(274a) Active Site Requirements and Elementary Steps in the Ketonization of Carboxylic Acids on TiO2 and ZrO2

Ketonization converts carboxylic acids into equimolar mixtures of alkanones, CO₂, and H₂O while forming a new C-C bond and removing O-atoms from the acid reactants [1], thus lengthening carbon chains and decreasing the oxygen content in biomass-derived streams [2]. Metal oxides, such as TiO₂ and ZrO₂, catalyze ketonization reactions of diverse carboxylic acids; the elementary steps involved and their kinetic relevance, as well the specific nature of the active sites involved, remain controversial [3-5]. Here, we combine kinetic, isotopic, spectroscopic, and density functional theory (DFT) methods to assess the pathways for the ketonization of C₂-C₄ carboxylic acids on anatase TiO₂ (TiO₂(a)), rutile TiO₂ (TiO₂(r)), monoclinic ZrO₂ (ZrO₂(m)), and tetragonal ZrO₂ (ZrO₂(t)). These studies show that the elementary step that forms the new C-C bond, via reactions between adsorbed 1-hydroxy-enolates and coadsorbed acids, limits ketonization rates on all these oxides. These steps require Lewis acid-base site pairs that allow the concerted stabilization of the required transition states both for the facile formation of 1-hydroxy-enolates and for their subsequent kinetically-relevant C-C coupling reactions. Consequently, both basic and acid strengths, probed here through DFT-derived binding energies of gaseous H⁺ and OH^- species, respectively, influence the stability of the transition states that mediate ketonization rates and of the relevant adsorbed 1-hydroxyl-enolate precursors to such transition states.

These studies also demonstrate the intrinsic mechanistic link between ketonization reactions of carboxylic acids and aldol condensations of carbonyl compounds; such a link is evident from the common requirement that an α-C-H bond be cleaved in one reactant molecule before a new C-C bond is formed at that position via reactions with another reactant molecule [3,6]. Not unexpectedly, the metal oxides that catalyze both reactions, TiO₂ and ZrO₂ in the present case, are also similar. The demonstrated involvement of enolates in aldol condensation on TiO₂ and ZrO₂ [6] makes it plausible that similar enolate-like species (1-hydroxy-enolates) are involved in ketonization; these species are then involved in nucleophilic attack with another acid to form 3,3-dihydroxy-carboxylic acids with a new C-C bond, which decompose via dehydration and decarboxylation reactions to form alkanones, in contrast with the water elimination that forms condensation products from aldols without the loss of a C-atom. These similarities in mechanisms and the common involvement of Lewis acid-base site pairs, supported by the theoretical treatments reported here, extends the use of titration methods developed to measure the number of active sites in condensation reactions to their ketonization counterparts [6]; such site counts are used here to measure intrinsic reactivities, in the form of ketonization turnover rates and activation free energies, thus allowing rigorous reactivity comparisons among metal oxides and the rigorous benchmarking of theoretical treatments against experimental results.

Acetic acid ketonization turnover rates (1.0 kPa; 523 K) were slightly higher on ZrO₂(m) (0.51 ks^-1) than on ZrO₂(t) or TiO₂(a) (0.44 and 0.25 ks^-1), but much lower on TiO₂(r) (0.024 ks^-1). These reactivity trends were also reflected on larger deactivation rate constants on the more active oxides. Infrared spectra during ketonization showed that di-adsorbed acetates (*AcO*), formed via slow dissociation on acid-base site pairs and acting as unreactive spectator species, decreased the concentration of undissociated acids (AcOH*), which are the reactive intermediates in 1-hydroxy-enolate formation, during steady-state catalysis. Cu co-catalysts, as physical mixtures with these oxides, and added H₂ markedly decreased deactivation rates and led to higher steady-state ketonization rates on TiO₂ and ZrO₂; these protecting effects appear to reflect the scavenging of trace levels of ketene species, present in equilibrium with *AcO*, to form acetaldehyde and ethanol, thus providing a mechanism for balancing the depletion and formation of unreactive bound carboxylates and thus inhibiting their blocking of active sites.

Ketonization turnover rates on TiO₂(a) increased monotonically with acid pressure for all C₂-C₄ carboxylic acids; this dependence weakened with increasing pressure, ultimately leading to pressure-independent (zero-order) rates (> 1 kPa; 503-533 K); these kinetic trends were also evident on ZrO₂-based catalysts. They reflect the strong bonding of acids at acid-base site pairs, which leads to densely-covered surfaces at all practical ketonization conditions. The weak H/D kinetic isotope effects measured (acetic acid-d₄, 1.1, 523 K) and the nearly saturated monolayers of AcOH* evident in infrared spectra during catalysis, indicate that the nucleophilic attack of co-adsorbed acids by 1-hydroxy-enolates (formed via α-C-H cleavage in the acid) is the sole kinetically-relevant step in ketonization catalytic cycles. These conclusions are consistent with the measured kinetic dependences, with the observed effects of alkyl substituents on ketonization rates, and with DFT-derived activation free energy barriers using periodic slab models. The zero-order ketonization rates occur on oxide surfaces fully covered by AcOH* and represent the highest attainable rates; they reflect free energy differences (Î?G^â?¡_CC) between C-C coupling transition states (TS) and two AcOH* species.

