Annulated and bridged tetrahydrofurans from alkenoxyl radical cyclization † ‡

4-Pentenoxyl radicals sharing two or more carbon atoms with a cycloalkane cyclize in a predictable manner stereoselectively and regioselectively to afford in solutions of bromotrichloromethane cycloalkylfused or -bridged 2-bromomethyltetrahydrofurans in up to 95% yield. Stereoselectivity in alkenoxyl radical ring closures arises from cumulative steric effects. The substituent positioned the closest to the alkene carbon, which is being attacked by the oxygen radical, exerts the strongest stereodirecting effect. This principal inductor guides 5-exo-cyclization 2,3-transor 2,4-cis-selectively. The substituent located further from the attacked π-bond is the secondary inductor. A secondary inductor in the relative transconfiguration enhances stereodifferentiation by the primary inductor; a cis-configured secondary inductor decreases this effect. A secondary inductor is not able to overrule the guiding effect of a similar sized primary inductor. Intramolecular 4-pentenoxyl radical additions to a cyclohexene-bound exo-methylene group or to endocyclic double bonds proceed cis-specifically, as exemplified by synthesis of a diastereomerically pure bromobicyclo[2.2.1]heptyl-annulated tetrahydrofuran from the verbenylethyloxyl radical. According to theory, the experimental 2,3-cis-specificity in alkoxyl radical cyclization to an endocyclic π-bond arises from strain associated with the 2,3-trans-ring closure.


Introduction
][3][4][5][6] Alkyl or ortho-substituted aryl groups in position 1 exert a stereodirecting effect, leading to 2,5-trans-configured tetrahydrofurans as principal products.8][9] Stereodifferentiation by alkyl or aryl groups arises from steric effects, which gradually increases as the distance between a controlling substituent and the attacked π-bond shortens, for example from a 15/85-cis/trans-ratio at room temperature to <2/98 by shifting a tert-butyl group from position 1 to position 3. 8 In synthesis, 5-exocyclized 4-pentenoxyl radicals are preferentially trapped by a heteroatom atom donor, 10,11 for introducing halogen, [12][13][14] alkylsulfanyl, 15 or other synthetically useful functional groups. 16he model to explain stereodifferentiation by a carbon substituent in 4-pentenoxyl radical cyclization predicts that the intramolecular addition proceeds via a distorted twistconformer of tetrahydrofuran as the favored transition structure (twist-model), 8,17 differing from the cyclohexane-based Beckwith-Houk-model for carbon radical cyclization. 18,19pplication of the alkoxyl radical approach to synthesis of more demanding targets, for example biologically active terpene-, acetogenin-, and fatty acid-derived cycloalkyl-fused tetrahydrofurans, [20][21][22] requires to extend the model in order to predict the selectivity for constructing bicyclic compounds. 7,23essons from carbon radical chemistry have taught that stereodifferentiation in synthesis of bicyclic compounds is difficult to extrapolate by transferring results from monocycle to bicycle formation, since transannular and other strain effects may superimpose in an unpredictable manner. 24,25To find out whether embedding two carbons of a 4-pentenoxyl radical into a cycloaliphatic framework conserves or changes guidelines for stereoselective tetrahydrofuran synthesis, we examined in this study bromocyclization of cis/trans-cycloalkyl-bridged alkenoxyl † Dedicated to the memory of Athelstan (Athel) L. J. Beckwith in recognition of his pioneering and creative contributions to the chemistry of free radicals in general and to free radical cyclizations in particular.We will miss his humor and his intellectual approach to chemical science.‡ Electronic supplementary information (ESI) available: Standard instrumentation, the protocol for ESI containing instrumentation, synthesis of 4-pentenols, O-pentenyl tosylates, carbon-13 spectra of selected compounds, calculated atomic coordinates and energies of transition structures and radicals.CCDC 1008593.For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c4ob01266f radicals, having the carbon-carbon double bond located in a conformationally flexible side chain (types A and B), in the exo-position of cyclohexane (type C), or incorporated into an alicyclic core (types D and E, Fig. 1).
