Aqueous Au-Pd colloids catalyze selective CH4 oxidation to CH3OH with O2 under mild conditions

See allHide authors and affiliations

Science  13 Oct 2017:
Vol. 358, Issue 6360, pp. 223-227
DOI: 10.1126/science.aan6515

A radical route from methane to methanol

The conversion of methane into chemicals usually proceeds through high-temperature routes that first form more reactive carbon monoxide and hydrogen. Agarwal et al. report a low-temperature (50°C) route in aqueous hydrogen peroxide (H2O2) for oxidizing methane to methanol in high yield (92%). They used colloidal gold-palladium nanoparticles as a catalyst. The primary oxidant was O2; isotopic labeling showed that H2O2 activated methane to methyl radicals, which subsequently incorporated O2.

Science, this issue p. 223


The selective oxidation of methane, the primary component of natural gas, remains an important challenge in catalysis. We used colloidal gold-palladium nanoparticles, rather than the same nanoparticles supported on titanium oxide, to oxidize methane to methanol with high selectivity (92%) in aqueous solution at mild temperatures. Then, using isotopically labeled oxygen (O2) as an oxidant in the presence of hydrogen peroxide (H2O2), we demonstrated that the resulting methanol incorporated a substantial fraction (70%) of gas-phase O2. More oxygenated products were formed than the amount of H2O2 consumed, suggesting that the controlled breakdown of H2O2 activates methane, which subsequently incorporates molecular oxygen through a radical process. If a source of methyl radicals can be established, then the selective oxidation of methane to methanol using molecular oxygen is possible.

In industry, CH4 can be used as a feedstock for methanol (CH3OH) production indirectly via the formation of synthesis gas (CO + H2) at high temperatures and pressures, an expensive and energy-intensive process (1). The direct oxidation of CH4 to CH3OH, which is challenging as overoxidation must be avoided, has been the subject of intensive study for many decades (2, 3). Cyclic gas-phase oxidation of CH4 with metal-exchanged zeolite catalysts with O2, N2O, or H2O requires high temperatures (200° to 500°C) to activate the oxidant and desorb produced CH3OH (48). Liquid-phase reactions typically use milder reaction conditions; however, a closed catalytic cycle is often not achieved. Periana and co-workers reported that electrophilic metals such as Hg and Pt complexes are active for methane oxidation (9, 10). These systems require high temperatures (180°C) and strongly acidic media, such as oleum, to facilitate the reaction. Oxidation products are trapped as methyl-bisulfate to protect against overoxidation, and the products are then hydrolyzed to release CH3OH and SO2. This catalyst was later heterogenized with the use of a solid carbon framework to anchor the Pt catalyst (11). Similarly, cationic Au in solution can oxidize CH4 in the presence of even stronger oxidizing agents such as selenic acid, which prevents the formation of metallic gold (12).

More benign oxidants, such as hydrogen peroxide (H2O2) with Fe complexes, have also been demonstrated to oxidize C–H bonds (13). Yuan et al. previously showed activation of CH4 with dissolved gold chloride and H2O2 at 90°C but observed agglomeration and precipitation of the catalyst from the solution (14). Pd2+ was also reported to catalyze CH4 oxidation to methyltrifluoroacetate using peroxytrifluoroacetic acid generated by H2O2 and trifluoroacetic anhydride (15). Similarly, heterogeneous catalysts based on Fe-ZSM-5 and Cu-modified ZSM-5 have been used for methane activation to methylhydroperoxide (CH3OOH), CH3OH, and formic acid (HCOOH) in aqueous media at 50°C (16). H2O2 is considered a desirable benign oxidant, second only to molecular O2 for this reaction, as the decomposition product is water.

We have also reported CH4 oxidation using supported gold-palladium nanoparticles (NPs) under mild aqueous conditions with H2O2 as an oxidant at 50°C. The reaction proceeded through a radical mechanism, as both methyl (CH3) and hydroxyl (OH) radicals were observed by electron paramagnetic resonance (EPR) spectroscopy (17). However, the relatively high cost of H2O2 for even stoichiometric oxidation of CH4 makes it difficult to envisage an economically viable process based on this chemistry.

