Elsevier

Journal of Catalysis

Volume 261, Issue 2, 25 January 2009, Pages 137-149
Journal of Catalysis

CO oxidation catalyzed by gold supported on MgO: Spectroscopic identification of carbonate-like species bonded to gold during catalyst deactivation

https://doi.org/10.1016/j.jcat.2008.11.005Get rights and content

Abstract

MgO-supported gold prepared by adsorption of Au(CH3)2(acac) (acac is acetylacetonate) on partially dehydroxylated MgO was activated for CO oxidation catalysis by treatment in flowing helium at 473 K. X-ray absorption spectra showed that the activation involved reduction of the gold and formation of clusters (with an average diameter <10 Å) in which the gold was essentially zerovalent. During CO oxidation catalysis in a batch reactor, at least some of the gold was oxidized, as evidenced by the appearance of an Auδ+single bondCO band at 2151 cm−1 in the infrared (IR) spectrum. During operation in a flow reactor, the catalyst underwent deactivation, accumulating species such as carbonate and bicarbonate on its surface, as indicated by IR spectra. The accumulation of such species on the MgO support took place only during the initial period of operation of the catalyst, whereas the accumulation of such species on the gold continued throughout the operation, consistent with the inference that these species blocked catalytically active sites on the gold. The catalyst was reactivated by decomposition of these species by treatment in helium at 473 K. After three activation–deactivation cycles, the average diameter of the supported gold clusters had increased to about 30 Å, and the catalytic activity increased. Thus, the results provide a resolution of the separate effects on the catalytic activity of gold aggregation and accumulation of species such as carbonates and bicarbonates.

Graphical abstract

Accumulation of carbonate-like species on gold led to the deactivation of MgO-supported gold catalysts for CO oxidation reaction.

  1. Download : Download high-res image (55KB)
  2. Download : Download full-size image

Introduction

Samples consisting of gold dispersed on metal oxide supports have been investigated extensively as catalysts for CO oxidation, motivated in large measure by Haruta's discovery [1] that some of these catalysts are highly active, even at temperatures much less than room temperature. Notwithstanding the extensive research, there are still many questions about the nature of the catalytically active sites, the reaction mechanisms, and catalyst deactivation.

The performance of these catalysts is sensitive to the support [2], the catalyst pretreatment conditions [3], the water content of the reactant mixture [4], and components such as chloride, which is present in commonly used precursors such as HAuCl4 [5]. There is a lack of quantitative results characterizing the performance of supported gold catalysts, which is complicated by the catalyst deactivation.

Deactivation of supported gold CO oxidation catalysts has been ascribed variously to reduction and aggregation of the gold [6], [7], other morphological changes in gold clusters [8], and reduction of the support surface [9]. The most commonly suggested causes of deactivation are sintering of gold clusters [10] and accumulation of carbonate-like species [4], [11], [12], [13] (carbonates, bicarbonates, carboxylates, and formates). IR spectra of these species on catalyst surfaces are characterized by bands in the 1800–900 cm−1 region, including bands representing νasym(OCO), νsym(OCO), etc. [14].

These deactivation processes may occur simultaneously. For example, Konova et al. [15], [16] found evidence of carbonate accumulation and gold agglomeration during CO oxidation catalyzed by Au/ZrO2 and Au/TiO2, observing that the two processes had different effects: sintering of gold caused irreversible deactivation (but the effect was usually small, with most supported gold catalysts retaining their activities as long as the average gold cluster diameter remained within a range of, typically, 2–5 nm); the accumulation of carbonate-like species, on the other hand, may lead to significant activity loss by blocking of active sites, although the effect can be reversed by removal of these species by treatment in inert gas at temperatures up to 573 K.

Deactivation caused by carbonate-like species can also be reversed by other treatments, such as addition of moisture. Kung et al. [17], [18] investigated the roles of moisture and of H2 on Au/Al2O3 catalysts, finding that both were effective in converting carbonate species to reactive formate species (which the authors proposed to be reaction intermediates), with regeneration of the catalyst. Daté et al. [4], [13] drew a similar conclusion regarding the effect of moisture on Au/Al2O3 and Au/TiO2.

