A Supramolecular Model for the Co-Catalytic Role of Nitro Compounds in Brønsted Acid Catalyzed Reactions.

Nitro compounds are known to change reaction rates and kinetic concentration dependence of Brønsted-acid-catalyzed reactions. Yet, no mechanistic model exists to account for these observations. In this work, an atomistic model for the catalytically active form for an alcohol dehydroazidation reaction is presented, which is generated by DFT calculations and consists of an H-bonded aggregate of two molecules of Brønsted acid and two molecules of nitro compound. The computed O-H stretching frequencies for the aggregate indicate they are stronger acids than the individual acid molecules and serve as predictors for experimental reaction rates. By applying the model to a chemically diverse set of potential promoters, it was predicted and verified experimentally that sulfate esters induce a similar co-catalytic effect. The important implication is that Brønsted-acid catalysis must be viewed from a supramolecular perspective that accounts for not only the pKa of the acid and the bulk properties of a solvent, but also the weak interactions between all molecules in solution.

Ground-State Chemical Reactivity under Vibrational Coupling to the Vacuum Electromagnetic Field.

The ground-state deprotection of a simple alkynylsilane is studied under vibrational strong coupling to the zero-point fluctuations, or vacuum electromagnetic field, of a resonant IR microfluidic cavity. The reaction rate decreased by a factor of up to 5.5 when the Si-C vibrational stretching modes of the reactant were strongly coupled. The relative change in the reaction rate under strong coupling depends on the Rabi splitting energy. Product analysis by GC-MS confirmed the kinetic results. Temperature dependence shows that the activation enthalpy and entropy change significantly, suggesting that the transition state is modified from an associative to a dissociative type. These findings show that vibrational strong coupling provides a powerful approach for modifying and controlling chemical landscapes and for understanding reaction mechanisms.

Nitro-Assisted Brønsted Acid Catalysis: Application to a Challenging Catalytic Azidation.

A cocatalytic effect of nitro compounds is described for the B(C6F5)3·H2O catalyzed azidation of tertiary aliphatic alcohols, enabling catalyst turnover for the first time and with a broad range of substrates. Kinetic investigations into this surprising effect reveal that nitro compounds induce a switch from first order concentration dependence in Brønsted acid to second order concentration dependence in Brønsted acid and second order dependence in the nitro compounds. Kinetic, electronic, and spectroscopic evidence suggests that higher order hydrogen-bonded aggregates of nitro compounds and acids are the kinetically competent Brønsted acid catalysts. Specific weak H-bond accepting additives may offer a new general approach to accelerating Brønsted acid catalysis in solution.

Breaking the dichotomy of reactivity vs. chemoselectivity in catalytic S(N)1 reactions of alcohols.

The inability to decouple Lewis acid catalysis from undesirable Brønsted acid catalysed side reactions when water or other protic functional groups are necessarily present has forced chemists to choose between powerful but harsh catalysts or poor but mild ones, a dichotomy that restricts the substrate scope of dehydrative transformations such as the direct SN1 reaction of alcohols. A systematic survey of Lewis and Brønsted acids reveals that the strong non-hydrolyzable Lewis acid B(C6F5)3 leads to highly chemoselective alcohol substitution in the presence of acid-sensitive alkenes, protecting groups and other functional groups without the typical compromise in reaction rates, substrate scope and catalyst loading.

Generative model based on junction tree variational autoencoder for HOMO value prediction and molecular optimization.

In this work, we provide further development of the junction tree variational autoencoder (JT VAE) architecture in terms of implementation and application of the internal feature space of the model. Pretraining of JT VAE on a large dataset and further optimization with a regression model led to a latent space that can solve several tasks simultaneously: prediction, generation, and optimization. We use the ZINC database as a source of molecules for the JT VAE pretraining and the QM9 dataset with its HOMO values to show the application case. We evaluate our model on multiple tasks such as property (value) prediction, generation of new molecules with predefined properties, and structure modification toward the property. Across these tasks, our model shows improvements in generation and optimization tasks while preserving the precision of state-of-the-art models.