Science 14 December 2007: Vol. 318. no. 5857, pp. 1757 - 1760 DOI: 10.1126/science.1150038

Reports

Activity of CeO_x and TiO_x Nanoparticles Grown on Au(111) in the Water-Gas Shift Reaction

The high performance of Au-CeO₂ and Au-TiO₂ catalysts in the water-gas shift (WGS) reaction (H₂O + COH₂ + CO₂) relies heavily on the direct participation of the oxide in the catalytic process. Although clean Au(111) is not catalytically active for the WGS, gold surfaces that are 20 to 30% covered by ceria or titania nanoparticles have activities comparable to those of good WGS catalysts such as Cu(111) or Cu(100). In TiO_2-x/Au(111) and CeO_2-x/Au(111), water dissociates on O vacancies of the oxide nanoparticles, CO adsorbs on Au sites located nearby, and subsequent reaction steps take place at the metal-oxide interface. In these inverse catalysts, the moderate chemical activity of bulk gold is coupled to that of a more reactive oxide.

¹ Chemistry Department, Brookhaven National Laboratory, Upton, NY 11973, USA.
² Center for Functional Nanomaterials, Brookhaven National Laboratory, Upton, NY 11973, USA.
³ Facultad de Ciencias, Universidad Central de Venezuela, Caracas, 1020-A Venezuela.

^* To whom correspondence should be addressed. E-mail: rodrigez@bnl.gov

Nearly 95% of the hydrogen supply is produced from the reforming of crude oil, coal, natural gas, wood, organic wastes, and biomass (1), but this reformed fuel contains 1 to 10% CO, the presence of which degrades the performance of the Pt electrode in fuel cell systems (2). To get clean hydrogen for fuel cells and other industrial applications, the water-gas shift (WGS) reaction (CO + H₂OCO₂ + H₂) is critical. Current industrial catalysts for the WGS (mixtures of Fe-Cr or Zn-Al-Cu oxides) are pyrophoric and normally require lengthy and complex activation steps before usage (3). Au-CeO₂ and Au-TiO₂ nanomaterials have recently been reported to be very efficient catalysts for the WGS (3–5). This is remarkable because neither bulk Au (6) nor bulk ceria and titania are known as WGS catalysts (3, 4).

The nature of the active phase(s) in these metal/oxide nanocatalysts and the WGS reaction mechanism are still unclear. For example, the as-prepared Au-CeO₂ catalysts contain nanoparticles of pure gold and gold oxides (AuO_x) dispersed on a nanoceria support. Each of these gold species could be in the active phase (3–5), and the ceria support may not be a simple spectator in these systems (7). Although pure ceria is a very poor WGS catalyst, the properties of this oxide were found to be crucial for the observed activity of the Au-CeO₂ nanocatalysts (4, 6, 8). Several studies dealing with metal/oxide powder catalysts and the WGS reaction indicate that the oxide plays a direct role in the reaction (3, 4, 9, 10), but because of the complex nature of these systems, there is no agreement on its role. Thus, we performed a series of experiments to test the chemical and catalytic properties of CeO₂ and TiO₂ nanoparticles (NPs) dispersed on a Au(111) template, as inverse model catalysts. Results of density-functional calculations point to a very high barrier for the dissociation of H₂O on Au(111) or Au(100) (11), which leads to negligible activity for the WGS process. Even gold NPs cannot dissociate water and catalyze the WGS reaction (11).