The most important finding from the study shows that cycloalkyl-bridged 4-pentenoxyl radicals cyclize in a predictable manner stereoselectively and regioselectively, to afford in solutions of bromotrichloromethane cycloalkyl-annulated or -bridged bromomethyltetrahydrofurans in up to 95% yield.The principal stereoinductor is the substituent attached the closest to the carbon-carbon double bond, which is being attacked by the radical oxygen.A principal inductor guides 5-exo-cyclization 2,3-trans-or 2,4-cis-selectively.The substituent bound further from the attacked π-bond is the secondary inductor.A trans-arranged secondary inductor enhances stereocontrol of the primary inductor, and a cis-configured secondary inductor decreases this effect.A secondary inductor of similar steric size, located further from the attacked π-bond is not able to overrule the directing effect of the principal inductor.Oxygen radicals attached via a methylene-or an ethylenespacer to cyclohexene cyclize cis-specifically, as exemplified by synthesis of a diastereomerically pure bromobicyclo[2.2.1]heptyl-annulated tetrahydrofuran from a type-E radical.The propensity of cyclohexenylethyloxyl radicals to cyclize 2,3-cisspecifically arises from strain associated with the 2,3-trans-ring closure, as derived by a Marcus analysis of density functionalcalculated reaction energies and barriers.
(i) Methods of alkenoxyl radical generation and product analysis.Photolyzing solutions of O-alkenyl thiohydroxamates 1a-c in benzene containing 10 equivalents (1.67 M) of bromotrichloromethane, using Rayonet® chamber apparatus equipped with 350 nm illuminants, quantitatively consume the starting material within 30 minutes, as determined by thin layer chromatography.Reaction mixtures from photochemical experiments tended to turn turbid and yellow.A gas chromatogram (GC) recorded by the end of the reaction time provided information on the original product pattern and distribution.Column chromatography furnished samples of purified 2-(trichloromethylsulfanyl)-4-methyl-1,3-thiazole (2), 5-exo-bromocyclized products 3a-c, and β-fragmented unsaturated 5-bromoaldehydes 4a-b, for collecting analytical data (Tables 1-4, and Experimental).Solutions from thermally initiated reactions were in addition charged with 15 mole percent of azo-α,α-bis-(isobutyronitrile) (AIBN) as the initiator.Such mixtures remained clear but tended to turn yellow by the end of the reaction.
(ii) Product pattern and kinetic interpretation.Reactions between O-(2-allylcycloalkyl) thiohydroxamates cis-1a-c and bromotrichloromethane furnish bromomethyltetrahydrofurans cis-3a-c, with the yields gradually decreasing for thermally initiated reactions from 61% for cis-3c through 34% for cis-3b to 8% for cis-3a (Table 1, entries 2 and 4; Table 3, entry 2).The reactions gave bromoaldehydes 4a-c as co-products in yields increasing from 12% for 4c, through 35% for 4b to 54% to 4a.Photolyzing or heating O-(2-allylcyclopentyl) thiohydroxamate trans-1a in the presence of bromotrichloromethane provided bromooctanal 4a, but no bromomethyltetrahydrofuran trans-3a as secured by independent analysis of an authentic sample of the compound (ESI ‡).The ratio of the bromocyclized product trans-3c and bromoaldehyde 4c obtained from O-cyclohexylallyl ester trans-1c is similar to the ratio of cis-3c and 4c obtained from the stereoisomer cis-1c (entry 2 in Tables 3 and  4).The pattern of products obtained from radical reactions conducted at 80 °C in summary is similar, except for bromoaldehyde 4c, which did not form in the room temperature experiments.None of the reactions furnished 2-allylcycloalkanols or 2-allylcycloalkanones in verifiable amounts (GC-MS).
In kinetically controlled reactions, the quotient between bromomethyltetrahydrofuran 3 and bromoaldehyde 4 is equivalent to the relative rate constant for the addition (k add ) versus β-fragmentation (k β ) (Scheme 3).Kinetic control for oxygen radical addition to terminal double bonds is documented. 8or the following reason we suggest that the sequence leading to bromoaldehyde 4 under conditions chosen in this study also is kinetically controlled.In 1.67 molar solution of bromotrichloromethane, the effective rate constant for bromine atom trapping by secondary alkyl radicals, such as III, is approximately 4.3 × 10 8 s −1 , based on k Br for the 6-hepten-2-yl radical (2.6 × 10 8 M −1 s −1 ; 26 °C) 48 as a reference.The rate constant k add for the 4-formylbutyl radical 5-exo-cyclization (8.7 × 10 5 s −1 ; 80 °C), 49 serving as a reference for the reaction III → I, is by almost three orders of magnitude slower than the effective rate of bromine atom transfer from bromotrichloromethane to the secondary carbon radical III.
