High throughput discovery of ternary Cu-Fe-Ru alloy catalysts for photo-driven hydrogen production
Maya Bhat, Zoe C. Simon, Savannah Talledo, Riti Sen, Jacob H. Smith, Stefan Bernhard, Jill E. Millstone and John R. Kitchin, “High throughput discovery of ternary Cu-Fe-Ru alloy catalysts for photo-driven hydrogen production”, Reaction Chemistry & Engineering 8(7), 1738-1746 (2023). https://doi.org/10.1039/d3re00059a
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
Multimetallic nanoparticles can be more active for the hydrogen evolution reaction (HER) than their parent metals, but the space of metal identities, loadings and ratios is large, and conventional work that synthesizes, isolates and characterizes each particle before testing it is slow. The authors asked whether combinations of Cu, Ru and Fe, formed in situ as colloidal particles during light-driven HER, show synergy, and whether a high-throughput, design-of-experiments workflow can find the active binary and ternary compositions (Bhat et al. 2023). Cu is a poor HER metal on its own but was expected to modify neighboring metals; Ru was hypothesized to enhance Cu; Fe was chosen as a cheap, earth-abundant group-mate of Ru that had formed particles and been active in the group’s earlier screens. Cu is immiscible with both Ru and Fe in the bulk.
Approach
Reactions ran in the group’s home-built 108-well photoreactor, with H2 detected by a hydrogen-sensitive tape imaged every ten minutes over 1000 minutes. Each 440 µL well held an iridium photosensitizer ([Ir(Fmppy)2dtbbpy]PF6) and metal salts in DMSO, triethanolamine as sacrificial donor, and thiolated poly(ethylene glycol) (PEGSH, 1 kDa) as stabilizing ligand; CuCl, FeCl2 and RuCl3 were the precursors. Most plates were dispensed by a liquid-handling robot. Response surface and fractional factorial designs, managed with the in-house gespyranto package, were used to survey and then sequentially sample toward active regions. Activity was reported as the maximum µmol H2 in a well, chosen over rates because derivatives of noisy time series depend on smoothing and differencing choices. Since particle area and composition were unknown, values were not normalized per area; instead each plate was scaled by six internal-standard wells (Eosin Y with a cobalt glyoxime molecular catalyst, within-plate variation under 2 µmol) and by the average activity of Cu wells present on every plate. This lowered the effective detection limit from 2.0 to 1.1 scaled µmol H2, with an upper bound near 25 µmol set by tape saturation. Synergy was defined as the maximum H2 of a mixture minus the sum of the pure metal activities at the same concentrations. Active wells were examined by TEM, HAADF imaging and STEM-EDS mapping.
Key findings
Monometallics (0 to 0.8 mM, 425 data points from eight experiments) behaved as in earlier work: Ru was inactive and did not form clear discrete particles, Cu had low activity with discrete particles, and Fe was active, rising with concentration and leveling off above about 0.3 mM, forming iron sulfide particles. All three binaries were active. CuRu was strongly active despite both parents being weak, with consistently positive synergy; STEM-EDS showed core-shell particles with a CuRu core and Cu shell plus Cu-rich species. CuFe was active, with some negative synergies at higher total loading, which the authors suggest may come from electrons going to reduce Cu instead of producing H2; it contained separate Cu and CuFe particle populations. RuFe gave positive synergies that grew with Ru content and contained Ru and RuFe particles but no Fe-only particles, surprisingly given that Ru alone formed no clear particles.
The ternary space was mapped at 0.8 mM total metal. Activity was highest toward the center of the composition triangle at lower Fe and somewhat higher Cu fractions, and lowest near Ru-rich corners. The best single well contained 0.1 mM Cu, 0.2 mM Ru and 0.5 mM Fe, and the ternary showed positive synergies, outperforming the corresponding binaries. A highly active ternary well (0.4/0.1/0.3 mM Cu/Ru/Fe) contained mostly Cu particles with some CuFe particles, which had less Fe than in Ru-free CuFe samples. Because each active well held a distribution of particle compositions, the authors could not say which particles were responsible for the activity.
Significance
The study identifies CuRu, CuFe and RuFe colloids and a new CuRuFe ternary as active photo-driven HER cocatalysts, with CuRu standing out as a synergistic pair of individually poor metals. It shows how in situ particle formation, robotic plate preparation, internal-standard normalization and statistical experimental design can screen binary and ternary composition spaces quickly. It also documents a central limitation: the molar ratio of salts added to a well differs from the stoichiometry and ordering of the particles that form, and adding a third salt can change particle composition, so linking activity to a specific alloy structure requires further work.