Photocatalytic aerobic oxidative carbonylation of methane by iron terpyridine catalysts

Employing methane as a C1 feedstock for commodity chemicals represents a sustainable and empowering opportunity for chemical catalysis. However, practical demands of industrial development have imposed serve challenges to utilize coeffective catalysts and reagents as well as ecologically benign conditions for highly efficient and selective C(sp)–H bond functionalizations, constantly driving the development of catalytic strategies. The aerobic oxidative carbonylation of methane represents an ideal approach for sustainable acetic acid synthesis, yet poses a longstanding challenge in selectivity for C2:C1 products. In recent years, metallaphotocatalysis in which photoredox catalysts operate in synergy with transition metal catalysts to facilitate both 1and 2-electron reaction steps, has provided intriguing opportunities to achieve selective cross-couplings otherwise unattainable. Here, we demonstrate the utilization of an iron terpyridine catalyst to facilitate multiple 1and 2-electron processes, including photoexcitation, radical generation, oxidative radical capture and carbonylation, in the form of methane aerobic carbonylation to acetic acid. High levels of catalytic efficiency (turnover numbers up to 31480) and selectivity (C2/C1 ratio up to 142:1) have been achieved with a robust iron terpyridine photocatalyst at ambient temperature under LED irradiation. Importantly, this practical catalytic manifold has been adapted to the selective oxidative carbonylation of various feedstock alkanes.

The catalytic carbonylation of methanol by rhodium and iridium complex, a significant milestone in transition metal catalysis, has dominated the industrial production of acetic acid and has been hugely inspiring for the development of novel carbonylation technologies [18][19][20] . Instead employing methane directly as a C1 feedstock in the catalytic aerobic carbonylation, exclusively converting inexpensive feedstock materials (i.e. CH 4 , CO, and O 2 ), has drawn considerable research attention [3][4][5][6][7][8] . Key challenges stem from the chemical inertness of CH 4 and high propensity for oxidation of activated C1 components 21,22 ; thus, controlling for aerobic oxidative carbonylation with poorly reactive CO over methane oxidation presents a longstanding challenge in C2/C1 selectivity. Elegant examples using RhCl 3 or Rh anchored on zeolite materials have demonstrated moderate to good efficiencies under high temperature conditions, however, the C2/C1 selectivities remain unsatisfactory 3,6-8 .
Recently, photoredox catalysis has been demonstrated to be a mild and effective platform for the functionalization of methane through catalytic hydrogen atom transfer (HAT) enabled by cerium and decatungstate photocatalysts 23,24 . Such an approach represents a new paradigm in methane C-H functionalization, operating through highly reactive open-shell methyl radical. However, under aerobic conditions required for practical oxidative carbonylation of methane, the nucleophilic methyl radical would be preferentially trapped by molecular oxygen (k = 4.7 ´10 9 M -1 s -1 ) 25 , as opposed to the sluggish and reversible coupling between methyl radical and CO (k = 2 ´10 6 M -1 s -1 ) 26,27 .
Nature provides inspiration for such a concept, as methane monooxygenase utilizes a radical rebound mechanism to capture incipient methyl radical, suppressing uncontrolled aerobic methane oxidation 28,29 . However, synthetic chemists have heretofore been unable to similarly control the reactivity of methyl radical, and as such radical-mediated carbonylation of methane has operated exclusively under deaerated atmosphere, necessitating the use of strong oxidizing agents [30][31][32][33] , making acetic acid under conditions impractical for large scale applications. Metal-centered oxidative radical capture, particularly synergistically employed with photoexcitation, has recently emerged as a powerful strategy to overcome reactivity and selectivity challenges in the realm of radical-mediated as well as transition metal catalyzedcoupling processes [9][10][11]14,17,[34][35][36][37] . We recently questioned whether a single metal center could facilitate both ligand-to-metal charge Fig. 1. Photocatalytic aerobic oxidative carbonylation of methane by iron catalyst. A radical-metal combination merged with LMCT-homolysis was proposed to elude the nonselective methane oxidation for highly selective carbonylation. transfer (LMCT) excitation for the generation of alkoxy radicals and oxidative radical capture by the reduced metal center for the generation of metal-methyl complex. Previously, the low-valent metal generated via LMCT-homolysis has only been invoked for single electron transfer (SET) events merely to achieve catalyst turnover 23 . In this hypothetical reaction manifold, the methyl radical generated through alkoxy radical-mediated HAT would be trapped by the low-valent metal complex in close proximity, in a mechanistically analogous manner to the radical rebound system, to allow diverse utilizations of methyl radicals for practical transformations, specifically methane aerobic carbonylation.
