A dual light-driven palladium catalyst: Breaking the barriers in carbonylation reactions

See allHide authors and affiliations

Science  17 Apr 2020:
Vol. 368, Issue 6488, pp. 318-323
DOI: 10.1126/science.aba5901

Lighting the way coming and going

Catalysts accelerate chemical reactions by breaking existing bonds and then forming new ones. Often, the factors that favor the first process can muddle the second one, constraining a catalyst's generality. Torres et al. found that visible light excitation of a palladium complex can facilitate both the breaking and making of carbon-halogen bonds (see the Perspective by Kathe and Fleischer). The reaction specifically forms acid chlorides by carbonylation of a wide variety of alkyl or aryl bromides and iodides. These products in turn can react further to form amides and esters.

Science, this issue p. 318; see also p. 242


Transition metal–catalyzed coupling reactions have become one of the most important tools in modern synthesis. However, an inherent limitation to these reactions is the need to balance operations, because the factors that favor bond cleavage via oxidative addition ultimately inhibit bond formation via reductive elimination. Here, we describe an alternative strategy that exploits simple visible-light excitation of palladium to drive both oxidative addition and reductive elimination with low barriers. Palladium-catalyzed carbonylations can thereby proceed under ambient conditions, with challenging aryl or alkyl halides and difficult nucleophiles, and generate valuable carbonyl derivatives such as acid chlorides, esters, amides, or ketones in a now-versatile fashion. Mechanistic studies suggest that concurrent excitation of palladium(0) and palladium(II) intermediates is responsible for this activity.

The capacity of transition metals to cleave and forge covalent bonds via the fundamental operations of oxidative addition and reductive elimination is a cornerstone of catalysis. Examples with palladium or nickel alone include such prominent reactions as cross coupling (1, 2), C–H bond functionalization (3), hydrogenation (4), reduction (5), or oxidation (6, 7) and are exploited across the spectrum of pharmaceutical and fine chemical synthesis. By manipulating the features that govern the basic steps (Fig. 1A), researchers over the past several decades have optimized catalysts for each of these reactions. Unfortunately, there are also intrinsic limitations to this cycle that diminish its utility. These arise from the forward and reverse sequences of the successive operations at the metal, in which steric or electronic features of the catalyst that favor one half-cycle often inhibit the other. Balancing these steps is thus a critical feature of reaction design that may lower overall catalyst activity or constrain substrate scope. Moreover, it is often not possible to perform catalysis that requires both difficult oxidative addition and reductive elimination steps (8), which blocks direct routes to many important classes of products.

Fig. 1 A light-based strategy for the oxidative addition–reductive elimination cycle.

(A) Opposing influences on these operations. L, ligand; M, metal. (B and C) Their limiting use in carbonylations. IMes, 1,3-dimesitylimidazol-2-ylidine; Ar, aryl. (D) Concept: Exploiting multiple photoevents to access potent, broadly applicable palladium catalysts. X, halogen; Nu, nucleophile.

These features are well illustrated in palladium-catalyzed carbonylation reactions. Carbonylation offers, at least in principle, one of the most efficient routes to assemble valuable carbonyl compounds from feedstock reagents, including such ubiquitous functionalities as esters, amides, or ketones (9, 10). However, the inhibitory influence of CO coordination to palladium on oxidative addition requires increased temperatures relative to the analogous cross-coupling chemistry and restricts the use of many key substrates, such as less-reactive, simple alkyl halides (11). These issues can be partially addressed by catalyst design. For example, efforts to promote the use of alkyl halides in carbonylations have been described by Alexanian and Ryu with electron-rich catalysts or ultraviolet (UV) light, respectively, but the factors strongly favoring oxidative addition often necessitate the use of strong nucleophilic partners for elimination (Fig. 1B) (12, 13). Carbonylations also often require nucleophiles that can readily associate to palladium for reductive elimination (1416). Research here has also been reported using sterically hindered ligands to favor the formation of acyl halides that react with an array of nucleophiles (Fig. 1C) (1719). However, the catalyst features needed to favor the challenging reductive elimination now restrict this chemistry to substrates that can undergo facile oxidative addition.

