A. Johns, J. A. Murphy and C. W. Patterson, J. Chem. Soc., Chem. Commun., 1986, 1419; J. A. Murphy, C. W. Patterson and N. F. Wooster, J. Chem. Soc., Chem. Commun., 1988, 924. 16 S. Kim and S. Lee, Tetrahedron Lett., 1991, 32, 6575. 17 P. Galatsis, S. D. Millan and T. Faber, J. Org. Chem., 1993, 58, 1215. 18 R. S. Davidson, W. H. H. Gunther, S. M. Waddington-Feather and B. Lythgoe, J. Chem. Soc., 1964, 4907. 19 D. M. Camaioni and D. W. Pratt, J. Am. Chem. Soc., 1972, 94, 9258; H. F. Walter, W. T. Beaudry, D. M. Camaioni and D. W. Pratt, J. Am. Chem. Soc., 1985, 107, 793.J. Chem. Soc., Perkin Trans. 2, 1999, 937–945 945 20 B. C. Gilbert, R. O. C. Norman and R. C. Sealy, J. Chem. Soc., Perkin Trans. 2, 1973, 2174; B. C. Gilbert and M. Trenwith, J. Chem. Soc., Perkin Trans. 2, 1975, 1083. 21 A. G. Davies and R. SutcliVe, J. Chem. Soc., Perkin Trans. 2, 1980, 819; A. G. Davies, J. A.-A. Hawari, B. Muggleton and M.-W. Tse, J. Chem. Soc., Perkin Trans. 2, 1981, 1132. 22 C. Corvaja, G. Giacometti and G. Sartori, J. Chem. Soc., Faraday Trans. 2, 1974, 70, 709. 23 J. C. Walton, J. Chem. Soc., Perkin Trans. 2, 1998, 603. 24 C. Chatgilialoglu, C. Ferreri, M. Lucarini, A. Venturini and A. A. Zavitsas, Chem. Eur. J., 1997, 3, 376. 25 C. Chatgilialoglu, K. U. Ingold and J. C. Scaiano, J. Am. Chem. Soc., 1981, 103, 7739. 26 D. Lal, D. Griller, S. Husband and K. U. Ingold, J. Am. Chem. Soc., 1974, 96, 6355. 27 A. Krief, Top. Curr. Chem., 1987, 135, 1; D. Labar and A. Krief, J. Chem. Soc., Chem. Commun., 1982, 564. 28 J. Heinzer, Quantum Chemistry Program Exchange, University of Indiana, No. 209, 1972. 29 A. C. Whitwood, University of York, personal communication. Paper 9/00377K