For comparing rates of 5-exo-cyclization to rates of β-fragmentation for intermediates Ia-c, we standardized reactant concentrations and used a tenfold molar excess of bromotrichloromethane.Under such conditions, the ratio of bromide 3 to 4 corresponds to the quotient k add /k β , gradually increasing along the series of radicals trans-Ia (k rel = 0), cis-Ia (0.2), cis-Ib (1.0) to cis/trans-Ic (k rel = 5-9).Dividing k add for the 4-pentenoxyl radical cyclisation (5.2 × 10 8 s −1 ; 26 °C) by k β for the cyclopentoxyl radical β-fragmentation (4.7 × 10 8 s −1 ; 80 °C) for calibrating the competition system with the aid of absolute rate constants leads to a similar order of magnitude for the k add /k β ratio. 2,49he propensity of cyclopentane-fused 4-pentenoxyl radicals to provide β-fragmented products, such as bromoaldehydes 4a-b, arises from strain, being ∼20 kJ mol −1 higher for cyclopentane than for cyclohexane. 50Substituting methyl for hydrogen at the terminal alkene carbon increases the fraction of the 5-exo-cyclized product from cis-3a to cis-3b, which we address to a rate enhancing polar effect of the methyl group in oxygen radical additions. 51iii) Stereochemical guidelines.1,2-Cycloalkyl-bridged 4-pentenoxyl radicals Ia-c cyclize 2,4-cis-selectively showing that the substituent in position 2 is the principal stereoinductor for 5-exo-cyclization of type-A radicals.A trans-arranged secondary inductor in position 1 enhances the directing effect of the principal inductor; a cis-configured secondary inductor decreases this effect.
(ii) Effect of methyl substitution at the terminal alkene carbon.Substituting two hydrogens at the terminal alkene carbon by methyl improves the stereoselectivity and regioselectivity in cyclization of type-B 4-pentenoxyl radicals (Tables 5 and 6).Terminal methyl groups furthermore improve the regioselectivity of the intramolecular addition, occurring with 80/20-selectivity for trans-Id, 94/6 for cis-Id, and 5-exo-specifically for cis/ trans-Ie (GC-MS; Table 5, entries 2 and 4, and Table 6).a Single diastereomer, according to proton-NMR-and GC-MS-data.a Stereodescriptor referring to the configuration of bridgehead carbons in products 3d and 5d.
(iv) On the origin of 2,3-trans-selectivity in 5-exo-cyclization of type-B 4-pentenoxyl radicals.Models built as instructed in section 2.1 show that type-B cyclohexyl-bridged 4-pentenoxyl radicals cis/trans-Id-e cyclize 2,3-trans-selectively, because steric constraints disfavor the 2,3-cis-mode of ring closure.In transition structures for 2,3-cis-cyclization, van der Waals repulsion between the (E)-positioned alkene substituent and the axially arranged hydrogens raises conformational free energy.The second aspect raising conformational free energy thus disfavoring a transition structure is eclipsing of hydrogens bound to carbons 4 and 5 (for TS 2 -trans-Id and TS 2 -cis-Id; Fig. 5).Extending the size of the (E)-substituent from hydrogen to methyl raises transannular repulsion, explaining the stereodirecting effect of a terminal substituent in cyclization of cis/ trans-Ie.
