ARTICLES Dynamic in situ observation of rapid size and shape change of supported Pd nanoparticles during CO/NO cycling MARK A. NEWTON 1 *, CAROLINA BELVER-COLDEIRA 2 , ARTURO MART ´ INEZ-ARIAS 2 AND MARCOS FERN ´ ANDEZ-GARC ´ IA 2 * 1 The European Synchrotron Radiation Facility, 6, Rue Jules Horowitz, BP-220, Grenoble, France 2 Instituto de Catalisis y Petroleoqu´ ımica, CSIC, C/Marie Curie 2, 28049 Madrid, Spain *e-mail: newton@esrf.fr; mfg@icp.csic.es Published online: 27 May 2007; doi:10.1038/nmat1924 Understanding and improving the behaviour of supported precious-metal catalysts for a vast array of environmentally and economically important processes is a central area of research in catalysis. The removal of toxic gases such as CO and NO, without forming others (such as N 2 O), is particularly important. By combining energy-dispersive extended X-ray absorption fine-structure spectroscopy with a vibrational spectroscopy (infrared) and mass spectrometry, at high time resolution, in a single in situ experiment, we dynamically observe and quantify CO-, and subsequent NO-, induced size and shape changes of Pd nanoparticles during CO/NO cycling. In doing so we demonstrate a novel, non-oxidative redispersion (for example, an increase in metal surface area) mechanism, and suggest a model to bridge the structural and reactive functions of supported Pd catalysts. Supported catalysts based on precious metals, such as palladium, platinum and rhodium, are used in a vast array of processes of environmental, practical and economic importance. These range from the reforming and cracking of hydrocarbon feedstocks to the use of such catalysts in the abatement of noxious gas emissions from petrol- or diesel-driven engines. Associated with these applications are many reasons to improve catalyst performance, such as to protect the environment, minimize the use of such expensive metals—through lower loadings, increased catalyst longevity or recyclability—or to design catalysts that can work under ever more demanding process conditions. A long-standing problem is the tendency of nanoscale particles of such elements to coalesce (‘sinter’) into much larger aggregates under process conditions 1 . This leads to deactivation of the catalyst caused by the reduction of the surface area of the precious metal available to the reactant species. In addition, the chemistry exhibited by these systems is often highly sensitive to the domain size of the active metal itself 2 . Therefore, sintering can also result in a catalyst exhibiting a different behaviour to that desired. It is known that such sintering processes can be reversed, at least in part, through the application of high-temperature oxidation methods and subsequent re-reduction 3–7 . However, this is not a panacea and, to be effective, requires the presence of chlorine (Cl) either in the catalyst formulation or in the gas phase. Ideally, the use of Cl should be avoided entirely, both on environmental and practical grounds. Although there has been considerable research into sintering and redispersion phenomena 3–7 , and adsorbate-induced morphological variation in metal nanoparticles, this has generally been limited to relatively static assessments of the systems under study; these range from the ex situ transmission electron microscopy (TEM) work of Harris 8 (refaceting of Pt particles after exposure to H 2 S) to in situ TEM studies of shape change in Cu particles of about 50 ˚ A in diameter under H 2 , CO and H 2 O (and mixtures thereof) 9 . The most precise observations of sintering (though not redispersion) have been made in model supported-Pd systems, using scanning tunnelling microscopy 10–14 . However, though elegant, this approach currently lacks the time resolution to see ‘inside’ all but rather slow processes. It also requires the use of planar model systems generally (though not always 10 ) maintained in an ultrahigh-vacuum environment, rather than catalysts operating under more realistic conditions. In the world of auto-exhaust catalysis, this can mean situations wherein the redox potential of the feedstock can change very rapidly indeed (between, for instance, feedstocks containing differing levels of CO and NO: so-called ‘Lambda’ cycling) 15 . Nonetheless, scanning tunnelling microscopy has shown in vacuo sintering by an Ostwald ripening mechanism of Pd particles commencing above 700 K (refs 12,13), CO-induced Pd mobility and sintering on planar SiO 2 substrates 11 , two routes to CO oxidation based on metallic and oxidic Pd (ref. 10) and that Pd nanoparticle oxidation can exhibit a significant size dependence 14 . Furthermore, recent density functional theory calculations 16 concerning very small (Pd 4 −Pd 9 ) Pd clusters supported on MgO have indicated that a low-temperature route to oxidation in O 2 may be possible. However, so far direct and real-time in situ observation of sintering and redispersion has not been achieved. Therefore, the detailed structural-dynamic information that could provide the basis for fully understanding these phenomena and, as a result, controlling them for practical application, has evaded elucidation. We have synchronously applied energy-dispersive extended X-ray absorption fine-structure (EXAFS) spectroscopy, diffuse reflectance infrared spectroscopy (DRIFTS) and mass spectrometry nature materials ADVANCE ONLINE PUBLICATION www.nature.com/naturematerials 1