A Decarbonylative Approach to Alkylnickel Intermediates and C(sp³)-C(sp³) Bond Formation

Abstract

The myriad nickel-catalyzed cross-coupling reactions rely upon formation of an organonickel intermediate, but limitations in forming monoalkylnickel species have limited options for C(sp³) cross-coupling. The formation of monoalkylnickel(II) species from abundant carboxylic acid esters would be valuable, but carboxylic acid derivatives are primarily decarboxylated to form alkyl radicals that lack the correct reactivity. Herein we disclose a facile oxidative addition and decarbonylation sequence that forms monoalkylnickel(II) intermediates via a non-radical process. The key ligand, bis(4-methylpyrazole)pyridine, accelerates decarbonylation, stabilizes the alkylnickel(II) intermediate, and destabilizes off-cycle nickel(0) carbonyl species. The utility of this new reactivity in C(sp³)-C(sp³) bond formation is demonstrated in a reaction that is challenging by purely radical methods: the selective cross-coupling of primary carboxylic acid esters with primary alkyl iodides.

One-Sentence Summary:

“Useful monoalkylnickel(II) intermediates can be accessed from alkyl carboxylic acid esters via a catalyst-controlled, non-radical decarbonylation and coupled with alkyl radicals to form valuable C(sp³)-C(sp³) bonds.”

Developing methods for selective C(sp³)-C(sp³) bond formation is an exciting frontier for modern organic synthesis (1) with a broadly recognized need (2, 3). The development of such reactions relies upon the discovery of catalytic mechanisms that allow the selective activation and coupling of two different alkyl starting materials. These reactions, in turn, are built from sequences of elementary steps, such as oxidative addition (OA), transmetalation, radical capture, and reductive elimination (4). A broad array of chemistry is enabled by relatively few steps and the development of new sequences can have profound implications for the field. Monoalkylnickel(II) species are general intermediates for C(sp³)-C(sp³) bond formation, but approaches to access them rely upon alkyl nucleophiles and electrophiles that are less abundant than alkyl carboxylic acids (Fig. 1A) (5–10).

Fig. 1. Breaking the 2e⁻ activation limitation of aliphatic acids by decarbonylation.

(A) 1e⁻ decarboxylation and 2e⁻ decarbonylation of aliphatic acids. (B) Cross coupling of carboxylic acid derivatives. (C) This work: Catalyst controlled decarbonylative 2e⁻ oxidative addition enables C(sp³)-C(sp³) cross-electrophile coupling of aliphatic acid esters. *Limited to cyclic anhydrides and difluoroalkyl substrates.

Aliphatic carboxylic acids are among the most abundant commercially available alkyl sources (11) for C(sp³) bond formation and valuable synthetic intermediates for complex molecule derivatization (Fig. 1A) (12). Recently, the application of carboxylic acids as alkyl donors has become widespread based upon the discovery of mild, general conditions for the generation of alkyl radicals. These approaches have relied upon 1e⁻ steps such as ligand-to-metal charge transfer (LMCT) processes with electrophilic metals (13), oxidation of carboxylates by chemical or photogenerated oxidants (14–16), and single electron transfer (SET) processes of the corresponding redox-active N-hydroxyphthalimide (NHP) esters (17). These mechanisms are believed to proceed through the generation of carboxyl radicals, which subsequently lose CO₂ and form free alkyl radicals via an outer-sphere process (18).

These radical processes have played a key role in enabling decarboxylative C(sp³)-C(sp³) cross-electrophile coupling and photoredox cross-coupling (16, 19). The focus to date has been on discovering fruitful combinations of alkyl radicals (20), generated photochemically (21–23) or electrochemically (24–26), that provide higher selectivity, for example when one of the radicals is slow to dimerize compared to the cross-coupling reaction. Nevertheless, it is frequently necessary to use one alkyl coupling partner in excess (typically 3 equivalents or more) as a sacrificial reagent to ensure that the limiting alkyl source is selectively trapped (20–26).

