Report

Atomic-layered Au clusters on α-MoC as catalysts for the low-temperature water-gas shift reaction

See allHide authors and affiliations

Science  28 Jul 2017:
Vol. 357, Issue 6349, pp. 389-393
DOI: 10.1126/science.aah4321

Low-temperature CO removal

Carbon monoxide deactivates fuel cell catalysts, so it must be removed from H2 generated from hydrocarbons on site. Yao et al. developed a catalyst composed of layered gold clusters on molybdenum carbide (MoC) nanoparticles to convert CO through its reaction with water into H2 and CO2 at temperatures as low as 150°C. Water was activated on MoC to form surface hydroxyl groups, which then reacted with CO adsorbed on the gold clusters.

Science, this issue p. 389

Abstract

The water-gas shift (WGS) reaction (where carbon monoxide plus water yields dihydrogen and carbon dioxide) is an essential process for hydrogen generation and carbon monoxide removal in various energy-related chemical operations. This equilibrium-limited reaction is favored at a low working temperature. Potential application in fuel cells also requires a WGS catalyst to be highly active, stable, and energy-efficient and to match the working temperature of on-site hydrogen generation and consumption units. We synthesized layered gold (Au) clusters on a molybdenum carbide (α-MoC) substrate to create an interfacial catalyst system for the ultralow-temperature WGS reaction. Water was activated over α-MoC at 303 kelvin, whereas carbon monoxide adsorbed on adjacent Au sites was apt to react with surface hydroxyl groups formed from water splitting, leading to a high WGS activity at low temperatures.

Efficient low-temperature catalysts for the water-gas shift (WGS) reaction, especially those operating under 423 K (17), are of interest for applications in fuel cells, especially those using H2 generated by hydrocarbon reforming processes that are contaminated with CO, which deactivates the catalysts. For heterogeneous catalysis, in addition to Cu-based catalysts, which display low activity at low temperature (8, 9), Pt-group noble metals and Au supported on reducible metal oxides, such as ceria (1) or FeOx (10), which contain oxygen vacancies, are commonly used. Flytzani-Stephanopoulos and co-workers demonstrated that noble metal catalysts dispersed on alkali-promoted inert supports can also be active for the WGS reaction, making a reducible oxide support no longer a requirement (4, 6). The alkali ion–associated surface OH groups are reactive toward CO in the presence of atomically dispersed platinum or gold, giving the catalyst superior metal atom efficiency in the WGS reaction. Metal carbide (e.g., hexagonal closest packing β-Mo2C)–supported noble metal catalysts provide similar functionalities and are more active for the reaction at low temperature (7, 11, 12). However, none of these systems displays an activity higher than 0.1 moles of CO per mole of metal per second (molCO molmetal–1 s–1) between 393 and 423 K (Table 1).

Table 1 Comparison of the activities of the representative catalytic systems for the low-temperature WGS reaction.

The operating pressure of the reactions was 1 bar.

View this table:

To achieve high WGS activity at low temperature, we searched for catalysts that could dissociate water efficiently and reform the generated oxygen-containing species (reaction of surface oxygen or hydroxyl with CO*) at low temperature. We report that Au confined over face-centered cubic–structured α-MoC is at least one order of magnitude more active than previous findings for the WGS reaction below 423 K. The α-MoC substrate facilitates epitaxially grown atomic Au layers with altered electronic structure for favorable bonding with CO. The synergy of atomic-layered Au clusters with adjacent Mo sites in α-MoC can effectively activate water at low temperature.

