Illuminating Palladium Catalysis

CONSPECTUS:

The past decade has witnessed significant advancements of visible-light-induced photocatalysis, establishing it as a powerful and versatile tool in organic synthesis. The major focus of this field has centered on the development of methodologies that either rely solely on photocatalysts or combine photocatalysis with other catalytic methods, such as transition metal catalysis, to address a broader and more diverse array of transformations. Within this rapidly evolving area, a subfield that we refer to as transition metal photocatalysis has garnered significant attention due to its growing impact and mechanistic uniqueness. A distinguishing feature of this subfield is the dual functionality of a single transition metal complex, which not only acts as a photocatalyst to initiate photochemical processes but also functions as a traditional catalyst, facilitating key bond-breaking and bond-forming events. As such, an exogenous photocatalyst is not required in transition metal photocatalysis. However, the implications of harnessing both the excited- and ground-state reactivities of the transition metal complex can extend beyond this simplification. One of the most compelling aspects of this area is that photoexcited transition metal complexes can exhibit unique reactivities inaccessible through conventional thermal or dual photocatalytic approaches. These distinct reactivities can be leveraged to accomplish novel transformations either by engaging an entirely different substrate pool or by unlocking new reactivities of known substrates.

In 2016, our group pioneered the use of phosphine-ligated palladium catalysts that can be photoexcited upon visible-light irradiation to engage diverse substrates in radical reactions. In our initial discovery, we showed that photoexcitation can redirect the well-established oxidative addition of a Pd(0) complex into aryl iodides toward an unprecedented radical process, generating hybrid aryl Pd(I) radical species. We subsequently extended this novel strategy to the formation of alkyl radicals from alkyl halides. These reactive radical intermediates have been harnessed in a wide variety of transformations, including desaturation, alkyl Heck reactions, and alkene difunctionalization cascades, among others.

Seeking to further expand this new avenue, we achieved the first example of asymmetric palladium photocatalysis in the context of allylic C–H amination, where the palladium catalyst now plays triple duty by additionally controlling the stereochemical outcome of the reaction. In parallel to reaction discovery, we have also established that diazo compounds, strained molecules, and electron-deficient alkenes can serve as alkyl radical precursors beyond organic halides and redox-active esters. Notably, the engagement of electron-deficient alkenes has been made possible by the photoinduced hydricity enhancement of Pd–H species, representing a new mode of photoexcited reactivity.

This Account presents our discovery and development of visible-light-induced palladium catalysis, organized by the type of transformations explored. Given the rapid progress in the field, we anticipate that this Account will provide readers with guiding principles and inspiration for designing and developing more efficient and novel transformations.

Graphical Abstract

1. INTRODUCTION

Aryl halides are among the most versatile aryl electrophiles in transition metal catalysis. A plethora of palladium-catalyzed transformations begin with Pd(0) undergoing oxidative addition with aryl halides to afford aryl Pd(II) intermediates (Figure 1, left). Our venture into palladium photocatalysis^5–9 began with the fundamental question of whether such a facile two-electron process can be altered upon visible-light irradiation. Specifically, we wondered whether the single-electron counterpart could be realized, as the resultant hybrid aryl Pd(I) radical species may exhibit novel reactivities and thus allow us to design new transformations (Figure 1, right). In this regard, we found encouragement in the pioneering works by Ryu on UV-induced palladium catalysis involving alkyl radicals,¹⁰ indicative of elementary steps involving single electron transfer (SET). Moreover, Caspar demonstrated in a series of photophysical studies that common Pd(0) complexes, such as Pd(PPh₃)₄, exhibit a broad UV absorption feature tailing into the blue-light region,¹¹ thus fulfilling the prerequisite of visible-light absorption. Taken together, we hypothesized that radical formation could be not only extended to the then-unknown aromatic systems but also achieved under milder and more practical visible-light irradiation.

Figure 1.

Reactivity of aryl halides with Pd(0) catalyst via classical two-electron vs photoinduced one-electron pathway.