The challenges to fulfill this design lie in the identification of a suitable transition metal species which can undergo LMCThomolysis in its high oxidation form and in turn combine with nucleophilic alkyl radical in its low valent state. Seminal reports on iron(III) porphyrins drew our attention to iron catalysts 38,39 . It has been established that the in situ coordinated iron(III) porphyrin alkoxide complex could undergo photoinduced-ligand homolysis to generate iron(II) and alkoxy radicals under irradiation 40,41 , although suppression of reported alcohol oxidation pathways would be required to achieve efficient HAT. Two additional critical precedents establish a rationale for investigating iron complexes. First, iron(II) porphyrin is reported to undergo rapid combination with methyl radical (rate constant, 3.9 ´10 9 M -1 s -1 ) 42 . Second, iron(III) alkyl complexes are well-established to undergo facile CO insertion to generate iron acyl species which could degrade into carboxylates 43,44 . Thus, iron complexes appeared to be an ideal platform for each requisite elementary step for LMCT-mediated aerobic carbonylation of methane.
Considering the enabling capacity of hexafluoroisopropoxy radical in HAT and the deactivation effect of two trifluoromethyl groups 23,45 , hexafluoroisopropanol (HFIP) was chosen as the solvent. Gratifyingly, under the irradiation of 400 nm LED (irradiation intensity: 0.37 W cm -2 ), commercially available iron(III) tetraarylporphyrin 1 indeed led to a small amount of acetic acid at ambient temperature, although with the opposite sense of C2:C1 selectivity desired. Peculiarly, trace amounts of hexafluoroisopropanyl acetate and methyl acetate were also observed. Notably, under the same partial pressure of three gas components, the use of cerium catalyst only resulted in methane oxidation with no C2 products observed. Decatungstate photocatalyst, previously demonstrated capable of generating methyl radical with 365 nm LED excitation 24 , was inactive under the current condition.
Intensive efforts of ligand modification eventually led us to identify iron(III) terpyridine complexes, which have not been exploited in LMCT catalysis, as more active and selective photocatalysts for this transformation. Iron(III) terpyridine boosts the production of acetic acid into 757 µmol, generates a total of 803 µmol C2 products with only 74 µmol C1 products formed, demonstrating a high C2/C1 selectivity of 11:1. Importantly, peroxide species, including peracetic acid and methyl peroxide, could not be detected via 1 H NMR or chemical reduction methods. The existence of other possible by products, such as ether from the coupling of methyl radical with hexafluoroisopropoxy radical, or chlorinated methanes, were thoroughly checked and precluded by GCMS analysis. No ethane could be detected either at an early stage or after 8 hours of irradiation. However, we did observe the formation of CO 2 (449 µmol), which was confirmed iii to be derived from aerobic oxidation of CO (vide infra). Considering that photodecarboxylation of acetic acid would be detrimental to the catalytic efficiency, control experiment swapping methane with acetic acid revealed a full recovery of acetic acid after irradiation (see Scheme S1), indicating that the photodecarboxylation is inhibited under the current non-basic setting, probably by virtue of using coordinating solvent.
Remarkably, tuning the electronic property of the terpyridine ligands via the installations of different substitutes has led to intriguing findings towards catalytic efficiency and selectivity. Improved catalytic efficiency was achieved by methoxy substituted iron complex 3, which promoted the production of C2 products and suppressed methane oxidation, leading to an improved 26:1 selectivity ratio with TONs up to 10680 (turnover frequency, 0.4 s -1 ). Iron catalysts equipped with electron withdrawing groups such as chloride (4), trifluoromethyl (5), and carboxylic ester (6), nonetheless resulted in somewhat reduced efficiencies for acetic acid and declined selectivity. Bis(terpyridine) iron(II) 7, a coordinatively saturated and stable iron(II) catalyst, was found to be much less effective in comparison to 2. The highest production of acetic acid was obtained with 0.05 µmol of 3 under anhydrous conditions, leading to a notable increase in TONs (up to 31480) and almost exclusive formation of C2 products (142:1 C2/C1 selectivity). Without the addition of water, the ratio of HFIP ester to acetic acid slightly increased (1: 3.8). Notably, this highly efficient iron catalyst further allows us to perform the aerobic oxidative carbonylation using 1 bar of methane to acquire 776 µmol of C2 products under the mild conditions. Importantly, a high level of selectivity was maintained and an 8% estimated conversion yield based on methane was obtained 6 .