An intriguing alternative approach to metal catalysis that is of growing prominence is to exploit external factors to drive these reactions. For example, seminal reports from a number of labs have shown that pairing a visible-light photocatalyst with a more conventional nickel or palladium catalyst can accelerate a challenging step in coupling reactions (2027). These can even be performed through the direct excitation of palladium catalysts (2832) and have been applied to carbonylative Suzuki coupling with alkyl halides (33). Although such strategies expand the scope of products available via catalysis, the other steps in the catalytic cycle do still rely on the classic steric and electronic factors of the metal and can be limited by the thermal barriers inherent to these operations.

In considering these features, we questioned if light might play a greater role in catalysis. For example, could visible light offer a pathway to completely eliminate the need to balance oxidative addition and reductive elimination steps and drive both in the same transformation? Because photoreactions can proceed with low barriers, such a system could offer the exciting opportunity to perform reactions without having to compromise with ligand effects in each step of the cycle and thus access exceptionally potent and versatile catalysts. The use of two different photoredox events to accelerate opposing operations such as these poses a substantial challenge of balancing two reverse redox events and their highly plausible mutual quenching (34). Nevertheless, light has been shown to have multiple influences on catalysis beyond photoredox chemistry that might be exploited in this design. We describe herein our studies toward such a system.

Our initial approach to this catalyst system focused on the use of light to favor the palladium-catalyzed formation of acid chlorides. As noted above, though the buildup of acid chlorides offers a route to apply carbonylations to an array of nucleophiles, the ligands needed to favor this challenging elimination (PtBu3 and CO; tBu, tert-butyl) limit oxidative addition to reactive substrates such as aryl iodides and bromides at high temperatures (Fig. 1C). In these, in situ halogen exchange with a chloride source leads to acid chlorides, but less reactive reagents such as alkyl halides are not viable substrates. We postulated that photoredox chemistry could oxidize in situ generated Pd(II)-acyl complexes to a transient Pd(III) intermediate to induce the reductive elimination, in analogy to recent reports in nickel catalysis (23, 24). Nevertheless, it was unclear if photoredox chemistry could drive reductive elimination from a stable Pd(II) center or if such a catalyst might also undergo light-driven oxidative addition.

To test for this potential, the carbonylation of o-tolyl iodide was performed in the presence of 1 equiv Bu4NCl, to allow the formation of acid chloride, and with the sterically hindered 2,6-diisopropylaniline as an in situ acid chloride trap (fig. S1) (18). After probing various catalyst systems and photocatalysts in the presence of blue-light irradiation, we were pleased to discover that the use of Xantphos with [Pd(allyl)Cl]2 and {Ir[dF(CF3)ppy]2(dtbpy)}PF6 (35) leads to the ambient temperature formation of acid chloride trapping product 1a in 34% yield (Fig. 2A and figs. S1 to S3 for full catalyst development; dF(CF3)ppy, 3,5-difluoro-2-[5-(trifluoromethyl)-2-pyridinyl-N]phenyl; dtbpy, 4,4′-bis(1,1-tert-butyl)-2,2′-bipyridine). However, to our surprise, when the reaction was performed in the absence of the iridium photocatalyst, amide was generated in similar yield. No reaction was observed in the absence of light, under green or red light irradiation, or without the palladium catalyst. Subsequent examination of the system without an added photocatalyst showed that whereas most other ligands lead to minimal reaction with visible light, [(2-diphenylphosphino)phenyl]ether (DPEphos) displayed enhanced activity and formed amide 1a in near-quantitative yield when performed in benzene solvent (Fig. 2B and fig. S3). The generation of acid chloride (2a) can be clearly seen by performing the same reaction in the absence of amine (Fig. 2C).

Fig. 2 Light and ligand effects on the Pd-catalyzed carbonylation of aryl iodides to acid chlorides.

