Directing-group-free catalytic dicarbofunctionalization of unactivated alkenes

Abstract

In the absence of directing auxiliaries, the catalytic addition of carbogenic groups to unactivated alkenes with control of regioselectivity remains an ongoing challenge in organic chemistry. Here we describe a directing-group-free, nickel-catalysed strategy that couples a broad array of unactivated and activated olefins with aryl-substituted triflates and organometallic nucleophiles to afford diarylation adducts in either regioisomeric form, in up to 93% yield and >98% site selectivity. By switching the reagents involved, the present strategy may be extended to other classes of dicarbofunctionalization reactions. Mechanistic and computational investigations offer insights into the origin of the observed regiochemical outcome and the utility of the method is highlighted through the concise syntheses of biologically active molecules. The catalyst control principles reported are expected to advance efforts towards the development of general site-selective alkene functionalizations, removing the requirement for neighbouring activating groups.

Multicomponent catalytic processes that can form multiple carbon–carbon bonds in a single step under mild conditions are highly sought-after for generating molecular complexity in chemical synthesis, particularly if the catalysts and substrates are inexpensive¹. One such class of reactions involves the transition-metal-catalysed installation of carbogenic units across readily available alkene feedstocks to access functionalized sp³-hybridized carbon frameworks^2–8, These entities are found within countless natural products, drug molecules and other important compounds of interest^9,10. A central challenge associated with these transformations is the ability to control the regioselectivity of olefin addition^6,11, especially in situations when the two newly introduced motifs possess similar electronic and/or steric attributes. This problem is further exacerbated with unactivated alkyl-substituted C=C bonds¹² which, by virtue of their diminished electronic and steric bias, are innately less reactive. Consequently, inseparable product mixtures arising from poor regiochemical control and other competing reactions are inevitable^13,14. To this end, two primary strategies have been adopted for promoting efficient and site-selective dicarbofunctionalizations across aliphatic olefins, both of which entail contrived substrate modifications^15,16 (Fig. 1a): (1) substrate control in which intramolecularity drives carbometallation to form a cyclic intermediate and (2) auxiliary control with a strategically positioned directing group that coordinates and stabilizes the metal centre in the putative organometallic species. In the context of three-component 1,2-diarylation, various chelating groups have been developed to direct additions across unactivated C–C π bonds, using catalytic amounts of a transition-metal complex in conjunction with a haloarene and an arylmetal(boron) reagent (Fig. 1b)^17–20. Despite these remarkable advances, conceiving a general approach to carry out regioselective dicarbofunctionalization that is compatible with both unactivated and activated alkenes, without relying on substrate modifications, is in crucial demand^15,21. To achieve this, we wondered if catalyst control could be exploited as a new strategy in reaction design to overcome substrate reactivity and selectivity limitations for diarylation (Fig. 1c). If successful, this would significantly enhance the way in which many biologically active molecules (for example 5 and 6) are prepared, by removing the requirement for pre-installation and removal of pendant-directing auxiliaries, which would otherwise incur additional reaction steps and costs^22,23.

Fig. 1 |. The significance and challenges of designing a catalyst-controlled blueprint for olefin dicarbofunctionalization.

a, Current approaches that induce high regioselectivity in alkene dicarbofunctionalization involve either substrate control or auxiliary control. b, Representative directing auxiliaries developed to promote site-selective 1,2-diarylation of unactivated alkenes. c, Catalyst control as a key strategy in the design of regioselective directing-group-free diarylation reactions. Of the two competing carbonickelation transition states, TS A is favoured over TS A′ due to greater steric repulsions between the alkene substituents and the sterically encumbered ligand (L) in the latter. Successful implementation of catalyst control offers easy access to complex molecules such as 5 and 6. R, R′ and G, functional groups; L, ligand; M, metal; Ar and Ar′, aryl groups; DG, directing group; LG, leaving group; TS, transition state.