To this end, we initiated our investigation of hybrid aryl Pd(I) radical using silyl ether 1 as a model substrate.¹ Previously, our group developed a program in silicon-based tethers for selective Pd-catalyzed C–H functionalization.¹² Although these tethers were efficient for C(sp²)–H activation, methods targeting C(sp³)–H bonds remained scarce.¹³ We envisioned that switching to a radical mechanism may open up new opportunities for C(sp³)–H functionalization. As illustrated in Figure 2, a plausible mechanism commences with the photoinduced generation of hybrid aryl Pd(I) intermediate A, which then undergoes thermodynamically favorable 1,5-hydrogen atom transfer (1,5-HAT) to produce the more stable hybrid alkyl Pd(I) species B. A subsequent β-H elimination furnishes silyl enol ether 2 and regenerates the Pd(0) catalyst in the presence of a base.

Figure 2.

Proposed catalytic cycle for the desaturation of silyl ethers via 1,5-HAT to an aryl radical.

To our delight, the feasibility of this process was validated using cyclohexanol derivative 3 as the substrate. Thus, Pd(PPh₃)₄ catalyzed the desaturation of 3 under irradiation with a 34 W blue LED (λ_max ≈ 450 nm) in 72% yield (Figure 3). Notably, thermal conditions did not furnish any desired product, underscoring the essential role of photoexcitation. Further screening led us to the optimal ligand L using Pd(OAc)₂ as the catalyst, with which the scope was examined. In general, both cyclic and acyclic alcohols were amenable to this transformation (5–9). Complex molecules, such as steroids, also underwent smooth desaturation (10). In a mechanistic study, cyclopropyl radical probe 11 was exclusively converted to 12 through radical ring-opening, thus ruling out ionic pathways involving β-H elimination (13) or β-C elimination (14) of the analogous alkyl Pd(II) intermediate. The radical nature was further supported by a deuterium-labeling experiment (d-3 → d-4). Together, these studies corroborated the generation of aryl radical under photocatalytic conditions.

Figure 3.

Photocatalytic desaturation of silyl ethers.

2. DESATURATION OF ALIPHATIC ALCOHOLS AND AMINES

Given the preference of the iodophenyl silyl tether for a 1,5-HAT process, a different tether was required to activate remote C(sp³)–H sites. Drawing inspiration from our previous work¹⁴ as well as others’,¹⁵ we employed an iodomethyl silyl tether to desaturate aliphatic alcohols at remote sites (Figure 4).¹⁶ Upon surveying the substrate scope, we found that HAT preferentially occurred at the γ-position over the β-carbon (18), which was in turn favored over the δ-position (19). Desaturation of challenging substrates such as 21 using the state-of-the-art methodology developed by the Baran group resulted in low regioselectivity.¹⁷ In comparison, our protocol involved a regioselective, Pd-mediated β-H elimination, thus enabling the desaturation of the same alcohol to form 25 as a single product. The exclusive regiochemical outcome further highlights the capability of palladium catalysts to both initiate photochemistry and mediate subsequent bond-forming step(s). More recently, the Yu group employed N-alkoxypyridinium salts toward δ/ε desaturation through 1,5-HAT to an O-centered radical.¹⁸

Figure 4.

Remote desaturation of aliphatic alcohols.

Beyond aliphatic alcohols, we were also interested in the analogous transformations of aliphatic amines. Since silyl-protected amines are hydrolytically unstable, we sought to explore amide-based tethers as practical alternatives. Thus, using carbonyl- (T¹) and sulfonyl-based (T²) tethers, we established streamlined protocols for the desaturation of primary and secondary amines in a one-pot fashion (Figure 5).¹⁹ Expectedly, T¹-protected amines demonstrated selectivity toward proximal desaturation due to favorable 1,5-HAT (28, 29), while those bearing T² were predisposed to 1,7- or 1,6-HAT,¹⁷ furnishing remote desaturation products (30–32, 36). The intramolecular nature of the HAT step could be leveraged to ensure site selectivity. Thus, a menthol derivative afforded 33 as the only product, as the other γ-C–H bond was spatially inaccessible.^16,17 Moreover, regiodivergent desaturation could be accomplished through the appropriate choice of tether (35 vs 36). Related amide desaturation protocols have also been developed by other groups.^20–22

Figure 5.