A way to form monoalkylnickel(II) intermediates from alkyl carboxylic acid derivatives by a non-radical method would enable a complementary approach to these fully free-radical approaches and set the stage for many new reactions. This goal could be realized by a 2e⁻ oxidative addition of a carboxylic acid ester followed by decarbonylation (27). While catalytic decarbonylation has been reported (28) and the decarbonylative cross-coupling of aryl carboxylic acid derivatives with organic nucleophiles and alkyl electrophiles (28–30) is known, decarbonylative cross-coupling with alkyl carboxylic acid derivatives has remained problematic (Fig. 1B) (30–34). One complication is that the decarbonylation of acylnickel complexes is thermodynamically unfavorable for most alkyl substrates (35–37), primarily due to unfavorable decarbonylation/CO dissociation and instability of the alkylnickel species (38). Therefore, most cross-coupling reactions using alkyl carboxylic acid ester derivatives result in ketone formation, even if evidence of decarbonylation is observed (39–44). Specialized substrates, such as cyclic anhydrides (45–47) and fluoroalkyl acid fluorides (31–33), can form nickelacycles by a chelation-promoted decarbonylation, but this reactivity is not general. A second challenge is that the resulting monoalkyl nickel(II) intermediates are susceptible to β-hydride elimination (48), isomerization (49), and/or disproportionation to form dialkylnickel species (50). Finally, CO formed under the reaction conditions will coordinate strongly to nickel(0), potentially inhibiting catalytic turnover, contributing to catalyst decomposition, and forming poisonous Ni(CO)₄ (30, 51, 52).

In this work, we report a non-radical pathway for oxidative addition/decarbonylation of carboxylic acid derivatives that form stable alkylnickel(II) complexes (Fig. 1C). The reactivity is enabled by using a tridentate bis(4-methylpyrazole)pyridine (^MeBpp) ligand that supports the disparate needs of individual steps in the catalytic reaction. These alkylnickel species are capable of reacting with alkyl radicals to selectively form C(sp³)-C(sp³) bonds instead of ketone products. The electronic differences between 2e⁻ OA precursors and 1e⁻ radical precursors avoid the necessity of using a large excess of one alkyl source in the C(sp³)-C(sp³) cross-electrophile coupling and present many opportunities for further reaction development.

Catalyst discovery.

We initially studied the decarbonylative cross-electrophile coupling between 2-pyridyl esters (1 equiv) and alkyl iodides (1.2 equiv) in the presence of different Ni catalysts (Fig. 2A, see Supplementary Material for additional details). As in all cross-electrophile coupling reactions, a terminal reductant, such as manganese metal powder (53), is needed (forming MnX₂ salts as a byproduct). This study revealed a ligand that enabled the desired decarbonylative coupling and shed light on the factors required for success. Standard bidentate ligands, such as bipyridine L1, result primarily in ketone product and low conversion of starting materials. This result is consistent with previous reports, where bipyridine nickel catalysts have been used for a variety of ketone-forming reactions (43). The low total conversion is consistent with strong CO binding to low-valent Ni intermediates (e.g. (bpy)Ni(CO)₂, Ni(CO)₄; warning, Ni(CO)₄ is highly poisonous and these studies were caried out on small scale with appropriate precautions!) (51, 52). In contrast, reactions with tpy (L2) nickel complexes consumed all starting materials, but only formed ketone and other dimeric products. Selectivity for cross-products could be obtained with the pybox ligand (L3). Replacing one of the pyridyl groups on tpy with an N-pyrazolyl group (pbp, L4) significantly improved the decarbonylative selectivity. When a second pyrazole was added (2,6-di(1-pyrazolyl)pyridine, bpp, L5), the yield of decarbonylated product more than doubled while maintaining high selectivity against ketone (Fig. 2A, see SM for details). We verified this was a decarbonylative reaction and that CO release was rapid at 70 °C by trapping the released CO_g in 83% yield in a connected vessel (see SM for details). This is notable because related aryl carboxylic acid decarbonylation reactions that rely upon bipyridine nickel catalysts generally require higher temperatures or reagents to trap CO (51, 54). Eventually, ^Mebpp (L6), which features minor tuning of the electronics on the pyrazoles, was identified as the optimal ligand, giving the decarbonylative cross coupling product 3 in 82% yield with only 3% of ketone 4 measured using gas chromatography (GC).

