Highly Selective Radical Relay 1,4-Oxyimination of Two Electronically Differentiated Olefins

Abstract

Radical addition reactions of olefins have emerged as an attractive tool for the rapid assembly of complex structures, and have plentiful applications in organic synthesis, however, such reactions are often limited to polymerization or 1,2-difunctionalization. Herein, we disclose an unprecedented radical relay 1,4-oxyimination of two electronically differentiated olefins with a class of bifunctional oxime carbonate reagents via an energy transfer strategy. The protocol is highly chemo- and regioselective, and three different chemical bonds (C–O, C–C, and C–N bonds) were formed in a single operation in an orchestrated manner. Notably, this reaction provides rapid access to a large variety of structurally diverse 1,4-oxyimination products, and the obtained products could be easily converted into valuable biologically relevant δ-hydroxyl-α-amino acids. With a combination of experimental and theoretical methods, the mechanism for this 1,4-oxyimination reaction has been investigated. Theoretical calculations reveal that a radical chain mechanism might operate in the reaction.

Graphical ABSTRACT

INTRODUCTION

The chemistry of radicals has progressed rapidly in recent years because of such powerful species allowing access to synthetic pathways which are previously considered to be highly challenging or unfeasible.^1–5 In this regard, the radical across olefins, especially the addition of two radicals, has emerged as an elegant and versatile tool for the straightforward construction of high-value, and often complex, molecular architectures.^6–8 Currently, the majority of such reactions focuses on the 1,2-difunctionalization of olefins, simultaneously introducing two functional groups on both sides of the double bond (Figure 1a).^6,7 Apart from this, another type of widely studied radical-involved addition reaction across olefins is radical polymerization (Figure 1a),⁸ being a key synthesis route for obtaining a wide variety of different polymers and composite materials. Despite significant progress, there is still enormous chemical space remaining to be explored in this addition field. For instance, the controllable radical-based remote 1,n-difunctionalization across more than one olefin is so far underdeveloped.

Figure 1.

(a) Addition of two radicals across olefins: 1,2-difunctionalization or polymerization. (b) This work: photochemical 1,4-oxyimination of two electronically differentiated olefins.

Recently, radical-based methods for the coupling of two distinct olefins have risen to prominence.^9–11 Typically, such olefin/olefin couplings are achieved through radical cascades, which involve an initial intramolecular cyclization or intermolecular radical addition to an electron-richer olefin to generate a C-centered radical. Subsequent intermolecular Giese type radical addition to the Michael acceptor furnishes the coupling product,^9,10 as exemplified by the pioneering works of Bluhm,^9a Hartung,^9b Knowles,^9c Studer,^9d Leonori,^9e Nagib,^9f and others. In another notable report,¹¹ Baran developed an iron-catalyzed reductive cross-coupling between two olefins through a radical relay strategy in which the nucleophilic C-centered radicals were formed by transfer of a hydrogen radical from an Fe–H species to an olefin. However, although these reactions are pioneering, they mainly focus on installing only up to one additional functional group during the olefin/olefin coupling. Very recently, Gutierrez and Molander reported a photochemical 1,2-carboimination of olefins, including one single example of a 1,4-trifluoromethylimination across an electron-rich and an electron-poor olefin using a bifunctional oxime ester.^7m Considering the well-developed radical-involved 1,2-difunctionalization of a single olefin,⁷ we questioned whether such transformations could be extended to the 1,4-difunctionalization of olefins, namely, the stepwise addition of two radicals to two electronically differentiated olefins. This hypothesis faces various severe challenges, including the identification of suitable radical precursors as well as overcoming chemo- and regioselectivity issues arising from competing monofunctionalization, 1,2-difunctionalization, polymerization, and other side reactions. However, if successful, it would unlock easy access to underexplored chemical space, and offer a conceptual complementary of olefin difunctionalization as well.

