Axially Chiral Enamides: Substituent E ﬀ ects, Rotation Barriers, and Implications for their Cyclization Reactions

: The barrier to rotation around the N -alkenyl bond of 38 N -alkenyl- N- alkylacetamide derivatives was measured ( Δ G ⧧ rotation varied between <8.0 and 31.0 kcal mol − 1 ). The most important factor in controlling the rate of rotation was the level of alkene substitution, followed by the size of the nitrogen substituent and, ﬁ nally, the size of the acyl substituent. Tertiary enamides with four alkenyl substituents exhibited half-lives for rotation between 5.5 days and 99 years at 298 K, su ﬃ cient to isolate enantiomerically enriched atropisomers. The radical cyclizations of a subset of N -alkenyl- N -benzyl- α -haloacetamides exhibiting relatively high barriers to rotation round the N -alkenyl bond ( Δ G ⧧ rotation >20 kcal mol − 1 ) were studied to determine the regiochemistry of cyclization. Those with high barriers (>27 kcal mol − 1 ) did not lead to cyclization, but those with lower values produced highly functionalized γ -lactams via a 5- endo-trig radical − polar crossover process that was terminated by reduction, an unusual cyclopropanation sequence, or trapping with H 2 O, depending upon the reaction conditions. Because elevated temperatures were necessary for cyclization, this precluded study of the asymmetric transfer in the reaction of individual atropisomers. for that


■ INTRODUCTION
In recent years, the bond rotational dynamics, asymmetric synthesis, and reactions of nonbiaryl atropisomers have received considerable attention. 1−3 The majority of work has focused on the chemistry of anilides 1a 4−9 and benzamides 1b, 10−14 where the amide group is perpendicular to the plane of the aryl system ( Figure 1). In 2,6-disubstituted anilide derivatives 1a (R 3 or R 4 ≠ H) rotation around the N-aryl bond is slow enough for atropisomers to be separated at room temperature, and axially chiral anilides 1a have been shown to undergo a range of reactions with transfer of chirality. 3 Secondary enamides are valuable synthetic intermediates, 15,16 but the chemistry of tertiary enamides 1c has received much less attention. 17,18 Systems with small substituents 2 (e.g., R 1 = R 2 = Me) may be planar in the ground state, but if substituents R 1 and R 2 are large enough or the alkene is substituted (R 3 = R 4 = R 5 ≠ H), then enamides 1c have the potential to exhibit axial chirality. Theoretically, if the rotation barrier around the N-alkenyl bond is high enough, then individual atropisomers may be separated and their chemistry studied.
The rotation dynamics of tertiary enamides can be complicated because of restricted rotation around both the amide N−CO [(E)-syn 2 → (Z)-syn 2] and the N-alkenyl bonds [(E)-syn 2 → (E)-anti 2] (Figure 1). The barrier to rotation around the amide N−CO bond for 2a (R 1 = R 2 = Me) has been measured at 14.0 kcal mol −1 and is slightly lower 19 than the general value for amides (15−20 kcal mol −1 ). 20 N-Cycloalkenyl-N-alkylacetamides such a 3 generally prefer the amide N−CO E-rotamer 21 and exhibit transient axial chirality on the NMR time scale. This makes analysis of the rotational dynamics around the N-alkenyl bond relatively easy to study by VT 1 H NMR. 21 In this paper we report studies into the effect of substitution (1c, R 1 , R 2 , R 3 , R 4 , and R 5 ) on the barriers to rotation around the N-alkenyl bond of tertiary enamides and show that it is possible to separate individual enamide atropisomers at room temperature with half-lives (t 1/2 ) of up to 99 years at 298 K. We probe the ability of a subset of sterically congested racemic α-halogenated tertiary enamides with relatively high barriers to rotation (ΔG ‡ 298 rot > 20 kcal mol −1 ) to undergo 5-endo-trig radical cyclization. Those with high barriers (ΔG ⧧ 298 rot > 27 kcal mol −1 ) do not lead to cyclization, but those with lower values (27 kcal mol −1 > ΔG ⧧ 298 rot > 20 kcal mol −1 ) mostly lead to highly functionalized γ-lactams (no β-lactam formation was observed), where the mode of termination is controlled in part by steric factors and in part by the method of cyclization.
■ RESULTS AND DISCUSSION Substitution at the Alkene. Compound 2b (R 1 = Me, R 2 = Bn) was prepared by acetylation of the N-benzylimine of acetaldehyde 22 (13%). Computational analysis of 2b using the TZVP basis set 23 and the B3LYP-D3(BJ) functional 24 using PC-GAMESS/Firefly 8.0 25 suggested that three out of the four possible conformations were planar [the exception being (E)syn-2b] with (E)-anti-2b and (Z)-anti-2b conformations predominating at equilibrium (Figure 2). Electronic energies and zero-point energies for all conformations are provided in the Supporting Information. Two diastereomeric versions of the nonplanar (E)-syn-2b conformation exist, where either the vinyl group or the phenyl group are angled either in front of the plane of the amide or behind it. The energy and relative equilibrium constant of the more stable conformer are shown. In all cases the minimized energy structures had imaginary frequencies of zero. The E+ZPE energy was used to calculate the relative populations at 298 K.
