On the Impact of Solvation on a Au/TiO Nanocatalyst in Contact with Water
Water, the ubiquitous solvent, is also prominent in
forming liquid-solid interfaces with catalytically active surfaces,
in particular with promoted oxides.
We study the complex interface of a gold nanocatalyst,
pinned by an F–center on titania support,
The ab initio simulations uncover
details of solvent-induced charge rearrangements at the metal particle.
Water is found to stabilize charge states
differently from the gas phase as a result of
structure-specific charge transfer from/to the solvent, thus
The metal cluster is shown to feature both “cationic”
and “anionic” solvation, depending on fluctuation and
polarization effects in the liquid,
which creates novel active sites.
These observations open up an avenue toward “solvent engineering”
in liquid-phase heterogeneous catalysis.
Highly dispersed gold nanoparticles supported on oxides have been shown to catalyze a number of important reactions, including low–temperature oxidation and the water gas shift reaction haruta (); boccuzzi (). Reducible oxides, in particular titania (), are ideal catalytic supports diebold (). The size of the gold particles substantially affects the catalytic activity, suggesting the key importance of metal/support interactions on a nanometer scale valden (). Reactions, and in particular oxidation, are believed to occur at specific active sites at the / interface liu1 (); hammer (); yates (); behm (). Although much is known regarding / catalytic activity in the presence of a gas phase, the complexity increases steeply when solvent is included.
Liquid–solid interfaces as such are relevant to many industrial applications of great significance, such as (photo-)catalysis, solar cells, gas sensors, or biocompatible devices. In heterogeneous catalysis, it has been shown that the presence of water increases the observed rate of CO oxidation rao (); date (). The degree of rate enhancement depends on the type of support used. In particular for the case of Au/TiO catalysts it has been shown that their activity at about 3000 ppm HO is so high that full conversion of CO is reached date (). Thus, the Au/TiO surface displays a pronounced catalytic activity toward the water-gas-shift (WGS) reaction. This fundamental reaction represents a key reaction to produce extra H fuel from steam reforming, which is reversible and exothermic, according to the following reaction: CO + HO CO + H. The so–called carboxyl and redox mechanisms have been proposed for the WGS reaction on metal/oxide surfaces; in the former mechanism the CO species reacts with terminal hydroxyl groups, whereas in the latter CO reacts with an O atom from OH dissociation or from the oxide support. In both mechanism proposed the starting point is a HO molecule that is initially adsorbed on the metallic cluster with its oxygen atom being attached to a metal atom.
Moreover, even large (as opposed to nanoscale) gold particles, which are usually catalytically inert, show considerable oxidation activity at aqueous conditions sanch (); zhi (). Because of its high dielectric constant, water may actively participate in chemical reactions for instance by stabilizing ionic species, thus speeding up reactions at liquid-solid interfaces, which opens up novel avenues to improve the performance of traditional heterogeneous catalysis at the gas-solid interface. Recently, it has been shown that the selective oxidation of alcohols in aqueous phase over (, )–supported catalysts is facilitated by high pH conditions davis (). This water-based approach to heterogeneous catalysis offers a sustainable, environmentally benign, and cheap alternative to traditional processes that rely on both toxic and expensive inorganic oxidants and organic solvents corma (). Despite these promising experimental findings, much remains unknown at the molecular level about the impact of water as a solvent on the reactivity of these modern catalyst systems.
Several ab initio molecular dynamics (AIMD) marx-hutter-book () studies have been recently reported focussing on the fundamentals of the water-titania interface. For a water film on it has been shown that at least two distinct layers form: molecules in the first layer are sluggish and bind strongly to fivefold coordinated sites, whereas those in the second layer interact only weakly with the substrate and diffuse rapidly liu (), thus yielding a highly anisotropic interface jphyslett (). Moreover, water was not seen to dissociate at the coverages examined. Upon studying the thermodynamics of de/protonation of the rutile-water interface using free energy perturbation methods, a value of 0.6 eV was found for the dissociation free energy of bulk water on defect-free rutile sprik (). These studies thus show that the ideal surface is rather inert with respect to solvation by water. However, little is known about the interaction between water and defective or metal-supporting interfaces, being relevant to catalysis and industrial applications.