Part of the experiments described below were carried out in ultrahigh-vacuum (UHV) chambers that have attached a batch reactor (12) or have capabilities for scanning tunneling microscopy (STM) (13). High-resolution x-ray photoelectron spectroscopy (XPS) spectra, probing only the near-surface region in the oxide/gold systems (5, 14), were acquired at the U7A beamline of the National Synchrotron Light Source. To prepare the TiO₂/Au(111) surfaces, Ti atoms were vapor deposited on a gold substrate covered with NO₂ at 100 K (15). The temperature was then raised to 700 K and the TiNO_x particles were transformed into TiO₂. STM images indicated that this methodology produces flat NPs of TiO₂ exhibiting a combination of rutile and anatase phases (15). The NPs of ceria were prepared according to two different procedures. In the first one, labeled CeO₂-I here, alloys of CeAu_x/Au(111) were exposed to O₂ (5 x 10^–7 torr) at 500 to 700 K for 5 to 10 min (13, 14). The CeO₂ NPs grew dispersed on the herringbone structure of Au(111) (Fig. 1A) and had a rough three-dimensional structure that did not exhibit any particular face of ceria (Fig. 1B). In the second procedure, labeled CeO₂-II here, Ce was vapor deposited onto Au(111) under an atmosphere of O₂ (5 x 10^–7 torr) at 550 K and then heated to 700 K. In these cases, the CeO₂ NPs grew preferentially at the steps between the terraces in the Au(111) substrate (Fig. 1C) and displayed regions with a CeO₂(111) orientation. When there was a deficiency of oxygen, groups or clusters of O vacancies were seen in STM images (Fig. 1D). This result is similar to that found for bulk CeO_2-x(111) (16). High-resolution XPS showed that Au atoms do not incorporate into the ceria lattice (14). Gold 4f spectra taken at photon energies of 240 to 380 eV, probing only two to three layers near the surface, showed the absence of the features expected for Au cations incorporated into ceria (5).

Fig. 1. (A) STM image taken after oxidizing a Ce-Au(111) alloy in O2 at 550 K and subsequent annealing to 690 K. Size: 200 nm by 200 nm; imaging parameters: –1.970 V and 0.03182 nA. (B) STM image also taken after oxidizing a Ce-Au(111) alloy in O2. Size: 30 nm x 25 nm; imaging parameters: –1.709 V and 0.02761 nA. (C) STM image acquired after depositing Ce in an atmosphere of 1.5 x 10–7 torr of O2 at 550 K followed by annealing to 700 K. Size: 200 nm by 200 nm; imaging parameters: 2.463 V and 0.02597 nA. (D) STM image also recorded after depositing Ce in an atmosphere of 1.5 x 10–7 torr of O2 at 550 K followed by annealing to 700 K. Size: 15 nm by 15 nm; imaging parameters: 0.806 V and 0.04404 nA [View Larger Version of this Image (172K GIF file)]

In Fig. 2, the WGS activity of a series of TiO₂/Au(111), CeO₂-I/Au(111), and CeO₂-II/Au(111) surfaces is shown. The fraction of the Au(111) substrate covered by the TiO₂ or CeO₂ particles was determined by means of ion-scattering spectroscopy (5, 7). In the kinetics measurements, the sample was transferred from the UHV chamber to the batch reactor at 300 K, then the reactant gases were introduced (20 torr of CO and 10 torr of H₂O) and the sample was rapidly heated to the reaction temperature (575 K). Product yields were analyzed by a gas chromatograph (17). The amount of product was normalized by the active area exposed by the front of the sample. The sample holder was passivated by extensive sulfur poisoning (exposure to H₂S) and had no catalytic activity (6). XPS spectra showed that there was no migration of S from the sample holder to the oxide/gold surfaces. The values reported in Fig. 2 were acquired after 5 min of reaction. In our batch reactor, a steady-state regime for the production of H₂ and CO₂ was reached after 1 to 2 min of reaction time. The kinetics experiments were done in the limit of low conversion (<5%).

Fig. 2. Production of hydrogen during the WGS reaction on a Au(111) surface partially covered with ceria or titania. The ceria NPs were prepared according to two different methodologies denoted as CeO2-I and CeO2-II (see text). Each surface was exposed to a mixture of 20 torr of CO and 10 torr of H2Oat 573 K for 5 min. [View Larger Version of this Image (19K GIF file)]

The data in Fig. 2 show a substantial increase in catalytic activity at small coverages of CeO₂ and TiO₂. Although clean Au(111) is not catalytically active, gold surfaces that are 20 to 30% covered by the oxide NPs had activities per unit geometric area similar to or larger than those of Cu(111) or Cu(100) under similar conditions (6, 17), noting that Cu is the best known metal catalyst for the WGS (11, 17, 18).

