Generation of α-Boryl Radicals by H• Transfer and their Use in Cycloisomerizations

Abstract

Carbon-centered radicals can be stabilized by delocalization of their spin density into the vacant p orbital of a boron substituent. α-Vinyl boronates, in particular pinacol (Bpin) derivatives, are excellent hydrogen atom acceptors. Under H₂, in the presence of a cobaloxime catalyst, they generate α-boryl radicals; these species can undergo 5-exo radical cyclizations if appropriate double bond acceptors are present, leading to densely functionalized heterocycles with tertiary substituents on Bpin. The reaction shows good functional group tolerance with wide scope, and the resulting boronate products can be converted into other useful functionalities.

Graphical Abstract

Carbon-centered radicals can be stabilized by delocalization of their spin density into the vacant p orbital of a boron substituent. α-Vinyl boronates, in particular derivatives of pinacol (Bpin), are therefore excellent hydrogen atom acceptors, generating α-boryl radicals from H₂ in the presence of a cobaloxime catalyst. These radicals undergo 5-exo radical cyclizations onto appropriate double bonds, eventually furnishing densely functionalized heterocycles with tertiary substituents on Bpin. The reaction shows good functional group tolerance, and the resulting unsaturated boronates can be easily converted into other useful functionalities.

Radical cyclizations are an efficient method for introducing molecular complexity in a single step.¹ Our groups have long been interested in the applications of H• transfer from transition-metal hydrides onto various olefins — especially those that carry our radical cyclizations.² We now report that H• transfer to α-vinyl boronates from H₂ in the presence of a cobaloxime catalyst can carry out intramolecular cycloisomerizations. Our method provides rapid access to functionalized five-membered heterocycles such as pyrrolidines, which are present in many natural products such as those shown in Figure 1.³

Figure 1.

Natural products with pyrrolidine moiety.

Organoboron compounds are intermediates for many useful transformations, including the frequently used Suzuki-Miyaura cross-coupling reaction for C-C bond formation.⁴ Vinyl boronates are easily accessible through transition-metal-catalyzed borylation of alkynes,⁵ vinyl halides⁶ or allenes,⁷ can react via ionic or radical pathways, and have recently attracted much attention. Attack on the vinyl boronates by either Pd(II) (Morken)^8a or an alkyl radical (Studer^8b and Aggarwal^8c) can induce 1,2 migration of R (alkyl or aryl) from the boron to the α-carbon, and bring about the formation of two C─C bonds and the difunctionalization of the original alkene.⁸

The empty p orbital of a boron atom can stabilize an adjacent carbon-centered radical through π-conjugation.⁹ Calculations indicate that the radical stabilization energy (RSE) of BH₂-CH₂• is about 11 kcal/mol, but the oxygens of a pinacolate (pin) ligand also interact with the empty boron orbital and decrease its ability to delocalize the unpaired electron; the RSE of Bpin-CH₂• is thus only about 5 kcal/mol (Scheme 1).¹⁰

Scheme 1.

Radical Stabilization by Adjacent Boron

α-Boryl radicals are typically formed as intermediates in radical chain reactions. Halogen-atom abstraction from α-haloboronates, as shown in Scheme 2A, is generally performed with Bu₃Sn• (which requires the use of a toxic tin hydride).¹¹ The related reaction in Scheme 2B, reported by Zard and co-workers, cleaves a C─S bond in a boryl xanthate by adding an initiating (or a chain-carrying) radical to thiocarbonyl of the xanthate.¹² The Baran group has generated boryl radicals by H• transfer to olefins, apparently from an in situ generated Fe hydride;¹³ the radicals then add to electron-deficient olefins, as shown in Scheme 2C.¹⁴

Scheme 2.