Here, we present large-scale AIMD simulations aimed to investigate solvent effects at a gold nanocluster pinned by an F–center on TiO(110) being in contact with liquid water. The aqueous solution is found to induce pronounced charge transfer and localization at the nanocatalyst–liquid interface and stabilizes structures different from the gas phase. The surface has been modeled by four –– tri–layer (42) supercell slabs separated by more than 15 Å. The most common point defects on the TiO(110) rutile surface are oxygen vacancies in the twofold coordinated rows diebold (). Therefore, a nanocluster was grown on TiO(110) where an F–center created by a surface vacancy acted as anchoring site for initial nucleation; note that this blocks the O defect to interact with water and thus prevents splitting of water molecules at this F–center. The lowest energy structure of adsorbed on a reduced rutile surface and employed in the present study is shown in Fig. 1(c)). In order to create the Au/TiO–water interface, the space between the slabs has been fully filled with 53 HO molecules (see Fig. 1(a)). It has been shown that is the smallest Au cluster on rutile with a measurable reaction rate for CO combustion combustion (). The model, although being a simplification of the nanometersized clusters on , appropriately mimics the active sites located at the nanogold/oxide interface where the oxidation process takes place hammer (); models1 ().
The binding of the Au cluster on the reduced (110) support entails a strong charge rearrangement at the metal/oxide contact (see Supporting Figure 1), the adsorption energy being eV. In the case of an isolated vacancy on the stoichiometric TiO(110) surface, the charge neutrality is maintained by the presence of two Ti ions thus creating an F–center. Spin density and bonding charge analyses reveal that the cluster, once adsorbed, leaves a reduced substrate with three Ti ions, and are transferred from the metal cluster to the oxide. This results into a slightly positively charged Au cluster supported by a reduced TiO(110) oxide surface.
In order to reveal the solvent-induced charge rearrangement at the metal-liquid contact, we have performed AIMD simulations of the Au/TiO nanocatalyst in aqueous solution. Additionally, corresponding gas phase simulations have been carried out to provide the solvent-free reference situation. In both cases the finite temperature dynamics preserves three reduced sites (see Supporting Figure 2). Snapshots of the AIMD simulations have been collected every ps thus generating a set of representative configurations for electronic structure analyses. In the following, we will refer to atoms of the supported nanocluster as top, middle and bottom sites, which are depicted as violet, black and yellow spheres in Fig. 1, respectively. A detailed investigation of the interaction between molecules and the bare oxide surface far from the supported metal cluster is out of the focus of this study , but we note in passing that our AIMD simulations are in agreement with those of liu (); jphyslett (); sprik () in the regions. In particular, water at the liquid–oxide contact is slow, strongly bonded to fivefold coordinated sites, and is not seen to dissociate.
The charge dynamics of the sites was extracted by computing, as a function of time, the Löwdin charges, see Fig. 1, where violet and black lines correspond to the charge evolution of top and middle atoms, respectively. As demonstrated by Fig. 1(b), two distinct charge patterns, corresponding to top and middle sites, can be clearly distinguished in aqueous solution; note that atoms in direct contact with TiO (yellow spheres in Fig. 1(a)) cannot be solvated by the liquid phase and thus do not exhibit a clear trend. The average charges on top and middle atoms are 11.028 0.006 and 11.220 0.006 , respectively (see Supporting Table 1 for statistical analysis). It is noted in passing that a very similar charge separation has been found for a gold cluster of different size and shape using an Au/TiO nanocatalyst in contact with water. Interestingly, at around 6 ps a considerable amount of charge is transferred to a specific top site due to solvent fluctuations, which reaches a value that is typical of middle atoms, see Fig. 1(b). In stark contrast, no such charge separation and fluctuation effects can be identified in the gas phase reference dynamics in Fig. 1(d). We conclude that the liquid phase promotes and stabilizes a net charge separation in the supported cluster, where top and middle atoms can be clearly distinguished on the basis of their electronic structure. As a result, water steers the sites towards either one or the other preferred charge value, depending on their specific location within the nanocluster and going hand in hand with different solvation patterns, vide infra.
