Branch-Selective Addition of Unactivated Olefins into Imines and Aldehydes

Abstract

Radical hydrofunctionalization occurs with ease using metal-hydride atom transfer (MHAT) catalysis to couple alkenes and competent radicalophilic electrophiles. Traditional two-electron electrophiles have remained unreactive. Herein we report the reductive coupling of electronically-unbiased olefins with imines and aldehydes. Iron-catalysis allows addition of alkyl-substituted olefins into imines through the intermediacy of free-radicals, whereas a combination of catalytic Co(Sal^t-Bu,t-Bu) and chromium salts enable a branch-selective coupling of olefins and aldehydes through the formation of a putative alkyl chromium intermediate.

Graphical Abstract

Hyfrometallation of unbiased alkenes with high branched selectivity by radical-anion crossover

Branch-selective reactions of alkyl-substituted olefins via carbocationic¹ or radical intermediates² benefits from an abundance of methods, but the analogous transformation into branched-carbanion equivalents remains underdeveloped (Figure 1). A common way to transform an olefin into a carbanion equivalent is via hydrometallation of a double bond. However, such branch-selective hydrometallation of alkenes is generally limited to styrenes, allenes, and dienes—all electronically biased systems that stabilize a developing carbon-metal bond.³ In the absence of electronic bias, canonical metal hydrides favor linear selectivity.⁴ To obtain branch-selectivity with electronically-unbiased alkenes, we have investigated M−H hydrogen atom transfer (MHAT) catalysis and subsequent capture of the nascent intermediates by a second metal complex.^5,6,7 For example, we recently established that nickel complexes intercept Co(Sal^t-Bu,t-Bu)-catalyzed MHAT cycles in a direct organocobalt to organonickel transmetallation.⁶ Similar alkyl transmetallations have been reported in non-catalytic systems between alkyl−Co(dmgBF₂)₂Py and inorganic nickel,⁸ and proposed for bioorganometallic⁹ and catalytic processes.¹⁰ This alkyl transfer does not appear limited to nickel: vitamin B12-mimetics (such as Co(salen) derivatives) can undergo alkyl transfer to palladium,¹¹ rhodium,¹² other cobalt,¹³ platinum,¹⁴ gold,¹⁵ chromium,¹⁶ and zinc¹⁷ salts and organometallic species. Yet despite the apparent generality of this transformation, there is a paucity of preparative cross-coupling methods that leverage this reactivity.

Figure 1:

Transformation of olefins into carbanion equivalents by a radical/polar crossover.

The capacity to form cobalt organometallics via MHAT followed by cage-collapse^6,18,19,20 prompted us to explore transmetallation partners that might lead to otherwise inaccessible branched products. Here we show that olefins can be added to imines and aldehydes to form sp³−sp³ bonds. The former reaction with an activated electrophile occurs under standard MHAT catalysis, whereas the latter reaction requires interception of MHAT intermediates with chromium salts (Figure 1e). This transformation expands the current scope of olefins as carbanion surrogates²¹ which has heretofore required the use of electronically-activated olefins, such as styrene, allenyl, or dienes. Alkyl-substituted olefins, in contrast, react with carbonyls at the least-substituted position through a Prins mechanism,²² or undergo iron-catalyzed hydromagnesiation reactions to form linear nucleophiles.^23,24,25

We initially investigated the Markovnikov addition of alkenes into carbonyl derivatives by utilizing the intermediacy of the free radicals and noticed that productive reactions were only obtained with standard radicalophiles, such as radical-stabilizing imines. Glyoxylimines are precedented as radical acceptors,^26,27 and chiral sulfinyl auxiliaries can be used to impart stereocontrol. Although the competitive reduction of these electrophiles by the metal hydrides or the stoichiometric silane was observed, this could be minimized by using a slight excess of the olefin and Fe⁺³ salts as the catalyst.²⁸ Several feedstock alkenes served as competent nucleophilic components and delivered unnatural amino acids derivatives with good to excellent diastereoselectivities. (Figure 2). The early transition state of radical reactions allows facile formation of sterically hindered unnatural α-amino acids bearing β-quaternary carbons, and reactive groups like free-hydroxyls (13) or two-electron electrophiles such as esters, epoxides, or aldehydes (8, 15, and 17) are tolerated. Complex feedstock terpenes can engage the sulfinylimines to deliver adducts 12, 14, and 15, and even glycans deliver amino esters with good diastereocontrol (16). Comparison of the optical rotation obtained from our reaction to that of t-butyl glycine derivatives shows that sulfinimes with the (S)-configuration affords the (S)-amine whereas the (R)-sulfinime affords the (R)-amine.²⁸ Better diastereoselectivity is obtained with the more hindered mesitylene or tri-isopropyl arene-derived sulfinamide, whereas the use of Ellman’s tert-butyl sulfinamide was not compatible with these radical conditions.²⁸ Given the ease with which these compounds are made, requiring no prefunctionalization prior to radical formation, we anticipate that this method will find application in the synthesis of unnatural amino acids.²⁹

Figure 2.