Incorporation of O2 into the primary oxidation products would represent substantial progress toward a feasible CH4-to-CH3OH process. Here we report this reaction under mild conditions, using colloidal Au-Pd NPs in the presence of both H2O2 and O2. We show that by removing the support material from the catalyst, a substantial improvement in activity is achieved and O2 is incorporated into the primary products, with the selectivity to primary products reaching >90% with minimal CO2 produced.

We prepared polyvinylpyrrolidone (PVP)–stabilized Au-Pd (1:1 molar ratio) colloids and supported catalysts by immobilizing the colloid on TiO2 (see supplementary materials for all experimental details) (18, 19). We investigated CH4 oxidation in water with H2O2 as an oxidant at 50°C and 30 bar of CH4 pressure for 30 min (19). With the supported colloid material, a 1% Au-Pd/TiO2 catalyst (Table 1, entry 1), minimal reaction products were observed (primary oxygenate selectivity: 26%), as determined by quantitative nuclear magnetic resonance (NMR) analysis (16, 19). In this experiment, most of the H2O2 was decomposed (73%). A gain factor, defined as mol of oxygenate produced/mol of H2O2 consumed, was calculated to be 2 × 10−3.

Table 1 Catalytic activity.

Comparison of catalytic activity of supported and unsupported colloidal Au-only, Pd-only, and Au-Pd catalysts for the liquid-phase oxidation of CH4 using H2O2. Test conditions: reaction time = 0.5 hours; stirring speed = 1500 rpm; CH4 pressure = 30 bar; reaction temperature = 50°C; stirred heating ramp rate = 2.25°C/min. Dash indicates not applicable.

View this table:

This high rate of H2O2 degradation was considered detrimental to the reaction, either by (i) the termination of reactive radical chains caused by radical concentrations being too high or (ii) the consumption of H2O2 at such a high rate that it does not allow sufficient interaction with the low amount of solubilized CH4. We investigated which component of the catalyst was responsible for the high H2O2 degradation rates. We observed (fig. S1) that the degradation of H2O2 at room temperature under atmospheric pressure was low in the presence of bare TiO2 and unsupported Au-Pd colloidal NPs. Only when the Au-Pd NPs were supported on TiO2 did the catalyst exhibit a high rate of H2O2 degradation, suggesting that either the interfacial sites at the support/metal interface (20) or a change in the morphology of the NP (21) upon immobilization led to the high H2O2 degradation rates (22).

This finding, coupled with the known ability of high–surface area TiO2 to quench radical reactions (23), led us to evaluate the ability of unsupported PVP-stabilized Au-Pd colloids to catalyze CH4 oxidation. Using 1000 μmol of H2O2 and the same amount of metal as in the supported catalyst, we observed substantially more product (15.7 μmol) compared with the supported catalysts (Table 1, entry 2), whereas much less H2O2 (38%) was consumed. Furthermore, the primary products, CH3OOH and CH3OH, were produced with 90% selectivity. The colloidal catalyst was an order of magnitude more efficient than the solid Au-Pd/TiO2 catalyst with respect to products generated per unit amount of H2O2 consumed, with a gain factor of 3 × 10−2 compared with 2 × 10−3.

The presence of CH3OOH, CH3OH, HCOOH, and CO2 in the product stream suggests a consecutive oxidation pathway, as previously observed (16, 17). Experiments with isotopically labeled 13CH4 confirmed that the carbon source was CH4 and not the organic stabilizer, PVP, present in the colloidal solution (fig. S2). Carbon-based products containing the label were present in quantities corresponding to the amount of labeled 13CH4 present; in addition, no carbon-based products were observed when CH4 was not present (table S1).