The 5-exo/6-endo-selectivity of radical If (53 : 47) at room temperature falls below the value reported for the 4-methyl-4pentenoxyl radical (69 : 31) and is higher than the regioselectivity determined for the 4-tert-butyl-4-pentenoxyl radical (46 : 54). 17Regioselectivity in 4-pentenoxyl radical cyclization originates from a balance between FMO attractions, torsional strain, and steric shielding.A carbon substituent in position 4 lowers the barrier for 6-endo-addition based on favorable frontier molecular orbital (FMO) interactions for the C,O-addition to the terminal carbon.Steric blocking of the incoming oxygen radical gradually lowers the rate of 5-exo-addition as the size of the carbon substituent in position 4 increases.The fraction tetrahydropyranyl radical IVf obtained from 6-endo-cyclization of If is in line with the general mechanistic interpretation. 17omolytic bromination of tetrahydropyranyl radical IVf occurs for steric reasons preferentially from the axial side (Scheme 4), similar to bromination of structurally related cyclohexyl radicals. 53ii) Stereochemical guideline.Methylenecyclohexylethoxyl radical If cyclizes 2,3-cis-specifically (Scheme 4).8).We address this phenomenon to magnetic anisotropy induced by the carbon-bromine bond, possibly in combination with three nonbonding electron pairs at bromine. 54ii) Stereochemical guideline.Cyclohexenylmethyloxyl radical 1g cyclizes 2,4-cis-specifically (Scheme 5).(iii) Verbenylethyloxyl radical chemistry.In extension to the chemistry summarized in this article, we propose that tricyclic products 6 and 7 arise from a sequence composed of intramolecular addition Ih → cis-IIh, ring-opening of cyclobutylmethyl radical cis-IIh, and bromine atom trapping by rearranged radicals V and VI (Scheme 7).1,2-Shifting of the methylene bridge releases cyclobutyl strain in radical cis-IIh, leading to the secondary carbon radical V.For steric reasons, we expect trapping of the bicyclic radical V by bromotrichloromethane to occur from the concave face due to shielding of the convex side with the vicinal exo-oriented methyl group.The minor product 7, according to the proposed model, results from 1,2-shifting of the dimethylmethylene bridge cis-IIh → VI and subsequent homolytic bromination.
(i) Density functional theory.For computing ground state energies of radicals and energies of transition structures, we used Becke's three parameter Lee-Young-Parr-hybrid functional (B3LYP) 57,58 and Becke's half and half Lee-Young-Parr hybrid functional (BHandHLYP) 59 in combination with 6-31+G**-and 6-311G**-basis sets. 56All selected density functional/basis set-combinations reproduce experimental stereoand regioselectivity for oxygen radical addition to carboncarbon double bonds with a precision coming close to the accuracy for determining experimental selectivity. 8,14,17,52,60ii) Theoretical approach.For calculating equilibrium structures of conformational flexible molecules and transition structures associated with radical addition to carbon-carbon double bonds we used an established strategy. 8,14According to theory, the 2-(cyclohexen-3-yl)-ethyloxyl radical Ii favors Both conformers served as starting points for modeling 5-exocyclizations.Equilibrium structures of propene, alkoxyl radicals Ii-k, cyclized radicals IIi-j, and the addition product VIII lack in negative eigenvalues of second derivatives of energy-minimized wavefunctions.Transition structures TS-I and TS-VII show one imaginary frequency i, describing the trajectory of oxygen radical addition to the inner alkene carbon (Table 9). 61tempts to localize a transition structure for the trans-5-exocyclization of conformer pa-Ii led to TS 1 -trans-Ii, already available from conformer pe-Ii.
(iii) Quality of the models.Computed wavefunctions characterizing equilibrium structures show expectation values for the spin operator 〈S 2 〉 close to 0.75 for oxygen and carbon radicals (ESI ‡), as expected for doublet states.Wavefunctions describing transition structures show 〈S 2 〉-values of ∼0.77 for B3LYPcalculated intermediates and 0.82-0.84for BHandHLYP-calculated transition structures (ESI ‡).The effect of spin contamination in BHandHLYP-calculated transition structures was discussed previously, but is not considered relevant for attaining reasonable precision in determining computed relative energies. 52iv) Methoxyl radical addition to propene.Theory predicts a lower barrier for methoxyl radical addition to the terminal carbon than for addition to the inner carbon of propene (ΔG 298 = −5.0 to −8.5 kJ mol −1 ; ESI ‡).The decision to compare structure and energetics from the disfavored mode of addition to data obtained for monocycle and bicycle formation was guided by structural similarity between TS-Ii-j and TS-VII on one side, and derived addition products IIi-j, VIII on the other (Table 9, Schemes 8 and 9).pathways strongly exothermic (Δ R E = −35 to −47 kJ mol −1 ), pointing to a notable barrier for the reverse reaction, the β-fragmentation.Computed energetics for the addition Ii → IIi are similar to the values calculated for the 4-pentenoxyl radical ring closure Ij → IIj (Δ R E = −41 kJ mol −1 ), and are less exothermic than the methoxyl radical addition to the inner carbon of propene (Δ R E = −53 mol −1 ).BHandHLYP-calculations provide similar trends for reaction energies, except for a stronger driving force for the intermolecular addition (Table 10).