Desaturation of aliphatic amines.

3. ALKYL HECK REACTIONS

In parallel with our endeavor in desaturation chemistry, we were interested in the possibility of intermolecularly trapping the hybrid alkyl Pd(I) radicals with alkenes, as the resultant C-centered radicals could then undergo β-H elimination to yield valuable alkyl Heck products. While Pd-catalyzed alkyl Heck reactions had been reported, certain limitations associated with the necessary harsh thermal conditions, such as limited functional group tolerance and moderate E/Z selectivity of the alkene products, were yet to be addressed.^23–26 With this in mind, we initially focused on various α-functionalized alkyl halides 38 as coupling partners. These substrates are not only underexplored but also versatile synthetic handles for further transformations (Figure 6).² Under the optimal conditions, a wide array of heteroatom-based functionalities were efficiently incorporated into alkenes 37 through the alkyl Heck reaction, furnishing diverse allylic systems 40–50. Additionally, simple unactivated alkyl halides were also amenable to this transformation (51).

Figure 6.

Alkyl Heck reaction of the α-functionalized alkyl halides.

The past few years have seen significant progress in the scope expansion of alkyl halides, now including iodides, bromides, and chlorides of all degrees of substitution.^27–29 Besides, redox-active esters, a popular class of alkyl radical precursors in photoredox catalysis, have been utilized in alkyl Heck reactions.^30,31 While diverse alkyl radical sources have become available, the alkene partner is still largely restricted to activated ones like styrenes and Michael acceptors.

Demonstrating its potential utility, we applied the alkyl Heck reaction in the development of positron emission tomography (PET) imaging probes (Figure 7).³² This methodology features a one-pot, two-step protocol that first converts alkenes 52 to β-fluoroalkyl iodides 53 through classical Markovnikov-selective iodofluorination. 53 can then be engaged in photoinduced Pd-catalyzed alkyl Heck reaction to produce ¹⁸F-enriched homoallylic fluorides 54. A diverse collection of alkenes 52 underwent radiofluoroalkenylation in the presence of HF/pyridine as a carrier in generally good radiochemical yields (55–57). To address the common issue of reduced molar activity when a carrier was used, we further optimized the reaction conditions and arrived at a carrier-free protocol with comparable efficiency (58).

Figure 7.

One-pot formal radiofluoroalkenylation of alkenes. RCC, radial chemical conversion; A_m, molar activity.

Having established alkyl Heck reactions using alkyl halides, we wondered whether the alkyl radical could instead be derived from a C–H bond. From a synthetic perspective, this approach can offer an attractive alternative, especially when alkyl halides are difficult to obtain. More importantly, site-selective C(sp³)–H alkenylation has been a longstanding challenge in the field of C–H functionalization. In our previous work on the selective desaturation of aliphatic alcohols, the silicon tethers proved effective in guiding radical formation at a targeted C(sp³)–H site.¹⁶ We therefore postulated that this strategy could provide an opportunity for radical relay C(sp³)–H alkenylation if the alkyl radical could be effectively intercepted with an alkene (Figure 8).³³ However, under the previous conditions featuring L as the ligand, the desired remote Heck product 62 was only obtained in trace amounts due to substantial desaturation (63) and reduction (65) side pathways. Consistent with the literature suggesting that ligands with larger bite angle tend to disfavor β-H elimination,³⁴ our ligand screening campaign identified Xantphos to be the superior ligand for promoting the Heck reaction. This protocol enabled site-selective tertiary C–H alkenylation at the γ- (66), β- (67), and δ- (68) positions. Additionally, less reactive secondary sites could also be functionalized (69). More recently, this radical relay concept has been further expanded by other groups employing amide tethers.^35,36 In addition to HAT, Dowd–Beckwith ring expansion³⁷ and 1,2-spin-center shift^38,39 have also been exploited to relay the initially formed alkyl radical to the adjacent carbon, further broadening the scope of radical relay Heck reactions.

Figure 8.

Radical relay Heck reaction of aliphatic alcohols.