Generation and Application of α-Boryl Radicals

We have also generated radicals by H• transfer to olefins, but we have obtained the H• from H₂. Substrates have included enol ethers,^2c acrylamides,¹⁵ N-vinyl indoles,¹⁶ and 1,1-disubstituted (with aryl and CF₃) olefins,¹⁷ and the transformations of these substrates have included cyclohydrogenations, cycloisomerizations, and hydrodefluorinations. In this article, we demonstrate the generation of boryl radicals by H• transfer from H₂ to to vinyl boronates. H₂ is a more practical, cheaper, and greener — especially for industrial applications — source of hydrogen atoms than the silanes which have become common in synthetic laboratories.

We have found Co(dmgBF₂)₂(MeOH)₂ (dmg = dimethylglyoximato) to be quite effective as a catalyst for this reaction, and Co(dmgBF₂)₂(THF)₂ even more so. The readily prepared substrate 1a gives the cycloisomerization product 2a (Table 1), surely via the formation of the radical A and its cyclization. No isomerization, hydrogenation or cyclohydrogenation side products have been observed.¹⁸ Control experiments (entries 2 and 4) show that both cobalt and H₂ are indispensable. (Using several atm of H₂ maximizes the generation of the intermediate responsible for H• transfer.¹⁹) Co(III)salen shows no catalytic activity, and CpCr(CO)₃H/H₂^2a gives a mixture of unidentified products. Screening of concentration, catalyst loading, and solvent gave the following conditions as optimal: 5 mol% Co(dmgBF₂)₂(THF)₂ and 4.8 atm H₂ in benzene (0.10 M) at 50 °C for 3 days.

Table 1.

Optimal Conditions for the Cyclization of the α-Vinyl Boronate 1a (eq 1)^a

^aConditions: 1 (0.2 mmol), 5.0 mol% catalyst C-1, 4.8 atm H₂, benzene (0.1M), 50°C, three days.

^bDetermined by ¹H NMR with internal standard.

^cUnder 6.1 atm H₂.

We have explored the scope of our cobaloxime catalyst with other aryl-substituted olefins as radical acceptors. The allylic boranes produced by these cyclizations are generally not stable enough for purification, so we have oxidized them to the allylic alcohols shown in Table 2. Neither steric hindrance (entry 2), nor an electron-donating substituent (entry 3), nor a heterocyclic substituent (entry 4) interfered with the reaction. Good results were also obtained when the nitrogen atom was replaced with an oxygen, making the tetrahydrofuran derivative in entry 5.

Table 2.

Scope of the Cyclization Reaction with an Aryl-substituted Olefin as Radical Acceptor^a

Conditions: 1 (0.2 mmol), 5.0 mol% Catalyst (C-1), 4.8 atm H₂, benzene (0.1M), 50°C, three days.

^aYield for cycloisomerization reaction determined by ¹H NMR with internal standard.

^bIsolated yield for the oxidized product based on the cycloisomerized boronate intermediate.

With alkyl substituents on the acceptor olefin, we also obtained cyclization; however, we found that our cobalt catalyst abstracted H• from the less hindered carbon of the cyclized radicals, leading to the cycloisomerized products in Table 3. (These were more stable than the aryl-substituted products, but decomposed upon storage in the refrigerator.)

Table 3.

Scope of the Cyclization Reaction with an Alkyl-substituted Olefin as Radical Acceptor^a

Conditions: 1 (0.2 mmol), 5.0 mol% Catalyst (C-1), 4.8 atm H₂, benzene (0.1M), 50°C, three days.

^aYield for cyclization reaction determined by ¹H NMR with internal standard

^bIsolated yield for the oxidized product.

^cratio determined by crude ¹H-NMR with internal standard.

As the cyclized radicals are less stable than the benzylic ones formed in Table 2, a slightly higher catalyst loading (10 mol%) and H₂ pressure (5.6 atm) are required to get complete conversion of the substrate. No second cyclization occurred with the unconjugated diene 1h or the unconjugated triene 1i, presumably because of the strain that would be introduced by a 4-exo cyclization. Varying the substituent on the nitrogen (1j–1l) had little effect on the reactivity of these substrates. Th bicyclic scaffold 2m was easily constructed. Sensitive functionality, including an aryl bromide (2n) ,iodide (2o), and a silyl enol ether (2p), was tolerated under these reaction conditions. The diastereoselectivity of these transformations will be discussed later (vide infra).