Evidence of build-up of carbonate-like species on catalyst surfaces has emerged from infrared (IR) spectroscopy. The most informative spectra are those characterizing working catalysts, as they may provide correlations between carbonate accumulation and catalytic activity. For example, Shubert et al. [19] claimed that IR spectra gave evidence of carbonate and carboxylate (single bondCOO) species during preferential oxidation of CO in a H2-containing atmosphere catalyzed by Au/α-Fe2O3. As the amounts of these surface species increased with time on stream, the authors inferred that they were responsible for the deactivation. Similarly, Daté et al. [13] observed bands in the carbonate region (at 1435 and 1230 cm−1) during CO oxidation catalyzed by Au/Al2O3. The change in intensity of the 1230-cm−1 band with time-on-stream correlated inversely with a measure of the catalytic activity for CO2 formation. When moisture was introduced into feed stream, both bands decreased in intensity and disappeared, and the catalytic activity increased, leading the authors to suggest that carbonate-like species near the active sites caused deactivation.

In work with Au/CeO2 [20], accumulation of carbonate-like species was indicated by IR bands at 1590, 1440, 1358, and 1270 cm−1. The authors [20] inferred that some of these species were on the support and not the gold—and thus not responsible for the activity loss—whereas the monodentate carbonate species (characterized by the 1358-cm−1 band), formed on the support in a reaction with hydroxyl groups and bicarbonate species, did cause deactivation.

To develop a better understanding of the roles of carbonate-like species on supported gold catalysts, especially when the supports are strongly basic and readily form such species from CO and from CO2 [14], [21], our goal was to assess where the species form and whether their formation is influenced by gold. Specifically, we sought to distinguish carbonate-like species on gold and on a strongly basic support; to determine their effects on catalyst performance; and to distinguish the role of these species from that of aggregation of the gold. Thus, we investigated the adsorption of CO and of CO2 separately on a basic support (MgO) and on Au/MgO. We report characterization of the surface species by IR spectroscopy and simultaneous measurements of the catalytic activity for CO oxidation, as well as characterization of fresh and used catalysts by extended X-ray absorption fine structure (EXAFS) spectroscopy to determine average gold cluster sizes, and by X-ray absorption near edge structure (XANES) spectroscopy to determine gold oxidation state(s). The results provide a basis for separating the effects of the formation of carbonate-like species, cluster growth, and modification of gold oxidation state(s) in the catalyst deactivation; the data demonstrate a correlation between carbonate-like species on gold and catalyst activity loss.

Section snippets

Materials

He (Airgas, 99.995%) and CO (Airgas, 10% mixture balanced in He, 99.999%) were purified by passage through traps containing reduced Cu/Al2O3 and activated zeolite 4A to remove traces of O2 and moisture, respectively. O2 (Airgas, 10% in He, 99.999%) was purified by passage through a trap containing activated zeolite 4A to remove traces of moisture. The MgO support (EM Science, 97%, 60 m2/g) was calcined in O2 at 673 K for 2 h, followed by evacuation at 673 K for 16 h (pressure < 10−2 Pa), and

EXAFS data analysis

Analysis of the EXAFS data recorded at the Au LIII edge was carried out with a difference file technique by use of the software XDAP [24]. The number of parameters used in fitting the data was justified statistically by the Nyquist theorem: n=(2ΔkΔr/π)+1, where Δk and Δr, respectively, are the ranges in k and r used in the data fitting (k is the photoelectron wave vector; r is the distance from the absorber Au atom) [25]. Criteria used to judge the appropriateness of a model tested in the data

Activity of Au/MgO catalyst for CO oxidation

Results of earlier work [22] showed that the as-prepared Au/MgO sample contained mainly site-isolated, mononuclear Au(III) complexes having a structure analogous to that of the precursor Au(CH3)2(acac), with the support surface being a bidentate ligand. The literature [28] indicates that the gold in these samples readily undergoes reduction and aggregation.

