Insertion of CO₂ Mediated by a (Xantphos)Ni^I–Alkyl Species

Abstract

The incorporation of CO₂ into organometallic and organic molecules represents a sustainable way to prepare carboxylates. The mechanism of reductive carboxylation of alkyl halides has been proposed to proceed through the reduction of Ni^II to Ni^I by either Zn or Mn, followed by CO₂ insertion into Ni^I-alkyl species. No experimental evidence has been previously established to support the two proposed steps. Demonstrated herein is that the direct reduction of (tBuXantphos)Ni^IIBr₂ by Zn affords Ni^I species. (tBu-Xantphos)-Ni^I-Me and (tBu-Xantphos)Ni^I-Et complexes undergo fast insertion of CO₂ at 22°C. The substantially faster rate, relative to that of Ni^II complexes, serves as the long-sought-after experimental support for the proposed mechanisms of Ni-catalyzed carboxylation reactions.

Carbon dioxide (CO₂) is a sustainable, inexpensive, and clean C1 source for chemical synthesis.^[1] These advantages underlie recent advances in developing Ni-catalyzed reactions to incorporate CO₂ into organic molecules.^[2] Dong and co-workers reported the Ni-catalyzed carboxylation of organozinc reagents, giving rise to a variety of carboxylic acids.^[3] Applying reductive conditions, the groups of Rovis,^[4] Martin,^[5] Tsuji,^[6] and others^[2] expanded the scope of carboxylation substrates to alkenes, aryl halides, and alkyl halides. The carboxylation of alkyl halides is proposed to initiate with the oxidative addition of Ni⁰ into an alkyl bromide, followed by the reduction of the Ni^II intermediate by either Mn or Zn to give a Ni^I-alkyl species (Scheme 1).^[5a,b] Carboxylation of the Ni^I-alkyl intermediate gives rise to the carboxylate product. Computational studies on the reductive carboxylation of chlorobenzene suggest a similar mechanism, where a Ni^I-phenyl intermediate is responsible for carboxylation.^[7] The reduction of Ni^II to Ni^I by either Mn or Zn and CO₂ insertion into Ni^I-alkyl intermediates have not been established before. In light of the rapid growth of Ni-catalyzed carboxylation to harness CO₂ as a chemical source, it is critical to obtain evidence for the proposed mechanism to inform future catalyst design.

Scheme 1.

Ni-catalyzed carboxylation of an alkyl bromide via a Ni^I-alkyl intermediate.^[4]

The insertion of CO₂ into organometallic molecules has been a long-time pursuit of fundamental studies.^[8] While CO₂ insertion into Ni^II-H complexes proved to be facile,^[9] only three examples of CO₂ insertion into Ni-carbyl complexes have been reported.^[10] PCP-pincer-ligand-stabilized Ni^II-Me molecules undergo slow CO₂ insertion at 150°C [Eq. (1)]. The

low reactivity contradicts the facile catalytic reactivity that has emerged in recent literature.^[2] Computational studies predict that the more electron-rich Ni^I-carbyl complex undergoes faster CO₂ insertion than Ni^II-carbyl complexes, as the carboxylation is nucleophilic in nature.^[7] This hypothesis has found support in the recent observation of CO₂ insertion into a Ni^I-aryl complex at room temperature.^[10c] The Ni carboxylate product, however, was not characterized. The scarcity of examples of CO₂ insertion into Ni^I-carbyl complexes is likely due to the difficulty of isolating well-defined Ni^I-carbyl and carboxylate molecules.^[11,12] The previous Ni^I-alkyl^[13] and Ni^Iaryl^[14] complexes have exceedingly bulky ligands to stabilize the open-shell molecules, and this is likely detrimental toward their reactivity with CO₂.^[14b,15] In light of the catalytic significance of Ni^I-carbyl complexes in mediating CO₂ insertion and the fundamental interest in this bond-formation process, herein, we report the chemical reduction of Ni^II complexes to Ni^I and the first example of insertion of CO₂ into Ni^I-alkyl complexes. This work serves as experimental support for the proposed mechanisms involving Ni^I species in Ni-catalyzed carboxylation reactions.