The analogous oxygen-containing substrates also worked well (1q-t in Table 4), indeed with enhanced diastereoselectivity. The reaction with 1s can be performed on a 2.0 mmol scale without significant loss of efficiency. The terminal TBS-protected (Z)-allylic alcohol 1u did not cyclize at all, giving only the isomerized substrate.

Table 4.

Scope of the Cyclization Reaction with Oxygen Instead of Nitrogen^a

Conditions: 1 (0.2 mmol), 5.0 mol% Catalyst (C-1), 4.8 atm H₂, benzene (0.1M), 50°C, three days.

^aYield for cyclization reaction determined by ¹H NMR with internal standard

^b2.0 mmol scale.

In an effort to understand the why 1u did not cyclize, we prepared a pair of isoelectronic substrates, 1v and 1w, differing only by the substitution of O for NTs. The nitrogen-containing 1v did not observably react under our standard condition, whereas its oxygen analogue 1w largely isomerized to 2w (Scheme 3). We presume that the radicals 1v’ and 1w’ are reversibly formed by the initial H• transfer (under D₂ 1v undergoes 85% deuterium exchange into its methylene, Scheme 4). However, the C–H of 1v’ is somewhat stronger than the C–H of 1w’: Density Functional Theory calculations (see SI for detail) give a BDFE of 47.6 kcal/mol for 1v’, compared to one of 45.0 kcal/mol for 1w’. (H• transfer to the cobaloxime from either 1w’ or 1v’ is downhill, as the BDFE of the cobaloxime–H bond is 49.6 kcal/mol.²⁰) Removal of H• by Co^II should thus be substantially easier for 1w’ than for 1v’, and it is understandable that isomerization will compete more successfully with cyclization for 1w’.

Scheme 3.

Isomerization vs Cycloisomerization

Scheme 4.

Deuterium Exchange into Substrate.

Treatment of 1x with two equiv of CpCr(CO)₃D for three days at 50 °C gave about 80% H/D exchange. Collectively, these D incorporation experiments support a reversible HAT mechanism for the generation of α-boryl radicals from vinyl boronates.²¹

The cyclized alkyl boronates, such as 2s, offer a platform for late-stage functionalization of 5-membered-ring heterocyclic products. For example, 1) 2s can be easily oxidized to an alcohol with peroxides;²² 2) Aggarwal’s lithiation method²³ can install a furan substituent on the scaffold of 2s; 3) 2s can be converted into its more stable BF₃K salt, which can be used for subsequent Suzuki cross-coupling reactions.²⁴

Beckwith and Houk have established a chair-like transition state for 5-exo radical cyclizations (Scheme 5),²⁵ with the bulkier substituent occupying the pseudoequatorial position. The Aggarwal group has recently quantified the A value for different boronic esters,²⁶ and concluded that the A value of Bpin is about 0.4 kcal/mol — much smaller than the A value of a methyl group (which is about 1.7 kcal/mol). The transition state TS1 should thus be favored in our cyclizations, which will lead to the cis diastereomer as the major product; the minor (trans) product has been identified by 2D-NOESY. A shorter C–X (X = O or NTs) bond will lead to a more congested transition state because of an enhanced 1,3-diaxial interaction,²⁷ which in turn will increase the diastereoselectivity. Our experimental results are consistent with this assumption.

Scheme 5.

Origin of diastereoselectivity.

In conclusion, we have developed a cobalt-catalyzed method for the HAT-initiated cycloisomerization of α-vinyl boronates. Our way of generating α-boryl radicals is simple and easy to use, with H₂ as the only stoichiometric reagent; it avoids the toxic Bu₃SnH and the waste associated with the use of silanes. Pyrrolidine and tetrahydrofuran derivatives are easily made, and the reaction can be done on a large scale. The B─C bond in the cyclized products provides a platform for many late-stage transformations. Finally, deuterium labelling experiments support an HAT-initiated radical mechanism.

Figure 4.

Synthetic Application of Cyclic Alkylboronates.