Regioselective aliphatic C-H functionalization using frustrated radical pairs

Abstract

Frustrated Lewis pairs (FLPs) are well documented for the activation of small molecules such as dihydrogen and carbon dioxide^1-4. Although canonical FLP chemistry is heterolytic in nature, recent work has shown that certain FLPs can undergo single-electron transfer to afford radical pairs⁵. Owing to steric encumbrance and/or weak bonding association, these radicals do not annihilate one another, and they have thus been named frustrated radical pairs (FRPs). Notable preliminary results suggest that FRPs may be useful reagents in chemical synthesis^6-8, although their applications remain limited. Here we demonstrate that the functionalization of C(sp³)-H bonds can be accomplished using a class of FRPs generated from disilazide donors and an N-oxoammonium acceptor. Together, these species undergo single-electron transfer to generate a transient and persistent radical pair capable of cleaving unactivated C-H bonds to furnish aminoxylated products. By tuning the structure of the donor, it is possible to control regioselectivity and tailor reactivity towards tertiary, secondary or primary C-H bonds. Mechanistic studies lend strong support for the formation and involvement of radical pairs in the target reaction.

Fig. 1 |. Background and introduction to FRPs.

Traditional FLP chemistry (a) and the extension of this concept to FRP chemistry (b). c, Approach developed in this work for the design of FRPs for the selective functionalization of C(sp³)–H bonds. The components of the redox pair were chosen on the basis of the following criteria: (i) the oxidant and reductant should undergo facile SET, generating a transient radical (HAA) and a persistent radical that do not recombine due to steric or electronic frustration, (ii) the HAA must be capable of cleaving strong aliphatic C–H bonds, and (iii) the persistent radical must rapidly capture the ensuing alkyl radical formed upon HAT. d, Proof co concept: aliphatic C–H aminoxylation using the FRP. For abundant hydrocarbon feedstocks such as cyclohexane (1), an excess of the substrate (5–20 equiv) was used in combination with 1 equiv of TEMPO⁺BF₄⁻ and 1.25 equiv of LiHMDS. For more precious complex substrates such as (–)-ambroxide (2), the alkane was used as the limiting agent with an excess of the two salts. In all cases, the addition of a small amount of TEMPO (0.2 equiv) was found to help improve the reaction yield, presumably by facilitating trapping of the alkyl radical at the initial stage of reaction with increased concentration of the radical coupling partner (see Supplementary Information Section 4.1 for optimization details). Footnotes: ^aYield determined by gas chromatography. ^bIsolated yield.

Fig. 2 |. Reaction development.

Substrate scope (a) and product diversification (b). For a, reactions were carried out on 0.2 mmol scale at 21 °C for 30 min, and the structures drawn were the only regioisomers observed, unless otherwise noted. Blue circles indicate the position at which the C–H aminoxylation took place. For products 1 and 7–36, reaction conditions were generally as follows: 5.0 equiv of substrate, 1.0 equiv of TEMPO⁺BF₄⁻, 1.25 equiv of LiHMDS, 0.2 equiv of TEMPO, and PhCF₃ (0.1 M). Underlined values are yields using alkane substrate as the limiting reagent (1.0 equiv) with 2.0–3.0 equiv of TEMPO⁺BF₄⁻ and 2.25–3.25 equiv of LiHMDS. For products 2 and 37–42, all reactions used the alkane substrate as the limiting reagent (1.0 equiv) with 2.0–3.0 equiv of TEMPO⁺BF₄⁻ and 2.25–3.25 equiv of LiHMDS, 0.2 equiv of TEMPO, and PhCF₃ (0.1 M). b, The aminoxylated products were readily converted to various further functionalized compounds (see Section 9 of Supplementary Information for details). Footnotes: ^a20.0 equiv of substrate. ^bYield determined by ¹H NMR. ^c1-mmol scale. ^dGram-scale (5 mmol). ^e15.0 equiv of substrate. ^fYield determined by gas chromatography.

Fig. 3 |. Divergent site selectivity via tuning of the HAA in the FRP.

a, A panel of three sterically distinct HAA radicals (^tBuO^•, HMDS^•, or HPDS^•) and their steric parameters. Sterimol parameters (B₁ and B₅) and percent buried volumes (%V_bur) were calculated using the geometries optimized at the M06–2X/6–31+G(d,p) level (See Supplementary Information Section 8). b, Divergent regioselectivity for C–H functionalization using FRPs containing ^tBuO^•, HMDS^•, or HPDS^•. See Section 5.7 of the Supplementary Information for experimental details.

Fig. 4 |. Mechanistic studies.

a, Cyclic voltammograms of TEMPO, LiHMDS, LiHPDS, and KO^tBu. Data were collected in an o-difluorobenzene solution (2 mM analyte) in the presence of TBAPF₆ as the electrolyte (0.2 M) using a Ag/AgNO₃ quasi-reference electrode at a scan rate of 100 mV/s. The half-peak potential E_p/2 is the potential where the current is half of the value of the peak current. The onset potential E_onset is defined as the potential where two tangent lines from the rising current and baseline current intersect (See Section 7.1 of Supplementary Information for experimental details). b, EPR spectra of TEMPO^• (red trace) and TEMPO⁺BF₄⁻ mixed with LiHMDS, LiHPDS, or KO^tBu (purple, blue, and gray traces, respectively). All spectra were collected at 100 K. It is clear that reactions between the oxidant/base pairs form FRPs containing TEMPO^•. c, Radical trapping experiments provided further evidence for the formation of FRPs comprising persistent TEMPO^• and transient disilylaminyl radical. d, A radical probe experiment confirmed that the FRP generated from LiHMDS and TEMPO⁺ is responsible for the observed C–H functionalization. e, Computationally predicted structures of TEMPO/HMDS and TEMPO/^tBuO adducts, showing severe steric repulsion between the constituents that favors dissociation into frustrated radical pairs. Gibbs energy changes are reported in kcal/mol. TEMP = 2,2,6,6-tetramethylpiperidin-1-yl.