To further characterize this puzzling process we have quantified the amount of charge transfer (CT) from atoms to the aqueous solution and back using Bader analysis bader (); bader1 (). For this purpose we have computed the difference between the Bader charges of the atoms in the fully solvated Au/TiO(110)–water system and those of the corresponding Au/TiO(110) gas phase system, where the latter is obtained upon removing the solvent while keeping all atomic positions fixed. As depicted in Fig. 2 we find that top sites transfer charge to the solvent, with an average CT of about 0.13 0.05 , while middle atoms attract about 0.06 0.03 per Au atom from water.
In view of these pronounced CT effects, we probe solvent-induced morphology changes of the metal cluster with respect to the gas phase. A set of trajectory configurations of the solvated nanocatalyst has been selected where water has been removed before quenching and optimizing the remaining Au/TiO system. In order to generate the proper gas phase reference, Au/TiO in the absence of water has been run at the same temperature of 450 K before applying the same quenching protocol; note that this is the relevant experimental temperature used for selective liquid–phase alcohol oxidation in aqueous solution using gold/titania catalysts. In case of the quenched solvated interface, the same local minima and no significant change in the Löwdin charges were observed as in the solvated state. However, the gas phase reference system yielded several local minima with significant changes of both the real-space structure and the Au charges (see Supporting Figures 3 and 4). These findings show that the liquid phase stabilizes structures and charge states within nanogold that are distinctly different from those observed in the gas phase, which eventually induces novel morphologies and active sites, respectively, due to solvation.
Let us now investigate the solvent-induced electronic charge redistribution at the nanocatalyst. In the following, we refer the electronic charge density of the solvated Au/TiO(110)–water system to that of the isolated one, Au/TiO(110), and the separated solvation water. We thus compute the difference, , between the charge densities of the combined system and the separated components while freezing all atomic positions.
A representative example of water-induced charge flow is depicted in Fig 3, where the positive (negative) component of the charge difference is shown by red (blue) isosurfaces. This reveals that charge depletion (blue) occurs mainly on top atoms, whereas additional electron charge (red) prefers to accumulate on middle sites. Furthermore, charge depletion and accumulation visible at the oxide-water contact can be traced back to the interaction between water and the TiO(110) oxide support. Here charge depletion occurs mainly on top of surface O atoms to which molecules point with their atoms, whereas charge accumulates close to five-fold coordinated atoms, being in turn coordinated by water O atoms. This is quantified by considering the charge redistribution perpendicular to the support, (see Fig. 3). Here is the sum of the positive and negative charge density components and integrated in planes perpendicular to the surface. This quantitative analysis reveals that, on the one hand, there is a net charge transfer of from top atoms to molecules (obtained by integrating within the shaded orange area in Fig. 3), leading to slightly positively charged atoms at the corresponding sites. On the other hand, there is a reverse charge flow of from water to middle atoms (see yellow area in Fig. 3), thus creating partially negatively charged atoms upon solvation. Similar results were obtained for all the representative structures sampled along the trajectory. This can be compared to the average negative and positive CT obtained from the Bader analysis. In total, this amounts to a structure-specific charge exchange between metal and water of and , in qualitative agreement with the charge difference analysis.
As a result, top and middle sites experience opposite charge flow upon interaction with the solvent, as clearly shown in Fig. 3, which connects to a peculiar solvation pattern.