The Sharpless oxyamination¹² represents one of the most powerful and straightforward strategies for the preparation of 1,2-amino alcohols which are widely found in naturally occurring compounds and bioactive molecules.¹³ Despite their versatility, such transformations usually only allow introducing amine and alcohol functionalities across olefins at the vicinal positions. Herein, we report a novel radical relay 1,4-oxyimination of two electronically differentiated olefins which is highly chemo- and regioselective, and involves the well-orchestrated formation of three different chemical bonds (C–O, C–C, and C–N bonds) in a single operation (Figure 1b). As an extension of the Sharpless oxyamination, this energy transfer¹⁴ (EnT)-enabled reaction offers rapid access to a variety of 1,4-oxyimination products with great structural diversity. Follow-up chemistry easily delivers valuable biologically relevant δ-hydroxyl-α-amino acids.¹⁵

RESULTS AND DISCUSSION

Reaction Development.

Considering that Han, Huo, and our group have recently identified oxime carbonates as a suitable bifunctional source for supplying both oxygen- and nitrogen-centered radicals via EnT catalysis,^16–18 we started our investigation toward the desired 1,4-oxyimination reaction by applying benzophenone O-ethoxycarbonyl oxime 1a as a radical precursor. To gain some preliminary understanding toward the reactivity of this bifunctional reagent with electronically differentiated olefins, a series of two-component reactions were conducted. As shown in Figure 2a, oxime carbonate 1a was separately treated with unactivated olefin A1, Michael acceptor A2, and 1,3-diene A3 under visible-light-sensitized conditions employing thioxanthone (5.0 mol %) as the organic photosensitizer in dry EtOAc (0.1 M) under an argon atmosphere at room temperature for 12 h. In line with our previous results,¹⁶ 1a reacted smoothly with A1, giving the 1,2-oxyimination product 2 in 45% yield. However, the reaction of 1a with A2 led to a complex mixture, and the desired product 3 was not observed. Instead, a byproduct, resulting from the addition of 1a across two molecules of A2, was detected by high resolution mass spectrometry (see Supplementary Information for the details). Moreover, there was no reaction at all between 1a and A3, and both starting materials could be recovered after the reaction.

Figure 2.

Reaction development and mechanistic proposal. (a) Two-component reactions of oxime carbonate 1a with A1, A2, or A3. Reaction conditions: oxime carbonate 1a (0.2 mmol), olefins A1, A2, or A3 (0.4 mmol), and thioxanthone (5.0 mol %) in EtOAc (0.1 M), irradiation with 18 W blue light emitting diodes (LEDs) (λ_max = 405 nm) under an argon atmosphere at room temperature for 12 h. Isolated yields are given. (b) Three-component reactions of oxime carbonate 1a, A1, and A2 with different equivalents. Reaction conditions: oxime carbonate 1a (0.2 mmol, 1.0 equiv), olefin A1 (0.4 mmol, 2.0 equiv), A2 (0, 0.25, 0.50, 0.75, 1.0, 1.25, 1.50, 1.75 or 2.0 equiv), and thioxanthone (5.0 mol %) in EtOAc (0.1 M), irradiation with 18 W blue LEDs (λ_max = 405 nm) under an argon atmosphere at room temperature for 12 h. Isolated yields are given. (c) Impact of other reaction parameters. (d) Condition-based sensitivity assessment. (e) Mechanistic proposal for the 1,4-oxyimination of oxime carbonate 1a, A1, and A2.

The 1,4-oxyimination product 4 was obtained in 17% yield when adding 0.25 equivalents of A2 to a mixture of 1a and A1. Further increasing the equivalents of A2 led to rapidly increased yields of product 4, simultaneously inhibiting the 1,2-oxyimination of A1 (Figure 2b). To our delight, the isolated yield of 4 reached 64% when using 2.0 equivalents of A2 as the Michael acceptor and only trace amounts of product 2 were detected. Notably, product 3 was not observed in these reactions. These results showed that—employing the optimized reaction conditions—this transformation is highly chemo- and regioselective.

