Selective functionalization of methane, ethane, and higher alkanes by cerium photocatalysis

See allHide authors and affiliations

Science  17 Aug 2018:
Vol. 361, Issue 6403, pp. 668-672
DOI: 10.1126/science.aat9750

Radically transforming light hydrocarbons

The methane, ethane, and propane in natural gas are mostly inert under ambient conditions. Mainly they are burned to produce heat. Hu et al. show that a simple cerium salt paired with an alcohol can catalytically transform these and other simple hydrocarbons into reactive radicals at room temperature (see the Perspective by Kanai). The reactions rely on light to photolytically cleave cerium alkoxide bonds, producing alkoxy radicals that strip H atoms from the hydrocarbons and regenerate the alcohol. The resultant alkyl radicals readily add to azo compounds, olefins, and aromatics.

Science, this issue p. 668; see also p. 647


With the recent soaring production of natural gas, the use of methane and other light hydrocarbon feedstocks as starting materials in synthetic transformations is becoming increasingly economically attractive, although it remains chemically challenging. We report the development of photocatalytic C–H amination, alkylation, and arylation of methane, ethane, and higher alkanes under visible light irradiation at ambient temperature. High catalytic efficiency (turnover numbers up to 2900 for methane and 9700 for ethane) and selectivity were achieved using abundant, inexpensive cerium salts as photocatalysts. Ligand-to-metal charge transfer excitation generated alkoxy radicals from simple alcohols that in turn acted as hydrogen atom transfer catalysts. The mixed-phase gas/liquid reaction was adapted to continuous flow, enabling the efficient use of gaseous feedstocks in scalable photocatalytic transformations.

Methane and other gaseous alkanes (C2 to C4) have traditionally been viewed more as fuels than as economical chemical feedstocks. However, the recent discovery of huge volumes of unconventional reservoirs and the ensuing soaring production of natural gas have made these gaseous hydrocarbons economically attractive and strategically important basic raw materials (1, 2). With the added economic and ecological benefits of reducing transportation costs and emissions, the direct transformation of gaseous hydrocarbons into value-added liquid commodity chemicals via the use of innovative homogeneous catalysis processes has received considerable attention in recent years (1, 38).

The intrinsic inertness of C–H bonds in methane and other gaseous alkanes has, however, brought extreme challenges for catalytic systems, not only in the activation step, but also in controlling chemoselectivity to avoid solvent functionalization and overfunctionalization under the frequently harsh conditions necessary (high temperature, superacid media, strong oxidants) (9). Moreover, the gaseous substrates’ low solubility in most solvents has raised substantial practical difficulties. Elegant catalytic systems using transition metals such as Pd, Ir, Rh, and Ru have been reported recently (5, 1012); however, the challenge remains to develop efficient Earth-abundant catalytic systems that operate under ambient conditions (1).

Photoredox catalysis has recently emerged as a powerful platform for the direct activation and functionalization of organic molecules via open-shell pathways (13). Hydrogen atom transfer (HAT) by photocatalytically generated amine and aminium (1417), thiyl (18), and other heteroatom-centered radicals (19, 20) has recently enabled myriad otherwise unattainable C–H functionalization transformations under mild reaction conditions (21). The unique reactivity and polarity-governed selectivity (22) inherent in the abstraction events by these heteroatom-centered radicals inspired us to question whether a photocatalytic HAT strategy could be extended to the most challenging C(sp3)–H transformations: the selective functionalization of methane and other gaseous alkanes. The nucleophilic character of alkyl radicals would provide strategic advantages for the development of diverse selective functionalizations, complementary to transition metal–catalyzed reactions, under mild conditions. Although gas-phase HAT events between methane and open-shell metal-oxo species have been extensively investigated for oxidative couplings of methane at elevated temperature (23), room-temperature light alkane functionalization via HAT photocatalysis in a homogeneous system remains largely unresolved.