Bulky bidentate ligands with large bite angles can stabilize Ni^I complexes.^[14b,15] We have recently demonstrated this concept in preparing (tBu-Xantphos)Ni^IAr complexes.^[16] In an effort to prepare Ni^I-alkyl complexes, we treated (tBuXantphos)Ni^IIBr₂ (1) with Zn in THF to generate a canary yellow solution (Scheme 2). X-ray crystallography established the product to be a 1:1 mixture of the ZnBr₂-bound (tBuXantphos)Ni^IBr (2) and (tBu-Xantphos)Ni^IBr (3; see Figure S2 in the Supporting Information). The ¹H NMR spectrum shows that no 3 is present in solution, indicating that the major species is 2. This single-electron reduction of Ni^II zinc supports a proposed intermediate step in catalytic carboxylation reactions (Scheme 1).^[5a,b] Clean 3, without ZnBr2, can be obtained in high yields by the reduction of 1 with Cp₂Co (Scheme 2). The EPR spectrum of 3 shows coupling to the bromide atom and both phosphines from the (tBu-Xantphos) ligand (see Figure S24).

Scheme 2.

Syntheses of (tBu-Xantphos)Ni-alkyl complexes and their reactivities toward CO₂. THF=tetrahydrofuran.

Alkylation of 3 with MeLi and EtLi afforded (tBu-Xantphos)NiMe (4) and (tBu-Xantphos)NiEt (5), respectively, as dark brown solids (Scheme 2). The low yield of 5 is attributed to fast decomposition, likely by b-hydride elimination, followed by ejection of a hydrogen radical to form (tBu-Xantphos)Ni(N₂).^[16] Attempts to use a Ph-Xantphos ligand resulted in rapid decomposition. The crystal structure of 4 shows a slightly distorted trigonal-planar geometry, with the O-atom of tBu-Xantphos weakly interacting with Ni (Ni(1)–O(1)=2.6018(11) Å; Figure 1A). The Ni-C(1) bond length of 2.055(7) Å is longer than that of (terpy)NiMe (1.95(13) Å),^[13a] and may be related to the redox activity of the terpy ligand. The structure of 5 is similar to that of 4, with a long β-H—Ni distance (3.18 c), suggesting the lack of a bagostic interaction (Figure 1B).^[17] Addition of lithium acetylide reagents to 3 generated the Ni^I-acetylides 6 and 7. The crystal structure of 6 shows a Ni(1)-C(1)-C(2) angle of 174.53(14)°, consistent with an sp-hybridized C(1) (Figure 1C). The decreasing Ni(1)-C(1) bond lengths of 4, 5, 11,^[16] and 6 reveal increasing bond strengths on the order of sp³ <sp² <sp (Table 1). The spin-density plots of 4–6 obtained by DFT calculations reveal that the unpaired electron is primarily localized on the Ni center, with a small portion distributed to C(1) (Figure 1).^[18] The spin-density is higher on C(1) for 4 and 5 than for 6 and 11, suggesting that the carbyl fragments on the former compounds have a more significant radical character. The EPR spectra of both 4 and 11 show strong coupling to the two phosphorus atoms of the (tBu- Xantphos) ligand and minimal coupling to the protons on the carbyl fragments (see Figures S25 and S26).

Figure 1.

X-ray structures^[19] and spin-density plots of 4 (A), 5 (B), and 6 (C) at 50% probability thermal ellipsoids. Hydrogen atoms are omitted and tBu groups are truncated for clarity.

Table 1:

Bond parameters for 4, 5, 6, and 11.

Bond length [Å] and angle [°]	4	5	11	6
C(1) hybridization	sp3	sp3	sp2	sp
Ni(1)-C(1)	2.055(7)	2.034(2)	1.9795(14)	1.9359(16)
Ni(1)···O(1)	2.6018(11)	2.6737(16)	2.5184(10)	2.5405(11)
Ni(1)-C(1)-C(2)		118.07(18)		174.53(14)

Introducing 1 atm of CO₂ to 4 immediately generated the Ni^I-carboxylate 9 as an orange compound [Eq. (2)]. The structure of 9 exhibits a trigonal-planar geometry with a secondary interaction between Ni and the Xantphos O(1) with a distance of 2.471 Å (Figure 2). The Ni-O(2) distance of 2.0145(13) Å is noticeably longer than that of the previously reported (PNP)Ni^II-formate (1.927(5) Å).^[10a] The longer NiO(2) distance reflects a more electron-rich Ni center. The long distance of 2.9003(17) Å between Ni and O(3) reveals

that the acetate is coordinated in an ɳ¹ fashion. The EPR spectrum of 9 shows that the unpaired electron is coupled to phosphorus atoms of the (tBu-Xantphos) ligand and possibly to the protons on the acetate group (see Figure S26). The reaction of 5 with CO₂ afforded 10 in 61% yield, as characterized by its analogous ¹H NMR spectrum to that of 9. It is noteworthy that the carboxylation reactions are quite rapid, finishing in seconds at room temperature. This fast rate contrasts with those observed for (PCP)Ni^II-Me,^[10] and is corroborated by previous DFT calculations.^[7]

Figure 2.