Fig. 2. Development of Decarbonylative C(sp³)-C(sp³) cross-electrophile coupling.

(A) Catalyst evaluation. Calibrated GC yields of 3 and 4 are reported. (B) Electronic and geometric effects in decarbonylation. Weaker π-accepting ligands with their higher π* energies result in increasing HOMO/LUMO energy gaps. Trends in π-acceptor capacities of tridentate ligands: L3>L4>L5. (C) Control experiment and structural analysis of CO binding to nickel complexes. Reaction as in (A) with L1 or L5 under standard conditions (balloon), in a sealed vial (with smaller headspace), and with added CO gas in a sealed vial. *THF:DMA (40:1). ^†ca. 30% recovery 1, 0% yield of 3, 21% yield of 4; ^‡trace recovery 1, 53% yield of 3, 9% yield of 4. THF, tetrahydrofuran; DMA, dimethylacetamide; dme, ethylene glycol dimethyl ether; Me, Methyl; tBu, tert-butyl.

Fig. 2 sheds light on how to design a catalyst that favors decarbonylation but at the same time disfavors CO binding, properties that would seem to be opposing and yet are crucial for general decarbonylative chemistry. A combination of electronic and geometric properties provides a solution. First, the rate of (L)Ni^II(acyl)X decarbonylation is accelerated by using a ligand (L5) that is a weaker π-accepting ligand than L3 and L4 (Fig. 2B). CO is a strong π-acceptor (via d→π* back donation). Thus, (L5)Ni^II would be expected to allow for stronger d→π* back donation from nickel to C≡O than (L3)Ni^II or (L4)Ni^II because L5 is a weaker π-acceptor than L3 or L4 (as measured by the calculated HOMO/LUMO energy gap); L5 competes less with CO for d electrons. This would stabilize the decarbonylated product and, by the Hammond postulate, could stabilize the decarbonylation transition state (depicted in Fig. 2B) (55). Stronger binding of CO could be deleterious to turnover, but the rigid, tridentate nature of L6 enhances CO release. This is because (L6)Ni^II(alkyl)(CO)⁺ is an 18e complex, but nickel(II) is more stable in a square planar, 16e configuration (see Fig. 3A for x-ray structure). Finally, CO binding to bipyridine nickel(0) complexes results in a stable, 18e, (L1)Ni⁰(CO)₂ complex that must dissociate a CO ligand to coordinate substrate, leading to CO poisoning (Fig. 2C, L1) (56). Faster associative substitution would decrease poisoning, but generally requires an open coordination site and/or less than 18e (57). However, Basolo noted that the presence of a redox-active ligand, nitrosyl, can also turn on associative substitution by accepting electrons and avoiding a 20e intermediate (58). Indeed, analysis of the electronic structure of (L5)Ni(CO) by DFT predicts it to a square-planar complex with an open coordination site and an (L5^•–)Ni^I(CO) ground state configuration (see SM for details). This is similar to the well-known redox-activity of ligand L2 in other nickel complexes (59–61). Consistent with these calculations, reactions with L5 were not poisoned by CO as strongly as reactions with L1 (Fig. 2C). These results suggest that redox active, tridentate ligands might offer a new approach to avoid CO poisoning that is complementary to reversible redox reactions, photolysis, or stoichiometric reagents (57).

Fig. 3. Mechanistic investigation of decarbonylation.