Ligand-to-metal charge transfer (LMCT) is a common photoexcitation manifold among coordination complexes of transition metals with an empty valence shell (24), which has nevertheless been underinvestigated and underutilized in synthetic organic transformations via modern photoredox catalysis (25, 26). We recently applied LMCT catalysis to the direct activation of alcohol feedstocks, enabling alkoxy radical–mediated skeletal rearrangement of cyclic alcohols (27) and remote C–H functionalization of primary alcohols via 1,5-HAT (28). Using absorbed light energy to promote targeted oxidation via the homolysis of the transiently coordinated Ce(IV)–alkoxide, challenging oxidations of a wide range of alcohols have been accomplished under mild and operationally simple conditions with cerium(III) salts as precatalyst. Recognizing that the ready abundance of cerium would make it an ideal catalyst for sustainable, large-scale photocatalytic systems (29, 30), we hypothesized that the synergistic merger of LMCT and HAT catalysis, via the use of inexpensive cerium salts in concert with simple alcohols such as methanol or 2,2,2-trichloroethanol, could productively activate and functionalize inert alkanes such as methane and ethane (Fig. 1). Indeed, highly electrophilic alkoxy radicals [CH3OH, O–H bond dissociation energy (BDE) = 105 kcal/mol] would be thermodynamically viable for the abstraction of C–H bonds from methane (C–H BDE = 105 kcal/mol) (31). Conceivably, the polarity-matching effect would render the hydrogen atom transfer with the ubiquitous, weak and acidic C–H bonds of solvent (CH3CN, C–H BDE = 93 kcal/mol) kinetically disadvantageous, thereby subverting the frequently encountered problem of competitive solvent functionalization in methane functionalization. Critically, the use of inexpensive and widely available alcohols as HAT catalysts would allow easy tunability of reactivity and selectivity (32). Here, we report the application of LMCT-enabled HAT catalysis as a general method for the catalytic and selective radical coupling of methane, ethane, and several other simple alkanes.

Fig. 1 Alkoxy radical–mediated HAT activation enables diverse selective functionalizations of gaseous alkanes.

The synergistic merger of LMCT and HAT catalysis would offer opportunities for the use of methane and other gaseous alkanes under economical ambient conditions. TCE, 2,2,2-trichloroethanol; TFE, 2,2,2-trifluoroethanol; FG, functional group.

We sought first to apply our LMCT/HAT functionalization to the amination of methane, a largely elusive transformation (33). The direct transformation of methane into liquid tert-butyloxycarbonyl (Boc)–protected monomethylhydrazine (MMH)—an important rocket propellant and valuable building block for heterocyclic compounds, currently produced in industrial quantities via oxidation of methylamine—would upgrade methane from an abundant feedstock into a useful commodity chemical. A detailed description of our proposed catalytic mechanism is outlined in Fig. 2. We envisioned that a Ce(IV)-alkoxy complex, catalytically generated in situ from simple alcohols and a Ce(IV) salt, would undergo photoinduced LMCT to generate a high-energy electrophilic alkoxy radical and a reduced Ce(III) species. The alkoxy radical would then abstract an H atom from a C–H bond in the substrate alkane to generate an alkyl radical species, which would then readily couple with di-tert-butyl azodicarboxylate (DBAD) to forge a new C–N bond with the formation of a stable captodative N-centered radical. Single electron reduction of this radical by the reduced form of the cerium catalyst would regenerate the active Ce(IV) and deliver the desired product after protonation.

Fig. 2 Proposed mechanism for the cerium-catalyzed C(sp3)–H functionalization of methane and other gaseous alkanes.

Highly electrophilic alkoxy radicals, generated from simple alcohols via photoinduced LMCT, are used for the HAT activation of challenging C–H bonds in light alkanes. di-Boc-MMH, di-tert-butyloxycarbonyl monomethylhydrazine. Me, methyl; Et, ethyl; Pr, propyl.