(vi) Transition structures.The distance d between the radical oxygen and the attacked carbon, as predicted by B3LYP theory for transition structures of cyclohexenylethyloxyl radical cyclization (2.04-2.08Å), 4-pentenoxyl radical cyclization (2.05 Å) and methoxyl radical addition to propene (2.06 Å), is marginally wider than those obtained from BHandHLYP-calculations (1.98-2.02Å; Table 9).Values for the angle α describing oxygen radical attack to the inner alkene carbon are grouped for all calculated transition structures in the range between 98 and 104 degrees, being more acute than the angle calculated for the highest in the energy transition structure TS-trans-Ii ( 3 T 4 ) (121-122 degrees; Table 9).Absolute values of improper torsion angles ω for transition structures TS-Ii, TS-Ij, and TS-VIII, according to B3LYP-and BHandHLYP-theory, are close to 160 degrees, indicating the hybridization change at the attacked carbon from sp 2 (ω = 180°for propene) toward sp 3 (122°for propane).
The computed relative Gibbs free energy of activation for the 2,3-trans-mode of cyclization is 55 kJ mol −1 above the value for the lowest in the energy pathway of 2,3-cis-ring closure (B3LYP; 58 kJ mol −1 for BHandHLYP calculations using either the 6-31+G** or the 6-311G** basis set; ESI ‡).A Gibbs free activation energy difference of 55 kJ mol −1 translates for a kinetically controlled reaction and a temperature of 298.15K into a relative rate constant of 4 × 10 9 in favor of the 2,3-cis-cyclization.Detecting a 2,3-trans-bromocyclized product with such a precision was beyond the capability of analytic instruments used in the study.
(viii) Marcus analysis.4][65] The intrinsic part describes contributions of strain and steric repulsion in a thermoneutral degenerated reaction to the barrier ΔE ‡ i in a transition structure located half way on the reaction coordinate (x ‡ = 0.5) between reactant(s) (x = 0) and product(s) (x = 1; Fig. 7).The thermodynamic part of the barrier ΔE ‡ TD describes energy changes arising from incipient bond forming and bond breaking in a transition structure.
(ix) Localizing transition structuresthe x ‡ -value.According to Marcus theory, reaction energies and barriers obtained from Table 9 Selected geometrical parameters of transition structures TS 1,2cis/trans-Ii, TS-j and TS-VII   .BHandHLYPcomputed energies lead to more negative thermodynamic barriers, but show otherwise similar trends.From the data we concluded that the thermodynamic barrier is not the key parameter for explaining the experimental 2,3-cis-specificity of verbenylethyloxyl radical cyclization.
(xi) The role of the intrinsic barrier.Intrinsic barriers modelled for 2,3-cis-cyclization of cyclohexenylethyloxyl radical Ii (ΔE ‡ i = 35-37 kJ mol −1 , B3LYP; for BHandHLYP-calculated values, refer to Table 10) and 5-exo-cyclization of 4-pentenoxyl radical Ij (37 kJ mol −1 ) are marginally smaller than the intrinsic barrier for methoxyl radical addition to the inner carbon of propene (43 kJ mol −1 ).An intrinsic barrier of 90 kJ mol −1 predicted for 2,3-trans-cyclization of the cyclohexenylethyloxyl radical Ii exceeds the value for the barriers of all other investigated oxygen radical additions in the study by far.From this information we concluded that the experimentally observed 2,3-cis-stereospecificity of the verbenylethyloxyl radical cycliza-    tion originates from a large intrinsic barrier associated with the 2,3-trans-ring closure.