AcOH* binds on Ti-O site pairs via concerted interactions between its carbonyl O-atom and the Ti center and the H-atom in its OH group and the basic O-atom in the site pair; *AcO*, in contrast, binds on two vicinal Ti-O pairs with its two O-atoms each interacting with one Ti center and the dissociated proton bound onto an O site. DFT-derived reaction energies and in situ infrared spectra during ketonization catalysis indicate that AcOH* is the prevalent species on TiO₂(a) at the nearly saturated surfaces that prevail during catalysis. The requirement for one Ti-O site pair for AcOH* and two Ti-O site pairs for *AcO* drives the strong preference to crowd surfaces with AcOH* species stabilized by lateral van der Waals and H-bonding interactions. The formation and spectroscopic detection of stranded carboxylates upon removal of gaseous acid molecules has led to the incorrect assertion that ketonization involves carboxylates as reactive intermediates [3-5].

The kinetic relevance of the C-C coupling step and the prevalence of AcOH* at coverages near saturation on TiO₂(a) surfaces were confirmed by DFT treatments of reaction and activation free energies on densely-covered anatase surfaces. These calculations give values for Î?G^â?¡_CC, for its enthalpy (Î?H^â?¡_CC) and entropy (Î?S^â?¡_CC) components, and for H/D isotope effects in quantitative agreement with measured values. The theoretical analyses reported here also illustrate how C-C coupling TS complexes become increasingly more stable relative to its relevant precursors (AcOH*) as the coverage of the latter increases. For example, the Î?G^â?¡_CC value is 181 kJ mol^-1 at 1/3 ML AcOH* but 160 kJ mol^-1 at 1 ML (523 K, 1 bar AcOH), as a consequence of H-bonding between the C-C coupling TS and vicinal AcOH* species. These observations and their mechanistic interpretation provide yet another demonstration about how densely-covered surfaces provide specifically beneficial environments for chemical reactions, by decreasing differences in free energies between TS complexes and their relevant kinetic precursors [7].

These mechanistic conclusions apply also to C₃-C₄ carboxylic acids and to ZrO₂-based catalysts, as evident from their kinetic rate equations for acetic acid ketonization on TiO₂(a) and from parallel DFT treatments. Acetic acid turnover rates on ZrO₂(m) give Î?G^â?¡_CC values smaller than on TiO₂(a) (136 ± 2 vs. 166 ±1 kJ mol^-1, 523 K), consistent with DFT-derived values on the low-index surfaces of these oxides (140 vs. 161 kJ mol^-1). These Î?G^â?¡_CC differences may reflect the different free energies for the formation of the 1-hydroxy-enolate reactive intermediate from AcOH* on a M-O (M=Ti, Zr) site pair or of the C-C coupling transition state from the 1-hydroxy-enolate-AcOH* pair precursor bound on two vicinal M-O site pairs. These two free energies reflect, in turn, the 1-hydroxy-enolate concentrations at steady-state and its intrinsic reactivity in reactions with AcOH* to form the C-C bond, respectively. While experiments reflect only their combined contributions, DFT treatments show that the more reactive nature of ZrO₂(m) surfaces arises from higher concentrations of 1-hydroxy-enolates bound on Zr centers. DFT-derived OH^- and H⁺ binding energies show that metal centers are weaker Lewis acids and O-atoms are more basic in ZrO₂(m) than TiO₂(a) surfaces, suggesting that the binding of 1-hydroxy-enolate anions is weaker on Zr than Ti centers, but that the respective proton moieties bind much more strongly with the O sites on ZrO₂(m) than TiO₂(a). These DFT-derived free energies lead us to infer that the higher 1-hydroxy-enolate concentrations on ZrO₂(m) benefit from the stronger basic O-sites that stabilize the proton moieties, which disfavor the reprotonation of 1-hydroxy-enolates to form AcOH*, consistent with the preference of ketonization reactions for acid-base site pairs containing strong basic sites [1,3].

The authors acknowledge BP for financial support through the XC² and ICC programs and XSEDE for access to computational facilities.

References

[1] T. N. Pham, T. Sooknoi, S. P. Crossley, D. E. Resasco, ACS Catal. 3 (2013) 2456â??2473.

[2] D. A. Simonetti, J. A. Dumesic, Catal. Rev. 51 (2009) 441â??484.

[3] G. Pacchioni, ACS Catal. 4 (2014) 2874â??2888.

[4] A. Pulido, B. Oliver-Tomas, M. Renz, M. Boronat, A. Corma, ChemSusChem, 6 (2013) 141â??151.

[5] T. N. Pham, D. Shi, D. E. Resasco, J. Catal. 314 (2014) 149â??158.

[6] S. Wang, K. Goulas, E. Iglesia, J. Catal. (submitted).

[7] D. Hibbitts, E. Iglesia, Acc. Chem. Res. 48 (2015) 1254â??1262.