(A) Preparation and crystal structure of 7 (H atoms and BArF⁻ are omitted for clarity). (B) CO (de)insertion of Ni(II) complexes in the presence of ¹³CO (g). COD, 1,5-cyclooctadiene; TMSCl, trimethylsilyl chloride; NaBArF, sodium tetrakis[(3,5-trifluoromethyl)phenyl]borate; NPhth, phthalimidyl; NaOPh, sodium phenolate.

Mechanistic investigation.

Mechanistic studies were conducted to better understand the key decarbonylation step and the origin of cross-selectivity in this reaction. First, we studied the decarbonylation by mixing equimolar amounts of Ni⁰(COD)₂, L5, and phthalimidyl acetic acid 2-pyridyl ester (5) in THF at room temperature (Fig. 3A). As the pyridone ligand caused some difficulties when isolating the alkylnickel complexes (see SM for details), TMSCl was used to swap to chloride (by formation of TMS-OPy) followed by anion exchange from chloride to BArF⁻ (tetrakis[(3,5-trifluoromethyl)phenyl]borate). The resulting stable, cationic alkylnickel(II) complex 7 was fully characterized, including by single-crystal x-ray diffraction. The Ni coordination environment is distorted square-planar, comparable to those in related (L₃)Ni^II(Ar)⁺ complexes (see SM for details). Use of an acid chloride instead of a 2-pyridyl ester in the above reaction resulted in rapid decomposition (see SM for details). As we were able to isolate 7 via an intermediate chloride complex, this last result suggests that the pyridone ligand plays a role in the decarbonylation step (51).

We next tested the reactivity of 7 with CO to better understand the decarbonylation step. Although CO coordination/insertion is thermodynamically favorable for reported alkylnickel(II) complexes (38), we could not directly observe insertion species (8) by NMR (Fig. 3B). The added ¹³CO_g (1 equiv) remained as uncoordinated, free CO (δ 184.5 ppm) and no peaks of carbonyl species or ketone byproducts between 280~200 ppm were observed (see SM for details). Only with 75 equiv of ¹³CO_g did the concentration of 7 decrease, but in this case to unidentified products. When 7 was treated with ¹³CO_g (1 bar) with a nucleophile trap (NaOPh, 1 equiv), the ¹³C-labeled ester product 9 was formed in 47% yield. Together, consistent with our hypotheses (vide supra), these results show that the [(bpp)Ni^II(CH₂NPhth)]⁺ (7) is more thermodynamically stable than the insertion complexes (8) but that these complexes are kinetically accessible.

Given the strong precedent for radical decarboxylation and the rarity of decarbonylative cross-coupling of alkyl esters with nickel catalysis, we sought further evidence that decarbonylation was the primary pathway in catalytic reactions. A distinguishing feature of 1e⁻ decarboxylation pathways is the intermediacy of a free alkyl radical. This necessarily leads to the ablation of stereochemistry when α-chiral carboxylic acids are used as substrates. In contrast, decarbonylation, a 1,1-migration, is known to be concerted and stereospecific (62). To shed light on the key decarbonylation step, we examined the stoichiometric and catalytic decarbonylative deuteration of N-Cbz-4-(tert-butylsilyloxy)-proline esters (Fig. 4A). Both diastereomers of the hydroxyproline derivative (10 and 12) react to form deuterated products 11 and 13 with near complete stereospecificity (>99% and 98% conservation of stereochemistry, respectively). To ensure that these stoichiometric results are relevant to catalysis, we then conducted a standard decarbonylative cross-electrophile coupling of 10 with alkyl iodide 2 in the presence of d₁-2,4,6-trimethylbenzoic acid (14) as an “alkylnickel trap.” While stereospecific cross-coupling is currently low-yielding due to fast β-hydride elimination (see SM for additional examples), the alkyl intermediate in reactions of 10 can be deuterated by 14 to form 11 in 63% yield (70% D, matching acid enrichment) and complete retention of stereochemistry. These results are in contrast to those reported to occur by radical routes, which provided coupling products with ~1:1 dr (26, 63).

