Motivation

Metal-metal oxide (M-MO) catalysts are central to petrochemicals, fine chemicals, pharmaceuticals,  and biomass upgrade reactions^1–5 owing to cooperative catalysis among multiple active sites. Inverse M-MO (M = Pt, Ir, Rh, etc.; MO = WO_x, MoO_x, ReO_x, etc.) catalysts where the oxide is on top of a metal core have demonstrated superior activity and selectivity than single metals or oxide catalysts for converting biomass-derived molecules to high-value C₃-C₆ terminal diols and aromatics via ring-opening or hydrodeoxygenation.^1–6 These multifunctional catalysts contain metal, redox, Brønsted, and Lewis acid sites. Establishing structure-activity relationships for these catalysts is complicated by the intimate coupling of Brønsted acidity and catalyst reducibility and their dynamic nature. Specifically, as the hydrogen availability increases for in situ Brønsted acid formation, partial reduction of the oxide also occurs, leading to redox centers (vacancy formation).⁷ Understanding the dynamics and coupling of multiple active sites under various reaction environments is essential to design better catalysts and processes.

Our Approach

We employ an integrated approach that combines multiscale modeling, advanced catalyst synthesis, in situ spectroscopies, reaction kinetics, and expert knowledge. We chose PtWO_x/C inverse catalyst synthesized by atomic layer deposition (ALD) as a showcase. This catalyst contains a near-monolayer of WO_x on the surface of Pt.

Modeling. We tested various oxide clusters on Pt(111) and found W₃O₉ to have a characteristic W L₃ X-ray absorption near edge structure (XANES) feature similar to in situ XANES spectra acquired at the synchrotron at the Argonne National Laboratory and Brookhaven National Laboratory (Figs. 1a-d). We took W₃O₉ as a reasonable approximation for computational work.

Fig. 1 | Model structures and phase behavior. a, WO₃ ­monolayer on Pt(111). b, W₃O₉ cluster on Pt(111). Color code: Blue, Pt; Grey, W; Red, O. 1R indicates 1 ALD cycle. c, Simulated W L₁ XANES spectra of model structures. d, Experimental W L₁ XANES spectra of WO₃ standard, WO_x/C-1R, and PtWO_x/C-1R. Reduction at 400 °C. e, ab initio phase diagram for the W₃O_x/Pt(111) model. Each square represents the most abundant state under the corresponding conditions. Color code: Black square, W₃O₉; Gray square, W₃O₅; Orange square, W₃O₇H₁; Red square, W₃O₇H₃. f, Microkinetic heatmap of Brønsted acid site density (number of OHs per W) at long times.

We predicted the inverse-catalyst equilibrium composition of various sites and the dynamics of their formation under working conditions. Specifically, we calculated the energies for H spillover, WO_x reduction, oxidation, WO_x hydration, and dehydration. The model predicts that the kinetics of WO_x reduction and protonation depends strongly on the oxidation state of WO_x, hydrogen and water concentrations, and temperature. Ab initio phase diagrams provide the acid site densities. However, since each elementary step proceeds at a different time scale, a static model is insufficient in predicting the densities of the active sites. We developed state-based microkinetic models to capture the dynamics of inverse catalysts. Depending on the reaction environment, intermediate (non-equilibrated) states can be substantially more active than the thermodynamically stable states.

Exposing the catalyst dynamics with probe molecules. To check the robustness of our model and to probe the dynamics of Brønsted acid sites, tert-butanol dehydration was performed over the PtWO_x/C inverse catalyst in a gas-phase continuous flow reactor. We found that by changing the pretreatment conditions and reaction cofeed, the activity of the inverse catalyst can be tuned by 1-2 orders of magnitude (Figs. 2a-d) and changes with time on stream. The effects of water, hydrogen, and pre-treatment temperature are in excellent qualitative agreement with our models.

Fig. 2 | Probing active sites with alcohol probe reactions. a, reaction of tert-butanol. Catalysts reduced in 0.1 bar H₂ for 1 h, carrier gas = 100 mL/min N₂/He (1:1) mixture. b, modulating surface OHs with annealing and water. Annealing time = 1 h, steaming time = 1 h with water partial pressure = 0.1 bar. c, effect of hydrogen partial pressure on the dehydration activity of fresh PtWO_x/C-1R. d, effect of oxidation state on the in situ formation of Brønsted acid sites by hydrogen. Reaction conditions for (a)-(d): 140 °C, 4 mol% tert-butanol, 10 mg catalyst. The dehydration rate accounts for the sum of isobutene and isobutane.

Tuning activity by periodic pulsing of H₂. Inspired by the modeling and probe chemistry insights, we designed a periodic H₂ pulsing strategy to increase the reaction rate by leveraging the dynamics of WO_x (Fig. 3a). Fig. 3b shows time on stream data for a reduced catalyst. Periodic H₂ pulsing allows the catalyst to reach alternate states with a much higher acid site density. The average dehydration rate over the last 4 h is about one order of magnitude higher than that with a constant hydrogen cofeed (Fig. 3c). One may expect more than one order of magnitude increase in catalyst activity upon optimizing the frequency of periodic pulsing. 

Fig. 3 | Impact of hydrogen pulsing on reaction rate. a, Schematic of periodic hydrogen pulse for sustaining high Brønsted acid site density. b, Time on stream data of the hydrogen pulse cofeed experiment. PtWO_x/C-1R is pre-reduced at 400 °C in 0.1 bar H₂ for 1 h prior to reaction. Reaction conditions: 140 °C, 4 mol% tert-butanol, 10 mg catalyst, carrier gas = 100 mL/min H₂/N₂ mixture (0.1 bar H₂) or N₂. c, Average dehydration rates from 10 h to 14 h.

Outlook

Our work demonstrates the power of integrating multiscale modeling, reaction kinetics, and in situ characterization to understand complex catalyst dynamics under reaction conditions. It paves a viable way for controlling this intriguing and important class of catalysts. We envision that catalyst performance enhancement would be achievable over complex core-shell M-MO catalysts for many other key reactions.

References

1. Védrine, J. C. Heterogeneous catalysis on metal oxides. Catalysts 7, 341 (2017).
2. Tomishige, K., Nakagawa, Y. & Tamura, M. Selective hydrogenolysis and hydrogenation using metal catalysts directly modified with metal oxide species. Green Chem. 19, 2876–2924 (2017).
3. Chia, M. et al. Selective hydrogenolysis of polyols and cyclic ethers over bifunctional surface sites on rhodium-rhenium catalysts. J. Am. Chem. Soc. 133, 12675–12689 (2011).
4. Wang, C. et al. Mechanistic Study of the Direct Hydrodeoxygenation of m-Cresol over WOx-Decorated Pt/C Catalysts. ACS Catal. 8, 7749–7759 (2018).
5. Wang, C. et al. A Study of Tetrahydrofurfuryl Alcohol to 1,5-Pentanediol Over Pt – WO x / C. Catal. Letters 148, 1047–1054 (2018).
6. Fu, J. et al. C-O bond activation using ultralow loading of noble metal catalysts on moderately reducible oxides. Nat. Catal. 3, 446-453 (2020).
7. Rasmussen, M. J. & Medlin, J. W. Role of tungsten modifiers in bimetallic catalysts for enhanced hydrodeoxygenation activity and selectivity. Catal. Sci. Technol. 10, 414–423 (2020).