Then, the impact of other reaction parameters was further investigated (Figure 2c). As expected, both thioxanthone and blue light (λ_max = 405 nm) were indispensable for this transformation (Figure 2c, entries 2–3). The reaction did not proceed when switching to blue light with a wavelength maximum of 450 nm (Figure 2c, entry 4). Replacing ethyl acetate with acetone, CH₂Cl₂, or MeCN still provided 4 in similar yields, showing that the reaction is relatively robust regarding the choice of solvent (Figure 2c, entries 5–7). In addition, other photocatalysts were examined, and 4CzIPN (1,2,3,5-tetrakis(carbazol-9-yl)-4,6-dicyanobenzene, 2,4,5,6-tetrakis(9H-carbazol-9-yl) isophthalonitrile) and the Ir-(dFCF₃ppy)₂(dtbbpy)PF₆ complex were both found to be suitable catalysts for this reaction with slightly diminished yields of product 4 (Figure 2c, entries 8–10). Moreover, a condition-based sensitivity assessment was performed (Figure 2d).¹⁹ The reaction proved to be relatively insensitive toward moisture, high light intensity, scale-up, and small changes in concentration and temperature. In contrast, the reaction was negatively affected by low light intensity and a high oxygen levels (see the Supplementary Information for the details).

Mechanistic Study.

The quantum yield of the reaction of 1a, A1, and A2 was determined to be 1.18, which suggested that a radical chain mechanism might operate in the reaction (see the Supplementary Information for the details).^7m Moreover, product 4 could be obtained in 49% yield when the reaction mixture of 1a, A1, and A2 was irradiated with a higher energy light source (λ_max = 365 nm) for 36 h in the absence of any photocatalyst, indicating that a radical polar crossover pathway is unlikely involved in the reaction (see the Supplementary Information for the details).²⁰ Based on the observations described above, and previous reports,^16–18 a mechanistic proposal for the selective 1,4-oxyimination of olefins A1 and A2 with oxime carbonate 1a was proposed (Figure 2e). The reaction is initiated by an EnT-mediated homolytic N–O bond cleavage of 1a to generate a persisent N-centered iminyl radical B and a transient O-centered alkoxycarbonyloxyl radical C, which are of ambiphilic and electrophilic properties, respectively.¹⁶ Because of the persistent radical effect (PRE)²¹ and favored polarity-matching,²² the transient O-centered alkoxycarbonyloxyl radical first adds to the C=C double bond of the electron-richer olefin A1 to form the adduct radical IM1. Subsequently, IM1 can undergo two different pathways. In the presence of Michael acceptor A2, the nucleophilic IM1 preferentially adds to A2 to generate another radical adduct IM2, which then reacts with B and/or 1a to furnish the desired 1,4-oxyimination product 4. In the absence of A2, IM1 reacts with B and/or 1a directly, giving 1,2-oxyimination product 2.

In order to further understand the mechanism and the origin of the observed reactivity and selectivity, we turned to dispersion-corrected density functional theory (DFT) calculations (see the Supplementary Information for the details). As shown in Figure 3, N–O bond cleavage takes place in the triplet state of 1a (49.9 kcal/mol uphill in energy which can be accessed by triplet–triplet EnT with excited thioxanthone) with a small energy barrier via TS (ΔG^‡ = 2.4 kcal/mol). This generates the ambiphilic N-centered iminyl radical B and the electrophilic alkoxycarbonyloxyl radical C (16.8 kcal/mol downhill from the excited state intermediate 1a*). Then, C can undergo irreversible radical addition to A1 via TS1 (ΔG^‡ = 7.1 kcal/mol) to form IM1 (Figures 2e and 3). Alternatively, the radical addition of C to A2 proceeds via a higher energy barrier (8.4 kcal/mol via TS6) to form intermediate IM3 (Figures 2e and 3). However, while IM3 is thermodynamically more favored than IM1, −17.0 versus −13.5 kcal/mol, respectively, the radical addition of C to A1 is predicted to occur ~10 times faster than the addition to A2 due to polarity matching.²² In addition, we considered alternative pathways for radical addition with the ambiphilic radical B, but the energy barriers were considerably higher (>16 kcal/mol, Figure S13).