High-angle annular dark field–scanning transmission electron microscopy (HAADF-STEM) imaging of the fresh Au-Pd–PVP colloids revealed that they primarily have multiply twinned icosahedral structures, although some cuboctahedral NPs were also detected (Fig. 1, A and B). The Au-Pd–PVP colloids ranged in size between ~2 and 12 nm and had a mean diameter of 3.7 nm (Fig. 1C). STEM–x-ray energy-dispersive spectroscopy compositional analysis showed the Au-Pd–PVP NPs to have a similar Au:Pd ratio, irrespective of their size (fig. S3). Transmission electron microscopy analysis of the sol-immobilized Au-Pd/TiO2 samples (fig. S4) showed the Au-Pd NPs to have a similar size, morphology, and composition to their colloidal counterparts. X-ray photoelectron spectroscopy (XPS) analysis of both the supported and colloidal samples showed that both Au and Pd were metallic in nature, with minor PdCl2 components detected in the colloidal samples (figs. S5 and S6).

Fig. 1 Catalyst characterization.

Representative HAADF images and particle size distributions for the unsupported Au-Pd–PVP sol in the fresh colloids (A to C) and after a CH4 oxidation reaction (D to F).

Previous EPR spin-trapping studies revealed the presence of both OH and CH3 radicals during reactions with a sol-immobilized Au-Pd catalyst, suggesting that the reaction mechanism is radical-based (17, 24). The observation that the primary product was CH3OOH implies that the primary termination step is either between CH3 and OOH radicals or from recombination of CH3 with dissolved O2 in the solution originating from the decomposition of H2O2 (25). In the case of the unsupported colloidal catalyst, which decomposes H2O2 at a much slower rate but makes substantially more products, we rationalized that CH3 was being produced over a longer time scale and that adding O2 to the reaction mixture would promote oxygen incorporation through the generation of CH3OO radicals. The reaction of CH3 with O2 has been reported to occur at high rates in gas-phase reactions (26). The addition of 5 bar of O2 pressure (Table 1, entry 3) to the reaction resulted in an increased product yield (26.8 μmol) compared with the H2O2-only reaction (Table 1, entry 2) while maintaining a high selectivity to primary oxygenates (95%). The gain factor also remained an order of magnitude higher than that of the supported catalyst and was more than double that of the colloidal catalyst with the H2O2-only reaction. These results suggest that the presence of additional O2 (beyond that originating from H2O2 decomposition) promoted the termination steps that generate primary products.

To demonstrate the incorporation of O2 into the primary products using the colloidal catalyst, 5 bar of 18O2 pressure was added to the reaction mixture. Solutions from isotopic labeling reactions were analyzed by gas chromatography–mass spectrometry (GC-MS), which resulted in the degradation of the CH3OOH primary oxidation product to CH3OH due to the high injection temperature. However, good agreement was observed between the combined amount of CH3OOH and CH3OH determined by NMR and the amount of CH3OH determined by GC-MS (fig. S7). Mass spectrometry analysis of the reactions using H216O2 and 18O2 revealed that CH3OH mass fragments containing the 18O label were responsible for 51% of the observed signal (fig. S8). Reactions in the absence of H2O2 at 50°C (Table 1, entry 4) showed no generation of oxygenated products, confirming that H2O2 was a necessary ingredient for the reaction to take place. These results show that under mild aqueous conditions, colloidal Au-Pd NPs can catalyze the reaction of H2O2 + CH4 + O2 with incorporation of O2 into the primary reaction products.

We investigated decreasing the amount of H2O2 to increase efficiency with the aim of generating more than one primary oxygenate species per molecule of H2O2 consumed. Figure 2A shows that by decreasing the amount of H2O2 from 2000 to 1000 μmol over 60 min while maintaining 5 bar of O2 pressure in the gas phase, the quantity of products formed initially increased from 18 to 43 μmol. This increase coincided with a reduction in the amount of H2O2 consumed (from 81 to 64%), indicating that a much greater efficiency was achieved with less H2O2 (a full breakdown of products is given in table S2).

Fig. 2 Methane oxidation reactions carried out over unsupported Au-Pd colloids.