(A) Unusual effect of light on in situ acid chloride generation. ditBupy, 2,6-di-t-butylpyridine; iPr, isopropyl. (B) Accelerated coupling with DPEphos instead of Xantphos in reaction A (with collidine base in benzene). Ph, phenyl; 1-Ad, 1-adamantyl; Bn, benzyl; Pr, propyl; BINAP, 2,2′-bis(diphenylphosphino)-1,1′-binaphthyl; dppe, 1,2-bis(diphenylphosphino)ethane; dppp, 1,3-bis(diphenylphosphino)propane; dppf, 1,1′-bis(diphenylphosphino)ferrocene; Bipy, 2,2′-bipyridine. (C) Catalytic synthesis of acid chloride with light.

Visible-light excitation of Pd(0) has been noted by Gevorgyan, Fu, Rueping, and others to accelerate oxidative addition via a postulated light-induced electron transfer from palladium to the organic halide (2832). However, the capacity to induce the room-temperature buildup of a reactive acid chloride, and to do so without an added photocatalyst, suggests that light plays a distinct role in this system, because the reductive elimination typically requires pressing thermal conditions (>100°C, high CO pressure). To more closely interrogate this product-forming step, the DPEphos-ligated Pd(II)-acyl complex 3b-Cl was generated (see supplementary materials for synthesis). Complex 3b-Cl absorbs blue light [absorption range (λabs) = 330 to 460 nm, fig. S4), and its irradiation even at low temperature leads to the near-quantitative reductive elimination of acid chloride 2b within 5 min (Fig. 3A). The reaction mixture reverts back to a near 1:1 equilibrium mixture of 3b-Cl and acid chloride 2b in the absence of light (fig. S4A). The excitation of 3b therefore appears to create an unusual photostationary state that drives the generation of acid chloride.

Fig. 3 Mechanistic insights into the role of visible light in catalysis.

(A) Photoinduced acid chloride reductive elimination, wherein trapping hints at the intermediacy of acyl radicals. OctSH, 1-C8H17SH. (B) Reversibility of light-driven reductive elimination. eq., equiv. (C) Light-driven aryl halide oxidative addition. (p-MeOC6H4I/Br was used because it is easily monitored by 1H NMR analysis.) (D) Alkyl iodide activation with light. (E) Potential mechanisms for the catalytic chemistry.

Stoichiometric reactivity studies provided further insights into this reductive elimination step. For example, the irradiation of complex 3b-Cl in the presence of 1 equiv (2,2,6,6-tetramethylpiperidin-1-yl)oxyl (TEMPO) leads to the rapid formation of the acyl-radical trapping product 4b (Fig. 3A). A similar transformation with a thiol leads to acyl-radical hydrogen-atom abstraction and aldehyde 5b generation even faster than the known thioester formation from the reaction at palladium (10). These and other control experiments (see fig. S4 for full mechanistic study) are consistent with the formation of acyl radicals upon excitation of 3, which could lead to acid chloride formation by subsequent chlorine abstraction (step A of Fig. 3E; see fig. S5 for other pathways to acyl radical reactivity and the potential role of the DPEphos ligand) (3638). Metal-acyl bonds have been reported to undergo light–driven homolysis, although not in palladium catalysis (39, 40). As shown in Fig. 3B, the accelerating influence of light on this radical-induced reductive elimination is sufficient to drive acid chloride formation even in the presence of a large excess of product, which quantitatively regenerates 3b-Cl within 1 hour when light is removed.