Besides alkenes, we also explored the alkylation of other π systems. We became particularly interested in imines, as the reported radical alkylation protocols are overall reductive, resulting in the loss of the synthetically versatile imine functionality.^40–42 We envisioned that the “oxidative” nature of palladium catalysis would enable a Heck-type reaction, thereby complementing these reductive transformations. Indeed, we found that oximes 70 could be efficiently coupled with alkyl halides 71 to furnish Heck-type products 72 (Figure 9A).⁴³ Notably, higher efficiency and improved reaction reproducibility were achieved by Lewis acid activation of the oxime with In(III) additive. Primary, secondary, and tertiary alkyl groups could be efficiently introduced (73–75). Furthermore, the oxime substituent could be varied, including easily removable protecting groups for the subsequent revelation of O-unprotected oximes (76, 77). We later expanded the imine scope to include α-ester hydrazones 78 (Figure 9B).⁴⁴ Importantly, the elusive E isomers were selectively obtained for primary and secondary alkyl groups (81, 82), which was leveraged in sequential N,C-alkylation with bishalides 85 to access tetrahydropyridazines (86–88).

Figure 9.

Alkyl Heck-type reactions of (A) oximes and (B) α-ester hydrazones.

4. ALKENE DIFUNCTIONALIZATION

Most of our efforts were initially devoted to the conceptualization of visible-light-induced hybrid Pd(I) radical chemistry and the development of different carbon electrophiles. After establishing a set of available radical precursors, we began applying them in transformations beyond desaturation and Heck reactions. Mechanistically, both desaturation and Heck reactions share a common feature, where the transposed alkyl radical undergoes facile β-H elimination to produce an alkene moiety. While alkenes are appealing targets, we wondered whether the transposed radical could instead be intercepted intermolecularly to install other functionalities. We found that alkenyl iodides 89 can be converted to a range of carbo- and heterocycles upon alkene carboiodination, such as dihydrobenzofurans (91) and cyclopentanes (92–94) (Figure 10).⁴⁵ This transformation likely proceeds via a radical relay initiated by the photoinduced formation of hybrid alkenyl Pd(I) intermediate A. This is followed by 1,5-HAT leading to tertiary alkyl radical B, meanwhile revealing an α-olefin for the subsequent 5-exo-trig cyclization to produce the primary alkyl radical C. Under the reaction conditions, intermediate C undergoes exclusive halogen atom transfer (XAT) with Pd(I)I, rather than β-H elimination, to furnish the difunctionalization product 90. While the overall transformation represents a HAT/atom-transfer radical cyclization cascade, a radical chain mechanism propagating through XAT from 89 to C is unlikely, given that the alkenyl–iodide bond is considerably stronger than the alkyl–iodide bond.⁴⁶ In their recent report on related systems involving a radical addition/cyclization cascade, Marchese, Lautens, and co-workers further showed that the XAT step is reversible under photocatalytic conditions.⁴⁷

Figure 10.

ATRC cascade of alkenyl iodides toward iodomethyl carbo- and heterocycles.

In the previous example, the translocated radical is intramolecularly trapped by the tethered alkene. Alternatively, one could intercept the radical by using an external trap. This approach offers the advantage of modularity but is considerably more challenging due to the propensity of β-H elimination as discussed before. Nonetheless, we found that isocyanides 97 can effectively interrupt an alkyl Heck reaction to afford synthetically versatile ketenimines 98 via radical intermediate A. 98 can then be readily diversified in a one-pot fashion to amides (100), cyanides (101), tetrazoles (102), and amidines (103) in generally good overall yields (Figure 11).⁴⁸

Figure 11.

Multicomponent carbofunctionalization of alkenes through one-pot diversification of ketenimines.