Figure 3.

Proposed mechanism supported by computational studies. Calculated free Gibbs energies [CPCM(EtOAc) uB3LYP-D3/def2-svp] are given in kcal/mol. For details see the Supplementary Information.

Next, we sought to calculate the possibilities for the two-component reaction of oxime carbonate 1a toward the olefins A1 and A2 (Figure 2a). The two-component reaction of 1a and A1 would initiate with the radical addition of C to A1 to form IM1 via TS1 as described above. Selective radical–radical cross-coupling of IM1 and B to form 2 would be feasible based on the PRE,²¹ but given that the relative concentration of 1a is much greater than the concentration of iminyl radical B, IM1 could alternatively add to 1a via TS3 with a barrier of 13.5 kcal/mol with respect to IM1, leading to radical intermediate IM9 (see Supplementary Information, Figure S9).^7m IM9 can then undergo rapid fragmentation to form product 2 under the concomitant release of an EtOCO₂• radical C. However, TS3 is in direct competition with a dimerization process via TS4 with a barrier of 14.4 kcal/mol (only 0.9 kcal/mol higher than TS3) and is potentially one of the reasons for the moderate yield of 2 in the two-component reaction (Figure 2a). In addition, we also evaluated the factors controlling the two-component reaction between 1a and A2. First, IM3 is formed through TS6 as described above. Then, the radical–radical cross-coupling of IM3 and B would lead to 3, but this product was not observed experimentally. However, IM3 may also undergo dimerization with a low energy barrier (10.1 kcal/mol via TS7) to generate IM4 (see Supplementary Information, Figure S10). This dimerization pathway could be the reason for the formation of a complex mixture observed experimentally (see the Supplementary Information for details).

Finally, the mechanism for the three-component reaction (Figure 2b) was evaluated. First, the radical addition of C prefers to occur over A1 via TS1 as mentioned above to generate IM1. Then, IM1 would undergo selective addition onto A2 via TS2 (ΔG^‡ = 9.5 kcal/mol) to form the radical IM2 (Figures 2e and 3). Selective radical–radical cross-coupling of IM2 and B to form 4 would be kinetically and thermodynamically feasible based on the PRE.²¹ However, due to the low concentrations of B, IM2 would likely undergo selective addition onto 1a via TS5 (ΔG^‡ = 14.8 kcal/mol) to form IM7 (see Supplementary Information, Figure S11). Lastly, IM7 quickly undergoes fragmentation to form the product 4 and concomitantly releases another radical C.

Substrate Scope.

To obtain some initial insights into the reaction’s compatibility with different functional groups and heterocycles, an additive-based robustness screening was carried out (see the Supplementary Information for details), demonstrating the reaction’s overall great functional group tolerance.²³ Next, with the optimized reaction conditions in hand, we explored the scope of unactivated olefins. As shown in Table 1, various symmetrical 1,1-disubstituted olefins smoothly underwent 1,4-oxyimination with A2, providing the desired products in moderate to good yields (4–12). The structure of 12 was further confirmed by X-ray crystallography (Table 1).²⁴ Notably, in these cases, three different chemical bonds (C–O, C–C, and C–N bonds) were formed in an orderly manner, and two quaternary carbon centers were easily created, highlighting the charm of radical chemistry in building complex molecules. A cyclic 1,2-disubstituted olefin and different kinds of electron-rich heteroatom-bound (including O, N, Si) olefins could also be engaged in this reaction (13–17). Subsequently, to avoid the unwanted diastereomers, a symmetrical electron-poor olefin, di-tert-butyl 2-methylenemalonate (A16), was selected as a Michael acceptor to examine the reactivity of monosubstituted unactivated olefins. To our delight, a wide range of unactivated olefins bearing functional groups, such as cyano, bromo, alkynyl, ester, amidyl, and even alkenyl, were suitable for the reaction to provide the corresponding 1,4-oxyimination products (18–31). In addition, terminal olefins embedded in complex molecules were well compatible with the protocol as well, delivering the desired products in satisfactory yields (32–36).

