The increasing emission of greenhouse gases represents a global environmental threat, and strategies to address this issue have been the focus of intense research in recent times1,2. From the sustainability point of view, the development of chemical processes that extend beyond the traditional degradations and repurpose such gaseous by-products as useful synthons to produce valuable chemical feedstocks is highly desirable. Whereas the revalorization of CO2 or CH4 as carbon sources for organic synthesis through catalytic strategies has received a great deal of attention12,13, much less interest has been focused on the chemical transformation of another major contributor to the global warming: N2O. Governmental reports and recent scientific evidence both warn of the consequences that result from the increasing presence of this undervalued gas in the atmosphere3,4,5. N2O exhibits a global warming potential >300 times that of CO2, with a decomposition half-time in the atmosphere of >100 years6. Human activities have accelerated the emissions, with an estimated rate of increase for N2O of 2% per decade. Yet, a detailed analysis through the lens of sustainable synthesis presents a unique opportunity for N2O revalorization, as it represents an excellent O-atom source: it is readily available, non-toxic (laughing gas) and releases benign N2 as a by-product on O removal. Conversely, N2O is an inert gas, whose activation requires high temperatures (140–350 °C) and pressures (50–200 bar), resulting in limited applications as an oxidant for organic synthesis7,8,14. Yet, the structure of N2O has captivated chemists, who studied in detail coordination modes and the reactivity thereof in a plethora of metal complexes9. However, few reports focused on its activation towards the formation of C‒O bonds10,15 (arguably among the most valuable bonds in organic synthesis), as it would permit access to highly coveted alcohols, ethers, epoxides and so on. Still, these few examples rely on traditional metal–oxo reactivity, which requires high temperatures (100–200 °C) and pressures (10 bar)11 or long reaction times (1 ton per week)16 (Fig. 1a). In a groundbreaking report17, a fundamentally different outcome was observed: on exposure to a N2O atmosphere, the O atom could be inserted into a Hf‒Ph bond of complex 1 forging the desired Hf‒O‒Ph (2) with extrusion of N2. However, regioselectivity issues arose, as the O atom was competitively transferred to the hydride ligand, thus also producing a Hf‒O‒H complex (3). Mechanistic studies on the O insertion step into various M‒C(sp2) bonds using N2O, peroxides or oxygen suggest that an organometallic Baeyer–Villiger (OMBV)-type mechanism is operating, whereby the anionic carbon migrates to the coordinated O atom, forging the M‒O‒C bond18,19,20,21,22,23. On the basis of this reactivity, we aimed at unlocking the potential of N2O as an O-atom source in a fundamentally different catalytic synthesis of phenols. In the canonical transition-metal-catalysed phenol synthesis from aryl halides24,25,26,27,28, th‑e C(sp2)‒O bond-forming step proceeds through the well-established ligand exchange with a nucleophilic O source. On reductive elimination with the aryl group, the desired C(sp2)‒O bond is formed while the metal center is twofold-reduced (Fig. 1b, left). The source of O in these cases is usually H2O or a protic O-based nucleophile in combination with a base that lead to the corresponding phenol29. In the alternative catalytic cycle proposed herein, the fundamental step for C(sp2)‒O bond formation capitalizes on the OMBV-type mechanism: on N2O coordination to the metal centre, the electrophilic O is eventually inserted into the M‒C bond, with concomitant formation of N2 (Fig. 1b, right). In contrast to the traditional synthesis of phenols, after C‒O bond formation the oxidation state of the metal centre remains intact, thus requiring an external reductant to close the cycle. To orchestrate this reductive process, we focused our attention on Ni and its demonstrated ability to manoeuvre between different oxidation states through single-electron transfer30,31,32. Here we demonstrate that a mechanistically guided approach for the activation of N2O with organometallic complexes results in the development of a mild and selective catalytic synthesis of high-value phenols from aryl halides using N2O as an electrophilic O source33. The mild conditions (25 °C and 1.5–2 atm) allow the accommodation of a variety of functional groups, including base-sensitive moieties, thus providing an orthogonal strategy to the current technologies (Fig. 1c).