As illustrated in Fig. 2a, an organonickel complex i, bearing a sterically encumbered ligand L, has to be identified in order to promote oxidative insertion with an appropriate aryl electrophile 2 to afford ii. This is followed by coordination with alkene 1 to trigger branched-selective nickel-aryl addition (that is, an arene is added onto the more substituted end of the olefin) to give alkylnickel species iii. Rather than relying on directing-group effects¹⁶, we postulated that the preferential formation of iii (versus regioisomeric iii′) is likely to be caused by the elevated steric repulsions between the olefin substituents and L in the energetically less favoured transition state TS A′ (compare with Fig. 1c) leading to iii’. At this stage, iii is poised to undergo transmetallation with a suitable arylmetal nucleophile 3 to form iv, before the ensuing reductive elimination regenerates i and delivers the final adduct 4 in high regioselectivity. Based on this mechanistic proposal, the projected sense of regioselectivity is opposite to those disclosed previously^13,14 (compare with Fig. 1a,b). To selectively obtain the alternative regioisomer of 4, the aryl moieties of 2 and 3 could be simply switched. In addition, the versatility of replacing 2 and 3 with different combinations of electrophiles and nucleophiles means that a broader assortment of dicarbofunctionalization adducts beyond diarylated molecules may be potentially generated.

Fig. 2 |. Mechanistic rationale and reaction optimization.

a, Proposed mechanism for a catalyst-controlled strategy to deliver 1,2-diarylated products with high regioselectivity. The reaction of i and 2 affords ii, which preferentially undergoes branched-selective arylnickelation to give iii (instead of iii′). Subsequent transmetallation with 3 and reductive elimination furnishes 4 via iv and turns over the catalytic cycle. β-H elimination to give 7 via v is a common side reaction that needs to be suppressed. b, Reaction conditions identified for efficient 1,2-diarylation without a directing unit. Poor conversion to 4a was observed with Ni(0) and Ni(II) complexes. G, functional group; L, ligand; Ar and Ar′, aryl groups; LG, leaving group; Tf, trifluoromethanesulfonyl; Bn, benzyl; cod, 1,5-cyclooctadiene; IPr, 1,3-bis(2,6-diisopropylphenyl)imidazol-2-ylidene; IMes, 1,3-bis-(2,4,6-trimethylphenyl)imidazol-2-ylidene; r.r., regioisomeric ratio.

However, implementation of this catalyst-controlled blueprint presents onerous challenges on several fronts. First, the catalyst must be sufficiently sizeable to induce high regiocontrol, yet sufficiently active to promote aryl additions to unactivated alkenes. With the sterically more demanding 1,1-disubstituted olefins, reaction efficiency may be further reduced as observed in previous reports^18,20. Second, in the absence of a chelating auxiliary to occupy a vacant metal site²⁴, iii may be susceptible to adventitious β-H elimination via v to afford Heck-type side products such as 7. Thus, L must be adequately large to preclude any agostic C_β–H bond coordination. Third, transmetallation (from iii to iv) and reductive elimination must outcompete β-H elimination and/or undesired hydroarylation pathways arising from reactions with trace moisture²⁵.

Results and discussion

Identification of an effective catalytic system.

The union of aliphatic 1,1-disubstituted olefin 1a with a phenylmetal nucleophile and 4-tolyl-substituted electrophile (both aryl substituents are electronically and sterically comparable) was selected as the model reaction to optimize conditions. Following an extensive survey of different reagents, Ni-based complexes, ligands, solvents and temperature (see Supplementary Section 5 for details), the best results were obtained when 1a was treated with 5 mol% of a dimeric N-heterocycliccarbene(NHC)-derived Ni(I) complex Ni-1 (refs.^26,27) in the presence of 4-tolyl triflate 2a (2 equiv.), phenylmagnesium bromide 3a (3 equiv.) and toluene as solvent at 40 °C for 5 h, furnishing 4a in 67% isolated yield and >98% site selectivity, along with 9% olefin isomerization by-product 8 (Fig. 2b). Of particular note, replacing 3a with phenylzinc chloride 3b afforded 4a as a single regioisomer in similar yield (63%), whereas changing the solvent or aryl electrophile only led to trace amounts of 4a (significant biaryl homocoupling and cross-coupling by-products were detected in most cases; see Supplementary Section 5 for details). This is note-worthy given the difficulty of generating quaternary stereogenic centres in previous dicarbofunctionalization protocols^18,20.