In addition to favoring reductive elimination, light also appeared to be involved in oxidative addition chemistry through the excitation of the Pd(0) complex (DPEphos)Pd(CO)2 (7) (Fig. 3) (41). Initial indications were noted upon closer analysis of the catalytic reaction mixture, where we observed the low-yield formation of a second product: the biaryl o-CH3C6H4-C6H5 (6a, 8%; see fig. S6A). The latter formally arises from o-tolyl radical addition to the C6H6 solvent (42). Pd(0) complex 7 is formed upon acid chloride elimination, can be independently generated (see supplementary materials for synthesis), and also absorbs blue light (λabs = 300 to 420 nm). Moreover, its irradiation in the presence of aryl iodide leads to the rapid reverse oxidative addition to form 3c-I together with biaryl p-CH3OC6H4-C6D5 (6c; Fig. 3C). Performing the reaction in the presence of a thiol as hydrogen-atom donor inhibits oxidative addition and leads instead to radical trapping product anisole (Fig. 3C). These data support the previously reported role of light-induced single-electron transfer from Pd(0) to aryl iodide opening a low-barrier, radical-induced oxidative addition pathway (step B in Fig. 3E; see fig. S7 for the full study of oxidative addition) (2832) and are similar to the radical carbonylations pioneered by Ryu (43, 44). Although aryl iodides can also undergo a slow thermal reaction with 7, similar light-induced reactivity is observed with aryl bromides and even alkyl iodides (Fig. 3, C and D), neither of which reacts thermally under these conditions, clearly demonstrating the role of visible light in oxidative addition.

Together, the data suggest a combination of two different roles of light acting in this system. The participation of Pd(0) and visible light in radical-induced oxidative addition has been described (32), but the capacity of light excitation to facilitate catalytic reductive elimination from Pd(II), and to do so in concert with Pd(0) photochemistry, is, to our knowledge, unknown. Evidence for the influence of light in catalysis can also be seen. As noted above, biaryl 6a is generated in addition to amide 1a (fig. S6A), suggesting the catalytic role of photoreduction of aryl iodide by Pd(0) in oxidative addition (Fig. 3A). Alternatively, performing the catalytic reaction in the presence of a TEMPO radical trap inhibits catalysis and leads to the formation of the acyl radical trapping product 4, which arises from the in situ excitation of the Pd(II) intermediate 3 (fig. S6B). Nuclear magnetic resonance (NMR) analysis of the catalytic reaction shows that complex 7 is the catalyst resting state (fig. S6C), which contrasts with thermal chemistry (18) but is consistent with the relative rates of the two individual photolytic steps (Fig. 3).

From a synthetic perspective, this visible light–based catalytic system offers a route to perform carbonylations in broad fashion, at ambient temperature, and with challenging electrophiles and nucleophiles. Catalysis proceeded with an array of aryl iodides, including those with electron withdrawing or donating substituents in the para, meta, and ortho positions (1a to 1i, Fig. 4). The reaction proved compatible with potentially coordinating groups such as aldehydes (1i), protected amines (1h), esters (1c and 1d), nitriles (1f), thioethers (1e), or even heterocyclic substrates (1j). Aryl bromides, which have to date required high temperatures for carbonylation to acid chlorides (100° to 120°C), could also undergo carbonylations to generate in situ acid chlorides at temperatures as low as −3°C (table S1). As with aryl iodides, a variety of functionalized aryl bromides could undergo carbonylation to generate in situ acid chlorides, including those with methoxy (1p), nitrile (1q), trifluoromethyl (1s), or ketone (1r) substituents, as well as heterocycles (1t).

Fig. 4 A broadly applicable, light-driven approach to carbonyl-containing products.

Performed with 1.5 equiv aryl iodide, 2 equiv aryl bromide, or 1 equiv alkyl halide, with 2.5% [Pd(allyl)Cl]2 and DPEphos or 5% 7 unless otherwise noted. *14-W light; †Aryl iodide as limiting reagent; ‡[Pd(allyl)Cl]2 and DPEphos; §1% 7; ¶Nucleophile added in a second step; #No NuH added; **10% 7; ††With Bu4NCl and Ph3BnPCl; ‡‡1mm is amide generated with 2,6-iPr2C6H3NH2 as in situ trap; §§0°C. Et, ethyl; d/r, diastereomeric ratio.