5. ALLYLIC SUBSTITUTION

While the XAT and isocyanide trapping strategies offer pathways to alkene difunctionalization, a more generalizable approach would involve further utilizing the transposed hybrid Pd(I) radical in Pd-catalyzed transformations. This is mechanistically intriguing, as it merges excited- and ground-state reactivities using a single palladium catalyst. Additionally, the synthetic implications can be profound, given the myriad reactions catalyzed by palladium. In our initial proof-of-concept study, we demonstrated that conjugated dienes 104 can undergo carbofunctionalization reactions with alkyl iodides 105 and diverse nucleophiles 106 (Figure 12).⁴⁹ The key enabling feature setting this transformation apart from previous alkyl Heck reactions is that the transposed allyl Pd(I) radical A can undergo recombination via a radical–polar crossover (RPC) path to provide the classical closed-shell π-allyl Pd(II) intermediate B. We reasoned that this intermediate is thermodynamically more stable than the analogous π-benzyl complex, rendering it more susceptible to interception by an external nucleophile through Tsuji–Trost-type substitution.⁵⁰ The triflimide additive plays a crucial role by suppressing the dimerization of A (109). In contrast to conventional hydrofunctionalization reactions, where site selectivity is strongly influenced by the diene stereochemistry,⁵¹ this transformation does not display such correlation due to exclusive alkyl radical addition to the terminal carbon. Thus, 1,2-addition products are generally obtained when terminal dienes are employed (110–114). However, tertiary alkyl iodides lead to the formation of regioisomers due to steric hindrance. Likewise, 1,4-addition is favored when the internal carbons are more sterically demanding (115). We also demonstrated that other nucleophiles commonly used in Tsuji–Trost substitution are competent reaction partners (116, 117). Based on their first report on the same topic,⁵² the Glorius group described analogous 1,4-difunctionalization of differently substituted dienes.⁵³ Importantly, this approach has since been extensively adopted by other research groups for the development of diverse transformations,^54–59 underscoring its generality and applicability.

Figure 12.

Carbofunctionalization of 1,3-dienes proceeding via radical–polar crossover.

Although the diene strategy enables access to the π-allyl Pd(II) complex from simple diene building blocks, translating this approach to complicated substrates may not be trivial from a synthetic perspective. Besides, conjugated dienes are not frequently encountered in complex molecules, rendering them not an ideal target for late-stage functionalization. To address these issues, we turned our focus to readily available alkenes, with the hypothesis that the allyl Pd(I) radical and thus the π-allyl Pd(II) intermediate could be directly derived from alkenes. Indeed, we found that aryl radical, photocatalytically generated from the corresponding bromide, serves as an efficient HAT agent by homolytically activating the allylic C–H bond (Figure 13).³ It is worth mentioning that in this case, the aryl bromide functions as an oxidant, which is unorthodox in palladium catalysis. Given the highly reactive nature of aryl radicals, ortho substituents were introduced at the aryl bromide to suppress undesired pathways, such as radical addition to and HAT from solvent molecules. Additionally, the use of dibromide not only improves the atom economy but also enhances the reaction efficiency. The outcome of this reagent design campaign is a highly chemoselective allylic C–H amination protocol that tolerates a wide range of other weak C–H bonds as well as terminal alkenes susceptible to aryl radical addition (121). Electron-deficient Michael acceptors are chemoselectively aminated at the allylic site, despite their propensity for conjugate addition (122). These examples demonstrate the complementary nature of palladium photocatalysis in comparison with its traditional ground-state chemistry, where only monosubstituted aliphatic alkenes can be activated.^60,61 Notably, 122 also contains a pendant tertiary amine unit, which could sabotage the reaction in the case of photoredox approaches.⁶² More excitingly, stereoselective processes can be accomplished when using a chiral BINAP ligand, including examples where a site-selective HAT step is incorporated (128, 129). Collectively, these results substantiate the feasibility of asymmetric palladium photocatalysis, in which a single palladium catalyst enables photochemistry, catalyzes bond formation, and controls stereochemistry.

Figure 13.

Allylic C–H amination of internal alkenes enabled by homolytic C–H activation.