Control experiments revealed an exclusive methane oxidation in the absence of CO, and a photocatalytic CO aerobic oxidation in the absence of methane (see Fig. S1). No reaction took place in the absence of iron catalyst, air or LED light. Furthermore, control experiment utilizing methanol or formic acid to replace methane as the C1 source in this catalytic system only revealed formic acid after irradiation, without any sign of acetic acid, precluding methanol and formic acid as viable intermediates for the carbonylation pathway (see Schemes S2 and S3). Considering that methyl halides have been identified as the active intermediates in transition metal catalyzed carbonylation processes, we then carried out the control experiment with methyl chloride and excluded it as a possible intermediate in this photocatalytic process (see Scheme S4).
To further elucidate the source of each carbon and oxygen atoms of the acetic acid product and gain insights of the mechanism, isotope labelling experiments were then conducted. As demonstrated in Scheme 1A and 1B, when 13 CH 4 was employed, isotope enriched methyl of acetic acid was observed, while isotope enriched carbonyl of acetic acid was observed when 13 CO was used. 13 CO 2 was overwhelmingly detected in the 13 CO experiment but negligible in the 13 CH 4 experiment, which further revealed that CO 2 is mainly generated from CO aerobic oxidation. 18 O isotope tracing experiments revealed the partial incorporation of 18 O into acetic acid either when 18 O 2 was employed as the oxidant (Scheme 1C), or excess H 2 18 O was introduced (Scheme 1D). In both experiments, the labelling ratios detected in acetic acid and water are in close range. Taken together, these results are indicative of the provenance of the components of acetic acid as follows: the methyl group derives from methane, the carbonyl carbon derives from CO, and the additional oxygen derives from water, either explicitly added to the reaction or formed in situ by O 2 reduction. Importantly, stoichmetric iron terpyridine complex could act as a viable oxidant to generate acetic acid under anaerobic condition, without the formation of methanol (see Scheme S9). Under this stoichmetric condition, without the interference of water generated from photocatalytic reduction of oxygen, the introduction of 18 O water has led to the overwhelmingly labelled acetic acid (98%) (Scheme 1E). These 18 O labelling experiments, together with the existence of HFIP ester, all lend support to a nucleophilic attack at the acyl carbonyl (either of acyl-iron complexes or acylium cation) by water or hexafluoroisopropanol as the predominant termination pathway for the formation of C2 products.
The in situ formed iron(III) alkoxide complex in HFIP could be isolated and characterized as complex 8 with three alkoxide ligands, which proved equivalently competent in the catalytic aerobic carbonylation of methane. When 8 was irradiated in the presence of neohexene (9) and DIAD, product 10 was observed; this product can be rationalized as deriving from HFIP-derived alkoxy radical addition across the olefin and subsequent trapping of the carbon-centered radical by DIAD.

Scheme 1. Isotope labelling experiments.
iv v Steady-state photolysis of complex 8, which exhibits a broad absorption band centered at 355 nm trailing to 500 nm, was carried out. Unexpectedly, the photolysis of the iron(III) complex via constant irradiation with a 375 nm laser (light intensity, 6.3´10 -8 Einstein s -1 ) rendered no appreciable spectral changes, indicating no isolated photoactivity of LMCT-homolysis (Fig.  4B). However, simply upon addition of neohexene as a radical trapping species, a fast LMCT-homolysis event was able to be detected. Upon irradiation, photoinduced decay of iron(III) can be assigned by the gradually declining of the broad absorption band centered at 355 nm (Fig. 4C), while the generation of reduced iron(II) species, deriving from Fe(III)-OR bond homolysis, could be detected. In particular, this metal-centered reduction can be more robustly measured when the irradiated solution was treated with an excess amount of terpyridine; then, an absorption band centered at 542 nm, the characteristic absorption band of the bis(terpyridine) iron(II) 7 could be observed (Fig. 4D). If instead, oxygen was introduced into the irradiated solution, the initial iron(III) complex was observed, suggesting oxygen could serve as the oxidant to achieve catalyst turnover. These data present a highly inefficient isolated LMCT-homolysis upon excitation of 8, which nevertheless is rendered efficient through the inclusion of an O-radical trap, thus delivering isolated Fe(II) species.