Experimentally, a 2:1 ratio of two conformations was observed in the 600 MHz 1 H NMR spectrum of 2b in CDCl 3 at 298 K. The major isomer was confirmed as (E)-anti-2b and the minor isomer (Z)-anti-2b upon the basis of their calculated theoretical 1 H NMR chemical shifts and NOE data. DFT ground-state structures were analyzed using the GIAO method Gaussian03, 26 with the mPW1PW91 functional 27 and the 6-311+G(2d,p) basis set and scrf = (solvent = chcl3,cpcm,read) radii = uaks nosymcav options. 28 NMR shifts were calculated using parameters specific to the functional, basis set, and option combination described Lodewyk et al. 29 The H-1 enamide proton resonates at 6.80 ppm in the major conformer and 7.50 ppm in the minor conformer, in good agreement with the theoretical calculations. This data is similar to that reported for the related N-benzyl-N-vinylformamide 2c (R 1 = H, R 2 = Bn) 30 and is consistent with a slow rotation around the amide bond and a fast rotation around the C−N bond of the alkene with a higher population of the (E)-anti conformer at equilibrium at room temperature. Heating 2b (298 K → 373 K) in toluene-d 8 caused a broadening of all signals and coalescence to a single set of peaks, consistent with rapid rotation around the amide bond.
In order to assess the effect of increasing alkene substitution on the barrier to N-alkenyl bond rotation, we prepared structures 4a−i ( Figure 3). We chose to study the N-benzyl derivatives, as measurement of rotation rates should be possible using variable-temperature (VT) 1 H NMR. Upon cooling, the benzylic CH 2 singlet should broaden and would be expected to decoalesce (caused by the two protons of the CH 2 group being

The Journal of Organic Chemistry
Article in an asymmetric environment) and ultimately form two doublets due to the diastereotopic nature of the benzylic protons. This behavior is not consistent with amide bond rotation, as this would cause the number of signals to double. Using the WINDNMR 7.1 line shape analysis program, 31 it was possible to determine the rotational rate constant at each temperature and the thermodynamic parameters via a standard Eyring plot. Compounds 4a, 32 4e, and 4f were prepared by acetylation of the known benzyl imines of acetone, 22 2-methyl propanal, 22 and 2-phenylcyclohexanone, 33 while 4b−d and 4g− i were prepared by benzylation (NaH, BnBr) or methylation (NaH, MeI) of the corresponding N-acetyl enamides 5 (16− 80%). 34−37 The barrier to rotation around the C−N bond for 4a and 4e was too low to measure by VT 1 H NMR (ΔG ⧧ 298 estimated to be <8.5 kcal mol −1 ). On cooling, the benzylic CH 2 singlet began to broaden as expected, but even at 179 K decoalescence was not observed. On the other hand, the values of ΔG ⧧ 298 for rotation for the 1,2-substituted derivatives 4b−d in toluene-d 8 were determined to be 16.3, 18.6, and 9.3 kcal mol −1 . As expected, increasing steric congestion around the alkene increases the barrier to rotation due to the inherent difficulty in passing through a hindered planar structure during rotation.
For the derivatives 4f−i, the barrier to rotation was too high to measure with conventional VT experiments. The 1 H NMR showed a sharp set of diastereotopic signals for the CH 2 benzyl protons, even at 373 K. It was not possible to fully resolve the enantiomers of 4f and 4g by chiral HPLC on a variety of columns, indicating that the atropisomers were either inseparable on the columns examined or that they were rapidly interconverting on the HPLC time scale; evidence for the latter is provided vide infra. While compound 4h was partially resolved, both enantiomers of 4i were fully resolved on a semipreparative Whelk-O column (30.1 and 37.4 min). The ΔG ⧧ 298 for rotation of 4i was determined to be 25.6 kcal mol −1 by measuring the kinetics of racemization at 82°C. 43 This suggests that a branching substituent at the sp 3 -hybridized α′position is required to generate suitable barriers to rotation for atropisomers to be successfully separated at room temperature. This is a similar structural motif to that found in anilides 6b (R 3 ≠ H, shown in red with an sp 2 branching point), where barriers to rotation were greater than 26 kcal mol −1 . 40 On the other hand, structures 4b, 4f, and 4g with intermediate barriers to rotation structurally resemble anilides 6a (R 3 = H) and as a consequence have lower barriers to rotation. 39 To further study the effects of branching at the α′-position, we prepared structures 7a−d, investigated their resolution by chiral HPLC, and calculated the barriers to rotation and half-lives of those that could be resolved ( Figure 4).