Catalysts made from Au supported by pure-phase α-MoC [2 weight % (wt %) Au/α-MoC] were synthesized by a precipitation method followed by sequential temperature-programmed ammonization and carburization. For comparative purposes, α-MoC, 2 wt % Au/β-Mo2C (13, 14), 2 wt % Au/SiO2 (15), and 2 wt % Au/CeO2 (1) catalysts were also prepared. The high dispersion of Au in the 2 wt % Au/α-MoC [hereafter, (2%)Au/α-MoC] catalyst was evidenced by the lack of x-ray diffraction (XRD) peaks associated with Au crystallites. Operando XRD studies (1% CO/3% H2O in He; 10 ml min–1) revealed that the bulk structure of the (2%)Au/α-MoC catalyst remained intact up to 523 K, beyond which the α-MoC was gradually oxidized by water (Fig. 1A). Ex situ XRD experiments [10.5% CO/21% H2O/20% N2 in Ar; gas hourly space velocity (GHSV) = 180,000 hour–1) confirmed that at higher water partial pressure, the bulk structure of catalysts is stable up to 473 K. Neither the oxidation of α-MoC nor the aggregation of Au was observed at temperatures up to 473 K (fig. S1).

Fig. 1 Catalytic properties and structural characterization of the (2%)Au/α-MoC catalyst.

(A) In situ XRD (wavelength, 0.3196 Å) of the (2%)Au/α-MoC catalyst under WGS reaction conditions at various temperatures. The y axes in the top and bottom panels are in arbitrary units. The rainbow color scheme in the middle panel spans from no signal (blue) to high-intensity diffraction peaks (red). MoO2 and α-MoC are shown on the right as ball-and-stick diagrams (red, O; purple, Mo; black, C). (B) CO conversion on different catalysts at various temperatures (10.5% CO/21% H2O/20% N2 in Ar; GHSV, 180,000 hour−1). (0.9%)Au/α-MoC was produced by NaCN leaching of (2%)Au/α-MoC. (C) The activity of different catalysts (moles of CO per moles of metal per second), measured at CO conversion below 15% in 11% CO/26% H2O/26% H2/7% CO2/30% N2. (D) Kinetic orders (n) of the reactants and products. (E) Apparent activation energy Eapp of various catalysts in 10.5% CO/21% H2O/20% N2-Ar balance.

The WGS activity was evaluated under product-free (10.5% CO/21% H2O/20% N2 in Ar) and full reformate (11% CO/26% H2O/26% H2/7% CO2 in N2) gas feeds. In the product-free gas (GHSV = 180,000 hour−1), α-MoC showed very low CO conversion (3.4%) at 393 K (Fig. 1B), and none of the reference catalysts achieved >5% CO conversion below 423 K. However, for the (2%)Au/α-MoC catalyst, CO conversion was >95% at 393 K and reached 98% at only 423 K. For reaction temperatures 523 K and above, CO conversion dropped, which may result from the thermodynamic limitation, as well as the gradual transformation of α-MoC to molybdenum oxide, as confirmed by the operando XRD results (Fig. 1A and fig. S2). The Au-normalized activity of the (2%)Au/α-MoC catalyst in product-free gas was 0.012, 0.13, 1.05, 1.66, and 3.19 molCO molAu–1 s–1 at 313, 353, 393, 423, and 473 K, respectively; this high activity at low temperatures compares favorably with other reported WGS catalysts (Table 1 and fig. S3; CO conversion below 15%). Because of the limitation of the water saturation vapor pressure, at low temperature, the composition of reactant gas was adjusted. We determined that 2 wt % is the optimal Au loading for the Au/α-MoC catalyst (fig. S4).

In full reformate gas feed under similar space velocity, the activity dropped slightly (62% activity at 393 K) because of product (H2 and CO2) inhibition (Fig. 1D). However, the activity of the (2%)Au/α-MoC catalyst remained as high as 0.62 and 2.02 molCO molAu–1 s−1 at 393 and 473 K, respectively. The apparent barrier energy (Eapp) of α-MoC itself is low (58 ± 10 kJ mol–1), and Eapp is even lower for the (2%)Au/α-MoC catalyst (22 ± 1 kJ mol–1). Thus, the addition of Au greatly enhanced the low-temperature reactivity of a good WGS catalyst (Fig. 1E). Its exceptional activity and high equilibrium CO conversion at low temperature can be exploited simultaneously (fig. S5), and the catalyst shows an excellent total turnover number, reaching up to 385,400 molCO molAu–1 in a single-run reaction (fig. S6 and table S1).