Alkyl radicals generated by MHAT add to chiral sulfinimines. d.r. of two major diastereomers reported. ^astereochemistry at the α-carbon is (S). ^bstereochemistry in the α-carbon is (R). ^cMn(dpm)₃ instead of Fe(acac)₃ was used. ^dcontains 15% unrearranged pinene product and its diastereomer. ^e3 equiv. of olefin used.

Addition of the free radical to aldehydes, however, proved challenging (see Table 1), as may be expected due to the lower stability of an O-centered radical relative to a C-centered radical, which is reflected by the more facile C–C bond scission than C–C bond formation.³⁰ Strategies to drive this energetically disfavorable addition include sequestering the unstable O-centered radical as an alkoxide (which cannot undergo homolytic β-scission) in an intramolecular setting³¹ or accessing excited-states via photochemistry.³² However, neither strategy may be used for intermolecular addition with alkyl-substituted olefins.³³ In light of the precedence for alkyl-cobalt complexes to transmetallate other metal species and the apparent facility with which organocobalt species can form from olefins,^6,20 we wondered if a two-electron nucleophile equivalent could arise from unactivated olefins via sequential one-electron reductions via interception of organocobalt with chromium species.

Table 1.

Conversion of C-centered radicals to 2-electron nucleophiles.

19 (1 equiv.)	20 (2 equiv.)	“standard conditions”	21 70%a (72%)b
Entry	Deviations from above		Yield (%)a
1	Fe(dpm)3, Fe(acac)3, Co(acac)2 or Mn(dpm)3 instead of Co(Salt-Bu,t-Bu)		< 10%
2	CrCl2 instead of CrCl3		trace
3	with Zn° or Mn°		11%
4	Co(salen)Cl and no [O]		45%c
5	no [O]		−
6	0.2 equiv. of CrCl3 instead of 1 equiv.		22%
7	0.2 equiv. of CrCl3 and TMSCl (1 equiv.)		−
8	in DMF instead of THF/CH3CN		−
9	without CH3CN		35%
10	under air		46%
11	with 1 equiv. of H2O		−
12	No Co(Salt-Bu,t-Bu)		−
13	No CrCl3		−d

^ayield determined by GC/FID using a calibrated internal standard

^bisolated yield with 20 mol % of Co(salen)Cl and CrCl₃(THF)₃

^c1 equiv. of NaBF₄ added

^disomerization and hydrogenation was observed; d.r. 1:1 in all cases.

We were drawn to chromium chemistry for several reasons: 1) organochromium reagents are known to add into carbonyls in a 1,2-fashion 2) Cr⁺² salts are proposed to intercept alkyl-radicals to form organochromium species with bimolecular rate constants on the order of 10⁷ M⁻¹s⁻¹,³⁴ 3) alkyl-cobalamines and -cobaloximes can also be intercepted by Cr⁺²,^16,35,36 and 4) chromium salts are inexpensive and of low toxicity in the +2 and +3 oxidation states.³⁷ Furthermore, the weak Brønsted acidity of organochromium complexes allows for a high functional group tolerance and for their use in late-stage functionalization for complex molecule synthesis.³⁸

Initially, attempts to merge MHAT catalysis and chromium chemistry met with poor results. β-diketonate complexes of Co and Mn were not productive, although iron salts afforded the product in low yield (Table 1, Entry 1).³⁹ We discovered, however, that use of Co(Sal^t-Bu,t-Bu) and equimolar amounts of 1-fluoro-2,4,6-trimethylpyridinium tetrafluoroborate in the presence of phenylsilane and CrCl₃ could couple the terminal olefin in 19 to 3-trifluoromethyl benzaldehyde in good yields. Given that Cr⁺² is typically the active species in the addition of alkyl halides into carbonyls, we initially explored the reaction using CrCl₂ or CrCl₃ alongside an external metal reductant only to discover that these conditions led to less product formed than the amount of [Co] pre-catalyst added (Entries 2 and 3). One explanation is that the external reductants impede the Co-cycle by unproductive reduction of Co⁺³ intermediates.⁴⁰ Our optimized conditions appear to circumvent this problem by reductive formation of Cr⁺² in situ (see below). Although it is possible to perform this reaction with the pre-oxidized Co(Sal^t-Bu,t-Bu)Cl, use of Co⁺² and an external oxidant generally afforded higher yields (Entry 4), thereby allowing use of the more convenient +3 and +2 oxidation states of Cr and Co, respectively. Control experiments indicate that both metals are necessary for product formation (Entries 12–13)⁴¹