It was possible to fully separate the enantiomers of 7b (containing both an α-methyl group and an α′-annulated phenyl group) on a semipreparative Whelk-O column (37.09 and 46.47 min), and a ΔG ⧧ 298 K for rotation of 27.5 kcal mol −1 was determined by measuring the kinetics of racemization of a 99:1 enantiomerically enriched sample at 82°C. Compound 7b is a restricted conformational analogue of the acyclic derivative 7a first prepared by Ahlbrecht (7a ΔG ⧧ = 21.3 kcal mol −1 ). 38 The difference in the measured barrier to rotation is significant, δΔG ⧧ = 6.2 kcal mol −1 , and the relatively low value for 7a indicates a likely cooperative gearing of the α′phenyl substituent rotation during the key rotation around the N-alkenyl bond in 7a (Figure 4). The related derivative 7d gave a single peak on the Whelk-O column, and its rotation dynamics were not further analyzed. On the other hand, 7c provided the highest barrier to rotation (ΔG ⧧ = 31.0 kcal mol −1 , t 1/2 = 99 years at 298 K, enantiomers separated at 29.0 and 31.9 min), and this barrier is comparable to published values for o-iodoacrylanilides 6b. 40 We solved the X-ray structures of 7b and 7d, which confirmed the preference for adoption of the amide N−CO E-rotamer geometry in the solid state ( Figure 4). The torsional angle of the key N-alkenyl bond CC−NC(O) in 7b is 74°, which is similar to that observed for related anilides 41 and other enamides. 21,42 The sum of the angles around the nitrogen atom was 359.9°, suggesting that the nitrogen was planar, as expected. For 7d the torsional angle was slightly smaller, 65°. The 400 MHz 1 H NMR of both 7a and 7d in toluene-d 8 showed one pair of mutually coupled sharp doublets (for 7a, J = 14.0 Hz at 5.27 and 3.48 ppm; for 7d, J = 14.0 Hz at 5.37 and 3.37 ppm) which did not broaden upon heating. This is consistent with the existence of largely a single E-amide rotamer in solution, with a high barrier to rotation around the N-alkenyl bond causing the benzyl protons to be diastereotopic, as seen in the crystal structure of 7d. These results suggest that enamides 1c (R 1 , R 2 , R 3 , R 5 ≠ H, R 4 = branched substituent) are likely to have sufficient barrier to rotation around the N-alkenyl bond to be resolvable for useful periods at room temperature.
Substitution at Nitrogen. The effect of the size of the nitrogen substituent 8a−e on the energy barrier for N-alkenyl bond rotation was assessed by VT NMR ( Figure 5). Deprotonation of N-1-cyclohexen-1-yl-benzeneacetamide 43 with NaH in THF followed by the addition of either MeI (8a, 84%), BuBr (8b, 70%), or i PrI (8d, 13%) furnished the desired enamide 8a, 8b, or 8d. 44 Compound 8c was prepared by acylation of the N-benzylimine of cyclohexanone with phenylacetyl chloride according to a literature procedure. 42 We also synthesized compound 8e by reaction of N-1-cyclohexen-1-yl-benzene acetamide with 3 equiv of 2,6-lutidine and TBSOTf ( Figure 5). 45 The moisture sensitivity of 8e required its preparation in situ in toluene-d 8 in an NMR tube, and it was not possible to isolate a pure sample. For compounds 8a, 8b, and 8d a characteristic change in the chemical shift of the alkene proton occurred upon N-alkylation (Δδ between 0.46 and 0.70 ppm), highlighting the loss of conjugation with the nitrogen lone pair in 8a−d. The sense and magnitude of this shift was also observed upon silylation in toluene-d 8 (8e 4.84 ppm), providing evidence that silylation to give 8e had occurred. As expected, the barrier to rotation increased, 8a (10.1 kcal mol −1 ) < 8b, 8c (11.5−11.7 kcal mol −1 ) < 8d (15.0 kcal mol −1 ), as the size of the alkyl group increased, in parallel with the trend observed in anilides 9a−d. 46 In both series there is a relatively small increase on moving from the N-methyl to N-1°alkyl substituent (enamides 8a → 8b δΔG ⧧ = 1.6 kcal mol −1 , anilides 9a → 9b δΔG ⧧ = 1.0 kcal mol −1 ), whereas there is a significantly larger increase for the isopropyl substituent (enamides 8a → 8d δΔG ⧧ = 4.9 kcal mol −1 , anilides 9a → 9d δΔG ⧧ = 5.1 kcal mol −1 ). This is likely due to the R group in the N−CH 2 R substituent being able to rotate away from the plane of the amide and alkene during C−N alkenyl rotation, while this is not possible for the methyl groups N−CH(Me) 2 in 8d. 47 No broadening of the geminal protons of 8e was observed on cooling to 193 K, indicating that rotation is rapid, presumably through the O-silyl imidate 8f ( Figure 5). Silylation of related anilides has been reported to proceed to give both Nsilylated and O-silylated structures that are in rapid equilibrium, 48 and this may be occurring for enamide 8e.
The addition of radicals onto tertiary enamides (derived from aryl bromides 10) has been reported to give tetrahydroisoquinolines 11 via a 6-endo-trig radical cyclization. 49 The bond rotation dynamics of N-2-halobenzyl derivatives has not been investigated. We briefly investigated the effect of o-halobenzyl substituents upon the rotation barrier around the N-alkenyl bond in related systems 13a−f (Table 1). While the effect of such substituents X and Y might be limited (as they are three atoms away from the nitrogen atom), the relatively large size of bromine and iodine atoms (typically used to initiate radical reactions or other metal-mediated cyclizations) may be significant. The compounds 13a−g were prepared by alkylating the known acetamide 12 34 with appropriately functionalized benzyl halides using the same approach as for 8a−e.