Fig. 4. Mechanistic study of decarbonylative C(sp³)-C(sp³) cross-electrophile coupling.

(A) Decarbonylative deuteriation of enantioenriched aliphatic acid 2-pyridyl esters. (B) Stoichiometric reaction of 7 and alkyl iodide 2. (C) Radical clock experiment. (D) Proposed catalytic cycle. Cbz, benzyloxycarbonyl; TBS, tert-butyldimethylsilyl; Py, 2-pyridyl;

We next investigated the reactivity and selectivity of alkylnickel(II) complex 7 toward alkyl iodide 2 (Fig. 4B). Alkylnickel(II) complexes have been invoked in C(sp³)-C(sp³) bond formation, but stoichiometric reactions are rare. Vicic’s group noted that reduced (tpy^•–)Ni^I(Me) reacted with alkyl iodides to form a C(sp³)-C(sp³) bond (6, 64), but Diao’s group recently reported that (L)Ni^II(CH₂TMS)Cl (L = phen, t-Bu-PyOx) did not react with a benzyl radical to form cross product (65). Stoichiometric reactions of 7 with 3-phenyl-1-iodopropane formed cross-product 15 in 54–61% yield (Fig. 4B), regardless of added reductant. Our reactions did not form much dimeric product 16, consistent with the literature data showing that tridentate alkylnickel(II) complexes are more resistant to alkyl ligand disproportionation/homocoupling than those with bidentate ligands (6, 50, 64, 65). Additionally, reaction with the iodocyclopropylmethane exclusively generated rearranged compound 17, consistent with a 1e⁻ radical mechanism (Fig. 4C). Further studies on the catalytic reaction with 1,1-diphenylethylene as a radical trap convincingly show that the alkyl iodide reacts via a free-radical intermediate (see SM).

Combining these data, we propose a catalytic cycle for the decarbonylative cross-electrophile coupling of alkyl 2-pyridyl esters with alkyl iodides (Fig. 4D). 2e⁻ Oxidative addition of the 2-pyridyl ester to (L6)Ni⁰ 20 (see SM for evidence supporting viability of Ni⁰) forms acylnickel(II) species 21. While we have evidence that 20 is accessible (see Fig. S22 in SM), we cannot rule out an alternative pathway where 19 reacts reversibly with the ester to form (L6)Ni^III(acyl)(OPy)(X) followed by reduction with Mn to 21 (66). Regardless of how it is formed, intermediate 21 can rapidly decarbonylate to complex 24, possibly via cationic species 22 and 23. Cationic alkylnickel(II) complex 24 (presumably in equilibrium with PyO⁻ coordinated species), can capture the alkyl radical (generated from reduction of alkyl iodide (67)) by oxidative ligation (i.e., radical capture) to form nickel(III) species 25. The dialkylnickel(III) complex can reductively eliminate to form the new C(sp³)-C(sp³) bond and nickel(I) species 19. We propose that the cross-selectivity arises analogously to C(sp²)-C(sp³) cross-electrophile coupling (68). The two alkyl electrophiles are distinguished by their different modes of activation, the 2-pyridyl esters acting as the equivalent of a non-radical alkyl electrophile and the alkyl iodides as a 1e⁻ alkyl radical source. Fast decarbonylation is induced by the 2-pyridone ligand and the weak-field bpp ligand. The intermediate alkylnickel(II) is stable enough to avoid dimerization and is capable of alkyl radical capture and reductive elimination.

Substrate scope.