In practice, the photocatalyzed reaction was carried out in a standard pressure reactor equipped with a sapphire window in the top plate to transmit 400-nm LED irradiation (see fig. S2). At ambient temperature under 5000 kPa of methane with DBAD as the limiting reagent, we observed that using 0.5 mole percent (mol %) CeCl3, in concert with catalytic 2,2,2-trichloroethanol, the desired Boc-protected MMH product was formed in 39% yield (table S6, entry 1; see tables S6 to S10 for detailed studies). Among several commercial cerium salts we evaluated in combination with exogenous soluble tetrabutylammonium chloride, cerium(IV) trifluoromethanesulfonate gave slightly higher efficiency (Table 1, entry 1); the premade Ce(IV) chloride complex (34) also proved effective in this catalytic system (table S6, entry 2). No overfunctionalization product was observed in any cases, and the product proved to be inert when subjected to the reaction conditions (fig. S7). Moreover, the alcohol HAT catalyst 2,2,2-trichloroethanol was left largely intact at the end of the reaction. The major unproductive side pathway was the reduction of DBAD to the corresponding hydrazine. An increase in absolute yield could be obtained by using D3-acetonitrile, affording di-Boc-MMH in 63% yield, as a smaller amount of reduced hydrazine by-product was formed (Table 1, entry 2). Moreover, no significant decrease in efficiency was observed when the reaction was performed at 0°C (table S9, entry 2), highlighting the highly reactive nature of the radical coupling system. Lowering the catalyst loading to 0.1 mol % or even 0.01 mol % increased the turnover numbers (TONs) to 320 and 2900, respectively, although this necessitated longer reaction times and resulted in minimally decreased yield (Table 1, entry 3, and table S8, entry 3). Control experiments indicated the necessity of cerium catalyst, alcohol catalyst, and LED irradiation, as omission of any of these resulted in no desired amination product (table S1).

Table 1 Cerium-catalyzed amination of alkanes.

Yields and regiomeric ratios (r.r.) of all products were determined by gas chromatography analysis with internal standard. Regiomeric ratios are presented as 1°/2° product ratios. See full details in supplementary materials. *With 40 mol % trifluoroacetic acid. †In D3-acetonitirle, in the absence of TBACl and trifluoroacetic acid. ‡With diisopropyl azodicarboxylate (DIAD). §At 1000 kPa of ethane. ||With 50 mol % alcohol catalyst. rt, room temperature; TBACl, tetrabutylammonium chloride; DBAD, di-tert-butyl azodicarboxylate.

Embedded Image
View this table:

The catalyst combination was also highly effective for ethane functionalization (see tables S11 and S12 for detailed studies). Because of ethane’s higher solubility in acetonitrile relative to methane, the reactions could be performed in regular vials under atmospheric pressure. Satisfactory yield and TONs were obtained using 0.5 mol % CeCl3 and 20 mol % 2,2,2-trichloroethanol (Table 1, entry 4). Furthermore, lowing the cerium catalyst loading from 0.05 mol % to 0.01 mol % led to catalytic TONs up to 6900 under ambient temperature and pressure conditions (table S12, entry 3). Additionally, increasing the pressure of ethane to 1000 kPa led to a remarkable rate acceleration, yielding 97% of aminated product in just 4 hours, at 0.01 mol % loading of cerium catalyst, corresponding to a TON of 9700 (Table 1, entry 5).

Propane and butane, the major components of liquid petroleum gas (LPG), a commonly used fuel, present interesting challenges with respect to selectivity of C–H bond functionalization (35, 36). We hypothesized that the regioselectivity of the C–H bond abstraction step could be tuned by modification of the alcohol HAT catalyst (22, 32, 37, 38). In both propane and butane, the methylene positions possess slightly weaker C–H bonds, whereas the methyl positions possess more numerous C–H bonds. We were pleased to observe that with low loading of CeCl3 and simple alcohols (2,2,2-trichloroethanol or 2,2,2-trifluoroethanol), propane could be functionalized to provide aminated products with high efficiency, although in a nonselective manner (Table 1, entries 6 and 7). Using methanol as the HAT catalyst, the regioselectivity improved to 1:3.9 (Table 1, entry 8), favoring the methylene position, although the reaction proceeded at a much lower rate. This selectivity could arise from the electronic difference in the electrophilic nature of methoxy radical relative to halogenated ethoxy radicals. In a similar vein, butane could be functionalized with high efficiency, and the regioselectivity could be tuned with different choice of alcohol catalysts to favor the methylene positions. When using isopropanol as the HAT catalyst, the 1°/2° product ratio could be increased to 1:4 (Table 1, entry 10); the lower efficiency was likely due to the decomposition of the isopropanol catalyst via alkoxy radical–mediated β-scission, as the side product of methyl radical coupling with DBAD was observed in the reaction (table S14, entry 5). Methanol proved to be a selective HAT catalyst, providing 1:8 selectivity for the methylene C–H bonds (Table 1, entry 11); this result validated the hypothesis that enhancement of regioselectivity could be accomplished through electronic tuning of the alcohol HAT catalyst. Pleasingly, liquid hydrocarbons such as cyclohexane could be aminated with high efficiency using this inexpensive photocatalytic platform at a low cerium loading (Table 1, entry 12).