The introduction of a single 2-fluorine (13b) or 2-iodine (13c) substituent had little effect on the rotation barrier; the Xray structure of the related 2-bromide 14 clearly showed the halide orientated away from the alkene group, where it does not interfere with the N-alkenyl bond rotation ( Figure 6). On the other hand, benzyl substituents containing two ortho substituents increase the barrier to rotation in line with their size (13d, F = +0.8 kcal mol −1 ; 13e, Cl = +1.6 kcal mol −1 ; 13f, Br = +2.9 kcal mol −1 ) despite their distance from the key Nalkenyl bond rotation. The size of the ring in which the enamide alkene was constrained also affected the barrier to rotation in the order 6-membered < 7-membered (compare 13f and 15 δΔG ⧧ = +1.7 kcal mol −1 ), which is a consequence of the bond angles of the different ring systems. 21 Substitution at the Acyl Group. Previous work with tertiary enamides containing electron-poor and electron-rich aromatic acyl substituents has shown that the electronic nature of the acyl substituent has a negligible effect on the N-alkenyl rotation barrier, suggesting that steric effects alone are important. 34 It has previously been reported that the size of the acyl group (R) in 3a−i only moderately affects the barrier to rotation in enamides ( Figure 7). 21,42 Hence, replacing an acetyl group (3a) (ΔG ⧧ 298 rot = 10.1 kcal mol −1 ) with the much    The Journal of Organic Chemistry Article larger trichloroacetyl group (3i) (ΔG ⧧ 298 rot = 14.2 kcal mol −1 ) increases the barrier to rotation by 4.1 kcal mol −1 . This steric effect is less significant for the acyl substituent than the nitrogen substituent [nitrogen substituent Me (8a) → i Pr (8d), δΔG = 4.9 kcal mol −1 ; acyl substituent Me (3a) → i Pr (3d), δΔG = 1.6 kcal mol −1 ]. 42 Similar behavior has also been reported for anilides. 47 For small R substituents (such as 3a) the barrier to N-alkenyl rotation is likely to be lower than amide N−CO rotation, but for larger substituents (such as 3i), the barriers are likely to be similar. 19 Consequently, a number of potential mechanisms for enantiomerization (E,M-3 → E,P-3) are possible, including simple enamide rotation (E,M-3 → E,P-3) or a cooperative coupled rotation of both amide and enamide, a geared process 50 On plotting ln k rot values for 3a− i against the cone angle (θ R values) 51 of the acyl substituent, we found a linear correlation (R 2 = 0.958), which suggests that the mechanism for rotation is likely to be the same for all of the series and that the cone angle of the acyl substituent may be a useful tool in predicting rotational barriers for a given series of enamides ( Figure 8).
In summary, the most important factors in controlling the rate of N-alkenyl bond rotation in enamides is the level of alkene substitution (particularly any branching substituent at the α′-position), followed by the size of the nitrogen substituent (and the distance of any branching from the nitrogen atom) and, finally, the size of the acyl substituent.
Radical Cyclization Substrate Dynamics. The radical cyclization of α-haloenamides 16−18 is well-documented and may proceed via a 4-exo or 5-endo cyclization, depending upon the substrate (Figure 9). 52−64 Rotational features of enamides can dictate the success or failure of radical cyclizations, with the E-amide rotamer being required. 30 In general, 1-substituted enamides 16 cyclize via a 5-endo pathway (R 2 = Ph or CO 2 R), 52−55 while 2-or 2,2-substituted enamides 17 proceed via a 4-exo pathway, 56−58 although electronic factors and temperature can also play a part in controlling the regiochemistry for these substrates. 58−60 By far the majority of cyclization reactions reported involve 1,2-substituted enamides 18, which generally proceed via a 5-endo pathway. 59−70 For cyclization of radicals derived from homolysis of the C−Br bond in substrates 3c and 3h, 66−70 it is necessary for a twisting to occur in the transition state so that the radical SOMO and alkene LUMO orbitals can overlap efficiently, and this has been confirmed by calculations. 71 As a consequence, the twisted ground state of molecules such as 3c and 3h likely facilitate cyclization with a movement toward planarity occurring during the reaction ( Figure 9). Far less is known about cyclizations of 1,2,2-substituted enamides, where movement toward planarity during cyclization may be hindered due to steric effects. 59,60,72 Cyclization of 19 was reported to be unsuccessful using Bu 3 SnH at 80°C, while cyclization of the analogue 21 gave both 4-exo 23 and 5-endo products 22, depending upon the temperature ( Figure 10). 72 This suggests that the 4-exo cyclization process was reversible, and at the higher temperature, 5-endo cyclization followed by irreversible loss of the phenylthiyl radical predominated to give 22. From our studies above it is apparent that these substrates are likely to have ground-state N-alkenyl bond rotation barriers of less than 20 kcal mol −1 .
To the best of our knowledge the cyclization of enamides containing further branching at the α′-position have not been studied in detail ( Figure 10). For these substrates, where barriers to rotation around the N-alkenyl bonds are likely to be significant (>20 kcal mol −1 ), it is unclear if cyclization would be efficient because steric interactions would develop as the radical and radical acceptor move toward planarity during the cyclization process. In order to address these issues, we prepared enamides 24−28, examined their barriers to rotation and investigated their radical cyclization reactions ( Figure 11).

