Direct Aroylation of Olefins through a Cobalt/Photoredox-catalyzed Decarboxylative and Dehydrogenative Coupling with α–Oxo Acids

Abstract

A photoredox/cobalt dual catalytic procedure has been developed that allows benzoylation of olefins. Here the photoredox catalyst effects decarboxylation of α–ketoacids to form benzoyl radicals. After addition of this radical to styrenes, the co-balt catalyst abstracts an H-atom. Hydrogen evolution from the putative cobalt hydride intermediate allows a Heck-like aroylation without the need for stoichiometric oxidant. Mechanistic studies reveal that electronically different styrenes lead to a curved Hammett plot, suggesting a change in product-determining step in the catalytic mechanism.

Graphical Abstract

Visible light promotes decarboxylative Heck-like benzoylations of olefins. This process leverages the synergistic interaction of a photoredox catalyst with cobaloxime catalysts to form enones via decarboxylation and hydrogen evolution. Mechanistic studies reveal different rate-limiting steps, depending on electronics of the olefin.

Introduction

Chemoselective methodologies for the synthesis of α,β–unsaturated aryl ketones, and chalcones in particular, are becoming increasingly desirable due to their role as key synthetic building blocks for many natural products and pharmaceuticals.^[1] Variations on aldol and Mukaiyama aldol condensations remain the most common reactions for preparing these classes of compounds, however issues with reactivity and product selectivity arise when coupling two different ketones,^[2a-d] and are very low yielding with diarylketones.^[2e] These substrates necessitate highly con-trolled couplings using stoichiometric reagents. For example, the Horner-Wadsworth-Emmons olefination^[3] and the Still-Gennari modification,^[4] require multistep preparation of the starting acyl-phosphonates, and Friedel-Crafts acylations^[5] and Lewis-Acid assisted aldol reactions^[6] produce stoichiometric amounts of metal waste. Consequently, a series of catalytic protocols have been developed.^[7–8] The carbonylative Heck reaction remains the most thoroughly explored reaction of this class^[9] despite requiring the use of toxic CO gas and precious metal catalysts. Glorius^[10] and Lei^[11] reported the first examples of direct dehydrogenative radical couplings of aldehydes and olefins, catalyzed by Rh and Cu respectively, which circumvent the need for CO gas, however these still require the use of peroxides as stoichiometric radical initiators. More recently, the use of photoredox catalysts as radical initiators has grown in popularity and has been successfully applied to the generation of acyl radicals.^[12] Aldehydes,^[13] acid chlorides,^[14] acyl silanes,^[15] carboxylic acids and anhydrides^[16] are all known to be reliable sources of acyl radicals under photolytic conditions when coupled with stoichiometric atom-transfer reagents or internal oxidants. With this in mind, α–oxo-acids provide an appealing alternative as they produce acyl radicals through a facile decarboxylation without the need for additional reagents,^[17] and thus have been used as powerful acylating agents.^[18] Herein, we describe a direct aroylation of olefins using a cobaloxime/photoredox dual-catalyzed coupling of olefins and α–oxo-acids. This method uses catalytic decarboxylation as a mechanism for radical formation and uses cobaloxime-catalyzed hydrogen evolution to obviate the need for in-ternal oxidants or stoichiometric additives that limit other methods.

In 2015, Shang and Fu reported the first example of trapping acyl radicals, generated from the decarboxylation of α–oxo-acids, with Michael acceptors under photoredox conditions to furnish γ–dicarbonyl compounds (Scheme 1A).^[19] This method has been adapted upon toward the synthesis of α,β–unsaturated aryl ketones by Zhu in 2017^[20] and Liu/Ngai in 2019, who utilized acid chlorides.^[21] Both protocols synthesize masked enones by first accessing the same carbon-centered radical (I) and intercepting by radical halogenation. The target compounds are then uncovered by a subsequent deprotohalogenation with stoichiometric base (Scheme 1B). Additionally, expensive iridium-based photocatalysts were required in both cases, in conjunction with Selectfluor as a stoichiometric oxidant in Zhu’s work. Thus, the development of a completely catalytic, oxidant-free method that utilizes easily handled, bench-stable α–oxoacids as acyl donors would be a significant advancement. Building upon the recent work by Lei,^{[22, 23]} Ritter,^[24a] Wu/Liu^[25] (Scheme 1C) and our own lab,^[24b] we envisioned that olefin formation could be achieved directly from the radical intermediate via a cobaloxime-catalyzed, hydrogen atom transfer (HAT) event (Scheme 1D). Cobaloximes have long been established as effective catalysts in radical chemistry^[26] and have continued to grow in popularity due to their elegant synergy with photocatalysts and proven viability in small molecule functionalization through hydrogen evolution.^[27] In 2017, Lei reported the first merger of cobaloxime and photoredox catalysis, when applied to olefin functionalization. Here, Lei showed that styrenes can be oxidized by excited state acridinium photocatalysts to generate radical-cation intermediates which were intercepted by a variety of oxygen/nitrogen nucleophiles,^[22] and later carbon-based nucleophiles,^[23] via dehydrogenative coupling. In 2018, a decarboxylative hydrogen evolution strategy was employed concurrently by our group and Ritter to achieve direct olefin formation from carboxylic acids.^[24] This strategy was adapted by Wu/Liu by intercepting the carbon-centered radical with olefins, to achieve a novel Heck-like coupling after the hydrogen evolution event (Scheme 1C).^[25] Here we implicate a similar strategy, in conjunction with α–oxo-acids as a source of acyl radicals, for the synthesis of a variety of substituted chalcones. In contrast to previous methods, the synthesis of α,β–unsaturated aryl ketones can be achieved without the need for oxidant, additives, or toxic CO gas and produces molecular CO₂ and H₂ as the only stoichiometric waste. It is worth noting that Zhong and co-workers have recently employed an analogous strategy in coupling α–oxo acids with acrylates to produce allylic ketones, though their system was shown to be incompatible with styrenes.^[28]