This light-driven system also enabled a distinctive carbonylative pathway to convert simple alkyl halides into valuable alkyl acid chloride products (2d to 2i). The latter are heavily exploited electrophilic building blocks in synthesis (45) and can be generated with both alkyl iodide and bromide reagents at catalyst loadings as low as 1 mol % palladium. The efficiency is presumably a result of the low reduction potential of alkyl halides (46). A diverse array of primary (1w and 1ff to 1hh), secondary (1u and 1dd), and even tertiary (1bb and 1ii) alkyl halides can be incorporated into this reaction, as well as those with various ester (1cc), ketone (1kk), nitrile (1gg), or alkyl chloride (1z) substituents. Alkyl chlorides, which are more challenging to reduce, are not reactive under these conditions. The reaction of cyclopropylmethyl iodide affords the ring-opened alkene product 1x, supporting the role of electron transfer in oxidative addition.

Because each of these transformations leads to the formation of acid chlorides, a broad array of nucleophiles can at the same time be incorporated into this chemistry, such as substituted anilines (1n to 1r), sterically hindered secondary amines (1k and 1l), tertiary alcohols (1dd), or even weakly nucleophilic N-heterocycles (1m). In reactions where the nucleophile is not compatible with the reaction conditions, it can be added after the catalytic buildup of acid chlorides. The latter has allowed the coupling of challenging substrate combinations (1y to 1aa, 1ii) and the generation of structurally more elaborate products (terpenes, 1kk; β-acetals, 1ff; and steroidal acid chlorides, 2i), none of which are viable via classical carbonylation chemistry.

The versatility of this dual light-driven catalyst suggests an opportunity to access various useful classes of products from carbon monoxide. For example, β amino acids are important reagents in the preparation of peptoids but are often generated by multistep protocols (47). By contrast, this visible light–based catalyst opens a direct route to β peptide derivatives via carbonylations (Fig. 5A). Alternatively, the electrophilic reactivity of acid chlorides can be applied to Friedel-Crafts acylations for synthesis of alkyl-substituted ketones from feedstock reagents such as arenes or heteroarenes, alkyl halides, and carbon monoxide (Fig. 5B) (48). This approach can also open routes to targeted synthesis, such as the cholesterol-lowering drug fenofibrate (Fig. 5C) (49). The power to convert both alkyl and aryl halides to acid chlorides is on display in this synthesis, where the initial functionalization of the 2-iodopropane to form an acid chloride for coupling with isopropanol, followed by an aryl halide carbonylative Friedel-Crafts reaction, affords 10 from 2 equiv carbon monoxide.

Fig. 5 Application of light-driven carbonylations to new classes of products.

(A to C) Applications include β amino acids (A), ketones via C–H bond functionalization (B), and targeted synthesis via sequential carbonylation (C). See supplementary materials for experimental details. DCE, 1,2-dichloroethane; rt, room temperature.

These results illustrate an alternative strategy to access versatile palladium catalysts for carbonylative coupling reactions, wherein visible-light excitation of palladium can drive oxidative addition and reductive elimination with low barriers. Considering the range of carbonylation and related catalytic transformations where this cycle plays a critical role, we anticipate that this approach will prove important as a pathway to create potent catalyst systems.

Supplementary Materials

Materials and Methods

Figs. S1 to S31

Table S1

References (5189)

References and Notes

  1. For an example of a postulated two photon role in nickel catalysis see (50).
Acknowledgments: We thank S. Bohle and R. Stein for helpful discussions on electron paramagnetic resonance analysis. Funding: We thank the Natural Sciences and Engineering Research Council of Canada, McGill University (James McGill Research Fund), and the Centre for Green Chemistry and Catalysis (supported by Fonds de recherche du Québec – Nature et Technologies) for funding this research. G.M.T. thanks the CONACyT (Mexican National Council of Science and Technology) for providing funding for doctoral studies. Author contributions: G.M.T. and Y.L. performed the research described in the paper and helped conceive the idea with B.A.A. B.A.A., G.M.T., and Y.L. prepared the manuscript. Competing interests: The authors declare no competing interests. Data and materials availability: All data are available in the main text or the supplementary materials.

Stay Connected to Science

Navigate This Article