As described in Fig. 2a, the absence of the minor regiosiomer 4a′ as well as Heck-type by-product 7′ (from linear-selective arylnickelation followed by β-H elimination) suggests that olefin arylnickelation is likely to be branched-selective and precedes transmetallation in the conversion from ii to iv via iii (see Fig. 3a for further discussion). By contrast, reactions using other Ni(0)-or Ni(II)-based NHC complexes were appreciably less efficient; ∼40% 4a (61–66% conversion) was obtained in the presence of 10 mol% Ni-2 or the Ni(0) species derived from Ni(cod)2 and an IPr ligand, whereas minimal reaction was detected with Ni-3 (<5% conversion). Notably, diarylation with the (IMes)Ni(0) complex derived from Ni(cod)₂ and the sterically smaller IMes ligand only afforded 4a in 15% yield and 90:10 regioisomeric ratio. The lower site selectivity (versus >98:2 regioisomeric ratio with IPr) is attributed to the weaker steric interactions between IMes and the alkene substituents, consequently leading to a smaller energy difference between TSA and TSA′ (compare with Fig. 1c) and poorer regiochemical outcome. These observations provide support for our catalyst control model presented in Fig. 1c.

Fig. 3 |. Mechanistic validation and application to complex molecule synthesis.

a, The underpinning factors that impact the efficiency and site selectivity of 1,2-diarylation. In cases where Ar and Ar′ have similar electronic and steric attributes, the exceptional regiochemical outcome is likely to arise from branched-selective olefin arylnickelation via monoarylnickel ii (instead of biarylnickel vi, which is prone to reductive elimination to give undesired 8). b, Computational studies were performed at the UB3LyP-D3/def2svp-CPCM(toluene) level of theory for the catalytic diarylation of 2-methylpropene (model system) and free energies are reported in kcal mol⁻¹ (298 K). The results show that branched-selective arylnickelation is kinetically favoured over linear-selective arylnickelation. X = Cl in the first cycle; X = OTf in subsequent cycles. c, The directing-group-free strategy enables the straightforward synthesis of regiodefined diaryl-containing building blocks such as 4bm–4bo, which are important precursors of biologically active compounds. Organozinc reagents were employed in these diarylation examples. See Supplementary Section 3 for details. G, functional group; L, ligand; M, metal; Ar and Ar′, aryl groups; LG, leaving group; Tf, trifluoromethanesulfonyl; TBS, tert-butyldimethylsilyl; Boc, tert-butoxycarbonyl; IPr, 1,3-bis(2,6-diisopropylphenyl) imidazol-2-ylidene; DeAD, diethyl azodicarboxylate; TBAF, tetrabutylammonium fluoride; TS, transition state; RT, room temperature.

In a separate set of experiments, subjecting a vinylcyclopropane substrate to 1,2-diarylation with Ni-1 delivered the desired adduct in 53% isolated yield (40% recovered substrate) with only trace amounts of cyclopropane ring cleavage and/or cyclopentene side products^28–30 (see Supplementary Section 6 for details). On the contrary, conducting the reaction in the presence of Ni-2 afforded inseparable ring-opened products in 70% yield along with 20% of a cyclopentene (9% recovered substrate), whereas the transformation with Ni-3 was unproductive. These markedly different outcomes with Ni-1, Ni-2 and Ni-3 suggest that adventitious formation of the corresponding Ni(0) or Ni(II) species is negligible in the Ni-1-mediated reaction. Control experiments also revealed that olefin intermediates are not involved during the course of the Ni-1-catalysed 1,2-diarylation, suggesting that the reaction does not proceed through an initial arylnickelation/β-H elimination followed by a second arylnickelation event (see Supplementary Section 6 for details).

Reaction scope.

