UC Berkeley

Thermodynamic constraints in alkane dehydrogenation and dehydrocyclization can be mitigated by the scavenging of H2 through reactions with O2 and CO2, an approach that promises significant increases in product yields but requires a detailed mechanistic understanding of H2-scavenging reactions to deploy them effectively and minimize undesired side reactions. The simplicity of the molecular rearrangements involved in these reactions and the ubiquitous nature of such rearrangements in catalysis also make them an attractive choice to probe fundamental concepts in surface catalysis. In this study, H2-O2 and H2-CO2 reactions are investigated using a combination of kinetic, isotopic, and spectroscopic measurements to address the kinetically relevant steps and site requirements for these reactions on transition metal nanoparticles. The resulting mechanistic inferences are benchmarked against theoretical assessments using models of active surfaces with relevant adsorbate coverages. The experimental and theoretical evidence presented herein indicate that O-O bond activation steps are kinetically-relevant in H2-O2 reactions on Pt and O atom removal and C-O bond activation steps are kinetically-relevant in H2-CO2 reactions that form CO and CH4 on Ru nanoparticles. O-O and C-O bonds are activated both through direct reactions with ensembles of uncovered surface metal atoms and through H-assisted pathways that form H-containing intermediates prior to bond activation. Significant effects of adsorbate coverages (H atoms in H2-O2 reactions; CO in H2-CO2 reactions) on the energies of the bound species and transition states that mediate O-O and C-O bond activation were observed, rendering traditional Langmuirian treatments an incomplete description of the reaction dynamics. Such conclusions underscore the importance of describing reaction energetics using theoretical models containing adsorbates at the appropriate coverages prevalent during steady-state catalysis.

H2 oxidation reaction rates on Pt nanoparticles at moderate temperatures (548-673 K) and suprastoichiometric H2/O2 ratios (above two) are much lower than H2-D2 isotopic exchange rates, indicative of quasi-equilibrated dissociative H2 adsorption steps to form H adatoms (H*). This equilibration, combined with kinetic trends that indicate negligible coverages of O2-derived species at steady state, allows H* coverages during H2-O2 reactions to be rigorously determined from H2 chemisorption isotherms independently measured at relevant reaction temperatures. The dependence of H2 oxidation rates on H2 and O2 pressures indicate that turnover rates reflect the reactive collision probabilities of gaseous O2 with catalytic surfaces, occurring via two kinetically-relevant pathways: (i) dissociation of O2 on bare Pt sites (*) to form two adsorbed O atoms and (ii) reaction of O2 with vicinal H*-H* pairs to form weakly-adsorbed hydrogen peroxide (*HOOH*) intermediates, which rapidly decompose to form H2O. The measured activation barriers for each of these two pathways are consistent with the barriers calculated using density functional theory (DFT) assessments on Pt(111) surfaces with varying H* coverages. The relative contributions of each pathway to the overall reaction rate depend on the relative surface coverages of H* and * and on the free energy barriers for each pathway; these two routes show similar enthalpic barriers and differ only in terms of the entropic penalties incurred upon forming the transition states from their respective precursors.