X-ray structure^[19] and spin-density plot of 9 at 50% probability thermal ellipsoids. Hydrogen atoms are omitted and tBu groups are truncated for clarity. Selected bond lengths [Å]: Ni(1)-O-(1)=2.0145(13), Ni(1)···O(2)=2.471, Ni(1)···O(3)=2.900.

The other (tBu-Xantphos)Ni^I molecules 6, 7, 8, and 11, however, did not react with CO₂ over 24 hours under either 1 atm or 4 atm of CO₂ (Scheme 3). Addition of Lewis acids, including LiCl, LiBF₄, BPh₃, and ZnCl₂ did not promote CO₂ insertion.^[10c] When the less bulky (Ph-Xantphos)Ni^IPh is formed in situ from the reaction of (Ph-Xantphos)Ni^IIBrPh and Zn under one atmosphere of CO₂, benzoic acid is obtained in 48% yield (see page S8 in the Supporting Information). This result suggests that the greater Ni–Csp² bond strength, relative to the Ni–Csp³ bond, is not the factor that hinders carboxylation of 11. Instead, we attribute the contrasting reactivity of Ni-alkyls with Ni-aryls/alkynyls to their different mechanisms of insertion (Scheme 4). Ni^I-alkyl complexes could undergo insertion by nucleophilic attack of the alkyl groups to CO₂, as proposed in previous computational studies (Scheme 4A).^[10a] In contrast, sp and sp² nucleophiles cannot access such a pathway, but can only undergo migratory insertion (Scheme 4B). The approach of CO₂ to these (tBu-Xantphos)Ni^I complexes is inhibited by the bulky substituents on the ligand.^[8c] The lack of reactivity of 8 with CO₂ can be attributed to the lower nucleophilicity of phenoxides compared with carbanions.

Scheme 3.

Ni^I-carbyl and phenoxy complexes that are unreactive toward CO₂.

Scheme 4.

Possible pathways for Ni^I-mediated CO₂ insertion.

Finally, to verify the catalytic relevance of (Xantphos)Ni complexes, we tested the catalytic carboxylation of BnZnCl, PhZnCl, and n-BuZnCl, as well as the reductive carboxylation of benzyl bromide.^[3] Either with no ligand or with tBu-Xantphos, no carboxylation reaction took place for zinc reagents, as the tBu-Xantphos ligand stabilizes Ni^I species and does not allow catalytic turnover. The use of Ph-Xantphos led to carboxylation of all three organozinc reagents under mild reaction conditions to afford the corresponding benzylacetic, benzoic, and valeric acids (Scheme 5A). (Ph-Xantphos)Ni^IIBr₂ could be reduced in situ to Ni^I and, in fact, (PhXantphos)Ni^IBr renders a better yield for the carboxylation of n-BuZnCl. The catalytic reactivity of PhZnCl contrasts with the lack of stoichiometric reactivity of 11, but is consistent with the reactivity of in situ formed (Ph-Xantphos)Ni^IPh (see page S8), highlighting the access to the migratory insertion pathway that comes with a smaller steric profile (Scheme 4B). (Ph-Xantphos)Ni^IIBr₂ also catalyzes reductive carboxylation of BnBr, albeit in low yield, but the bulkier (tBu-Xantphos)Ni^IBr does not effectively turn over the reaction (Scheme 5B).

Scheme 5.

(Xantphos)Ni-catalyzed carboxylation of organozinc reagents (A) and reductive carboxylation of BnBr (B). DMF=N,N’-dimethylformamide.

In summary, the direct reduction of (tBu-Xantphos)Ni^IIBr₂ by Zn generates (tBu-Xantphos)Ni^IBr. (tBu-Xant-phos)Ni^I-Me and (tBu-Xantphos)Ni^I-Et complexes undergo fast insertion of CO₂ at 228C, whereas (tBu-Xantphos)Ni^Iphenyl, acetylide, and phenoxide complexes gave no insertion products. This observation represents the first characterization of CO₂ insertion into Ni^I–alkyl bonds and verifies the nucleophilic addition mechanism proposed by previous DFT calculations. The fast reaction rate corroborates the facile catalytic reactions and the predicted importance of Ni^I intermediates relative to Ni^II in carboxylation. Characterization of these stoichiometric reactions provides evidence to support mechanistic proposals for Ni-catalyzed carboxylation reactions of organic and organometallic reagents.