The generality of our established conditions was first evaluated using a wide array of functionalized terminal alkenes (Table 1). 1,1-Disubstituted aliphatic olefins, styrenes and 1,3-dienes were found to serve as effective substrates, furnishing diarylation adducts 4b-4z containing fully-substituted quaternary carbon centres in 43–93% yield. Exceptional site selectivity (>98:2 regioisomeric ratio) was attained across the board, irrespective of the electronic or steric features of the alkene. The products include those generated from activated acyclic aryl (4b-4g) and heteroaryl (4h) olefins, as well as exocyclic C=C bonds (4j). Bisdiarylation could be effected as exemplified by 4i, the structure of which was ascertained by X-ray crystallographic analysis. The reaction of alkyl-substituted 1,3-diene occurred preferentially on the olefin terminus to afford 4k. This result merits mention given the propensity of such dienes to form allylmetal intermediates that undergo rearrangement to give isomeric mixtures from 1,2- and 1,4-additions³¹. Transformations with unactivated acyclic alkenes were similarly efficient, delivering products including ethers (4n–4p), an indole (4m), a sulfide (4q) and a silane (4r), as well as those containing unsaturated C–C bonds (4s and 4t). Aliphatic exocyclic substrates underwent diarylation to afford medicinally valuable piperidines³² (4u and 4v) as well as carbocyclic compounds (4w and 4x). Reactions are also compatible with C=C bonds derived from complex bioactive compounds as demon-strated by the synthesis of 4y (from antioxidant d-α-tocopherol) and 4z (from lipid-lowering agent gemfibrozil).

Table 1 |.

Directing-group-free diarylation with various unactivated and activated alkenes

Regioisomeric ratios and diastereomeric ratios were determined by gas chromatography analysis of unpurified mixtures; yields are for isolated and purified products. PhZnCl was used for the synthesis of 4e, 4y and 4z. Compound 4s was obtained as a 2:1 diastereomeric mixture. R and G, functional groups; Ar, aryl group; Tf, trifluoromethanesulfonyl; Bn, benzyl; TBS, tert-butyldimethylsilyl; Boc, tert-butoxycarbonyl; r.r., regioisomeric ratio. See Supplementary Section 2 for details.

Monosubstituted olefins also participated in diarylation to generate 4aa–4al in 40–70% yield and 84:16 to >98:2 regioisomeric ratios. Small quantities of 1,1-disubstituted olefin side products (∼20–30%) arising from competitive Heck-type reactions^13,14 (compare with v→7 in Fig. 2a) were detected in some cases. Site selectivities were somewhat lower (versus 1,1-disubstituted alkenes), presumably owing to less severe steric interactions and hence a smaller energy difference between the two competing transition states (compare with TS A and TS A′ in Fig. 1c). Products appended with a variety of non- (or weakly at best) coordinating functionalities such as arenes and heteroarenes (4aa, 4ac–4ag), a silane (4ab), an amide (4ah), an alkyne (4ai) and a long-chain alkoxy unit (4aj) could be obtained with good regioselectivities. Styrenes (4ak) and electron-rich olefins (4al) were amenable substrates in the catalytic method, providing access to molecules with diaryl- and nitrogen-substituted stereogenic centres, respectively. However, 1,2-disubstituted and trisubstituted olefins were found to be incompetent substrates in our protocol (low conversion to product).

Employing either phenylmagnesium bromide 3a or phenylzinc chloride 3b, an assortment of substituted aryl triflates (obtained in one step from abundant and inexpensive phenols) was examined for their ability to promote diaryl addition across alkene 1a (Table 2). Arenes and heteroarenes with different electronic and steric properties could be incorporated, affording 4am–4ax as the sole regioisomers in up to 77% yield. For aryl moieties bearing functional groups (for example methoxy, fluoride or ester) that are prone to reaction with 3a (4ap–4as, 4au and 4av, 4ax), use of the less nucleophilic reagent 3b allowed diarylation to proceed smoothly. Heterocycles such as Lewis basic pyridines (4ax) were also tolerated under the standard conditions.

Table 2 |.