Having established the conceptual ground of homolytic activation via aryl radical, we continued to expand the repertoire of this new reactivity mode with diverse oxygen-based nucleophiles toward allylic C–H oxygenation (Figure 14A).⁶³ While linear internal alkenes proved to be competent substrates, we primarily focused on cyclic alkenes, as they were minimally explored in the previous work. Additionally, cyclic substituted alkenes offer the opportunities of rapidly accessing complex cyclic scaffolds if the regio- and diastereoselectivity of this process could be controlled. To our delight, we found that systems possessing multiple allylic C–H sites can be regioselectively functionalized (133). The site selectivity likely arises from the preferred HAT from the less hindered and weaker C–H bond. Moving the substituent to the aliphatic site, a diastereoselective transformation can be achieved with a wide range of nucleophiles (134). More interestingly, we showed that site-selective HAT and diastereoselective C–O bond formation can be coupled to furnish complex frameworks, which is readily translated to more complicated settings (135). The intramolecular variant is also feasible, in this case affording bicyclic lactones (136, 137). In a more recent work, we further showcased the generality of this aryl radical-based strategy by developing an allylic C–H sulfonylation protocol using sulfinate salts as nucleophiles (Figure 14B).⁶⁴ Likewise, a range of electronically diverse cyclic, and acyclic alkenes proved to be viable substrates (138–140). We also demonstrated that sulfinate salts readily derived from drug or biologically active molecules are effective coupling partners (141).

Figure 14.

Allylic C–H functionalization with diverse (A) oxygen-based and (B) sulfinate nucleophiles.

Besides allylic C–H activation, we have also developed an alternative approach toward allylic functionalization through the in situ assembly of allyl electrophiles (Figure 15).⁶⁵ In the first half of the reaction sequence, dielectrophiles 143 undergo a photoinduced alkyl Heck reaction with alkenes 142 to produce intermediate 145. The second half involves a ground-state Tsuji–Trost substitution of 145 with amine nucleophiles to furnish allylic amine products. Through identification of the appropriate ligand, various leaving groups in 143 can be utilized, including trimethylamine (151, 152), acetate (153, 154), and bromide (155, 156). While there seems to be a matching effect between the ligand and leaving group, other reaction parameters, such as the solvent, may also contribute to the observed outcomes. As in our previous work, switching to BINAP ligands renders the process stereoselective (157, 158). Overall, this transformation represents an interesting merger of two catalytic cycles, one operating through photoinduced Pd(0/I/II) catalysis and the other through traditional Pd(0/II) chemistry, again with a single palladium catalyst.

Figure 15.

Homologative amination of alkenes through the merger of light and dark catalytic cycles.

6. TOWARD NEW RADICAL PRECURSORS

Aiming at further expansion of the scope of electrophiles beyond halides and redox-active esters, we recently revisited the development of new radical precursors. In the first example, we described the generation of aryl radicals from aryl triflates 159 through a formal C–O bond homolytic cleavage (Figure 16).⁶⁶ Given the abundance and availability of phenols, this method represents a substantial addition of aryl radical precursors to the existing substrate pool. 159 contains an amide linkage that tethers an alkyl fragment for 1,5-HAT via intermediate A, eventually affording oxindoles or isoindolin-1-ones depending on the linker connectivity. Introducing meta substituents typically leads to a mixture of regioisomers (161), where the regiochemical outcome is likely a consequence of the relative radical stability upon cyclization. This protocol enables access to complex tricyclic systems (162) and tolerates enolizable carbonyl groups by virtue of its mildly basic conditions (163).^67,68 In the case of 2° C–H sites, the current method is limited to systems lacking a β hydrogen due to a competing desaturation pathway (164 vs 165). Similarly, benzamide derivatives yield the corresponding isoindolin-1-ones (166, 167). Deuterium-labeling studies corroborated the formation of aryl radicals. Thus, 168 produced 170 with 48% deuterium incorporation, which aligns with a radical mechanism through quantitative deuterium transfer in intermediate B followed by a statistical loss upon cyclization. In contrast, a concerted metalation–deprotonation (CMD) pathway would lead to a complete loss of deuterium. Likewise, the statistical loss of deuterium enrichment using 169 argued against the CMD mechanism.

Figure 16.

Aryl triflates as aryl radical precursors in heterocycle synthesis via radical C–H arylation.