To investigate the feasibility of the M‒C(sp2) oxidation, we drew inspiration from previous work17, in which N2O was demonstrated to react with certain phosphine–Ni(II) complexes34,35. To this end, we synthesized the product of oxidative addition 4, and studied its reactivity with N2O (Fig. 2a). As expected, 4 rapidly decomposes mainly towards homocoupling (5) when dissolved in DMA under argon, with only traces of protodemetallation (6) detected (path a). This reactivity is exacerbated by the presence of reducing agents such as Zn (path b). Yet, when the argon atmosphere is replaced by N2O, the bright red colour of the solution of 4 remains, thus pointing towards a slower decomposition rate. After acidic workup, a 15% yield of phenol 7 is observed. However, when the same reaction is performed in the presence of a reducing agent, substantially higher yields of 7 were observed, with a 73% yield obtained when a combination of Zn and NaI was used (path d). These results point to the feasibility of developing a reductive catalytic protocol based on Ni catalysis using aryl halide precursors. From our extensive ligand survey (see Supplementary Information), it was evident that tridentate nitrogenated ligands with the general pattern of 2-substituted bipyridine were crucial to obtain catalytic activity, with terpyridine (L18) and 6-pyrazolyl-2,2′-bipyridine (L50) affording the highest yields of 9 (Fig. 2b and Supplementary Information). Analysis of the R group revealed three key features of the ligand for catalytic activity: replacing the N atom with C‒H or S prevents catalytic activity (L48 and L61); steric encumbrance next to the N of the pyrazole unit inhibits productive catalysis (L55 and L58); and electron-deficient substituents on the pyrazole markedly reduce the yield of phenol. To confirm Ni–ligand ligation in the reaction mixture, complex 10 was prepared and structurally characterized. A 75% yield of 9 was obtained using 10 as a catalyst, thus confirming that the pre-ligated complex is catalytically competent. L50 represents a new ligand platform in Ni catalysis, with virtually no examples. As Fig. 2a, b suggests that formal Ni(I)‒C(sp2) species might be involved, we prepared terpyridine–Ni derivatives such as 11 (Fig. 2c), as greater stability of the corresponding (terpy)Ni(I)‒Ar has been noted36,37,38. As for 4, reaction of 11 in the absence of reducing agent led to no phenol formation. Yet, despite the presence of two Me groups in ortho, reaction performed under N2O in the presence of Zn and NaI afforded the desired mesitol (12) in 49% yield on acidic workup. To further confirm the involvement of a formal Ni(I)‒C(sp2) in the tridentate system, the (tBu-terpy)Ni(I)‒I (13) was reacted with Ph2Zn under N2O. Despite the reported instability of (terpy)Ni(I)‒Ph (ref. 37), a 20% yield of phenol (14) was obtained. These findings suggest that reduction of Ni(II) species to formal Ni(I) and the presence of iodide salts in the system are of importance to forge the desired C(sp2)‒O bond. Mechanistic investigations on M‒Ar oxy insertions for late transition metals (Pd, Ni and Fe) reveal that subtle changes in the ligand environment also lead to differences between the metal‒oxo/oxyl or concerted Baeyer-Villiger pathways for Ar‒O bond formation18,19,20,21,22,23. In this case, we suggest that in the continuum between the two extreme possibilities offered in the OMBV reaction, the oxy insertion of N2O in a d9 complex lies towards the formation of the M‒O bond and N2, before Ar migration39.