To illustrate the different selectivity mode for our reaction, we elected to focus on coupling two structurally similar primary alkyls that would be challenging for methods that proceed through radical mixtures (Fig. 5 and SM for an analysis of these reactions). Similar to other cross-electrophile coupling approaches, functional-group compatibility is high and encompasses methyl ethers, silyl protected alcohols, and halogens (F, Cl and CF₃) (27–31). Sulfones, nitriles, naphthyl ethers, and esters were also compatible (32–35). Protected amines were also suitable substrates (36–37). Notably, a terminal alkene was tolerated despite the potential for coordination and isomerization (38, 48% yield, with 3% E-2-alkene, see SM for details). More complicated alkyl iodides are easily prepared by iodination from the corresponding alcohols, and the alkylation of galactose, thalidomide, and a Corey lactone derivative provided the expected products (39–41). In reactions that produced lower yields of product, the balance of the mass was primarily the homocoupling of the alkyl iodide, decarbonylative reduction of the ester, and ketones.

Fig. 5. Reaction scope.

Isolated yields are shown. Conditions: 0.2 mmol aliphatic acid 2-pyridyl esters, 0.24 mmol alkyl iodides, 1 mL solution, 70 °C, reaction time ranging from 6 to 12 h. See the supplementary materials for details. *NMR yield. ^†0.1 mmol aliphatic acid 2-pyridyl esters, 0.12 mmol alkyl iodides, 0.5 mL solution. ^‡using L7 instead of L6, THF:DMA (40:1, 0.2 M). ^§0.3 mmol aliphatic acid 2-pyridyl esters, 0.2 mmol alkyl iodides. ^¶Calibrated GC yield vs mesitylene, alkyl iodide was added via syringe pump over 2 h. OMe, methoxy; Boc, tert-butyloxycarbonyl; Bn, benzyl.

Aliphatic acids are one of the most abundant alkyl sources, and the corresponding 2-pyridyl esters are conveniently prepared and isolated (see SM for details). A variety of amino acids proved suitable substrates, including glycine (15, 42–45), β-alanine 46, homoproline 47, and γ-aminobutyric acid 2-pyridyl esters (48). Succinic acid mono esters are widely used in synthesis, and the corresponding 2-pyridyl esters showed decarbonylative reactivity (49 and 50). Although heteroaryl rings can be challenging due to the potential for coordination and catalyst inhibition, two complex heterocyclic substrates successfully underwent decarbonylative cross-electrophile coupling (51 and 52). The reaction for forming 52 favored oxidative addition of the pyridyl ester over the aryl chloride and the reaction also tolerated different reducible functionality (53).

At this time, one of the major side reactions we observe is β-hydrogen elimination from the alkylnickel(II) intermediate. We studied reactions with the 2-pyridyl ester of 3-phenylpropanoic acid because the corresponding (L)Ni^II(CH₂CH₂Ph)(OPy) intermediates should be prone to β-H elimination to generate styrene. After further ligand development (see SM for details), we found that reactions with ligand L7 provided higher yields of product by speeding up both 2e⁻ OA of pyridyl esters and 1e⁻ radical generation of alkyl iodides. This result suggests that further ligand development will be fruitful.

The yields in Fig. 5 are comparable to those reported for dual-radical approaches to 1°/1° couplings of alkyl carboxylic acid derivatives (see SM Table S11) while generally using a smaller excess of one coupling partner (1.2 equiv vs 2.5–4.0 equiv) (20, 22, 24, 26). Related couplings of activated carboxylic acids with an alkylzinc reagent (2 equiv of Alkyl₂Zn) give a slightly higher average yield (59 vs 50%) at the expense of using an excess of the less abundant alkylzinc partner (69). In cases where both coupling partners are valuable or where separation of the product from large amounts of dimeric products is challenging our new approach could offer a considerable advantage.

Outlook.

By establishing how to achieve rapid decarbonylation of acylnickel(II) complexes to alkylnickel(II) complexes, this work sets the intellectual foundation for myriad alkyl–C and alkyl–X bond forming reactions that will complement purely radical approaches by offering distinct modes for cross-selectivity and stereocontrol.

Data and materials availability:

X-ray structural data with access code CCDC 2328709 for 7 is available free of charge from Cambridge Crystallographic Data Centre (www.ccdc.cam.ac.uk/data_request/cif). All other data are available in the supplementary materials.