Sulphur Poisoning of Water-Gas Shift Catalysts: Site Blocking and Electronic Structure Modification
Nilay İnoğlu and John R. Kitchin, “Sulphur Poisoning of Water-Gas Shift Catalysts: Site Blocking and Electronic Structure Modification”, Molecular Simulation 35(10-11), 936-941 (2009). https://doi.org/10.1080/08927020902833129
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 Cu/ZnO catalysts also carry out methanol synthesis, where WGS and its reverse run alongside. Many CO2 feedstocks contain sulphur impurities (for example trace H2S) that bind strongly to Cu and deactivate it even at very low concentrations. The usual picture of poisoning is that adsorbed S simply occupies active sites. This paper asks how S adsorption on the three low Miller index Cu facets depends on facet, site and coverage, and whether S also deactivates the surface by changing its electronic structure, so that poisoning is more than site blocking (İnoğlu 2009). Reconstructed sulphided Cu surfaces seen in experiments are explicitly left out of scope.
Approach
The authors used plane-wave DFT in DACAPO with Vanderbilt ultrasoft pseudopotentials, a 340 eV cutoff and the PW91 GGA functional. The computed Cu lattice constant is 3.64 Å (experiment 3.61 Å). Cu(111), (100) and (110) were modeled with five-, six- and eight-layer slabs respectively, each about 9 Å thick, with 10 Å of vacuum, S on one side, and the top two layers plus adsorbates relaxed to forces below 0.05 eV/Å. Monkhorst-Pack grids of 12 × 12 × 1 (1 × 1 cells) and 6 × 6 × 1 (2 × 2 cells) were used, giving an estimated uncertainty below 0.05 eV/S; a dipole correction was tested and found unnecessary. In all, 43 structures were computed at 0.25, 0.50, 0.75 and 1.00 ML in hollow (fcc and hcp on (111), fourfold on (100) and (110)), bridge and atop sites. Adsorption energies are reported per S atom relative to gas-phase H2S and H2, corresponding to dissociative H2S adsorption releasing H2. The surface d band was obtained by projecting Kohn-Sham states onto atom-centred orbitals, with the d-band center as the first moment and the width from the second moment, and the results were interpreted with the Hammer-Nørskov model and a rectangular d-band model (İnoğlu 2009).
Key findings
S binds most strongly to the most open facet and most weakly to the most close-packed one, in the order (110) > (100) > (111), and on every facet it prefers the most highly coordinated site. On Cu(111) fcc and hcp sites differ little, with fcc slightly more favorable, consistent with earlier DFT reports. Threefold hollow adsorption is preferred up to 0.50 ML; above 0.50 ML adsorption on Cu(111) becomes endothermic relative to H2S/H2, and atop adsorption is unfavorable even at 0.25 ML. On Cu(100) adsorption stays favorable on fourfold sites up to 1.00 ML but on bridge sites only at 0.25 ML. On Cu(110) bridge and atop S relaxed into neighbouring hollows at low coverage, so only high-coverage energies are reported for those sites, and the fourfold hollow is the most stable site over the whole range on both (100) and (110). Specific adsorption energies are given only in figures, which are not legible in the extracted text, so they are not reproduced here (İnoğlu 2009).
On all facets the S adsorption energy generally becomes weaker as coverage increases. Integrating the surface-projected d states gives about 9.9 d electrons per Cu atom at every coverage (fractional filling near 0.99), so the d-band filling is essentially conserved while the band broadens with coverage. At fixed filling a rectangular band model predicts that the square root of the second moment varies linearly with the d-band center with a slope of -1.16; the DFT data for all coverages and facets fall on a line with slope -1.09. Broadening therefore pushes the d-band center down, and for fcc, hcp and bridge sites on (111), hollow and bridge sites on (100), and fourfold sites on (110), the adsorption energy weakens almost linearly as the d-band center becomes more negative. Atop sites show no correlation, which the authors attribute to weak overlap between S orbitals and the surface d band. At 0.75 ML, S on bridge and atop sites of (100) and (110) forms dimers and trimers that stabilize the adsorbates and break the d-band correlation, since such correlations are expected only for geometrically similar structures. On Cu(110) the fourfold-site energy is unchanged between 0.25 and 0.50 ML despite slight d-band broadening, which the authors suggest reflects the large spacing between sites (İnoğlu 2009).
Significance
The study reinforces experimental evidence that Cu WGS catalysts are easily poisoned by ppm levels of H2S and gives a two-part mechanism: S occupies the most active, highly coordinated sites, and it also intrinsically deactivates the remaining metal by broadening and down-shifting the surface d band. Facet-dependent binding implies that sulphur may stabilize facets differently and alter particle morphology, building on the same authors’ atomistic thermodynamics study of Cu surfaces under WGS conditions. The constant-filling width-center relation used here is the same one developed for bimetallic surfaces and applied to coverage effects on Pd(111) (İnoğlu 2009).