Directing-group-free diarylation with various triflates and organometallic reagents

Regioisomeric ratios were determined by gas chromatography analysis of unpurified mixtures; yields are for isolated and purified products. Synthesis of opposite regioisomers can be accomplished by exchanging the aryl unit between 2 and 3, as exemplified by 4ar and 4bk. M, metal; Ar and Ar′, aryl groups; Tf, trifluoromethanesulfonyl; Bn, benzyl; r.r., regioisomeric ratio. See Supplementary Section 2 for details.

To further showcase the diversity of aryl components that can be installed, the diarylation of 1a was carried out with various Grignard or zinc reagents 3 (commercially available or readily prepared from halides) in conjunction with triflate 2a. As expected, the transformations were compatible with arenes containing electron-rich or electron-deficient groups located on the ortho-, meta- and/or para-positions and 4ay–4bj were furnished in 50–77% yield and >98:2 regioisomeric ratios. Furthermore, repeating the reaction by exchanging the aryl substituent on the triflate and organozinc nucleophile offers direct entry to regioisomeric 4ar and 4bk in 70 and 73% yield, respectively. Effectively, the ease of synthesis associated with 2 and 3 means that different permutations of diarylated building blocks can be expeditiously accessed in either regioisomeric form³³, which is advantageous for developing compound libraries in drug discovery programmes.

Mechanistic evaluation.

The mechanistic underpinnings influencing the efficiency and regioselectivity of diarylation are further deliberated in Fig. 3a. With respect to the results shown in Fig. 2b, the alternative regioisomer 4a′ was never detected in the remarkably regioselective diarylation of 1a and biaryl cross-coupling of 2a and 3a was a common side reaction. This suggests that formation of a hypothetical biarylnickel species vi by transmetallation of ii with the arylmetal reagent is plausible in our system. However, in such a scenario, addition of vi to 1a is unlikely to be site-selective because both aryl units are insufficiently distinct (electronically and sterically) to favour preferential generation of iv or its regioisomer. Rather, vi presumably undergoes facile reductive elimination to give biaryl by-product 8 as seen in a previous report³⁴ and supported by quantum mechanical calculations.

Unrestricted, dispersion-corrected density functional theory (DFT) calculations lend further credence to the proposed catalyst-controlled mechanistic model based on steric arguments (Fig. 3a,b) (see Supplementary Section 8 for full computational details). Specifically, as shown in Fig. 3b, substrate- or solvent-assisted formation of Ni(I) chloride species I from the dimeric Ni-1 pre-catalyst³⁵ is followed by oxidative addition with the triflate to afford II. Compund II will then undergo a branched-selective 1,2-migratory insertion (via TS A) across the alkene substrate to generate III. Notably, the opposite linear-selective insertion (via TS A′) is significantly less favoured than branched-selective insertion (ΔG^‡ = 25.8 kcal mol⁻¹ versus 15.8 kcal mol⁻¹), presumably due to severe steric repulsions between the alkene substituents and the sizeable IPr ligand bound to the Ni centre in TS A′, which can be avoided by addition of the nickel to the less substituted end of the C=C bond. Indeed, the structure of TS A′ is significantly distorted compared to that of TS A, which is likely to be as a consequence of this steric hindrance. Following branched-selective arylnickelation, irreversible transmetallation with the organometallic reagent will lead to species IV. Compound IV is primed to undergo reductive elimination to furnish the desired product 4bl with high regioselectivity and to regenerate the catalytically active Ni(I) species I. Additionally, we were unable to locate the biarylnickel species vi and all attempts led to formation of the corresponding biaryl by-product 8. This result supports the earlier notion that vi prefers to undergo reductive elimination instead of olefin addition.

Representative applications.

The ability to incorporate aryl groups across C=C bonds with high efficiency and site selectivity eliminates the need for pre-installation and post-reaction removal/conversion of neighbouring directing auxiliaries. This provides attractive advantages in terms of cost savings, step economy and waste reduction as showcased in Fig. 3c. In one instance, the reaction of commercially available 4-penten-1-ol 9 and phenol 10 affords alkene 11 in 95% yield. This was then subjected to the standard Ni-catalysed 1,2-diarylation with the requisite reagents to deliver 4bm, a key intermediate used in the synthesis of sphingosine 1-phosphate receptor modulator 12, in 43% yield and >98:2 regioisomeric ratio. The overall route compares favourably with a previous report (the longest linear sequence is two steps versus four steps)³⁶.