We designed a two-step temperature-programmed surface reaction (TPSR) experiment to explore the reaction route. After preactivation of the catalysts, 2% H2O in Ar (100 ml min–1, 10 min) was introduced into the reactor at 303 K. Production of H2 was immediately observed on both (2%)Au/α-MoC and α-MoC catalysts, indicating the presence of a low-temperature water dissociation center on α-MoC that led to the formation of H2 and surface OH species (Fig. 2, A and B, and figs. S7 and S8). In contrast, no H2 production was observed on (2%)Au/SiO2 or (2%)Au/β-Mo2C catalysts (Fig. 2C and fig. S9). After purging with Ar (100 ml min–1), the system was switched to 2% CO in Ar (100 ml min–1) at 303 K, kept at that temperature for 10 min, and then increased to 523 K at 5 K min–1. For the (2%)Au/SiO2 catalyst, only water desorption was observed at ~403 K. In sharp contrast, CO2 and H2 were detected simultaneously on the (2%)Au/α-MoC catalyst at around 308 K, and their intensities reached the maxima at 367 K. Thus, the reaction of CO with surface OH could occur at very low temperature (308 K) to form CO2 and additional H2. The reforming reaction could also happen on the α-MoC catalyst, but initiating at a much higher temperature (347 K).

Fig. 2 Mechanism study and electron microscopy characterization.

Water adsorption (at 303 K) followed by CO TPSR, using (A) (2%)Au/α-MoC, (B) α-MoC, and (C) (2%)Au/SiO2. Signals of H2 [mass/charge (m/z) = 2], H2O (m/z = 18), CO (m/z = 28), and CO2 (m/z = 44) were detected. (D and E) High-resolution high-angle annular dark-field (HAADF)–STEM images of fresh (2%)Au/α-MoC, with single atoms of Au marked in blue circles and layered Au structures highlighted in yellow. The Au clusters were further identified by elemental analysis (figs. S18 and S19). (F) HAADF-STEM image of the (2%)Au/α-MoC catalyst after reaction, in which the sample still contains both single-atom Au and layered Au clusters. (G) HAADF-STEM image of the (2%)Au/α-MoC catalyst after NaCN leaching, showing predominantly single atoms of Au, most of which overlap with Mo sites in the support lattice. The very bright features in this image are caused by overlapping MoC particles, as confirmed by elemental mapping (figs. S18 and S19).

The coexistence of a low-temperature water dissociation center on α-MoC and the low-temperature reforming center on the (2%)Au/α-MoC catalyst is the key to the exceptional activity of this catalyst. The Au L3-edge extended x-ray absorption fine structure (EXAFS) fitting (table S2 and fig. S10) shows a low Au-Au first shell coordination number (CN) of 6.9, indicating that the average size of Au species is ~1.5 nm for a hemispherical morphology (16). The Au-Mo CN of 1.6 is striking, considering that Au nanoparticles (NPs) tend to undergo sintering because of the low Tammann temperature (668 K) of bulk Au. (17, 18) Given that this sample was activated at 973 K for more than 2 hours, a strong metal-support interaction must exist between Au and α-MoC. X-ray photoelectron spectroscopy (XPS) (fig. S11) revealed that the Au 4f binding energy shifted 0.6 eV to higher energy with respect to bulk gold (19), indicating that the electronic structure of the Au species is perturbed by the substrate. The reaction order of CO of –0.16 (Fig. 1D) also indicated that CO was already relatively strongly adsorbed on the electronically modified Au surface.