Article
The most commonly used protocols for mediating radical cyclizations of α-haloenamides are (i) Bu 3 SnH/ AIBN, 52−54,62−65 where chlorides provide higher yields of cyclized products than bromides or iodides, 70 and (ii) copper(I) complexes of bipyridine, 57 hexamethyltriethylenediamine, 66 or tripyridylamine, 56,57,67−69 which work for trichloroacetamide or tertiary halide derivatives only (Bu 3 SnH, R = CH 2 Cl, CH 2 Br, CH 2 I; Cu(I), R = CCl 3 , CMe 2 Br). 64 Consequently, we tested α-haloenamides 24−28 under a range of conditions and determined the regiochemistry of their radical cyclization reactions ( Figure 11). It was not possible to fully resolve the atropisomers of 24−26, despite the large acyl substituents. We believe this is indicative of individual isomers interconverting on the HPLC time scale at room temperature. Evidence for this assumption can be obtained from the HPLC chromatogram for compound 26a, which is indicative of the individual enantiomers interconverting at room temperature ( Figure 12d). The ΔG ⧧ value for rotation for 26a was estimated to lie between 21 and 23 kcal mol −1 based upon a half-life of minutes at 298 K. In all these cases, the cooperative gearing of the phenyl substituent during N-alkenyl bond rotation is likely to lower the observed barrier to rotation. 73 For relatively large acyl substituents (25b, 26b, R = CCl 3 ; 25c, 26c, R = CMe 2 Br) the barrier for N−(CO) amide bond rotation was found to be lower than that of N-alkenyl bond rotation ( Figure 12b indicating that the population of the two rotamers was dictated by the acyl substituent. Heating either 25b,c or 26b,c at 373 K led to coalescence of the four sets of doublets to two broad singlets, indicative of a rapid interconversion on the NMR time scale between the (E)-and (Z)-amide rotamers with a slower rotation around the enamide N−C bond (Figure 12a). The barrier to rotation around the N−(CO) amide bond is known to be influenced by the acyl substituent with sterically demanding (t-Bu) groups lowering the barrier by 4−5 kcal mol −1 compared to simple acetamide derivatives. 74 The X-ray crystal structure of 25c clearly shows the (Z)-amide rotamer in the solid state (Figure 12c). A similar doubling of signals was also observed in the 1 H NMR of the related structure 27c (1:0.17). Thus, for enamides containing four alkenyl substituents, the (E)-geometry is favored in both solution (CDCl 3 ) and the solid state for relatively small primary acyl groups (R = Me, CH 2 Cl, CH 2 Br, cone angles 112°−130°), but mixtures of both (E)-and (Z)-amide rotamers can be detected in solution if larger acyl groups or strongly electron withdrawing groups (R = CMe 2 Br, CCl 3 , cone angles 153°−160°) are present.
It was possible to fully separate the enantiomers of 27a−c on a semipreparative Whelk-O column, allowing for barriers to rotation to be calculated (27a, er = 99:1, ΔG ⧧ = 27.8 kcal

The Journal of Organic Chemistry
Article mol −1 , t 1/2 = 163 days at 298 K; 27b, er =98:2, ΔG ⧧ = 29.3 kcal mol −1 , t 1/2 = 5.6 years at 298 K; 27c, er = 87:13, ΔG ⧧ = 25.6 kcal mol −1 , t 1/2 = 4 days at 298 K), suggesting that these compounds would make interesting substrates with which to investigate chirality transfer during 5-endo-trig radical cyclization reactions at room temperature. If cyclization is significantly more rapid than N-alkenyl bond rotation, then chirality transfer from individual atropisomers is theoretically possible. 1 Radical cyclizations of related axially chiral o-haloanilides have been shown to proceed with high levels of chirality transfer from the chiral axis to the newly formed stereocenter. 39,40,75−79 If, however, elevated temperatures are necessary for cyclization (e.g., 80°C), then only bromide 27b or iodide 27d are likely to exhibit a suitable barrier to rotation (27b t 1/2 = 18.5 h at 353 K) compared to 27a (2 h) and 27c (7 min). It is interesting to observe that 27c has a lower barrier to N-alkenyl bond rotation than either 7b or 27a,b despite having a larger acyl substituent. The fact that it exists as a mixture of (E)-and (Z)-amide isomers in solution suggests that the lower N-alkenyl barrier may be due to a cooperative gearing with rotation of the amide functional group. However, in anilides it has been reported that the steric repulsion between bulky substituents causes pyramidalization of amide nitrogen and twisting of the amide bond (destabilization of the ground state) to bring about the decrease in the rotational barrier around an N−C chiral axis. 73 While we cannot disprove that this is the reason for the lowered barrier of 27c, it is unlikely, as the related substrate 25c shows no such pyramidization in its X-ray structure (Figure 12c).
