Elementary Steps in Aldol Condensation on Solid Acids: Insights from Experiments and Theory
The efficient conversion of oxygen-rich feedstocks derived from biomass to valuable chemicals and fuels requires an understanding of the oxygenate chemistry that can lengthen carbon chains and decrease oxygen content. Aldol condensation of alkanones and alkanals provides a pathway to form new C-C bonds, while also removing O-atoms as H¬2O. In this study, condensation reactions are investigated experimentally, through spectroscopic, H/D isotopic, and kinetic measurements, in combination with theoretical treatments (density functional theory (DFT) and coupled cluster methods (CCSD)) for a diverse range of solid Brønsted acid catalysts. These acids include H-Al-FER, H-Al-TON, H-M-MFI (M = Al, Ga, Fe, or B at framework locations), H-Al-BEA, H-Al-FAU, and H-Al-MCM-41. Condensation rates on these solid acid catalysts decrease rapidly with time on stream as a result of secondary reactions of the quasi-equilibrated pool of unstable dimer intermediates, leading to the formation of large, unsaturated (C9+) products that block active sites. A hydrogenation metal function, present as a separate component in physical (extracrystalline) mixtures or within the solid acids (intracrystalline), effectively scavenges these unsaturated species. In doing so, it (i) essentially eliminates deactivation; (ii) removes thermodynamic bottlenecks in C-C bond formation; (iii) avoids secondary β-scission reactions that decrease C-chain length; and (iv) allows rigorous mechanistic studies and the elucidation of the consequences of acid strength and confinement within zeolite voids on the kinetically-relevant rate constants and transition states.
The selective titration of protons by non-coordinating bases (2,6-di-tert-butyl-pyridine) during acetone condensation, together with infrared (IR) spectra also collected during acetone condensation, indicate that condensation reactions (453-483 K, 0.1-10 kPa acetone) occur on protons nearly saturated with H-bonded acetone without detectable contributions to condensation rates from Lewis acid sites. These results are consistent with DFT-simulated energies and frequencies of reaction intermediates and transition states (T12-site, Al-MFI unit cell, VASP, RPBE, D3BJ). Such theoretical treatments suggest strong acetone binding at reaction temperatures (ΔG = -80 kJ mol-1, 473 K) present as H-bonded species and confirm the measured shift in the infrared spectra of the acidic O-H stretch in zeolites upon addition of acetone (0.4 kPa acetone, 473 K) from 3600 cm-1 to 2440 cm-1. Measured condensation turnover rates are proportional to acetone pressure on all samples, and the resulting first-order rate constant reflects the free energy differences between bimolecular C-C bond-formation transition states and protons saturated with H-bonded acetone and gaseous acetone. The kinetic relevance of C-C bond formation is consistent with the DFT-derived free energy barrier of this step, which lies at the highest value along the reaction coordinate at T12 sites in H-Al-MFI structures.
These first-order condensation rate constants can be compared on solid Brønsted acids with different acid strength and confining environments to determine their consequences on the stability of the kinetically-relevant transition states and their relevant precursors. The effects of acid strength are evident from comparisons among catalysts with different acid strength (Al-, Ga-, Fe-, B- framework heteroatoms) and similar void structure (MFI). Here, rate constants increase exponentially with decreasing deprotonation energy (DPE), a theoretically-accessible measure of catalyst acid strength. These measured effects of acid strength on rate constants reflect the stability of the conjugate anion, which influences the stability of the ion-pair transition states more strongly than for the less charged H-bonded acetone precursors.
Microporous aluminosilicates with different framework structures provide protons of similar local composition and local coordination (acid strength), but very diverse confining void environments. Measured condensation rate constants reach a maximum value for voids similar in size to the C-C bond formation transition states (MFI, BEA); these frameworks contain voids that preferentially stabilize these transition states over smaller H-bonded acetone precursors through van der Waals interactions. The relative size and shape of these species and their strong effects on reactivity require theoretical methods that accurately account for attractive dispersion forces in energy minimizations and that can assess, posteriori, the quantum mechanical and dispersion components of DFT-derived Gibbs free energies (VASP, RPBE + D3(BJ)). These treatments are supplemented by computational screening methods developed here, which use Lennard-Jones potentials to determine interaction energies between each framework oxygen atom and all atoms in the organic moieties, thus rigorously capturing the consequences of both the size and the shape of the relevant organic moieties and the inorganic frameworks surrounding the proton. These interaction energies are then ensemble-averaged for each transition state and precursor species over all crystallographically-distinct proton locations within each microporous framework. The resulting energies act as rigorous descriptors of the “fit” for the kinetically-relevant species in a manner that accounts for the most favorable placement of all structures within the accessible confining void environments. These descriptors extend those based only on the independent properties of the species (proton affinities) and the catalyst (DPE) and their ability to reorganize charge to include parameters that describe how organic and inorganic components in transition state and precursors interact, which depends on the size and shape of both the organic and the inorganic moieties.
We have also addressed the enduring challenge of deactivation and secondary reactions that break C-C bonds in metal-free solid acids. All solid acids show significant selectivities to isobutene (C4) and acetic acid (formed in equimolar amounts), which form via β-scission of the species from the quasi-equilibrated C6-product pool (diacetone alcohol, mesityl oxide, and isomesityl oxide, for acetone reactants). These secondary reactions depend strongly on the size of the void environment and are most effectively catalyzed in MFI. These reactions require the presence of protons (mesityl oxide/isomesityl oxide/H2O mixtures do not react on Si-MFI (SIL-1)), but β-scission selectivities at a given acetone conversion increase sharply as Al (and proton) densities decrease, even though acetone condensation turnover rates remained unchanged. These counterintuitive trends, where lower densities of the purported active sites preferentially enhance secondary reactions, cannot reflect diffusional constraints and require the concurrent presence of protons and a vicinal confining environment that, even without active sites, can convert highly reactive intermediates into more stable β-scission products. A detailed examination of the effects of H2O, which determines the relative concentrations of mesityl oxide/isomesityl oxide and their ketol precursors, shows that two β-scission routes with different site requirements are involved. One route involves the β-scission of ketol species (diacetone alcohol); it occurs on all zeolites and is exclusively catalyzed by protons. A second and previously unrecognized route involves β-scission events of mesityl oxide/isomesityl oxide or their tautomer C6-alkenol products. These pathways involve the formation of an initial product on protons and its subsequent reaction within vicinal confined spaces that provide stabilization for chain propagation transition states, even in the absence of a proton. In this manner, β-scission rates become proportional to both the number of initiation sites (protons) and of propagating locations (voids), a dependence that cannot be described by conventional bifunctional catalytic formalisms. Such physical catalysis by voids free of active sites was recently shown to account for the reactivity of pure-silica zeolites in reactions of molecules with radical character; its plausible involvement here is consistent with coupled cluster (CCSD) calculations of the free energy of formation of potential chain initiators and radical propagators from mesityl oxide/isomesityl oxide and their C6-alkenol analogs.