We next applied this photocatalytic protocol to radical alkylation reactions using methane and ethane as alkylating feedstocks. As shown in Fig. 3, the alkoxy-mediated methylation and ethylation of electron-deficient alkene 1 using methane and ethane delivered the desired product with high efficiency (56% and 90% yield, respectively). Furthermore, methane and ethane proved viable pronucleophiles in Minisci arene alkylation reactions. The acidic conditions made use of diminished product overfunctionalization by electronically deactivating the benzylic C–H bonds through protonation of the N-heterocycle (39). With ammonium persulfate as an economical oxidant, methylated and ethylated isoquinoline could be produced at room temperature.

Fig. 3 Photocatalytic alkylations using methane and ethane.

Photocatalytic reactions in batch reactors are typically limited to small-scale applications because of the attenuation effect of superficial light penetration (40). In addition to the increase of the light utilization efficiency through glass microreactors, the development of continuous-flow photocatalytic systems for large volume reactions could further benefit from the ease of handling gaseous reactants and the enhanced mass transfer between gas and liquid phases, which are crucial for the utilization of gaseous alkanes. Given the attractiveness of flow chemistry for mixed-phase gas/solution reactions, we investigated the C–H functionalization of gaseous alkanes in continuous-flow microreactors (Fig. 4). An acetonitrile solution of reagents and catalysts was pumped as a single stream and then directly combined and mixed with the gaseous alkane feedstock in the microreactor. Using a flow setup of 10 parallel microreactors (4.5 ml internal volume in total), at a pressure of 1500 kPa for the ethane gas and a flow rate of 0.75 ml/min for the liquid solution stream, amination product was furnished in 90% yield with a residence time of 6 min and a production throughput of 2 mmol/hour. Under similar conditions, propane and butane were functionalized with good yield (76% and 56%, respectively) and productivity (1.7 mmol/hour and 1.3 mmol/hour). In our attempts to use methane in this mixed liquid/gas flow reaction setting, we were hampered by the pressure limits of our commercial microreactors (1800 kPa), leading to a somewhat diminished, but promising, 15% yield. As for higher alkanes such as cyclohexane, the scaled-up amination could be conveniently performed with simple liquid injection mode, affording remarkable productivity of 4.2 mmol/hour.

Fig. 4 Photocatalytic amination of alkanes in continuous-flow reactors for scaled-up applications.

*At 0.01 M with 40 mol % trifluoroacetic acid. †With DIAD. ‡In liquid mode at 0.2 M.

Further studies were performed to provide additional evidence for the intermediacy of alkoxy radicals under this catalytic manifold. As shown in fig. S11, methoxy, trichloroethoxy, and trifluoroethoxy radicals could each be trapped by styrene to generate 1,2-alkoxyamination products (80%, 25%, and 59% yield, respectively). In the case of 2,2,2-trifluoroethanol, a small amount of trifluoromethylated product was also observed, indicative of a β-scission pathway and trapping of trifluoromethyl radical.

This photocatalytic platform has enabled several direct transformations of methane and other simple hydrocarbons, including amination, alkylation, and arylation, and offers intriguing opportunities for further functionalizations of feedstock alkanes.

Supplementary Materials

Materials and Methods

Figs. S1 to S11

Tables S1 to S15

GC data and NMR spectra

References and Notes

Acknowledgments: Funding: We thank the National Natural Science Foundation of China (21772121) and the “Thousand Plan” Youth program for financial support. Author contributions: Z.Z. conceived and directed the project; Z.Z., A.H., J.J.G., and H.P. designed the experiments; A.H., J.J.G., and H.P. performed and analyzed the photocatalytic batch reactions; H.P. performed continuous-flow reactions; and Z.Z., A.H., J.J.G., and H.P. prepared the manuscript. Competing interests: The authors declare no conflicts of interest. Data and materials availability: Data are available in the supplementary materials.
View Abstract

Stay Connected to Science

Editor's Blog

Navigate This Article