(A) Gain factor (blue), selectivity (red), and total amount of products (green) as a function of the different amounts of H2O2 used. (B) GC-MS spectra of CH3OH formed (mass = 32 and 34 for CH316OH and CH318OH, respectively) during methane oxidation with a Au-Pd colloid via H216O2 + 16O2 (upper spectrum) or H216O2 + 18O2 (lower spectrum). For CH4 oxidation with 18O2, >70% of 18O2 molecules were incorporated in the CH3OH product. m/z, mass/charge ratio. (C) Time-on-line and Au-Pd colloid reuse study for the methane oxidation reaction employing 50 μmol of H2O2 and 5 bar of O2 pressure and showing no induction period. This panel shows total products (black) and individual products [CH3OOH (red), CH3OH (blue), HCOOH (pink), and CO2 (green)] generated as a function of time on line. The black dotted vertical line at 120 min indicates a subsequent second addition of H2O2. (D) Oxidation of CH4 performed at 23° and 50°C with 50 μmol of H2O2. Blue and red bars represent reactions performed with and without 5 bar of O2 pressure, respectively. Black squares indicate their respective gain factors.

For a radical mechanism, an excess of H2O2 could likely cause termination of radical reactions and limit product formation. A further reduction in the amount of H2O2 from 1000 to 500 μmol resulted in a slight increase in reaction products from 43 to 50 μmol, suggesting that the reaction had become limited by the availability of a reactant other than the radical species generated by H2O2—that is, either O2 or CH4, as determined by their solubility in the aqueous reaction media. Decreasing the amount of H2O2 further resulted in lower consumption of H2O2 but also led to a decrease in product formation, probably because of a reduction in the concentration of radicals generated. For the optimized amount of H2O2 (50 μmol) (Table 1, entry 5), the gain factor reached 1.2 and represents an increase by three orders of magnitude over the corresponding 1% Au-Pd/TiO2 sol-immobilized catalyst, together with a higher selectivity to primary oxygenates. Furthermore, 18O2 isotopic labeling experiments carried out under these optimized conditions again revealed substantial incorporation of 18O2 (~50%) into the primary products (fig. S9).

In all of the isotopically labeled reactions, CH3OOH was decomposed to CH3OH during the analysis. To ensure that thermal decomposition did not result in a loss of the isotopic label, we quantitatively (confirmed by NMR analysis) reduced CH3OOH to CH3OH with NaBH4 before performing GC-MS analysis. When we used this procedure to prevent thermal decomposition of CH3OOH in the presence of atmospheric oxygen, more than 70% of the CH3OH detected showed incorporation of the 18O label (Fig. 2B). In the previous isotope labeling experiments, radical decomposition of labeled CH3OOH in the presence of air and any remaining H2O2 resulted in an underestimate of the degree of 18O incorporation (27).

We conducted control experiments with unsupported monometallic Au and Pd colloids, along with the corresponding precursor metal chlorides using the same metal concentration of 6.6 μmol under identical reaction conditions (Table 1, entries 6 to 9). The monometallic Au and Pd colloids, as well as the metal chloride precursors, showed no activity for CH4 oxidation. In fact, the HAuCl4 and Pd colloid precipitated during the reaction. Hence, alloyed Au-Pd NPs show a synergistic effect in terms of activity and stability for this reaction. Recently, leaching of precious metals into reaction solutions to generate clusters containing three to five atoms has been implicated in catalytic reactions and characteristically shows an induction period at the start of the reaction as the clusters form (28). HAADF-STEM analysis of the used colloid revealed some limited particle growth, but no evidence of subnanometer clusters or isolated Au atoms was found either before or after 30 min of reaction under optimized conditions (Fig. 1, D to F).

Furthermore, no substantial differences in the concentrations of cationic Au or Pd content were observed by XPS measurements on pre- or postreaction Au-Pd–PVP colloids (fig. S5). Time-on-line analysis indicates that there is no induction period associated with this reaction system and that products were generated even before the 50°C reaction temperature was reached. The catalyst being colloidal rather than a solid powder likely removed some mass transfer limits (Fig. 2C). This observation, coupled with the inactivity of the monometallic Au and Pd sols and metal chloride precursors, indicates that the bimetallic colloidal NPs are the active catalyst. The presence of reaction products at the start of the reaction (i.e., just after completion of the heating ramp at 2.25°C/min) suggests that there is some activity at even lower temperatures. As the reaction proceeded, the total amount of product generated plateaued at ~120 min (Fig. 2C and table S3), which coincided with the depletion of H2O2 from the reaction mixture. After adding more H2O2, more products were generated and a similar oxidation rate was achieved, indicating that the Au-Pd colloid was stable during this time (Fig. 2C and table S4).

