Atomistic Thermodynamics Study of the Adsorption and the Effects of Water-Gas Shift Reactants on Cu Catalysts Under Reaction Conditions
Nilay İnoğlu and John R. Kitchin, “Atomistic Thermodynamics Study of the Adsorption and the Effects of Water-Gas Shift Reactants on Cu Catalysts Under Reaction Conditions”, Journal of Catalysis 261(2), 188-194 (2009). https://doi.org/10.1016/j.jcat.2008.11.020
This summary was written by a language model from the paper and has not been checked line by line. Rely on the published paper for details.
Question
Copper is the best pure metal for the low-temperature water-gas shift (WGS) reaction, and WGS and its reverse also run alongside methanol synthesis. Real feeds contain H2O, CO, CO2 and H2 together with trace H2S, so a working Cu catalyst might be partly oxidized, covered with CO, or poisoned by sulfur, while the large H2 pressure could strip S and O back off as H2S and H2O. Because supported particles expose several facets that may interact differently with the gas, adsorption could also change the particle shape. The paper asks which adsorbate structures are thermodynamically stable on Cu(111), (100) and (110) under realistic temperatures and pressures, and how those structures change the equilibrium shape of a Cu particle (INOGLU 2009).
Approach
The authors built a first-principles atomistic thermodynamics framework in the style of Reuter and Scheffler. The surface free energy of each structure is written as the clean-surface energy plus an adsorption term and an environment term, with adsorbate chemical potentials tied to gas-phase reservoirs: S to H2S minus H2, O to H2O minus H2, and CO to CO. Gas chemical potentials are treated as ideal gases using tabulated enthalpies and entropies, so each chemical potential maps onto a temperature and a partial-pressure ratio (for example H2S/H2). Adsorbate vibrational free energies (estimated at 2 to 6 meV/Ų over 10 to 650 K) and configurational entropy (at most about 4 meV/Ų) were neglected, so free energies reduce to DFT total-energy differences. Upper limits on each chemical potential are set by condensation of bulk phases, and bulk CuS and Cu oxide stability are marked on each diagram (INOGLU 2009).
The DFT calculations used DACAPO with PW91, ultrasoft pseudopotentials and a 340 eV cutoff; the Cu lattice constant was 3.64 Å (experiment 3.61 Å). Slabs of five, six and eight layers (about 9 Å thick) represented (111), (100) and (110), with 10 Å vacuum and estimated adsorption-energy uncertainty below 0.05 eV, or about ±4 meV/Ų in surface energy. S was studied in 43 structures (0.25 to 1.00 ML in hollow, bridge and atop sites), O in fcc sites and CO in atop sites over the same coverage range, with H and CO2 adsorption checked for comparison. Equilibrium particle shapes were obtained by Wulff construction with the WINXMORPH program (INOGLU 2009).
Key findings
S binds strongly to all three facets at low coverage and prefers the most highly coordinated sites. The authors note that S at such sites broadens the surface d band, lowering the d-band center and weakening binding with coverage (not seen for atop sites). The computed formation energy of bulk CuS is -0.40 eV, so above a S chemical potential of -0.40 eV the sulfide is favored. Below about -1.25 eV clean surfaces are most stable. Cu(111) is covered by 0.25 ML S in fcc sites between -1.25 and -0.40 eV, Cu(100) by 0.25 ML S in fourfold hollows between -1.85 and -0.40 eV, and Cu(110) by 0.50 ML S in fourfold sites from -1.65 eV (INOGLU 2009).
Avoiding S adsorption requires very small H2S/H2 ratios: on Cu(111) at 400 K with 1 ppm H2S, roughly 10³ atm of H2 would be needed. With 25 atm H2 and 1 ppm H2S, S-covered surfaces are more stable than clean or H-covered ones at 400 K, but the clean surface is most stable at 650 K, so high temperature and H2 pressure can regenerate a poisoned surface, at the risk of sintering. For O, a 0.25 ML phase is stable only in a very narrow window on (111) and a wider one on (100) before bulk oxide forms, while (110) favors a 0.75 ML partially oxidized surface. The calculated CO binding at 0.25 ML atop on Cu(111) is -0.65 eV, compared with an experimental -0.5 eV, and CO phases of 0.25 and then 0.50 ML are stable. H and CO2 phases were less stable than O phases at 400 K (INOGLU 2009).
For a representative WGS feed (50 atm total; 10% CO, 51% H2, 11% CO2, 28% H2O; 1 ppm H2S), the surfaces are partly covered by CO rather than O, and S-covered surfaces take over at higher H2S levels; mixed adsorbate phases were not computed. Because each facet’s surface energy depends differently on the S chemical potential, the Wulff shape changes from one dominated by (111) and (100) facets at very low S chemical potential to one dominated by (110) at high S chemical potential (INOGLU 2009).
Significance
The work shows quantitatively why Cu WGS catalysts are poisoned by ppm H2S at low temperature, sets out conditions under which hydrogen can regenerate them, and argues that reaction environments can reshape catalyst particles and so change the number and type of active sites. It carries the Reuter-Scheffler atomistic thermodynamics approach into a multicomponent WGS environment, and its brief d-band remarks are taken up in the companion study of S adsorption on Cu facets (INOGLU 2009).