The activity of this sample was tested for CO oxidation catalysis at 303 K and atmospheric pressure, with a total feed flow rate of 100 mL

Interactions of CO and of CO2 with MgO

After adsorption of CO and CO2 on the MgO sample we have established formation of surface uni- and bidentate carbonates as well as of bicarbonates. All these species were observed by various authors with other MgO preparations [35], [39], [40], [41]. However, Smart et al. [35] and Di Cosimo et al. [40] did not observe bands near 1217 cm−1 during exposure of their samples to CO or CO2.

Daté et al. [13] observed a band at 1230 cm−1 during CO2 adsorption on Al2O3, which they inferred to originate

Conclusions

The Au/MgO catalyst made by bringing Au(CH3)2(acac) in contact with partially dehydroxylated MgO was not active initially for CO oxidation at room temperature, but the sample was activated by treatment in flowing helium at 473 K, during which the original mononuclear Au(III) species decomposed as methyl ligands were lost and the gold was reduced and aggregated. XANES and EXAFS characterizations indicate that the treated sample contained clusters of gold that was essentially zerovalent.

The

Acknowledgments

This work was supported by U.S. Department of Energy (Grant FG02-04ER15513), the Alexander von Humboldt Foundation (AvH fellowship to M.M.), and a NATO Collaborative Linkage Grant (PST.CLG 980289). K.H., E.I., and M.M. also acknowledge support from the Bulgarian Scientific Foundation (project VUX-303). We acknowledge the National Synchrotron Light Source (NSLS) and the Stanford Synchrotron Radiation Laboratory (SSRL) for access to beam time. We thank the staffs of NSLS and SSRL for assistance.

References (58)

  • M. Haruta et al.

    J. Catal.

    (1993)
  • M.M. Schubert et al.

    J. Catal.

    (2001)
  • F. Boccuzzi et al.

    J. Catal.

    (2001)
  • H.-S. Oh et al.

    J. Catal.

    (2002)
  • J. Steyn et al.

    Catal. Today

    (2007)
  • F. Moreau et al.

    Catal. Today

    (2006)
  • M. Azar et al.

    J. Catal.

    (2006)
  • M. Daté et al.

    Catal. Today

    (2007)
  • F. Boccuzzi et al.

    Surf. Sci.

    (1985)
  • P. Konova et al.

    Catal. Commun.

    (2004)
  • P. Konova et al.

    J. Mol. Catal. A

    (2004)
  • C.K. Costello et al.

    Appl. Catal. A

    (2002)
  • C.K. Costello et al.

    Appl. Catal. A

    (2003)
  • M. Vaarkamp et al.

    Physica B

    (1995)
  • G. Spoto et al.

    Prog. Surf. Sci.

    (2004)
  • J.I. Di Cosimo et al.

    J. Catal.

    (1998)
  • G. Busca et al.

    Mater. Chem.

    (1982)
  • F. Boccuzzi et al.

    Surf. Sci.

    (2000)
  • M. Manzoli et al.

    Appl. Catal. B

    (2004)
  • N. Weiher et al.

    J. Catal.

    (2006)
  • M. Daté et al.

    Angew. Chem. Int. Ed.

    (2004)
  • J.C. Fierro-Gonzalez et al.

    J. Phys. Chem. B

    (2004)
  • F. Moreau et al.

    Top. Catal.

    (2007)
  • K.Y. Ho et al.

    Gold Bull.

    (2007)
  • M.M. Shubert et al.

    J. Catal.

    (2004)
  • M.M. Shubert et al.

    Catal. Lett.

    (2001)
  • M. Moreau et al.

    Appl. Catal. A

    (2008)
  • L.H. Little

    Infrared Spectra of Adsorbed Species

  • J. Guzman et al.

    Angew. Chem. Int. Ed.

    (2003)
  • Cited by (0)

    View full text