As the reaction time was increased further, some overoxidation of the primary products was observed, but the level of CO2 production remained at <4% of the total products formed over the entire 240-min duration (table S4, entry 4). We also conducted reactions at room temperature, both with and without the addition of 5 bar of O2 pressure as part of the reaction mixture, and observed activity with similar levels of primary oxygenate selectivity (96%). A gain factor of 1.35 was observed at room temperature in the presence of O2 (Fig. 2D), similar to the reaction performed at 50°C in the presence of O2 (table S5), producing 5.4 μmol of products with only 4 μmol of H2O2 consumption, corresponding to a productivity of 10.4 moloxygenates kgcatalyst−1 hour−1. This productivity compares favorably with that reported for methane monooxygenase from Methylococcus capsulatus (Bath) (5.05 molmethanol kg−1 hour−1) (29), a biological system that selectively oxidizes CH4.

Preliminary kinetic analysis shows a first-order dependence on colloid, CH4, and H2O2 concentration (fig. S10), indicating a rate-determining step that includes CH4 and H2O2 with an observed activation energy of 39 kJ mol−1. Because the reaction proceeds only when H2O2 is present in the reaction mixture, the initial activation of CH4 to CH3 is likely to occur through a radical mechanism (Fig. 3), a process that is suppressed by the presence of a catalyst support such as TiO2 (see table S6 for this comparison). These CH3 radicals can react quickly with dissolved O2, which results in incorporation of >70% O2 into the primary reaction products under optimized conditions. Some products containing 16O were formed through radical reactions between CH3 with either •16O16OH or 16O2 generated from the decomposition of H216O2. The CH3 radicals were generated via hydrogen abstraction by OH from H2O2, and this initiation step activates CH4 (Fig. 3). In the optimized reaction, 10 μmol of H2O2 and 5 bar of 18O2 pressure were required to generate 20 μmol, where the primary products contained 70% 18O (14 μmol) and 30% 16O (6 μmol). These isotopic ratios and the reaction scheme proposed are broadly in line with the total amount of H2O2 consumed (16 μmol), where 10 μmol was used to generate CH3 radicals and 6 μmol was used in 16O products via decomposition. In terms of O2 incorporation, greater efficiency was achieved by utilizing H2O2 to activate CH4 rather than using it to supply oxygen into the primary products. Although CH4 activation via OH was required for formation of CH3 radicals, once formed, they readily reacted with O2 to form CH3OH.

Fig. 3 Proposed reaction scheme for methane oxidation in the presence of H2O2 and molecular O2.

With this mechanism in mind, we tested an iron-based Fenton’s type catalyst with H2O2 to investigate whether OH in the absence of the Au-Pd colloid could activate methane, but negligible product formation was observed (table S7). Therefore, the presence of the Au-Pd colloidal NPs are also essential for CH4 activation. We propose that, in principle, other routes for the generation of CH3 radicals could be used to facilitate this chemistry. For example, rather than using H2O2, it would be highly desirable to devise a method of coupling the Au-Pd colloidal catalyst with a photochemical (30, 31) or electrochemical fuel cell (32, 33) to generate OH for H abstraction to facilitate CH3 radical formation.

Supplementary Materials

Materials and Methods

Figs. S1 to S10

Tables S1 to S7

References and Notes

  1. Materials and methods are available as supplementary materials.
Acknowledgments: We acknowledge Cardiff University for financial support as part of the MAXNET Energy Consortium. C.J.K. acknowledges funding from the NSF Major Research Instrumentation program (grant MRI/DMR-1040229). S.M.A. thanks the Saudi Arabian government for his Ph.D. scholarship. All results are reported in the main text and supplementary materials.
View Abstract

Navigate This Article