Frustrated Radical Pairs for Selective Functionalization of Aliphatic CH Bonds
Frustrated Lewis pairs (FLPs) are well documented in the activation of small molecules such as H2 and CO2 . Although typical FLP chemistry is outcrossed in nature, recent studies have shown that some FLPs can generate radical pairs via single-electron transfer. Due to steric burden and / or weak bonding, these radicals do not annihilate each other, hence they are named as frustrated radical pairs (FRPs) . Noteworthy preliminary results suggest that FRPs may be useful reagents in chemical synthesis, although their application remains limited. Here, the authors demonstrate that functionalization of C( sp 3 )-H bonds can be accomplished with a class of FRPs generated from diazide donors and N – oxoammonium acceptors . Together, these species undergo single-electron transfer, generating transient and persistent radical pairs capable of cleaving unactivated CH bonds to provide aminated products. By tuning the structure of the donor, regioselectivity and reactivity toward tertiary, secondary, or primary bonds can be controlled. Mechanistic studies strongly support the formation and participation of free radical pairs in targeted reactions.
Transformation of aliphatic carbon (C)-hydrogen ( H) bonds has become an increasingly powerful tool for the expedient upgrading of chemical feedstocks and polymers and the late-stage modification of bioactive compounds. However, the strong and ubiquitous C( sp 3 )-HSelective functionalization of bonds remains challenging. Although many elegant strategies exist for efficient functionalization of aliphatic C-H bonds, systematic tuning of site selectivity is only possible in selected cases. Recent pioneering reports in this regard have shown that domain differentiation of C( sp3 )-H functionalization can be achieved using novel hydrogen atom transfer (HAT) agents, transition metal catalysts, or directing groups. Despite these advances, there remains considerable interest in developing strategies to achieve tunable site selectivity and overcome intrinsic substrate bias, especially in reactions of complex molecules. Importantly, the study of selective, non-directional activation of primary CH bonds ( typically the least reactive site in organic compounds ) is still primitive, with the few known examples focusing only on borylation and carbene insertion chemistries.
Frustrated Lewis pairs (FLPs) are a well-established class of complexes comprising a strong electron acceptor ( such as borane ) and a strong electron donor ( such as phosphine ) ( Fig. 1a) . Due to steric hindrance or mismatched orbital energies, FLPs retain their Lewis acidic and basic features and can act synergistically in a heterofission manner to break strong covalent bonds [ such as HH and C- oxygen (O) bonds ] . Therefore, these species have been used for small molecule activation in organic synthesis. It is worth noting that recent studies have shown that some typical FLPs can also form frustrated free radical pairs (FRPs) with one-electron reactivity through a single-electron transfer (SET) reaction ( Fig. 1b) . Although studies of FRPs to date have focused on elucidating their structure and formation mechanisms, Stephan , Melen , and OoiIt was found that radical pairs generated by established FLPs ( phosphine / borane or amine / borane ) can promote the homocleavage of organostannyl tin (Sn)–H bonds, the benzmethylation of alkenes and alkynes, and the α- alkylation of silyl-substituted amines, respectively. These pioneering studies have focused on the activation of relatively weak chemical bonds; however, an FRP capable of cleaving strong, non-activated aliphatic CH bondsStill elusive. More broadly, the development of novel FRPs for organic synthesis is largely uncharted territory and represents a unique opportunity to discover strategies to address challenging problems in synthesis, such as site selection. functionalized hydrocarbons. Here, the authors describe a FRP that consists of a transient hydrogen atom acceptor (HAA) and a persistent free radical trap generated by an oxidation – reductant pair ( Fig. 1c) . HAA is able to cleave strong aliphatic CH bonds, and persistent free radicals can quickly capture alkyl radicals subsequently formed on HAT . This reaction provides a versatile intermediate that can be further derivatized into synthetically useful products.