With the optimized catalytic system in hand, a preliminary scope of the aryl halide counterpart was interrogated. As shown in Fig. 3, aryl iodides bearing other halogens in both para (9, 15 and 16) and meta (17–19) positions smoothly afforded the corresponding phenol in excellent yields. The presence of electron-withdrawing groups such as CF3 (7), ketone (20), ester (21 and 24) or nitrile (22–23) posed no difficulty for the C‒O bond formation. Electron-donating substituents such as alkyl (25), aryl (26), or even methoxy and thiomethyl (27 and 28) delivered phenol in good yields. Moreover, a fluorene derivative (29) featuring benzylic C‒H bonds was also amenable for phenol synthesis, albeit in 38% yield. A classical feature of reductive couplings is that steric hindrance at the ortho position can impede reactivity. Indeed, C‒O bond formation from indanone (31) and 1-chloro-2-iodobenzene (30) derivatives afforded slightly diminished yields. In contrast to 30, 32 was obtained in 79% yield, illustrating a possible beneficial chelating effect of the ortho OMe and the Ni centre. A silylated benzylic alcohol (33) or a diethylphosphonate (34) were also tolerated in this protocol. Heterocycles such as indole (35), quinoline (36), carbazole (37) or dibenzothiophene (38) also afforded the corresponding phenol in good yields. Substrates prone to rapid oxidation after C‒O bond formation could be further functionalized in situ, as exemplified by the 68% yield obtained for the pivaloyl derivative 39. An iodide derivative of the biologically active agent clofibrate could be converted to phenol 40 in 78% yield despite the presence of a tertiary α-oxy ester. This protocol does not require the use of nucleophilic alkoxy surrogates, and hence base-sensitive functionalities such as esters or sensitive amides, can be tolerated. An example of this chemoselectivity is observed in the derivatization of a substrate containing pinacol boronate. In this case, phenol 41 was still obtained in 56% yield, thus providing an orthogonal tool to classical oxidation. The observation of 7% yield of sulfoxide in the reaction of 28 (ref. 40), and the low yield obtained for fluorenol 29, suggest that the oxy-insertion step lies towards the oxo/oxyl–pathway in the continuum postulated for OMBV-type reactions. N2 was detected using a gas chromatography–thermal conductivity detector in the headspace after the reaction had finished for 7, 9, 18, 25 and 34 (Fig. 3). When the oxygen on the solvent was labelled ([18O]DMF, 25% 18O), no 18O was incorporated in 9. On the other hand, when N15N18O was used (45% 18O), 22% ± 1 of the O in 9 was labelled (Supplementary Information). Together, these data point to N2O as the source of O.
The same optimized reaction conditions for aryl iodides permitted C‒O bond formation of more accessible and commercially available aryl bromides. Yet, electron-withdrawing substituents were required to allow C(sp2)‒Br cleavage to occur. In this sense, phenols bearing CF3 (7), Ac (20) and CN (22) in the para position, as well as paraben (21), could be obtained in high yields (Fig. 4a). Medicinally relevant phthalide was also smoothly converted to the phenol (42), thus providing a method to synthesize this building block with three fewer steps compared with the reported method41. Phenols derived from π-extended or conjugated systems such as naphthoate 43 or cinnamate 44 were also obtained in 65% and 78% yields, respectively. In contrast to current light-mediated processes, no isomerization of the double bond in 44 was observed30. Finally, another base-sensitive group such as the aryl methyl sulfone could be tolerated and the corresponding phenol 45 was obtained in 82% yield. Heterocyclic bromides are not compatible with the current protocol.
Complex aryl halides functionalized with sensitive moieties were then tested. For example, an empagliflozin derivative, which contains a plethora of weak C‒H bonds prone to HAT, was smoothly converted to the corresponding phenol (46) in excellent yield (Fig. 4b). An ester derivative of the natural product eugenol afforded the desired phenol (47) in an 84% yield, highlighting the high chemoselectivity of this process over alternative oxidation through metal–oxo pathways. Despite the triphasic nature of the protocol, synthesis of 47 could be scaled up to 5 mmol with only a slight reduction in the yield (66%). Substrates containing saturated N-heterocycles such as piperazinone 48, azetidine 49, pyrrolidinone (aniracetam intermediate) 50 and nortropinone derivative 51 are well tolerated. The requirement of an electron-withdrawing group to activate the aryl bromide can be turned into a synthetic advantage, thus permitting regioselective control on the activated aryl bromide (52, 78%). Finally, a derivative ezetimibe, the drug used to treat high blood cholesterol, could be smoothly converted into the corresponding phenol (53), without altering the chiral and unprotected secondary alcohol, the ester and the strained β-lactam. Similar chemoselectivity can be observed in the conversion of paroxetine derivative 54. Finally, Fig. 4c illustrates a proof-of-concept of the potential for the revalorization of greenhouse gases for organic synthesis. It is now possible to combine N2O and CO2 revalorization strategies and obtain metaxolone (59), in which 66% of the oxygen atoms originate from waste gaseous feedstock42. A more striking example is illustrated in the synthesis of bazedoxifene (68), a drug candidate against breast and pancreatic cancer. The three phenolic building blocks could be rapidly obtained from the parent halides in good yields (64–66). Subsequent Fischer-indole synthesis allows access to indole 67, enabling the synthesis of bazedoxifene (68) with all O atoms originating from N2O (refs. 43,44,45,46). Whereas a 42% yield could be obtained for precursor 63 with 1 mol% catalyst loading, a <10% yield of 14 was observed with the same catalyst loading, with substantial protodehalogenation of the parent iodide 61, which highlights the subtle differences between aryl iodides and aryl bromides in this system.