We thus conclude by discussing the solvation pattern around the Au/TiO(110) nanocatalyst, see Fig. 4(a), which is found to feature both so-called “cationic” and “anionic” solvation depending on fluctuation and polarization effects. This is similar to what we have observed previously for a single (aq) cation and neutral atom, (aq), in aqueous bulk solution pccp (). In particular, the aqua ion forms a quasi-linear molecular structure, --, in which two water molecules are firmly bound through their oxygens to the site, as expected for a cation, yielding an - distance of Å. Upon adding an electron to to yield Au, we found that its solvation pattern changes distinctly. The latter features two distinct regions, supporting a Janus-type behavior. In one region a typical anionic solvation pattern prevails where two water molecules are pointing with their atoms toward the metal. In other regions water binds via its site to the (aq), thus giving rise to a typical cationic solvation pattern.
Around the nanocatalyst, two distinct solvation patterns corresponding to different charge states can be identified. The solvation around middle atoms consists of molecules pointing with their atoms toward the atoms like in an anionic solvation shell. In addition, several water exchange processes are observed as for (aq) in the bulk pccp (). At variance with those sites, the solvation around top atoms features a typical cationic solvation pattern, where a single molecule point to a top atom via its site. A projected density of states analysis reveals that the interaction between gold atoms and solvating water molecules results from the overlap between the Au(d) and O(p) states of the respective gold and oxygen sites. Interestingly, no exchange processes are observed in this case again in line with the (aq) cation case pccp (). As depicted in Fig. 1, the average Löwdin charges of top and middle sites are in qualitative agreement to those computed for a single gold cation (10.924 ) and neutral gold atom (11.210 ) in aqueous bulk solution. Based on this comparison, we are in a position to explain the sudden switch of the Löwdin charge involving a specific top atom and taking place at ps in Figs. 1(b) and 2. In Fig. 4(b) we plot the distances between the top atoms with typical “cationic” solvation pattern and the atom of its solvating molecule. This demonstrates that two atoms are always “bonded” to the the same molecule along the full trajectory, the time-averaged distances between and sites being 2.25 and 2.35 . The situation is very different at the third top site. This atom is initially cationically solvated yielding a – distance around 2.3 . At ps the molecule rearranges and points to the same gold atom with one of its atoms, a fluctuation which induces the change of charge state of this Au site discussed earlier. This reveals the intimate connection between the creation of excess charge at specific nanocluster sites and the preferential orientation of the interfacial water molecules. As discussed in the introduction for the carboxyl and redox mechanisms proposed for the WGS reaction on metal/oxide surfaces, the first step involves HO molecules that are initially adsorbed on the metal cluster such that their O atoms are coordinated to an Au site. Therefore our simulations suggest that top site Au atoms, which feature a typical “cationic” solvation pattern at the water-gold interface, are the most promising candidates for being the active site for CO oxidation via the WGS reaction, whereas “anionically” coordinated Au sites are expected to be chemically inert. The very same argument will hold for other liquid-phase catalytic reactions – thus providing a novel mechanism to create active sites at catalyst–water interfaces in much more general terms.
In conclusion, finite temperature simulations have been employed to investigate solvent-metal interactions on the gold/titania nanocatalyst in liquid water. Comparing our liquid phase to gas phase and bulk solvation reference data demonstrates that interfacial water alters both the structure and electronic properties of the supported metal cluster in a significant way, including the creation of active sites. The fundamental phenomena observed will have a profound impact on understanding the role of solvent in heterogeneously catalyzed liquid-phase reactions, where de- and re-solvation processes play a key role.
All calculations have been performed using spin–polarized pbe (), and ultrasoft pseudopotentials vanderbilt () as implemented in CPMD cpmd () and Quantum Espresso QE (). In line with our previous work cite-also-our-basic-PRB-2009 (); PRL (); PRB_ma (), the value of = 4.2 eV was adopted. We have carefully checked the charge distribution dependence on by re-computing the Löwdin charges of the top and middle atoms of the supported gold nanocluster in liquid water for selected snapshots along the trajectory shown in Fig. 1(b) using four different values and found no significant change (see Supporting Material for data).