Scheme 1.

Results and Discussion

We initiated our studies by an investigation of commercially available phenylglyoxylic acid (1a) and α–methyl-4-fluorostyrene (2a) as the model substrates for the olefin aroylation reaction, in the presence of cobaloxime and photo-redox catalysts with blue LED irradiation. After thorough assessment (Table 1 and Tables S1-S3) we determined that using the acridinium photocatalyst PC5 in conjunction with Co(dmgH)₂Cl(N-Me-imidazole) C3 with Cs₂CO₃ (10 mol%) in a DME/Water solvent system was optimal for this reaction, producing the desired cross-coupled product 3a in a 71 % isolated yield. The transformation is regioselective for the Zaitsef elimination product and is selective for the E isomer also. Using N-Methyl-Imidazole (NMI) as the axial-base ligand C3 gave superior yields, though the lower yielding cobaloximes were often more selective for the E-isomer (Table 1, entries 5–6).

Table 1.

Selected Optimization Results

^[a]Yields determined by quantitative 1H NMR analysis. Numbers in parenthesis are isolated yields.

^[b]Diastereomeric ratios determined by 1H NMR analysis of the crude material

The acridinium tetrafluoroborate PC5 was established as the most effective to carry out this transformation and notably, is the most oxidizing of the photocatalysts used (Figure 1: E_1/2(P*/P⁻) = +2.17 V). Although all the photocatalysts screened are calculated to be sufficiently oxidizing to generate the carboxyl radical needed for decarboxylation (E_1/2(CO₂⁻/CO₂^.) = 0.95 – 1.3 V vs SCE),^[30] there is a clear correlation between yield and their oxidation potentials (Table 1, entries 7–9). The most reducing photocatalyst screened, the iridium-based PC1, saw an inversion of diastereoselectivity, leading to the Z isomer being the major product. This is a known phenomenon with this class of photocatalyst, where a difference in quenching rates of the excited state photocatalyst by the two isomeric products leads to a build-up of the thermodynamically disfavored isomer.^[31] The use of a catalytic quantity of Cs₂CO₃ had a noticeable positive effect on yield in comparison to when the base was absent (Table 1, entry 10) whereas using stoichiometric base greatly inhibited reactivity, which correlates with Wu/Liu’s observations^[25] (Table 1, Entry 11).

Figure 1.

Redox potentials of screened catalysts.²⁹

The selection of organic bases screened were incompatible in this system, potentially due to their affinity to coordinate to cobalt (see supporting information: Table S2).^[32] Adding a small amount of water as a co-solvent to DME was observed to aid reagent solubility and thus led to improved yields (entry 12). Due to the fact styrenes are also able to quench the excited state photocatalyst, products of dehydrogenative coupling with solvents can be significant if the water concentration is too high, or if protic solvents are used. This was particularly prominent with electron rich olefins, as previously observed by Lei.^{[22, 23, 33]} (see supporting information Scheme S1).

