(387e) Distinct Catalytic Reactivity of Sn Substituted in Framework Locations and at Defect Grain Boundaries in Sn-Zeolites

Zeolites with some of their framework silicon atoms substituted with Lewis acidic heteroatoms, notably tetravalent Sn⁴⁺, catalyze a variety of oxygenate conversion reactions, including sugar isomerization [1], Baeyer-Villiger oxidation [2], and aldol condensation [3]. The structural diversity among different host zeolite crystal topologies and polymorphs, and defect sites within them, provide a diverse array of local coordination environments around Sn active sites that influence their catalytic reactivity. Spectroscopic characterizations of Sn sites in Sn-Beta zeolites detect two different Sn configurations using Lewis base probe molecules (CD₃CN IR [4], trimethylphosphine oxide ³¹P NMR [3]) or without probe molecules (¹¹⁹Sn NMR [1]), commonly attributed to framework Sn sites in closed (Sn-(OSi≡)₄) and hydrolyzed-open ((HO)-Sn-(OSi≡)₃---HO-Si) configurations; however, these assignments are debated because hydrolyzed-open sites convert to closed configurations during the high-temperature (>800 K) pretreatments used before characterizations [5]. Here, we combine in situ pyridine titrations during a gas-phase reaction with spectroscopic characterizations, targeted syntheses and post-synthetic treatments, and theoretical calculations to quantitatively probe the structural features that lead to different Sn configurations and reactivity in Sn-zeolites.

Measurements of turnover rates of gas-phase bimolecular ethanol dehydration to diethyl ether (404–438 K) on a suite of hydrophobic and hydrophilic Sn-zeolites (Sn-Beta, Sn-BEC, Sn-MFI) of varying Sn content, together with quantitative titration of active Sn sites by pyridine during catalysis, identify two types of Sn sites with reactivity differing by more than an order of magnitude (>20×). Activation entropies to form bimolecular dehydration transition states from ethanol monomer-covered sites are less negative (ΔΔS^‡_app = 56 ± 22 J mol^-1 K^-1) at the more reactive subset of Sn sites, present in amounts equivalent to 17–26% of the Sn sites quantified by the peak centered at 2308 cm^-1 in CD₃CN IR spectra (Sn₂₃₀₈), but not correlated with that at 2316 cm^-1 (Sn₂₃₁₆).

Synthetic and post-synthetic treatments to prepare Sn-zeolites containing Sn sites hosted within diverse local coordination environments suggest that Sn₂₃₁₆ sites are not associated with Sn bound to residual fluoride anions, or Sn sited at external crystallite surfaces, amorphous domains, or among the diverse T-site locations contained within CHA, MFI, BEC, and STT frameworks. Treating Sn-Beta in HF or NH₄F solutions, which dissolve zeolitic domains preferentially at defect grain boundaries, decreased the number of Sn₂₃₁₆ sites but not Sn₂₃₀₈ sites. These data indicate that Sn₂₃₁₆ sites are preferentially located at stacking faults in zeolite Beta, which provide tetrahedral coordination environments for Sn in defect-open configurations ((HO)-Sn-(OSi≡)₃) with proximal Si-OH groups that do not permit condensation to tetrahedral closed configurations.

A computational model was developed for stacking fault defect-open Sn sites, which result in apparent activation free energies for bimolecular ethanol dehydration that are 65–74 kJ mol^-1 higher than at framework closed Sn sites that are capable of stabilizing transition states via Sn site-opening and closing as part of the catalytic cycle, consistent with the lower experimentally measured reactivity for Sn₂₃₁₆ sites. In contrast, defect-open sites possess Si-OH groups that preferentially stabilize hydride shift transition states involved in glucose-fructose isomerization turnovers. These findings highlight the ability of a given zeolite framework to confer structural diversity to nominally site-isolated Lewis acid sites, thus generating sites with distinct reactivity for different chemical transformations.

References

[1] Bermejo-Deval et al., Proc. Natl. Acad. Sci. 109 (2012) 9727–9732.

[2] Boronat et al., J. Catal. 234 (2005) 111–118.

[3] Lewis et al., ACS Catal. 8 (2018) 3076–3086.

[4] Harris et al., J. Catal. 335 (2016) 141–154.

[5] Yakimov et al., J. Phys. Chem. C. 120 (2016) 28083–28092.