In Fig. 2, the catalytic activity of the surface reaches a maximum, then decreases and disappears once there is no Au exposed and multi-layers of ceria or titania cover the metal substrate (13–15). As we discuss below, small oxide NPs exhibit the highest chemical reactivity toward CO and H₂O. In general, the effects of ceria on the catalytic activity are greater than those of titania, with TiO₂/Au(111) being as active as Cu(100). In Fig. 2, we compare the relative WGS activities of the CeO₂-I/Au(111) and CeO₂-II/Au(111) surfaces. The procedure followed to prepare the CeO₂-I/Au(111) surfaces yielded a greater dispersion for the oxide NPs on the gold template (compare Fig. 1, A and C), and the NPs had a more disordered structure (compare Fig. 1, B and D). These morphological differences could account for a superior catalytic activity for the CeO₂-I/Au(111) systems.

After collecting the kinetics data, the gases were pumped out from the reaction cell and the surfaces were characterized with XPS, which showed adsorbed CO_x groups with a C 1s binding energy of 289.6 to 289.9 eV. This binding energy matches well those found for formate (HCOO) and carbonate (CO₃) groups bonded to oxides (19, 20). The XPS data also indicated a lack of oxidation of the Au substrate. The catalysts exhibited Au 4f spectra that were practically identical to those of Au(111) and very different from those expected for AuO_x species or Au incorporated into the ceria lattice (4, 5). The lack of oxidation of the Au substrate seen in the XPS data are consistent with in situ measurements of near-edge x-ray absorption spectroscopy for high–surface area catalysts (3, 5), which show that Au⁺ species are not stable under typical WGS conditions.

For the surfaces that have the highest activity in Fig. 2 (the fraction of Au covered < 30%), the Ce 3d and Ti 2p core-level XPS spectra showed a significant reduction of the oxides upon exposure of CeO₂/Au(111) and TiO₂/Au(111) to the reactants of the WGS. After curve-fitting (7, 14) the corresponding Ce 3d and Ti 2p spectra, we found that 14 to 17% of the O atoms in the CeO₂ and TiO₂ NPs were removed to produce partially reduced ceria and titania. The degree of reduction of the CeO₂ and TiO₂ NPs substantially decreased when the fraction of gold covered by the oxide increased above 35%. As an example, we show in Fig. 3 the percentage of O vacancies in ceria after performing the WGS on the CeO₂-I/Au(111) catalysts: The systems with a high catalytic activity also have a significant concentration of O vacancies and associated Ce³⁺ cations.

Fig. 3. (Left) Ce 3d XPS spectra taken before (trace a) and after (trace b) exposing a CeO2-I/Au(111) surface to the WGS reaction. The fraction of the Au surface covered by ceria was 30%. For comparison, we show the corresponding spectrum for a Ce2O3/Au(111) system. (Right) Percentage of O vacancies in the CeO2-I/Au(111) catalysts after the WGS reaction (20 torr of CO, 10 torr of H2O at 573 K for 5 min). The amount of oxygen removed from the ceria was calculated by curve-fitting (5, 7, 14) the corresponding Ce 3d spectra and obtaining the ratio of Ce3+/Ce4+ cations. [View Larger Version of this Image (16K GIF file)]

The high catalytic activity for low coverages of ceria and titania can be attributed to special chemical properties of the oxide NPs and cooperative effects at oxide-metal interfaces. However, it is very difficult to quantify the number of active sites in the oxide/gold catalysts of Figs. 1 and 2. The reaction likely involves O vacancies that are near oxide-gold interfaces. Because O vacancies tend to form groups or ensembles (Fig. 1D) (16, 21), uncovering in this way the metal substrate, oxide-gold interfaces can exist within the oxide NPs. We estimate that the best catalysts in Fig. 2 have a turnover frequency (TOF) that is 40 to 50 times as large as that of Cu(100) (22).