Gratifyingly, the described transformation is tolerant towards a broad range of functionalized styrenes and α–oxo-acids, providing β,β–disubstituted aryl vinyl ketones in good to moderate yields (Scheme 2). Notably, some labile functionalities in traditional Heck couplings were well-tolerated; aryl chlorides, bromides and iodides (3a-d) underwent smooth reaction without the observation of extensive dehalogenation. The transformation can also be applied to substrates bearing both electron-withdrawing and donating groups. Although electron deficient olefins generally provided higher yields (3a; 71 % E/Z = 69:31), electron-rich substrates exhibit better E/Z selectivities (3e; 38 % E/Z = 93:7). In contrast, a-oxo acids with electron donating substituents were superior (4b; 84 %), potentially due to the increased nucleophilicity of the acyl radical.^[34] Substrates with strong binding affinities to transition metals can also be troublesome for traditional Heck-like couplings, though compounds containing thioethers, 3f (53 %) and 4d (68 %), thiophenes 3m (67 %) and 4i (69 %), pyridine 3s (48 %), pyrrole 4j (40 %) and indoles 4l (45 %) were all observed to be tolerated in this reaction system. The highest yields were observed with a-aryl substituted styrenes (4a, 3o-t), particularly the product of coupling with 1,1-bis(fluorophenyl)ethylene (3q) being obtained in 92 % yield. The aroylation was observed to successfully functionalize 4-aryl coumarins at the 3-position (3n; 71 %) and was also successful with an N-protected, diphenylethylene fragment derivative of Tamoxifen (3t; 61 % yield), a common treatment for breast cancer. Unfortunately, alkyl-oxo acids were not observed to be compatible in this system. The reaction with pyruvic acid (R = Me) only provided the desired vinyl ketone product in 14 % yield, whereas glyoxylic acid (R = H) was unreactive. When substituted alkyl acids were used (R = ^tBu) a subsequent decarbonylation was observed, yielding the products of alkyl radical addition to diphenylethylene in a combined 23 % yield. Significant decarbonylation was also observed with the thioester and amido derivatives of the a-oxo-acids, with thiophenol (31 %) and phenyl disulfide (34 %), and aniline (39 %) composing the major products in their respective cases (see Supporting Information Scheme S2).

Scheme 2.

Scope of Olefin and Acid Coupling Partners

[a] Reactions were run on a 0.2 mmol scale with respect to α–oxo acids. [b] Yields reported are isolated yields. In most cases, the minor isomers could not be isolated independently and were observed in trace quantities upon co-elution with the major isomer. [c] Diastereomeric ratios were determined by analysis of the crude 1H NMR.

With the goal of further studying the mechanism and chemoselectivity of this aroylation reaction, we conducted a Hammett study of the 4-substituted styrenes shown in Scheme 3, with a range of σ values from −0.268 to 0.540.^[35] The competition studies were carried out under the standard conditions reported in Table 1, although a larger excess of α–methylstyrene and the competing 4-substituted–α–methylstyrenes were added to the reaction mixture (5 equivalents each) to ensure that reactant concentration did not affect partitioning. The relative reactivities of the olefin substrates were determined by comparing the relative yields of the cross-coupled substituted product vs 3h via analysis of the crude ¹H NMR spectra. These yields are reported as a ratio k_X/k_H in Scheme 3 (E/Z isomer yields were included in the total combined yield). From the competition studies, slightly inductively withdrawing substituents, like 4-Cl out-performed the neutral styrene, though as the electronics approached the extremities in both the donating and withdrawing directions, yields dropped off significantly.

Scheme 3.

Competition Analysis between α–methylstyrene and 4-X-α–methylstyrenes

To emphasize this observation via Hammett analysis, the data points [log(k_X/k_x)] were plotted against their respective σ values resulting in a concave curve (Figure 2), strongly indicating a change in product-determining step when going from electron donating to electron withdrawing substituents. The potential mechanistic steps that could account for the nonlinearity observed in the Hammett plot are depicted in Figure 2. Acyl radicals are generally accepted to be nucleophilic in nature,^[34] so a logical conclusion could be that the cross-coupling event, either through radical addition or carbometallation would be significantly slower with electron rich olefins. As the olefins employed become more electrophilic, the aptitude for coupling to these substrates increases steadily. However, once the electron withdrawing capabilities exceeds that of 4-Cl, the product selectivity reverts toward favoring the neutral coupling partner. Due to this observation, we can propose that the acyl radical addition step must be reversible to some extent, that becomes more prominent with strongly electron deficient olefins. Though the reversibility of radical additions has been suggested since the 1980’s,^[36] recent work by Glorius^[37] and Li^[38] provide evidence for its existence. Li has reported a series of radical olefin acylation reactions by generating acyl radicals from aldehydes in the presence of an iron catalyst. Li’s observations suggest acyl radical addition can be reversible, particularly in cases where there is a high energy barrier for the formation of non-stabilized or destabilized alkyl radicals, which may be at play with the strong electron withdrawing groups in our system.

Figure 2.

Hammet Analysis of olefin aroylation reaction and potential product determining steps.