Evaluation of the scope (Figure 3) revealed that both aromatic and alkyl aldehydes are competent electrophiles, and a wide range of electronic variation is tolerated. In general, electron withdrawn substrates afford higher yields than electron-rich electrophiles, yet even vanillin-derived aldehydes such as 33 and 34 react in high yield. Various heteroaromatic aldehydes may be employed (37 − 39), as well as terpene-derived substrates (41). Esters (45), tosylates (44), and chlorides (46) are orthogonal electrophiles, but competitive reduction of bromides, and iodides was observed. A switch in solvent from THF to DME allows 1,2-disubstituted olefins to be engaged (54 − 57), although trisubstituted olefins are not yet competent. Modest diastereocontrol is imparted by a chiral directing group (49), and sterically bulky substrates (35, 43, 50 and 52).⁴²

Figure 3.

Conversion of alkenes into carbanion surrogates with branched-selectivity. d.r. at the formed bond is close to 1:1 unless otherwise noted. See Supporting Information for the d.r. of the isolated compounds. ^a20 mol% of [Co] and 20 mol% of [F⁺] were used.

Although we currently do not have a complete mechanistic model, several observations are worth noting. First, the yield of the product formed does not vary as a function of delayed Cr⁺³ addition, which is consistent with intermediate formation of a stable organocobalt species that is engaged by the Cr, and inconsistent with an alternative hypothesis that the cobalt cycle continuously generates a C-centered radical whose reactivity would favor formation of side products prior to addition of the CrCl₃.⁴³ These observations draw analogy to our previously reported Ni/Co hydroarylation⁶ and the mechanistic studies of Espenson and coworkers.³⁵ Stoichiometric experiments support a transmetallation and suggest reaction with Cr⁺² rather than the Cr⁺³ species. In these experiments, a sec-alkyl cobalt was formed in situ by displacement of 2-bromopropane by Co^I(Sal^{t-Bu, t-Bu})(py)₂; addition of CrCl₃ and aldehyde 20 yields no product, whereas CrCl₂ produces around 50% of adduct 58 based on the yield of the alkyl-cobalt.²⁸ Control experiments with the alkyl-halide during the same time period yields no product under these conditions. We suspect reduction of Cr⁺³ to Cr⁺² occurs via the stoichiometric silane reductant necessary for the MHAT catalytic cycle.⁴⁴ By analogy to the proposal of Espenson and coworkers in similar systems,³⁵ a possible mechanism for the alkyl transfer could involve electron transfer from a Cr⁺² to an alkyl–Co⁺³ intermediate to form an unstable alkyl–Co⁺² species which is known to homolyze to afford an alkyl radical that could escape the solvent cage and capture a second Cr⁺² species, a kinetically facile process (k = 10⁷ M⁻¹s⁻¹).^35,45,46

In summary, we have discovered divergent reactivity available to alkenes that enables branch-selective (Markovnikov) addition to radicalophilic and non-radicalophilic electrophiles. First, carbon-centered radicals generated by MHAT are competent to add to chiral sulfinimines, which stabilize the incipient N-centered radical and impart stereocontrol. The products of these reactions are valuable, unnatural amino acid derivatives. Second and complementarily, although these same radicals do not productively add into aldehydes, the addition of Cr⁺³ salts allows coupling to occur. This latter method circumvents the poor reactivity of free radicals towards carbonyl intermediates while maintaining the Markovnikov reactivity and chemoselectivity of MHAT. Overall, this work enables cross-coupling of abundant chemical feedstocks (aldehydes and olefins) without the need for pre-functionalization. Mechanistic experiments and analogy to the literature is consistent with alkyl–Co⁺³ transmetallation to alkyl–Cr⁺³, mediated by Cr⁺². This second example^5,6 of catalytic MHAT organocobalt transmetallation calls attention to the potentially general use of these alkyl-cobalt complexes as catalytically-generated organometallic species capable of transferring their alkyl ligands to various other transition metals (including Ni and Cr) for previously inaccessible branch-selective bond-forming processes from olefins. This reactivity complements catalytically-generated organocuprate species which can also engage in hydrometallation/transmetallation, but with linear selectivity.⁴⁷

Figure 4.

Delayed addition and stoichiometric reactions suggest transmetallation of alkyl−Co⁺³ to alkyl−Cr⁺³.