The second application involves the synthesis of alcohol 4bn, a precursor of the serotonin 2 receptor antagonist 5 (ref. ³⁷). Conversion of 4-penten-1-ol 9 to silyl ether 13 sets the stage for a Ni-catalysed diarylation and in situ desilylation to furnish 4bn, which was isolated in 37% overall yield and 95% regioselectivity. It is worth mentioning that a lengthier process involving catalytic diarylation of a pre-synthesized 4-pentenamide¹⁸, followed by amide hydrolysis and further functional group manipulations, would otherwise be required to obtain both 4bm and 4bn. Generation of azetidine 4bo, containing a sterically demanding quaternary carbon centre, from commercially available exocyclic olefin 14 was easily accomplished in 60% yield and complete regioselectivity through diarylation. This method is more concise, higher yielding and less time consuming than the alternative six-step sequence³⁸ and facilitates preparation of glycine transporter 1 inhibitor 6 and other related analogues of the medicinally valuable azetidine scaffold³⁹. To summarize, our Ni-catalysed method provides an enabling platform for efficient chemical synthesis to access various compounds of interest.

Besides 1,2-diarylation, weshow that the present directing-group-free strategy is robust and amenable to other dicarbofunction-alization transformations, offering a direct entry to a myriad of sp³-hybridized carbon scaffolds (Table 3). Aryl-alkylation could be achieved by switching to aliphatic organometallic nucleophiles to furnish the desired products 4bp–4bv in 46–82% yield and >98% site selectivity. On the other hand, organozinc compounds were utilized to promote the corresponding aryl-alkenylation and aryl-alkynylation reactions under the established conditions, affording 4bw–4ca in 40–58% yield and >98% regioisomeric purity. The exceptional regioselectivity, even in the presence of differently hybridized carbogenic reagents, highlights the fidelity of these catalyst-controlled processes with Ni-1.

Table 3 |.

Extension to other classes of dicarbofunctionalization reactions

Regioisomeric ratios were determined by gas chromatography analysis of unpurified mixtures; yields are for isolated and purified products. R and G, functional groups; Ar, aryl group; LG, leaving group; M, metal; Tf, trifluoromethanesulfonyl; Bn, benzyl; TMS, trimethylsilyl; r.r., regioisomeric ratio. See Supplementary Section 4 for details.

Conclusion

The Ni-catalysed system we have developed is capable of mediating 1,2-diarylation across both unactivated and activated olefins with exceptional efficiency and site selectivity, without the requirement of a neighbouring auxiliary. These versatile catalytic reactions offer a convenient avenue to assemble valuable diarylated compounds for complex molecule synthesis and can be readily extended to other important olefin difunctionalizations. Key to these kinetically controlled transformations is the use of a sizeable (NHC)Ni(I)-based complex to induce high regioselectivity through steric effects. We expect the present study to pave the way for future development of general alkene addition processes that no longer rely on activating units within the substrate for high efficiency and selectivity.

Methods

General procedure for alkene 1,2-dicarbofunctionalization: in a N₂-filled glove box, an oven-dried 4 ml vial equipped with a stir bar was charged with Ni-1 (5.0 mol%) and toluene (0.5 ml). The alkene substrate (0.1 mmol, 1.0 equiv.) and aryl triflate (0.2 mmol, 2.0 equiv.) were then added to the system. After that, the organometallic reagent (0.3 mmol, 3.0 equiv.) was slowly added into the system. The vial was sealed and the reaction mixture was stirred at 40 °C for 10 h. After cooling to ambient temperature, the crude mixture was purified by silica gel chromatography.

Data availability

All data supporting the findings of this study are available within the Article and its Supplementary Information. Crystallographic data for compound 4i have been deposited at the Cambridge Crystallographic Data Centre, under deposition number CCDC 2069191. Copies of the data can be obtained free of charge via https://www.ccdc.cam.ac.uk/structures/.