The AIMD simulations marx-hutter-book () used Car–Parrinello propagation, the canonical ensemble was established at 450 K with a Nos-Hoover chain thermostat, and about ps trajectories have been generated for analysis (see Supporting Material for details).
We thank Martin Muhler for fruitful discussions. Partial financial support from Research Department “Interfacial Systems Chemistry” and the Cluster of Excellence RESOLV (EXC 1069) funded by Deutsche Forschungsgemeinschaft is gratefully acknowledged. Computational resources were provided by NIC (Jülich), Bovilab@RUB (Bochum), RV–NRW as well as PRACE (FERMI at Cineca).
Detailed descriptions of methods used for the calculations, model system and and much additional analyses. This material is available free of charge via the Internet at http://pubs.acs.org.
- Haruta, M.; Yamada, N.; Kobayashi, T.; Iijima, S. Gold Catalysts Prepared by Coprecipitation for Low-Temperature Oxidation of Hydrogen and of Carbon Monoxide. J. Catal. 1989, 115, 301-309.
- Boccuzzi, F.; Chiorino, A.; Manzoli, M.; Andreeva, D.; Tabakova, T. FTIR Study of the Low-Temperature Water-Gas Shift Reaction on Au/FeO and Au/TiO Catalysts. J. Catal. 1999, 188, 176-185.
- Diebold, U. The Surface Science of Titanium Dioxide. Surf. Sci. Rep. 2003, 48, 53-229.
- Valden, M.; Lai, X.; Goodman, D. W. Onset of Catalytic Activity of Gold Clusters on Titania with the Appearance of Nonmetallic Properties. Science 1998, 281, 1647-1650.
- Liu, Z.-P.; Gong, X.-P.; Kohanoff, J.; Sanchez, C.; Hu, P. Catalytic Role of Metal Oxides in Gold-Based Catalysts: a First Principles Study of CO Oxidation on TiO Supported Au. Phys. Rev. Lett. 2003, 91, 266102-266105.
- Wang, J. G.; Hammer, B. Role of Au in Supporting and Activating Au on TiO(110). Phys. Rev. Lett. 2006, 97, 136107-136110.
- Green, I. X.; Tang, W.; Neurock, M.; Yates Jr. J. T. Spectroscopic Observation of Dual Catalytic Sites During Oxidation of CO on a Au/TiO Catalyst. Science 2011, 333, 736-739.
- Widmann, D.; Behm, R. J. Active Oxygen on a Au/TiO Catalyst: Formation, Stability, and CO Oxidation Activity. Angew. Chem. Int. Ed. 2011, 50, 10241-10245.
- Ojifinni, R. A.; Gong, J.; Froemming, N. S.; Flaherty, D.; Pan, M.; Henkelman, G.; Mullins, C. B. Carbonate Formation and Decomposition on Atomic Oxygen Precovered Au(111). J. Am. Chem. Soc. 2008, 130, 11250-11251.
- Daté, M.; Okumura, M.; Tsubota, S.; Haruta, M. Vital Role of Moisture in the Catalytic Activity of Supported Gold Nanoparticles. Angew. Chem. Int. Ed. 2004, 43, 2129-2132.
- Sanchez-Castillo, M. A.; Couto, C.; Kim, W. B.; Dumesic, J. A. Gold-Nanotube Membranes for the Oxidation of CO at GasâWater Interfaces. Angew. Chem. Int. Ed. 2004, 43, 1140-1142.
- Shang, C.; Liu,Z.-P. Origin and Activity of Gold Nanoparticles as Aerobic Oxidation Catalysts in Aqueous Solution. J. Am. Chem. Soc. 2011, 133, 9938-9947.