Concluding remarks
Cycloalkyl-fused and -bridged 4-pentenoxyl radicals provide bicyclic and tricyclic tetrahydrofurans by 5-exo-cyclizations.The selectivity determining step is the intramolecular oxygen radical addition to the carbon-carbon double bond, occurring in most instances with notable stereochemical preference.From the observed stereoselectivities we concluded that a system exists, which can be summed up by two new directives for predicting the stereochemical outcome of similar cyclizations not exemplified in this article.The new guidelines supplement the set of existing directives, developed for predicting the major products in synthesis of monocyclic disubstituted tetrahydrofurans by the oxygen radical method. 7,14,67he first of the new guidelines ranks the hierarchy of two similarly sized stereoinductors by the distance between the alkyl group and the alkene carbon which is being approached by the oxygen radical.This guideline states that the substituent positioned the closest to the attacked alkene carbon is the principal ( primary) inductor, guiding 5-exo-cyclization 2,3trans-and 2,4-cis-selectively.The substituent bound further from the attacked π-bond is the secondary inductor, enhancing stereodifferentiation exerted by the principal inductor in the case of the trans-configuration, and decreasing this effect in the case of the cis-configuration.A secondary inductor is not able to overrule the guiding effect of a similarly sized primary inductor.The first guideline applies to 5-exo-cyclization of type-A and type-B 4-pentenoxyl radicals (Fig. 1).
The second new directive states that 4-pentenoxyl radical 5-exo-cyclization to a cyclohexene-bound exo-methylene group or an endocyclic double bond occurs cis-specifically.The second guideline refers to intramolecular addition of type-C-E radicals (Fig. 1).
From the hierarchy of similar-sized inductors we expect a substituent located in position 3 to also control the stereoselectivity in 5-exo-cyclization of 4-pentenoxyl radicals having similar sized substituents attached to carbons 1, 2, and 3.According to the first new guideline, a group in position 2 will be secondary and a group in position 1 the least effective, the tertiary inductor.From today's point of view we expect the stereoisomer of a 1,2,3-substituted 4-pentenoxyl radical corresponding to an all-trans-configured type-A and type-B radical to cyclize with notable 2,3-trans-, 2,4-cis-, and 2,5-trans-selectivity, possibly providing a single diastereomer.In the same model, a sterically more demanding substituent in position 2, for example tert-butyl, should be able to overrule the effect of a smaller group in position 3, such as methyl.Stereochemical questions of this kind attracted our attention and are being pursued at the moment in our laboratory, with the aim to provide new solutions to synthesis of functionalized ethers from oxygen radical addition to alkenes.       4 Hz, 3 H), 2.71-2.80(m, 1 H), 4.15 (d, J = 6.8 Hz, 2 H), 5.35 (ddt, J d = 9.8, 3.0, J t = 1.5 Hz, 1 H), 6.13 (q, J = 1.4 Hz, 1 H). 13

Scheme 3
Scheme 3 Competing reaction pathways for mechanistically interpreting the origin and yields of 5-exo-cyclization-versus β-fragmentationproducts from 2-allylcycloalkyloxyl radicals Ia-c (R = H, CH 3 ; for details associated with rate constants k β , k Br , and k add , see the text).

Fig. 5
Fig. 5 Transition structure models for explaining the origin of 2,3trans-stereoselectivity in 5-exo-cyclization type-B pentenoxyl radicals, exemplified by favored intermediates TS 1 -cis/trans-Id and disfavored intermediates TS 2 -cis/trans-Id.Hydrogen atoms drawn as black circles give rise to 1,3-diaxial repulsion, ecliptically arranged hydrogens are drawn as open circles.a Stereodescriptor referring to the configuration at carbons 2 and 3 in cyclized radical IId (cf.Fig. 3).b Stereodescriptor referring to the configuration at carbons 3 and 4 in TS-Id.

Scheme 9
Scheme 9 Structure formulas of radicals and intermediates associated with the 4-pentenoxyl radical 5-exo-cyclization (top) and methoxyl radical addition to the inner carbon of propene (bottom; for discussion of TS 1 -VII, refer to the text and the ESI ‡).

Fig. 7
Fig. 7 Potential energy curves E(x) for reaction associated with an energy change Δ R E across a barrier ΔE ‡ , having an intrinsic barrier ΔE ‡ i , according to Marcus-theory, and harmonic potentials of the identical curvature for the initial (x = 0) and final state (x = 1).