In their quest for FRPs to achieve CH -bond activation chemistry , the authors identified the sterically burdened hexamethyldiazide anion (HMDS − ) and the N – oxoammonium ion 2,2,6,6- tetramethyl – 1 – oxo – piperazine (TEMPO + ) as promising precursors ( Fig . 1c) . In fact, HMDS – can be oxidized to generate nitrogen (N) -centered free radicals, due to its strong NH bond dissociation energy (BDE = 109 kcal·mol -1, while the BDE of cyclohexane = 98 kcal mol -1 ) , which is an effective HAA ; density functional theory (DFT) calculated the BDE and strong electrophilicity induced by the hyperconjugation of single-occupied molecular orbitals with low-lying σ*C-Si or Si(3d) orbitals. Notably, the disilylaminyl radical has not had any synthetic applications other than its initial discovery in the context of radical chain halogenation. Meanwhile, N – oxoammonium ions are widely recognized as mild oxidants and generate persistent aminooxyl radicals that react with alkyl radicals near the diffusion limit. The authors envision that SET from HMDS- to TEMPO + will generate FRP consisting of transient HMDS • and persistent TEMPO • . Transient radicals can induce homolysis of aliphatic CH bonds, while persistent TEMPO will react with early carbon-centered radicals to formally split CHkey. Another advantage of this reactivity is that the resulting product alkyl -TEMPO adducts have weak CO and NO bonds ( BDEs of cyclohexyl -TEMPO are 49 kcal mol −1 and 51 kcal mol −1 , respectively, according to DFT calculations ) , allowing subsequent synthetic diversification through radical or polar transformations.
The authors found that the conjugation of readily available lithium hexamethyldiazide (LiHMDS) and TEMPO + BF 4- to alkanes in trifluorotoluene solvent indeed resulted in the amination of aliphatic CH bonds ( Fig. 1d) . After system optimization, two protocols were developed for efficient CH amination of abundant hydrocarbon feedstocks such as cyclohexane ( compound 1) and more complex substrates such as (-)- ammonium bromide ( compound 2) at ambient temperature and under simple conditions. Subsequent treatment of the resulting reaction mixture with m-chloroperbenzoic acid (mCPBA) or zinc powder resulted in the corresponding ketones ( compounds 3 and 4) or alcohols ( compounds 5 and 6) as final products, respectively . Thus, this approach allows for distinct carbon-hydrogen oxidations of carbonyl or alcohol products with high chemical fidelity, which can be challenging to accomplish using related oxidation methods.
Next, the authors explored the substrate scope of the method for a variety of simple and complex compounds containing inert and activated CH bonds ( Fig. 2) . Aminoacylation of a group of cycloalkanes ( compounds 1 , 7-11 , S13 and S14) went well. Linear alkanes including ethane ( compounds 12-14) were converted to the desired products in situ with minimal steric hindrance. The successful functionalization of α- or β-CH bonds in silanes ( compounds 15 , 16 and S17) provides a potential approach for the chemical modification of silicone polymers. Various electronically different arenes with phenyl CH bonds are also good substrates ( compounds 17-32 and S16-S19) . In particular, the reaction tolerates useful functional groups such as halides ( compounds 18-20 , 29and S20) , nitriles ( compounds 21 and S18) , trifluoromethyl ( compound 22) , methyl esters ( compound S16) , nitro ( compounds 29 and S17) , and aryl boroesters ( compound 23) , as well as heteroaromatics including 2- methylfuran ( compound 30 ) , dimethylpyridine ( compound 31) and 1,3- dimethylpyrazole ( compound 32) . Allyl and propyl substrates ( compounds 33 , 34 and S20) yielded higher yields. Furthermore, in saturated heterocyclic rings, the α to heteroatomThe CH bonds were selectively oxidized ( compounds 35 and 36) . Selected simple substrates ( compounds 1 , 7 , 8 , 23 , S16 , 33 , 34 , and 35) were also performed under alkane-limiting reagent conditions, and the desired products were synthesized in useful yields. The authors’ method also successfully functionalized cyclododecane compound 10 on the gram scale with an increased yield of 77% .
Finally, the authors tested the reaction on structurally more complex substrates. For example, (-)- ambromide derivatives ( compound 37) , (+)- longifolene ( compound 38) and protected neomenthol ( compound 42) were converted to functionalized products with high selectivity, often at the most sterically accessible secondary sites. In addition, (-)- aminobromide ( compound 2) , methyl dehydroabietate ( compound 39) and biflavonoid analogs ( compound 40) were selectively aminated at the α- ether or phenyl position with excellent selectivity. The antihistamine loratadine ( compound 41) produced two phenylaminated products in a ratio of 1.3:1 , with a total yield of 52% .