- Zope, B. N.; Hibbitts, D. D.; Neurock, M.; Davis, R. J. Reactivity of the Gold / Water Interface During Selective Oxidation Catalysis. Science 2010, 330, 74-78.
- Corma, A.; Garcia, H. Supported Gold Nanoparticles as Catalysts for Organic Reactions. Chem. Soc. Rev. 2008, 37, 2096-2126.
- D. Marx and J. Hutter, Ab Initio Molecular Dynamics: Basic Theory and Advanced Methods, (Cambridge University Press, Cambridge 2009).
- Liu., L-M.; Zhang, C.; Thornton, G.; Michaelides, A. Structure and Dynamics of Liquid Water on Rutile TiO(110). Phys. Rev. B 2010, 82, 161415-161418(R).
- Kimmel, G. A.; Baer, M.; Petrik, N. G.; VandeVondele, J.; Rousseau, R.; Mundy, C. J. Polarization - and Azimuth - Resolved Infrared Spectroscopy of Water on TiO(110): Anisotropy and the Hydrogen-Bonding Network. J. Phys. Chem. Lett. 2012, 3, 778-784.
- Cheng, J.; Sprik, M. Acidity of the Aqueous Rutile TiO(110) Surface from Density Functional Theory Based Molecular Dynamics. J. Chem. Theory Comput. 2010, 6, 880-889.
- Lee, S.; Fan, C.; Wu, T.; Anderson, S. L. CO Oxidation on Au/TiO Catalysts Produced by Size-Selected Cluster Deposition. J. Am. Chem. Soc. 2004, 126, 5682-5683.
- Kim, H. Y.; Lee, M.; Henkelman, G. CO Oxidation Mechanism on CeO Supported Au Nanoparticles. J. Am. Chem. Soc. 2012, 134, 1560-1570.
- Kowalski, P. M.; Farnesi Camellone, M.; Nair, N. N.; Meyer, B.;;Marx, D. Charge Localization Dynamics Induced by Oxygen Vacancies on the TiO(110) surface. Phys. Rev. Lett. 2010, 105, 146405-146408.
- Farnesi Camellone, M.; Kowalski, P. M.; Marx, D. Ideal, Defective, and Gold-Promoted Rutile TiO(110) Surfaces Interacting with CO, H, and HO: Structures, Energies, Thermodynamics, and Dynamics from PBE+U. Phys. Rev. B 2011, 84, 035413-035430.
- R. F. W. Bader, Atoms in Molecules A Quantum Theory, (Oxford University Press, Oxford, 1990).
- Henkelman, G.; Arnaldsson, A.; Jnsson, H. A Fast and Robust Algorithm for Bader Decomposition of Charge Density. Comput. Mater. Sci. 2006, 36, 354-360.
- Farnesi Camellone, M.; Marx, D. Solvation of Au Versus Au in Aqueous Solution: Electronic Structure Governs Solvation Shell Patterns. Phys. Chem. Chem. Phys. 2012 , 14, 937-944.
- Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient Approximation Made Simple. Phys. Rev. Lett. 1996 , 77, 3865-3868; 78, 1396-1396(E).
- Vanderbilt, D. Soft Self-Consistent Pseudopotentials in a Generalized Eigenvalue Formalism. Phys. Rev. B 1990, 41, 7892-7895.
- J. Hutter et al., CPMD, www.cpmd.org.
- Giannozzi, P. et al., QUANTUM ESPRESSO: a Modular and Open-Source Software Project for Quantum Simulations of Materials. J. Phys.: Condens. Matter 2009, 21, 395502; Quantum Espresso, www.pwscf.org.
- Kowalski, P. M.; Meyer, B.; Marx, D. Composition, Structure, and Stability of the Rutile TiO(110) Surface: Oxygen Depletion, Hydroxylation, Hydrogen Migration, and Water Adsorption. Phys. Rev. B 2009, 79, 115410-115425.