Using density-functional theory (DFT) (11, 23, 24), we investigated the WGS reaction on Au(100) or Au(111), a free Ti₂O₄ cluster [see structure in (25)], a free TiO₂ single chain, and over a model TiO₂/Au(111) catalyst that contains chains of TiO₂ over the gold substrate in a (3 x 1) array (Fig. 4). This model catalyst exposes not fully coordinated Ti centers, as expected for a TiO₂ NP, and allows the study of the oxide-metal interface. A very high barrier for the dissociation of water on Au(111) or Au(100) was seen (Fig. 4), but once OH formed, subsequent steps for the WGS process occurred readily on the gold substrate. Indeed, experimental studies show that CO adsorbs and is chemically active on gold surfaces first covered with oxygen or other chemical species (26, 27). On a free Ti₂O₄ cluster or on a nonsupported TiO₂ single chain, the dissociation of water is not difficult, and is even easier than on Cu(100) (Fig. 4), but the reaction of the formed OH with CO leads to the formation of a stable formate species that prevents the production of H₂ and CO₂.

Fig. 4. (A) Calculated reaction profile for the WGS on Au(100), a free Ti2O4 cluster, and TiO2/Au(111) model catalysts. Transitions states are denoted as TS1, TS2, and TS3. (B) Optimized structures for the different steps of the WGS on TiO2/Au(111). Large yellow spheres, Au; large gray spheres, Ti; small red spheres, O; small white spheres, H; small dark gray spheres, C. Adsorbed species are denoted by asterisks (*). [View Larger Version of this Image (20K GIF file)]

The DFT calculations for the model TiO₂/Au(111) catalyst show a system that can readily perform the WGS process (Fig. 4). The reaction pathway with the minimum-energy barriers involves the following steps:

(1)

(2)

(3)

(4)

(5)

(6)

The adsorption and dissociation of water take place on the oxide, whereas CO adsorbs on sites of the gold substrate located nearby (bifunctional catalyst). All of the subsequent steps occur at an oxide-metal interface. The DF calculations show that the activation energy for the dissociation of water on TiO₂/Au(111), 0.6 eV, is also much smaller than on Cu(100), 1.1 eV (11), so TiO₂/Au(111) should be a better WGS catalyst than Au(100) and Cu(100), as found above. The intermediate that precedes the formation of CO₂ and H₂ in the WGS process is a HOCO species. CO_x species were observed experimentally on the surface of the catalysts after the WGS, and they could be simple spectators when strongly bound to the oxide nanoparticles.

Our results imply that the high performance of Au-CeO₂ and Au-TiO₂ catalysts in the WGS (4, 6) relies heavily on the direct participation of the oxide-metal interface in the catalytic process. The oxide helps in the dissociation of water, a reaction that extended surfaces and NPs of gold cannot perform (11). Experiments in our laboratories have verified that TiO_2-x/Au(111) and CeO_2-x/Au(111) easily dissociate water, and no decomposition of this adsorbate is seen when no O vacancies exist in the oxide nanoparticles (14). Exposure of small coverages of TiO₂ and CeO₂ to CO at 575 K leads to the appearance of O vacancies in the oxide NPs, and these systems become active for the dissociation of water. For the WGS, it is critical that the properties of the oxide facilitate H₂O dissociation, and we have found that this is the case for NPs of CeO_2-x, TiO_2-x, MoO_3-x, and ZnO_1-x.

Previous studies indicate that overlayers of Au can be catalytically active for the oxidation of CO, if they are nanosized in one dimension or interact strongly with an oxide support (28). In contrast, the situation for the WGS on TiO_2-x/Au(111) and CeO_2-x/Au(111) takes advantage of the moderate chemical activity of bulk gold by coupling it to that of a more reactive oxide material.

References and Notes

 

转载本文请联系原作者获取授权，同时请注明本文来自戴启广科学网博客。
链接地址：https://blog.sciencenet.cn/blog-3913-12639.html

上一篇：【Science 】碳固定第五种途径——A Fifth Pathway of Carbon Fixation
下一篇：【催化牛人简介】有机催化--Benjamin List