Diazo compounds and N-tosylhydrazones are versatile synthons but are only sporadically engaged in radical reactions.^69–74 We therefore wondered whether they could serve as alkyl radical precursors under palladium photocatalysis. Using the alkyl Heck reaction as a standard transformation to probe alkyl radical reactivities, we established Brønsted acid (HX)-assisted protocols enabling the coupling between alkenes 171 and diazo compounds 172 or N-tosylhydrazones 173 (Figure 17).⁷⁵ For 172, a piperidinium salt (BA) was found to be the optimal source of HX. Notably, basic conditions completely shut down the reaction. In the case of 173, for which deprotonation is required to reveal active diazo intermediates, a combination of DIPEA and NaI proved to be the most effective. Diverse diazo compounds and alkenes can be used (177–180) except for ortho-substituted styrene derivatives, for which little to no product was obtained (181). Nevertheless, this limitation can be circumvented by using N-tosylhydrazones as the coupling partner (183). Radical probe experiments using vinyl cyclopropane 184 established the radical nature of this transformation. Labeling studies were performed to elucidate the source of hydrogen. When AcOD was used as a Brønsted acid, product 187 was obtained with 91% deuterium incorporation, suggesting that sequential proton transfer/electron transfer (PT/ET, path a) or a concerted mechanism (PCET, path b) can be operative. Perhaps more interestingly, we found that an independently synthesized palladium deuteride complex (40% D) delivered product 188 with an appreciable level of deuterium transfer. This result points to an alternative Pd–H-promoted pathway involving the rearrangement of carbene intermediate D to photoactive alkyl Pd(II) species E (path c).

Figure 17.

Diazo compounds and N-tosylhydrazones as alkyl radical precursors in the alkyl Heck reaction.

The finding that Pd–H is potentially involved in our previous work was intriguing and prompted us to investigate it further with other substrate classes. Cyclopropenes and bicyclo[1.1.0]butanes (BCBs) are both strained molecules known to undergo strain-release functionalizations.^76–79 Along this line, we showed that 190 and 191 are readily hydropalladated (A, A′) and converted to cyclopropyl and cyclobutyl radicals (B, B′) upon photoexcitation (Figure 18).⁸⁰ Subsequent trapping of the latter with alkenes affords a range of alkenyl cyclopropanes and cyclobutanes (194–197). Interestingly, we discovered that certain gem-difluorocyclopropenes 198 undergo cascade hydroalkenylation/diastereoselective rearrangement to furnish densely substituted cyclopentenes (200, 201). Mechanistic studies revealed that the Pd–H species is generated in situ under the reaction conditions. Additionally, the synthesized Pd–H complex can catalyze the reaction without an external HX source. These results further reinforce the involvement of Pd/H intermediates in these photoinduced transformations.

Figure 18.

Strained molecules as alkyl radical precursors in the alkyl Heck reaction.

Having established that strained π systems serve as alkyl radical precursors via Pd/H catalysis, we sought to develop more general protocols applicable to other alkenes. Inspired by the Miller group’s discovery of visible-light-induced hydricity enhancement of Ir–H complexes,^81,82 we hypothesized that the alkene scope could be substantially broadened if Pd–H species displayed similar photoinduced behavior. Indeed, we found that photoexcited Pd–H complex can engage electron-deficient alkenes 203 in radical reactions with alkenes 202, where the overall transformation represents an unusual head-to-tail cross-dimerization of alkenes (Figure 19).⁴ The key mechanistic feature of this work is the photoinduced polarity reversal of the Pd–H complex, which gears it from the conventional preference for electron-rich/neutral alkenes (A) to hydridic hydropalladation with electrophilic 203 (B). A wide range of differently substituted Michael acceptors (205–207) and alkenes (208–209) can participate in this transformation. Notably, 210 was obtained as the exclusive isomer with the terminal aliphatic alkene moiety remaining intact, showcasing the complete selectivity switch upon photoexcitation. In addition, electron-deficient vinyl (hetero)arenes can also be employed as radical precursors (211, 212). The latter example is remarkable, as it demonstrates a chemoselective cross-dimerization of two different vinylpyridines. The effect of photoexcitation on hydricity enhancement is supported by mechanistic studies using the trityl cation 213 as a hydride trap. Thus, we observed hydride transfer from Pd–H complex only in the presence of light irradiation. Additionally, control studies showed that the thermal reaction between styrene and butyl acrylate yielded only trace amounts of styrene homodimer 215. Expectedly, the desired cross-selectivity was restored upon irradiation (216). In retrospect, hydricity enhancement, in addition to strain release, might have also played a role in promoting the reaction in the case of cyclopropenes and BCBs.⁸⁰

Figure 19.