The authors were intrigued to find that for certain substrates with multiple types of CH bonds ( compounds 12 , 13 , and 25) , primary sites were preferentially functionalized over thermodynamically favored secondary and tertiary sites. The authors hypothesize that the selectivity arises from the steric hindrance of HMDS • and thus it can be tuned by changing the steric hindrance profile of HAA ( Fig. 3) . To test this hypothesis, the authors synthesized a designed lithium dinitride library ( Fig. 3a) and indeed found that when the bulkier lithium hexaphenyl dinitride (LiHPDS) was used instead of LiHMDS , the primary hydrocarbon Compound selectivity. For example, LiHMDS mainly achieves the secondary hydrocarbon functionalization of isopentane ( compound 54) (50% selectivity ) , while LiHPDS mainly achieves the amination product (93% selectivity ) .. Even between the two main sites, the less hindered site was preferentially activated ( 2.4:1 yield selectivity , 4.8:1 CH bond number normalized selectivity ) . The authors also used potassium t -butoxide (KO t Bu) instead of disilazide , a potent but less sterically hindered HAA . In fact, TEMPO + is able to oxidize tert-butanol even though the electron transfer is moderately thermodynamically uphill ( Ep /2 (KO t Bu) = 0.70 V and Ep /2 ( TEMPO) = 0.46 V , Fig. 4 , Ep /2 is the half-peak potential ) . Use of KO t Bu in the functionalization of compound 54The highest selectivity (68% selectivity ) is for the sterically hindered but thermodynamically favored tertiary CH bonds .
Using this series of sterically distinct HAAs ( Fig. 3a) , the authors evaluated various substrates containing primary, secondary, and tertiary CH bonds. The structure – selectivity relationships were consistent across substrates , with HPDS • showing high selectivity for the most accessible CH bonds, and tBuO • showing high selectivity for the weakest CH bonds ( Fig. 3b) . When branches are introduced into the hydrocarbon chain, such as isopentane ( compound 54) , 2,2 -dimethylbutane ( compound 55) , 2,2 -dimethylpentane ( compound 56) and s- butylbenzene ( compound 59) , HPDS •The ability to distinguish secondary and primary bonds is most evident. Notably, functionalization of statistically favorable tert-butyl primary CH bonds was minimal ( compounds 55 and 56) , and even weak phenyl CH bonds were replaced by more accessible sites ( compounds 57 , 59 and 60) , again highlighting the steric sensitivity of HPDS compared to known HAT agents .
For complex bioactive substrates, a clear shift in site selectivity was observed depending on base choice, suggesting broader utility of this strategy for discriminating CH bonds ( Fig. 3b) . For example, the methyl ester of dehydroabietic acid ( compound 39) interacts with HMDS only at the secondary phenyl position •The CH amination reaction occurs , while tBuO • and HPDS • generate products by activating the most favorable thermodynamic site and the most accessible distal ring system, respectively. Interestingly, when the natural products nootkatone ( compound 61) and sclareolide ( compound 62) were combined with LiHMDS , acidic α- carbonyl CH functionalization was observed, which the authors attribute to an alternative polar pathway initiated by LiHMDS via α- deprotonation. Switching the base to the bulkier LiHPDS greatly suppresses this two-electron pathway, likely by slowing down the deprotonation kinetics and leading to a more inert functionalization of the long-range CH bonds. Finally, while the substrate loratadine ( compound 41) preferentially reacts with HMDS to produce aniline products, the amine is only obtained in the presence of tBuO, probably at the weakest allylicHAT on the CH bond and then eliminated.