Taking these experimental studies into consideration, the following mechanism for this cobaloxime/photoredox catalyzed olefin aroylation reaction can be proposed (Scheme 4). Photocatalyst PC 5 can be elevated to the excited state Mes-Acr-Ph⁺* upon irradiation with blue light (E_1/2(PC*/PC⁻) = +2.17 V). A single electron transfer (SET) from the deprotonated carboxylic acid to the highly oxidizing Mes-Acr-Ph⁺* would generate the carboxyl radical intermediate. This would then undergo a facile decarboxylation to furnish the acyl radical. The reduced photocatalyst Mes-Acr-Ph^. can be oxidized by the Co^III catalyst to close the photocatalytic cycle. The generated Co^II species can reversibly intercept the acyl radical (Int. 1) which subsequently undergoes a radical addition to produce the Co^III alkyl complex (Int. 2b). Homolytic cleavage of the Co-C bond results in a Co^II species and the carbon centered radical (Int. 2a), which readily yields the desired olefin product after the hydrogen atom transfer event. The Co^III-H may then react with a proton from another molecule of phenylglyoxylic acid to evolve H₂ gas, and return cobalt to its catalytic cycle.^[27b]

Scheme 4.

Potential Mechanism

Conclusion

The development of a direct olefin aroylation using cobaloxime/photoredox co-operative catalysis to cross-couple a-oxo acids and simple styrenes has been reported. This method allows for access to chalcone cores in a fully chemoselective fashion in good to moderate yields and diastereoselectivities. Here, molecular CO₂ and H₂ gas are produced as the only waste by-products and the use of stoichiometric additives or oxidants are not required. This methodology is also tolerant to a broad range of functional groups, with over 30 examples being prepared.

Experimental Section

General Information:

Purification was accomplished with column chromatography using silica gel (60 Å porosity, 230 × 400 mesh, standard grade) which was purchased from Sorbent Technologies (catalog # 30930M-25). TLC analysis was performed (fluorescence quenching and potassium permanganate acid stain) with silica gel HL TLC plates with UV254 purchased from Sorbent Technologies. 1H and 13C NMR spectra were obtained on a Bruker ADVANCE 500 DRX equipped with a QNP cryoprobe. These spectra were referenced to residual protio solvent signals. HRMS data was obtained on an ESI LC-TOF Micromass LCT (Waters). HRMS data was collected using ESI mass spectrometry. Melting points were obtained with Digimelt MPA 160 SRS (# 111278) and samples were loaded with borosilicate glass Kimble tube capillaries (# 34505–9a). Phenylglyoxylic acid (1a) is commercially available and was used without further purification. All other α–oxo acids were prepared from aryl ketones in correspondence with known literature procedures. α–methyl styrene (2h) and 4-methyl-α–methyl styrene (2j) were purchased from Sigma-Aldrich. All other styrenes were synthesized from the corresponding aryl ketones using methyltriphenylphosphonium bromide purchased from Sigma-Aldrich. All aryl ketone derivatives are commercially available and were used without further purification. Cesium carbonate was purchased from Combi-Blocks. Photocatalysts 9-mesityl-10-phenylacridinium tetrafluoroborate and 9-mesityl-2,7-dimethyl-phenylacridinium tetrafluoroborate were purchased from Sigma-Aldrich. Photocatalyst 9-mesityl-10-methylacridinium and 9-mesityl-2,7-dimethyl-methylacridinium were purchased from TCI. Cobaloxime catalysts were prepared from anhydrous cobalt chloride, dimethylglyoxime and the corresponding nitrogen heteroaromatic ligand, all purchased from Sigma-Aldrich. Anhydrous DME was purchased from Sigma-Aldrich. Final decarboxylative coupling reactions were run in a screw threaded tube from Chemglass (CLS-4208). One Kessil PR160 Blue LED grow light was used in the set up shown below, which provided 40 W and 456 nm light. A 2.0 mL solution of DME/H₂O (19:1) had an internal temperature of 40 ^oC after 1 hour under standard reaction conditions.

General Procedure for Cobaloxime/Photoredox-catalyzed Olefin Aroylation:

A flame dried 10 mL screw threaded glass tube equipped with stir bar was charged with Co(dmgH)₂Cl(NMI) (C2, 10 mol%, 0.02 mmol), Mes-Acr-Ph⁺ (5 mol%, 0.01 mmol), Cs₂CO₃ (10 mol%, 0.02 mmol), α–oxo-acid (1a-1l 0.2mmol, 1 equiv.) and styrene (2a-2t, 0.4 mmol, 2 equiv.). The vial was sealed and placed under an N2 atmosphere (3 cycles evacuation/back fill with N₂). DME (1.9 mL) and H₂O (0.1 mL) was added to the reaction vessel under positive nitrogen pressure. The sealed vessel was stirred and irradiated by one 40 W Kessil PR160 Blue LED Grow Light for 18 hours, which reached an internal temperature of 40 ^oC. After such time, the reaction is removed from the light, the tube opened, and the solvent removed under reduced pressure. The resulting product is purified via silica gel chromatography in 1:5–1:40 Et₂O/Hexanes.