TABLE 1.

Scope of Unactivated and Electron-Rich Olefins^a,b

^aReaction conditions: oxime carbonate 1a (0.2 mmol), unactivated olefins (0.4 mmol), Michael acceptors (0.4 mmol), and thioxanthone (5.0 mol %) in EtOAc (0.1 M), irradiation with 18 W blue LEDs (λ_max = 405 nm) under an argon atmosphere at room temperature for 12 h.

^bIsolated yields are given. The d.r. values were determined by ¹H nuclear magnetic resonance (NMR) analysis.

To our surprise, the reaction could be smoothly extended to trisubstituted and even tetrasubstituted olefins. As shown in Figure 4, 3-ethylpent-2-ene (A37) and 2,3-dimethylbut-2-ene (A38), despite lower reactivity, also reacted with 1a and A2 to give the highly substituted 1,4-oxyimination products in decent yields (38 and 40). Notably, in the absence of Michael acceptor A2, the 1,2-oxyimination of 1a with A37 or A38 was not observed. This observation can be rationalized with the fact that the reaction of N-centered iminyl radical B and/or 1a with the C-centered radical (generated by the addition of alkoxycarbonyloxyl radical to olefins A37 or A38) was inhibited because of significant steric hindrance as well as the polarity mismatching of these radical species. In the presence of A2, steered by polar effects, the nucleophilic adduct C-centered radical could add to A2 to form another relatively neutral adduct radical (stabilized by the electron withdrawing groups), and its reaction with the N-centered iminyl radical B and/or 1a was therefore kinetically favored due to polarity matching.

Figure 4.

Reactivities of tri- or tetrasubstituted olefins in the two or three component reactions.

Next, the scope of Michael acceptors was investigated. As summarized in Table 2, a wide range of electron-deficient olefins including acrylonitrile (41), acrylates (42–49), vinyl ketone (50), acrylamide (51), and vinylphosphonate (52) could smoothly react with 1a and A1 to give the desired products in moderate yields. It is worth noting that a boronate group was also tolerated by this method, albeit with lower reactivity, providing a versatile synthetic handle for further functionalization of the product (53). Moreover, various styrene derivatives could be used as Michael acceptors, as documented by the successful construction of products 54–62. Recently, the construction of cyclobutane-containing organic molecular skeletons has attracted significantly increased interest, since such moieties are not only common in natural products,²⁵ but are also more frequently embedded in pharmaceuticals.²⁶ To our delight, a series of cyclobutene-containing products were smoothly delivered by using commercially available methylenecyclobutane (A4) as the starting material (63–70). Adamantane-containing product 71 was obtained in moderate yield as well.

TABLE 2.

Scope of Michael Acceptor^a,b

^aReaction conditions: oxime carbonate 1a (0.2 mmol), unactivated olefins (0.4 mmol), Michael acceptors (0.4 mmol), and thioxanthone (5.0 mol %) in EtOAc (0.1 M), irradiation with 18 W blue LEDs (λ_max = 405 nm) under an argon atmosphere at room temperature for 12 h.

^bIsolated yields are given. The d.r. values were determined by ¹H NMR analysis.

Generally, oxime carbonate 1a was completely consumed in all the above reactions, given its high reactivity to generate a pair of N-centered iminyl radical and O-centered alkoxycarbonyloxyl radical under the standard reaction conditions. The protonation and homocoupling reactions of the persistent N-centered iminyl radical were the main side-reactions observed in the reaction mixtures. Moreover, the 1,2-oxyimination of the olefin was only a minor side reaction (see the Supplementary Information for a detailed discussion).