The authors further demonstrated the synthetic utility of their method by using a set of known and novel transformations to derive alkoxyamine products ( Fig . 2b) . For example, previous studies have shown that redox-active TEMPO substituents can be transformed into a variety of functional groups. Oxidation with mCPBA or magnesium monoperoxyphthalate converted secondary carbamic acid adducts to ketones ( compounds 50-52) in good yields. Zinc-mediated reductive NO bond cleavage proceeds smoothly to generate the corresponding alcohols ( compounds 48 and 49) . The authors also developed three methods for converting alkoxyamine products. First, formal desaturation of hydrocarbons was achieved by electrochemically driven elimination to give olefinic compounds 45 and 46 . The mechanism may be the oxidation-induced carbocation formation of compound 46 through the cleavage and oxidation of the CO bond . Additionally, halogenation of phenyl TEMPO adducts using selective fluorine or trichloroisocyanuric acid (TCCA) affords fluorinated ( compound 43) and chlorinated ( compound 44)product. Finally, radical deuteration was also achieved by thermally induced homolysis of weak phenyl CO bonds using deuterated thiophenols ( compound 47) . It is worth noting that literature 46 ( bottom left of Fig . 2b ) and literature 49 also explored alkyl -TEMPO species in different photo- or electrochemically activated substitution reactions.
Finally, the authors performed a series of experiments to confirm the formation of FRPs and their role in the observed CH activation. Cyclic voltammetry analysis showed that LiHMDS , LiHPDS , and TEMPO were oxidized at similar potentials ( E p/2 ≈ 0.44-0.46 V vs. Ag + /Ag) , suggesting a bimolecular SET is thermodynamically feasible ( Fig. 4a) . Meanwhile, although the oxidation of tBuO – by TEMPO + has a certain uphill thermodynamics ( E p/2 ≈ 0.70 V) , it may be driven by the fast irreversible downstream reaction of the instantaneous tBuO • . In fact, the in situ formation of TEMPO was detected by electron paramagnetic resonance (EPR) spectroscopy after mixing the FRP precursors •, electron transfer between all three base donors and TEMPO + can be clearly seen ( Fig. 4b) . The co-generated dianilinyl or t – butoxyl radicals were not observed due to their short lifetimes, consistent with previous reports on FRPs in which only the more persistent radicals of the pair were detected. However, using styrene ( compound 63) as a radical trap, trapping of HMDS • and TEMPO • was observed to form the nearly bifunctionalized product 64 ( Fig. 4c) .
To support the role of free radicals in CH functionalization, the authors performed a free radical clock experiment ( Fig . 4d) . The ring-opened product of cyclopropane compound 65 ( compound 66) was observed , but no ring-integrated product was detected, strongly suggesting the formation of a carbon-centered radical via HAT . Furthermore, DFT analysis of HAT site selectivity for selected substrates using HMDS • , HPDS • or tBuO • as HAAs provided predictions that were largely consistent with experimental data . In addition , in the intermolecular competition experiments of cyclohexane and cyclohexane – d 12 , the kinetic isotope effects of HMDS • , HPDS • and tBuO • were 5.0 , 3.8 and6.9 , consistent with the calculated predicted value of the HAT mechanism.
Finally, the authors use DFT calculations to help understand the properties of FRPs . Two possible covalent complexes composed of TEMPO/HMDS pairs were identified ( Fig. 4e) . In addition to a simple adduct (A) with a NO bond between TEMPO and HMDS , a helical ring structure with a four-membered heterocycle ( adduct B) is predicted to be energetically feasible, reminiscent of the authors’ previous observation TEMPO -N 3 complex. In addition to having weak chemical bonds ( i.e., NO and NN) , these structures exhibit strong steric repulsion between the two components, with their closest H···H contacts within 2.2 Å , smaller than the van der Waals radius of two hydrogen atoms Sum. Therefore, if formed, these adducts are expected to have very short lifetimes. In fact, both adducts A and B are predicted to dissociate spontaneously into HMDS • /TEMPO •Free radical pair ( the changes in Gibbs free energy of dissociation are Δ G dis = -6.2 kcal·mol -1 and Δ G dis = -4.7 kcal·mol -1 , respectively ) . The TEMPO/ tBuO complex was also computationally localized ( adduct C) , characterized by a newly formed NO bond between the two components . Although tBuO is smaller than HMDS , its complex with TEMPO will also be sterically unstable ( the closest H··H contact is within 2.1 Å ) , and will spontaneously decompose to form tBuO • /TEMPO • FRP (Δ G dis = -7.5 kcal·mol-1 ) . Insights from these mechanisms will serve as the basis for the strategic design and application of FRPs to address various other synthetic challenges.