CO2-H2 reactions on Ru nanoparticles form CO and hydrocarbons via interlinked steps on catalytic surfaces. Kinetic trends and infrared spectra measured during CO2-H2 and CO-H2 catalysis on Ru nanoparticles are remarkably similar, thus providing compelling evidence for the inextricable mechanistic connections between CO2 and CO methanation reactions; after CO2 activation steps, these reactions proceed via an identical sequence of elementary steps occurring on Ru surfaces that contain significant coverages of adsorbed CO (CO*). During CO2-H2 reactions, activation of the first C-O bond in CO2 occurs via quasi-equilibrated steps, as evidenced by fast C16O2-C18O2 isotopic exchange rates and by the kinetic trends observed for CO2-H2 reactions; these CO2 activation steps form bound CO and O species (CO* and O*). The first H-addition step to O* is kinetically-relevant in CO2-consuming reactions and is followed by a more facile second H-addition step to form H2O. The activation of the second C-O bond in CO2 (now present in CO*) occurs via a H-assisted pathway involving the sequential addition of two H adatoms, consistent with the observed kinetic trends and previous theoretical assessments, as also shown for CO-H2 reactions on Ru, Co, and Fe catalysts. CO* species readily desorb and re-adsorb during CO-H2 and CO2-H2 reactions, leading to CO(g) pressures that are equilibrated with bound CO* species. In addressing the analysis of CO2 hydrogenation reactivity and selectivity data, the use of integral reactor methods is essential to derive accurate mechanistic interpretations, because of the ubiquitous strong inhibition by CO and the equilibrated nature of CO adsorption-desorption steps. Such conclusions underscore the importance of a detailed understanding of mechanistic features before any attempts at developing structure-function relations that seek to connect intrinsic catalytic properties with measured turnover rates and selectivities.

These mechanistic conclusions for H2-O2 and H2-CO2 reactions allow their use in tandem with C3-C4 alkane dehydrogenation reactions so as to scavenge the H2 formed as a byproduct of dehydrogenation, thus shifting equilibrium yields. H2-O2 reactions provide a stronger thermodynamic driving force (Gibbs free energy change of reaction: -215 kJ per mole H2, 573 K) than CO2-H2 reactions (-15 kJ per mole H2 for methanation, 18 kJ per mole H2 for reverse water-gas shift; 573 K), thus removing dehydrogenation thermodynamic constraints when O2 is concurrently introduced at the required stoichiometric H2/O2 ratios. The coupling of dehydrogenation with H2 oxidation reactions in practice requires that H2/O2 ratios be maintained at suprastoichiometric levels to prevent large yield losses due to parasitic hydrocarbon oxidation. However, even when O2 introduction is carefully staged such that sufficient H2/O2 ratios are maintained throughout the reactor bed, undesired hydrocarbon combustion reaction cannot be eliminated entirely. Encapsulation of the Pt metal function (for H2-O2 reactions) within small-pore zeolites was thus explored in order to exploit the sieving properties of these materials to increase the selectivities of H2-O2 reactions in the presence of C3-C4 hydrocarbons. The larger size of isobutane (0.49 nm kinetic diameter) and propane (0.43 nm) and of their respective alkene products relative to H2 (0.29 nm) and O2 (0.35 nm) allow LTA structures (0.4-0.5 nm apertures) to hinder access by the hydrocarbon molecules to the intracrystalline Pt function. In doing so, these materials lead to up to 30-fold increases in selectivity for H2 oxidation over alkane oxidation compared to the unencapsulated metal function (Pt supported on amorphous SiO2).

The challenges associated with the coupling of selective H2 oxidation catalyzed by Pt/LTA and alkane dehydrogenation catalyzed by Pt/SiO2 were further explored. Selectivity enhancements, defined as the ratio of the H2 oxidation selectivity measured on Pt/LTA to the H2 oxidation selectivity measured on Pt/SiO2, that are conferred by Pt encapsulation are diminished as O2 conversions increase, with selectivity enhancements approaching unity as O2 pressures approach zero. Such effects, resulting from the attenuation of intracrystalline concentration gradients of alkanes as O2 pressures decrease, require operation at sufficiently high O2 pressures in order to maintain significant intracrystalline hydrocarbon gradients. Selectivity enhancements of encapsulated Pt also depend on prevalent alkene concentrations; alkenes are much more reactive than alkanes, and alkene oxidation was detected even on Pt-free zeolites (LTA, FAU, MFI), amorphous SiO2, and other nominally inert support surfaces. Such observations reflect the formation of carbonaceous deposits, when catalytic surfaces are exposed to alkenes, that react with O2 and may impose additional transport restrictions. These findings highlight the challenges in the application of seemingly straightforward size-exclusion concepts to predict rates and selectivities on metals encapsulated within 8-MR zeolites without a concomitant analysis of the relevant dynamics of reaction and transport processes.