Subsequently, the efficiency of other bifunctional reagents (1b-1i) was evaluated with A1 and A2. As summarized in Figure 5, various oxime carbonates (1b-1g), the analogues of 1a, smoothly underwent this reaction, forming the corresponding products in moderate to good yields (72–77). To our delight, a menthol-derived bifunctional reagent 1h could be engaged in this method with good yield of the product 78, highlighting the compatibility of this protocol with complex molecules. Likewise, a benzoic-acid-derived oxime ester 1i was also suitable for the reaction, affording product 79 in 36% yield. On the other hand, oxime carbonates 1j-1l failed to give the desired 1,4-oxyimination products. Specifically, 1j and 1k were completely decomposed after the reaction, neither the 1,2- nor 1,4-difunctionalization reaction were detected. As for the case of 1l, 1,2-carboimination of olefin A2 was observed as the major transformation in the reaction.

Figure 5.

Scope of bifunctional reagents. Reaction conditions: oxime carbonate 1b-1l (0.2 mmol), unactivated olefins (0.4 mmol), Michael acceptors (0.4 mmol), and thioxanthone (5.0 mol %) in EtOAc (0.1 M), irradiation with 18 W blue LEDs (λ_max = 405 nm) under an argon atmosphere at room temperature for 12 h. Isolated yields are given. d.r. values were determined by ¹H NMR analysis. Fmoc: fluorenylmethoxycarbonyl.

Synthetic Applications.

To further demonstrate the utility of this method, gaseous ethylene (2.0 atm) was employed as the starting material, providing the desired product 80 in a synthetically useful yield (Figure 6a). In addition, a remote 1,6-oxyimination via intramolecular cyclization was realized by using diethyl 2,2-diallylmalonate (A71) (Figure 6b). In order to further highlight the synthetic practicality of this protocol, a gram scale reaction of 1a, A2, and A5 was performed (Figure 6c), providing the desired product 6 in 67% yield (1.75 g). Product 6 was then deprotected under different conditions to selectively give rise to products 82 or 83 both in excellent yields. Finally, the efficient conversion of a series of 1,4-oxyiminated products into unprotected biologically relevant δ-hydroxyl-α-amino acids with previously inaccessible structural features was demonstrated by simple refluxing in 6 N HCl (84–89, Figure 6d).⁷ⁿ

Figure 6.

Synthetic applications. (a) Reaction of 1a, A2, and ethylene (A70, 2.0 atm.) under the standard conditions. (b) 1,6-Oxyimination of 1a, A2, and A71 under the standard conditions. (c) Gram scale synthesis and deprotection of product 6. (d) Deprotection of selected products to produce the corresponding free amino acids.

CONCLUSIONS

In summary, we have demonstrated a rare example of a radical relay 1,4-difunctionalization reaction across two electronically differentiated olefins. The presented approach relies on the combination of a class of bifunctional radical precursors with EnT catalysis, and allows for the rapid build-up of highly complex molecular scaffolds from simple starting materials in a single operation. Three different chemical bonds (C–O, C–C, and C–N bonds) are formed in an orchestrated and highly chemo- and regioselective manner to give rise to valuable 1,4-oxyiminated products. This mild method features an exceptionally broad substrate scope and excellent tolerance of sensitive functional groups. Olefins ranging from simple ethylene to sterically encumbered tetra-substituted olefins could smoothly participate in the reaction. Furthermore, the products were easily converted into valuable biologically relevant δ-hydroxy-α-amino acids. A series of experimental and computational studies were carried out to elucidate the mechanism of this transformation. Overall, these studies are consistent with an EnT event to promote the homolytic N–O bond cleavage of a bifunctional oxime carbonate reagent to form O-centered alkoxycarbonyloxyl and N-centered iminyl radicals. In turn, the alkoxycarbonyloxyl radical undergoes a chemoselective double Giese type addition to electron-rich olefin/electron-deficient olefin followed by C–N bond formation with an oxime carbonate reagent that functions as a radical chain mediator. Taking all this into account, we believe that this work would pave the way toward further developments in the area of radical 1,4-difunctionalization reactions.