Radical Cyclization Reactions. We have previously shown that a slow syringe pump addition of Bu 3 SnH/AIBN to the chloride 3j produces mainly 33 (92%) from reduction of the cyclized radical 30 ( Figure 13). 66 Reactions of the bromide 3c and iodide 3k proceed differently. While the bromide 3c gave 33 as the major product (55%), the alkene regioisomers 34 (11%) and 35 + 36 (11%) were also isolated. The iodide 3k gave the uncyclized material 32 (68%) as the major product with 34 (11%) and 35 + 36 (13%). The different product ratios were explained via a competing electron transfer from the intermediate radical 30 to the starting halides 3c and 3k, giving the acyl iminium ion 31. Elimination of a proton from 31 produces the three alkene regioisomers 34−36. The higher ratio of reduced product 32 from the iodide 3k was shown to be due to a competing nonradical deiodination process. 70 By analogy, we initially chose to investigate the Bu 3 SnHmediated cyclization of the primary chlorides 26a, 27a, and 28 representing varying levels of hindrance to rotation around the N-alkenyl bond. The chlorides were chosen to suppress products arising from electron transfer from the cyclized radical to the starting chloride.

The Journal of Organic Chemistry
Article mol −1 ) failed, presumably due to the increased difficulty in moving to planarity during the cyclization. Reaction of the chloro derivative 27a with Bu 3 SnH (0.01M) and Et 3 B 79 at room temperature led to recovered starting material 27a (44%) and the reduced substrate 7b being isolated (70% based upon recovered starting material). On the other hand, addition of 1.5 equiv of Bu 3 SnH and 20 mol % AIBN (via a syringe pump addition for 2 h, initial concentration 0.01 M) led to oxidation to the naphthalene 38a in 19% yield (47% based upon recovered starting material). Complete oxidation to the naphthalene 38a could also be accomplished if 27a was reacted with 2 equiv of AIBN in the absence of Bu 3 SnH, indicating that the radical initiator was responsible for the oxidation and that this process was more rapid than homolytic fission of the C−Cl bond. 80−82 The same oxidation to give 38b was observed for the iodide 27d. In order to enhance the rate of radical initiation over that of oxidation, we investigated the reaction of the bromide 27b. Over 18 h, 1.5 equiv of Bu 3 SnH and 0.2 equiv of ACN were added via a syringe pump to a 0.12 M solution of the bromide 27b under nitrogen in dry toluene at reflux. Thin layer chromatography revealed that significant amounts of starting material remained after 24 h, indicative of a slow initiation, so further Bu 3 SnH/ACN was added. Three further aliquots of the initiator (1 equiv in total) over 48 h were required for complete consumption of the starting material. Upon workup, three products were isolated: the uncyclized acetamide 7b (42%), the naphthanilide 38c (7%), and the ketone 39a (10%), arising from oxidative cleavage of the tetralone ring ( Figure 15). Repeating the reaction using degassed toluene, in a Schlenk tube under argon, led to suppression of 39a (trace amounts), and only the naphthanilide 38c was isolated (along with recovered starting material). Although the mechanism for the formation of 39a remains unclear, the oxidative α-C−C bond cleavage of 2-substituted 1tetralones to give 40 under radical conditions (TEMPO) has been reported. 83 The mechanism has been reported to involve addition of oxygen to give an α-hydroperoxide. Radical or anionic fragmentation, Baeyer−Villiger/Criegee fragmentation, or base-induced fragmentation of the peroxide can all potentially lead to the observed products (although a theoretical study indicates that a radical fragmentation is most likely). 83 In order to remove the complication of undesired tetralone cleavage, the reaction of 28 was investigated where the analogous oxidation was not possible. Reaction with Bu 3 SnH and ACN led to formation of uncyclized 7c in 88% yield as the sole product.
While it is most common to carry out Bu 3 SnH-mediated reactions at elevated temperatures (typically benzene or toluene at reflux), efficient 5-endo radical−polar crossover reactions (3h → 35 and 36 R = Me) mediated by Cu(I) complexes have been reported at room temperature for tertiary halides 3h ( Figure  13). 66 The ability to conduct such cyclizations at lower temperature is attractive if chirality transfer during reaction of individual atropisomers is to be studied. Another advantage of generating the reactive radical from homolysis of a C−X bond via a Cu(I)-mediated atom transfer is that this initiating process is reversible, effectively providing longer "lifetimes" and less reduction of precyclized radicals than alternative Bu 3 SnHmediated processes. As before, the cyclization of substrates with ΔG ⧧ 298 rot < 26 kcal mol −1 proceeded smoothly, albeit at elevated temperatures. Reaction of 24a with either 30 mol % of Cu(Me 6 -tren)Cl or Cu(TPMA)Cl in toluene or DCM at room temperature only led to trace amounts of 41a (<5%) after 48 h. On the other hand, heating of 24a with 1 equiv of Cu(TPMA) Cl in toluene at reflux led to complete conversion to 41a in 70% yield. We next investigated the cyclizations of 25b and 26b using the optimum conditions determined for 24a [100 mol % Cu(TPMA)Cl], Figure 16.