Aberration-corrected scanning transmission electron microscopy (STEM) analysis on the (2%)Au/α-MoC catalyst showed that the catalyst supports were porous assemblies of small α-MoC NPs (3 to 20 nm in diameter; fig. S12). High-resolution STEM Z-contrast imaging (Fig. 2, D and E) revealed two types of Au species on the surface of α-MoC: (i) small layered Au clusters epitaxially grown on the α-MoC support and (ii) atomically dispersed Au. The epitaxial Au clusters had an average diameter of 1 to 2 nm and thickness of 2 to 4 atomic layers (<1 nm), as measured from edge-on clusters occasionally found in profile view (Fig. 2E and fig. S13). Detailed crystal structure analysis (fig. S12, C and D) also showed that these epitaxial Au clusters strongly aligned with the (111) planes of the α-MoC support, with some exposed (200) facets. There were no larger Au NPs present in this sample (fig. S14). No obvious structural difference was observed between the fresh and used catalyst samples (Fig. 2F), and both types of Au species were retained in the tested sample, which is also consistent with the relatively good stability of the catalyst noted in the catalytic reaction.

We used NaCN solution to selectively leach the layered Au clusters from the (2%)Au/α-MoC catalyst (1, 20). The Au loading decreased to around 0.9 wt %, leaving predominantly the atomically dispersed Au atoms, as confirmed by both STEM and XAFS results (Fig. 2G; fig. S12, E to G; and table S2). The Au-normalized WGS activities of (0.9%)Au/α-MoC (leached by NaCN) at 393 and 423 K decreased to around 1/11th and 1/6th of their original values, respectively, but were still higher than those of NaCN-leached α-MoC catalyst (with Eapp similar to that of fresh α-MoC; Fig. 1E). This result indicated that atomically dispersed Au species were indeed catalytically active (1), but the catalytic efficacy of the layered Au clusters on the α-MoC support for low-temperature WGS was even higher than that of the atomically dispersed Au. Furthermore, the Eapp increased to 41 kJ mol–1 after NaCN leaching (Fig. 1E), suggesting some degree of blocking of the low-temperature reaction route after the removal of layered Au clusters. Thus, we attribute the low-temperature WGS activity mainly to the epitaxial Au clusters decorating the α-MoC support.

We carried out density functional theory (DFT) calculations to investigate the WGS reaction path on the Au/α-MoC catalyst. Three catalyst models (fig. S15) of Au(111), monolayer Au/α-MoC(111), and cluster Au15/α-MoC(111) were constructed to represent the different sites on Au/α-MoC, in which Au(111) and monolayer Au/α-MoC(111) simulate large Au NPs and electronic property–modified Au NPs, respectively. Au15/α-MoC(111) represents the interface model of our atomic-layered Au cluster over α-MoC(111). Similar to experimental observations, the Au cluster in Au15/α-MoC has a layered structure, with (111) and (200) exposed facets.

As shown in fig. S16, H2O is thermodynamically and kinetically hard to dissociate on Au(111) and monolayer Au/α-MoC(111), with barriers of 1.91 and 1.66 eV, and the reactions are endothermic by 1.57 and 1.15 eV, respectively. In contrast, when we investigated the first step of the WGS reaction (namely, water dissociation) on Au15/α-MoC(111), we found that at lower coverage (Fig. 3A), two H2O molecules could be easily dissociated and form two H atoms and two OH species with the effective barrier of 0.77 eV (CO + 2H2O → CO + 2OH + 2H), and the two OH species could immediately react (and without a barrier), forming a surface O atom (CO + 2OH + 2H → CO + H2O + O + 2H; exothermic by 0.38 eV). These results indicate that some surface domains of α-MoC could be oxidized by water during the reaction, which has been confirmed by XPS and 18O nuclear magnetic resonance (NMR) experiments (figs. S7 and S8). After the surface was partially decorated with oxygen (Fig. 3B), the calculations showed that surface O atoms could further promote water dissociation. The successive O-assisted water dissociation (CO + 3O + H2O → CO + 2O + 2OH) on the boundary of Au15 and α-MoC(111) had a much lower barrier of 0.22 eV, indicating that the first O–H bond of water could be easily broken at low temperature by this bifunctional catalyst.

Fig. 3 The reaction paths for the WGS reaction on Au15/α-MoC(111).