With an understanding of the external factors controlling LMCT-homolysis efficiency, we next turned our attention to the metal complex characteristics which influence this key step. Initial rates of photoinduced LMCT-homolysis of iron(III) terpyridine alkoxides in the presence of neohexene were estimated based on the production of Fe(II) (Fig. 4E). Complex derived from 3, which exhibits a slightly more intensive absorption in visible-light region among these catalysts (see Fig. S7), demonstrates a relatively slow bond-homolysis rate compared to other iron(III)-terpyridine complexes equipped with electron-deficient substitutes. As the rates all showed first-order dependence on olefin concentration (see Fig. S8), it can be concluded that the relative rates of Fe(II) generation of each complex show an inverse dependence on the rate of radical recombination of the fleeting Fe(II) and alkoxy radical species. Thus, it can be concluded that the most electron-rich, reducing Fe(II) species, that derived from 3, has the fastest rate of radical recombination with hexafluoroisopropoxy radical upon homolysis.
While the observation that the most efficient, selective catalyst for methane aerobic carbonylation has the least efficient bond homolysis step may seem counterintuitive, we posited that this effect is actually crucial to rationalizing the excellent C2:C1 selectivity observed. First, if the rate of Fe(II)-alkoxy radical recombination is very high, a negligible concentration of alkoxy radical will exist in solution, inherently requiring HAT to occur in proximity to the Fe(II) center, setting the stage for radical methyl radical capture at iron. Reinforcing this, the more reducing Fe(II) species derived from 3 would be optimal for oxidative methyl radical capture. Accordingly, taking the kinetics previously measured with structurally similar iron(II) complex into account, the formation of iron-methyl complex would prevail over the coupling of methyl radical with triplet oxygen, thus facilitating acetic acid formation via iron-mediated carbonylation.
A catalytic model was tentatively proposed based on these investigations and insights (Fig. 4F). Under irradiation, the constant photoexcitation of iron(III) alkoxide renders alkoxy radical homolysis and radical recombination on a loop until being interrupted by methane in close proximity. Methyl radical is gener-ated via HAT with hexafluoroisopropoxy radical, and subsequently trapped by the resultant nearby iron(II) complex to produce an iron(III) methyl complex, which would promote CO insertion to generate an acyl iron(III) complex. The degradation of the resultant intermediate, possibly through nucleophilic attack by water or HFIP and an oxidation of the iron center, would generate acetic acid and complete the catalytic cycle.
With a thorough understanding of the reactivity of this catalytic system, we further applied it towards the oxidative carbonylation of various alkane feedstocks (see Scheme 2). Light alkanes including ethane, propane, and butane could be directly transformed into the corresponding one-carbon homologated carboxylic acids with high selectivity for C-H carbonylation vs. C-H oxidation. Importantly, good levels of selectivities of methylene vs. methyl groups have been observed in propane, butane and pentane. Cyclic alkane feedstock, including cyclohexane and norbornane, could also be employed in this iron catalytic system to produce valuable carboxylic acid building blocks. This demonstrates that the Fe photocatalytic system presented herein could find wide applicability in a variety of alkane conversions.
In summary, we have developed a practical photocatalytic protocol for the aerobic oxidative carbonylation of methane. Under LED irradiation at ambient temperature, acetic acid can be produced at high levels of efficiency and selectivity by abundant and inexpensive iron photocatalysts. We anticipate that further operando spectroscopic studies will provide more mechanistic details and guide the development of more efficient catalysts. We envision that the exploitation of the photoinduced iron-LMCT system and radical-metal combination strategy will provide new avenues towards the development of diverse alkane conversions under practical conditions.