Head-to-tail hydroalkenylation of alkenes enabled by the photoinduced hydricity enhancement of Pd–H.

The implications of photoinduced hydricity enhancement go beyond access to alkyl radicals from alkenes. For instance, we showed that this new reactivity can be coupled with allylic substitution chemistry (cf. Figure 12) to accomplish difunctionalization of dienes 217, which would otherwise undergo facile hydrofunctionalization via the ground-state Pd–H catalysis (Figure 20).⁸³ Notably, acrylic acid (218) serves a triple role in this transformation: it contains an acidic proton and an alkene for the generation and capture of Pd–H species while possessing a (pro)nucleophile for the subsequent allylic substitution step. 218 encompasses a diverse range of substrates, including acrylic acids (220–222), acrylamides (223, 224), and Baylis–Hillman adducts (225, 226). Remarkably, complex bis-spiro frameworks can be readily accessed via this protocol (227, 228). In addition, conjugated enynes can be employed to deliver the corresponding alkynyl lactones and lactams (229, 230).

Figure 20.

Pd–H-enabled carbofunctionalization of conjugated dienes and enynes.

7. CONCLUSION AND OUTLOOK

Nine years ago, by the generation of aryl radicals from aryl iodides,¹ we demonstrated that visible-light irradiation redirects the well-established two-electron palladium chemistry to unorthodox radical reactivity. We subsequently showed that this can be extended to alkyl halides toward alkyl radical formation.² Since then, we and others have unlocked similar reactivity beyond aryl or alkyl halides, from redox-active esters and benzotrifluorides to our recent discoveries of diazo compounds, strained molecules, and electron-deficient alkenes. While the majority of palladium photocatalysis is driven by photoinduced SET from a Pd(0) catalyst (Figure 21A), it has become clear that photoexcitation can also facilitate other elementary steps, such as reductive elimination (Figure 21B)^47,84 and migratory insertion (Figure 21C)⁴ as well as oxidative addition involving Pd(II).^85,86 Such diversity has in turn enabled a range of transformations beyond desaturation and Heck reactions, including C–N cross-coupling,⁸⁶ carbonylation reactions,⁸⁴ and hydroalkenylation chemistry.⁴ In addition, we and others have demonstrated encouraging results substantiating the feasibility of asymmetric palladium photocatalysis.^3,57–59 Together, these discoveries could serve as design principles for the exploration of novel reactivities and the design of new photochemistry-enabling ligands toward the development of new photoinduced transformations.

Figure 21.

Diverse radical reactivities of palladium complexes facilitated by visible-light photoexcitation.

Biographies

Kelvin Pak Shing Cheung earned his B.Sc. and M.Phil. in Chemistry from the Chinese University of Hong Kong. In 2018, he joined the Gevorgyan group at the University of Illinois at Chicago as a Ph.D. student and later at the University of Texas at Dallas, where he received his Ph.D. in Chemistry in 2024. His research focused on the development of visible-light-induced transition-metal-catalyzed transformations.

Vladimir Gevorgyan received his Ph.D. from the Latvian Institute of Organic Synthesis. After postdoctoral research at Tohoku University in Japan with Prof. Y. Yamamoto as a JSPS and then Ciba Geigy International Postdoctoral Fellow, in 1996 he accepted a faculty position there. In 1999, he moved to USA to join UIC (Associate Professor, 1999; Professor, 2003; LAS Distinguished Professor, 2012). In 2019, he joined UTD as a Robert. A. Welch Distinguished Chair in Chemistry, with a secondary appointment at the UT Southwestern Medical Center. His group is interested in the development of novel synthetic methods.