Cyclization of both substrates was complete in 2 h and produced the expected dichlorides 42a and 43a along with the monochlorides 42b and 43b as single diastereomers (assigned from the 1 H NOE difference spectrum of 43b). Both sets of products arose from trapping of the intermediate acyl iminium ion by water (as seen for the Bu 3 SnH reaction of 26a). The monochloride 42b is likely formed by reduction of the α-amide radical obtained from a second atom transfer from 42a to Cu(TPMA)Cl or by electron transfer from the cyclized radical to 42a (the driving force being the relief of the eclipsing

The Journal of Organic Chemistry
Article interactions between the gem-dichloride group and the neighboring quaternary center).
Reaction of 25c with 1 equiv of Cu(TPMA)Br in toluene at reflux for 10 h produced three products, 44, 45, and 46, in 36%, 15%, and 4% isolated yields, respectively ( Figure 17). The major product 44 arises via the intermediate acyl iminium ion 47, as previously observed for 25b, while the reduced product 45 may arise from abstraction of a hydrogen atom from toluene by the cyclized radical. This would indicate that electron transfer in the radical−polar crossover step was slower than in the trichloroacetyl derivatives 25b and 26b. The formation of the cyclopropyl compound 46 deserves comment. There is a significant steric clash between the C-3 and C-4 gem-dimethyl groups in the acyl imimium ion intermediate 49. In order to relieve these clashes, torsion of the C-3 to C-4 C−C bond can occur, which places the C-4 pseudoaxial methyl group almost parallel to the p-orbital of the acyl iminium ion 49, initiating a rearrangement to form a protonated cyclopropane intermediate 48. Formation of a cyclopropane by loss of a proton during a 1,2-migration of a methyl group has been previously reported, 84,85 and this process would give rise to the observed product 46. More of the cyclopropyl derivative was isolated from the analogous reaction of 26c (50, 51, and 52 being formed in 33%, 21%, and 5% yields, respectively). Presumably, the more-electron-rich nature of the methylene group in 26c compared to the methyl group in 25c is responsible for the greater yield of trapped migration product 51 ( Figure 18).
Reaction of the tetralone derivative 27c was relatively slow compared to the other substrates (25c and 26c) and required heating at reflux for 44 h with 1.2 equiv of Cu(TPMA)Br before all the starting material was consumed. This substrate exhibited the highest barrier to N-alkenyl bond rotation of the substrates cyclized using Cu(TPMA)Br. Two products were isolated, the expected cyclized product 53 and the oxidatively ring-opened compound 39b (in a 2:1 ratio). Repeating the reaction in a Schlenk tube with degassed solvent (three freeze− thaw cycles) suppressed the formation of 39b to trace levels (<2%), indicating that dissolved O 2 was most likely responsible for mediating the unusual ring-opening process to give 39b. The reaction was more efficient, requiring less copper reagent (0.6 equiv) and a shorter reaction time (15 h). In addition, two further products, 54 and 55, were isolated in 31% and 7%, respectively (although it was not possible to obtain 55 completely free of impurities) ( Figure 19). Unfortunately, attempts to conduct the reaction of 27c at a lower temperature (353 K) provided trace levels of cyclized products only, while the barrier to rotation around the N-alkenyl bond of 27c at 383 K was too low (t 1/2 = 18 s) to warrant investigating whether chirality transfer from one atropisomer to the cyclized products was possible.
As with the formation of 39a, from the Bu 3 SnH-mediated reaction of 27b, the mechanism for the formation of 39b remains unclear, although the oxidative α-C−C bond cleavage of 2-substituted 1-tetralones with either CuCl or CuCl 2 and amine bases in the presence of O 2 has been reported. 83

■ CONCLUSIONS
In conclusion, we have prepared 38 different enamides 1c varying in the substitution around the acyl group 1 (R 2 ≠ H), the nitrogen substituent (R 1 ≠ H), and alkene substituents (R 3 , R 4 , R 5 ≠ H). The majority show slow rotation around the Nalkenyl bond of the enamide with ΔG ⧧ 298 rot barriers varying between <8.5 and 31.0 kcal mol −1 . The most important factor in controlling the rate of rotation is the level of alkene substitution (R 3 , R 4 , R 5 ), followed by the size of the nitrogen substituent (R 1 ) and, finally, the size of the acyl substituent (R 2 ). Electronic effects are small for substituents positioned on the acyl group (R 2 ), and the rate of rotation was linearly correlated with the cone angle of the substituent for the studied series 3a−i, indicating that the mechanism of rotation was likely to be the same for all the compounds. For N-benzyl groups 13a−f, o-halo-substitution has little effect on the barrier to rotation, except where two ortho substituents are present, indicating that cooperative gearing of the nitrogen substituent is likely during the N-alkenyl rotation. A similar gearing for freely

The Journal of Organic Chemistry