(A) H2O dissociation and CO reforming at lower coverage, (B) O-assisted H2O dissociation on the boundary oxidized by three O atoms, and (C) CO reforming at high coverage. The energies of gaseous molecules include the zero-point energy and entropy correction at 423 K. Au, Mo, C, O, and H atoms are shown in gold, cyan, gray, red, and white, respectively; to distinguish the C atom in CO, it is represented in black. The subscripts (g), b, and t denote gas phase, bridge site, and top site, respectively. TS, transition state.

The surface OH species formed on the Mo site were apt to react with CO adsorbed on the adjacent Au surface, which has the right geometry (triangular) to enable a low reaction barrier. Indeed, at low CO coverage (Fig. 3A), the effective barrier for the reforming of CO on Au and OH on α-MoC(111) was 0.72 eV, including a migration barrier of 0.22 eV and reaction barrier of 0.50 eV. At high CO coverage (Fig. 3C), the reforming barrier was even lower—0.52 eV—demonstrating that the reaction between adsorbed CO and surface OH species on the peripheral interface of Au and α-MoC (CO + OH = CO2 + ½H2) was apt to proceed. Although the reforming process was facile, it still had a higher barrier than the first step of the WGS reaction (water dissociation on partially oxidized α-MoC). Thus, the rate-determining step of the WGS reaction on Au15/α-MoC is the reforming process, which is in good agreement with our TPSR observations (Fig. 2). The interfacial nature and optimum bonding of this α-MoC–confined Au nanostructure confers the catalyst with outstanding WGS reactivity at low temperature.

Supplementary Materials

www.sciencemag.org/content/357/6349/389/suppl/DC1

Materials and Methods

Supplementary Text

Figs. S1 to S32

Tables S1 and S5

References (2538)

References and Notes

  1. Acknowledgments: This work received financial support from the 973 Project (grants 2017YFB0602200, 2013CB933100, and 2011CB201402), the CAS Pioneer Hundred Talents Program, and the Natural Science Foundation of China (grants 91645115, 21473003, 21222306, 21373037, 21577013, and 91545121). The electron microscopy work was also supported in part by the U.S. Department of Energy (DOE), Office of Science, Office of Basic Energy Sciences, Materials Sciences and Engineering Division (to W.Z.), and through a user project at Oak Ridge National Laboratory's Center for Nanophase Materials Sciences, which is a DOE Office of Science User Facility. The x-ray absorption spectroscopy experiments were conducted at the Shanghai Synchrotron Radiation Facility and the Beijing Synchrotron Radiation Facility. We also acknowledge the National Thousand Young Talents Program of China, the Shanxi Hundred Talents Program, and the Fundamental Research Funds for the Central Universities (grants DUT15TD49 and DUT16ZD224). The research done at Brookhaven National Laboratory (BNL) was financed under contract no. DE-SC0012704 with the DOE. Some of the theoretical calculations were done at the Center for Functional Nanomaterials on the BNL campus. The Advanced Light Source is supported by the Director, Office of Science, Office of Basic Energy Sciences, of the DOE under contract no. DE-AC02-05CH11231. Y.Y. acknowledges the support of the Advanced Light Source Doctoral Fellowship. C.J.K. gratefully acknowledges funding from the U.S. National Science Foundation Major Research Instrumentation program (grant MRI/DMR-1040229). D.M. thanks L. Peng and M. Wang for help with 17O NMR experiments. All data are reported in the main text and supplementary materials. C.S. and X.Z. are inventors on a patent application (ZL 2015 1 0253637.6) held by Dalian University of Technology that covers the preparation of Au/α-MoC. D.M., C.S., and J.A.R. designed the study. X.Z. and S.Y. performed most of the reactions. W.Z., L.Lu, C.J.K., and L.G. performed the electron microscopy study. R.G., X.W., P.L., and Z.Z. finished the DFT calculations. S.Y., W.X., and W.L. performed the x-ray structure characterization and analysis. S.Y., D.M., W.Z., and J.A.R. wrote the paper. Other authors provided reagents, performed certain experiments, and revised the paper.
View Abstract

Navigate This Article