Article rotatable α-substituents (Ph in 7a) is also indicated. Tetrasubstituted alkenes possessing rigid α,α′-substituents (7b,c) have high enough barriers to rotation at room temperature to be separated by chiral HPLC with half-lives of up to 99 years at 298 K. This should theoretically enable future investigations into asymmetry transfer from enantiomerically pure enamides in a range of synthetic processes. For enamides 25b,c, 26b,c, and 27c containing large or electron-withdrawing acyl groups (cone angles θ R > approximately 130°) both (E)-and (Z)amide rotamers were detected in solution at room temperature. Interconversion of these two rotamers was found to be fast on the NMR time scale at 80°C for all the examples studied. An examination of the barriers to rotation for the series 7b and 27a−c shows that while there is a steady increase between with increasing size of the acyl substituent, as expected (7b, ΔG ⧧ While radical cyclization of enamides 3h−j with relatively low barriers to rotation around the N-alkenyl bond (ΔG ⧧ 298 rot = 13−14 kcal mol −1 ) can be accomplished at room temperature, 66 in this study those with higher barriers (ΔG ⧧ 298 rot > ∼20 kcal mol −1 ) required elevated temperatures, presumably due to the extra steric crowding hindering these molecules movement toward planarity during cyclization. Molecules with very high barriers to rotation (ΔG ⧧ 298 rot > ∼26 kcal mol −1 ) did not undergo radical cyclization with Bu 3 SnH; instead, precyclization reduction or alternative reaction pathways predominated. For molecules with intermediate barriers to rotation (ΔG ⧧ 298 rot ∼ 20−26 kcal mol −1 ), highly functionalized γ-lactams were produced via a 5-endo-trig radical−polar crossover process and terminated either by reduction, an unusual cyclopropanation sequence, or trapping with H 2 O, depending upon the reaction conditions. Steric congestion in cyclized products derived from trichloroacetamide derivatives 25b and 26b was relieved by a competing second atom transfer and reduction leading to replacement of one α-chloro substituent with a hydrogen atom (42b and 43b). Unfortunately, because elevated temperatures were necessary for cyclization of substrates, in this study it precluded the examination of asymmetric transfer in the reaction of individual atropisomers such as 27d. However, this report does indicate that enantiomerically enriched atropsiomeric tertiary enamides should be regarded as potential asymmetric building blocks for reactions that can be accomplished at room temperature or below and may be useful functional groups in molecular machines and gears. 1 ■ EXPERIMENTAL SECTION General Methods. 1 H NMR spectra were recorded at 300, 400, 500, or 700 MHz and 13 C NMR spectra were recorded at 75.5, 100, 125, or 175 MHz with residual solvent as standard; infrared (IR) spectra were recorded as neat solutions or solids; and low-and highresolution mass spectra were recorded using the electrospray ionization technique and a TOF mass analyzer.

H NMR was used to check the purity of all the compounds.
General Procedure for the Formation of Enamides 2b, 24a,b, 25b,c, 26b,c, via Imines, Method A. Ketone (1 equiv) and benzylamine (1.0 equiv) with or without TsOH were dissolved in dry toluene and heated to reflux under Dean−Stark conditions for 4−16 h. The reaction mixture was then cooled to 0°C. Triethylamine (1.2 equiv) was then added slowly, followed by the dropwise addition of the appropriate acid chloride (1.1 equiv). The reaction mixture was allowed to warm to room temperature and stirred for 12 h. NaHCO 3 (∼50 mL) was added, and the layers were separated. The aqueous phase was extracted with Et 2 O (3 × 50 mL), and the organic layers were combined, dried over MgSO 4 , filtered, and concentrated in vacuo to give the crude product. Products were purified by column chromatography. N-Benzyl-2-bromo-2-methyl-N-(2-methyl-3,4-dihydronaphthalen-1-yl)propanamide (27c). 2-Methyl-1-tetralone (1.00 g, 6.24 mmol), benzylamine (0.818 mL, 7.49 mmol), and titanium isopropoxide (2. 77 mL, 9.36 mmol) were heated to 80°C for 48 h. The reaction mixture was then cooled to room temperature and toluene (12 mL) was added. The resulting solution was then cooled further to 0°C, triethylamine was added (1. 30 mL, 9.36 mmol), followed by the dropwise addition of 2-bromoisobutyryl bromide (0. 93 mL, 7.49 mmol), and the reaction mixture was stirred at room temperature for 14 h. Saturated ammonium chloride (15 mL) was then added, and the phases were separated. The organic phase was diluted with toluene (20 mL) and washed with 2 M HCl (3 × 30 mL), dried over MgSO 4 , filtered, and concentrated in vacuo to give the crude product as a brown oil (1.26 g). The crude product was purified by column chromatography (14:1, pet. ether:EtOAc) to give the product as a light brown oil (144 mg, 6%) as a 6:1 mixture of amide rotamers. R f 0. 17  General Procedure for the Formation of Acetamides 5a−j. To the oxime (1 equiv) was added acetic acid (3 equiv), acetic anhydride (3 equiv), and iron powder (2 equiv) in anhydrous toluene. The mixture was heated at 70°C for 4−16 h, cooled to room temperature, and filtered through Celite. Dichloromethane was added, and the organic extracts were washed with 2 M NaOH (3 × 30 mL) and brine (30 mL). The organic phase was dried over MgSO 4 , filtered, and then concentrated in vacuo. Products were purified by column chromatography or recrystallization as reported. (E)-And (Z)-N-(but-2-en-2yl)acetamide (5a, X = H, R 1 = R 2 = Me, R 3 = H; 5b, X = H, R 1 = R 3 = Me, R 2 = H), 34