Kyohei Kawashima

Identifying reaction coordinates (RCs) is a key to understanding the mechanism of reactions in complex systems. Deep neural network (DNN) and machine learning approaches have become a powerful tool to find the RC. On the other hand, the hyperparameters that determine the DNN model structure can be highly flexible and are often selected intuitively and in a non-trivial and tedious manner. Furthermore, how the hyperparameter choice affects the RC quality remains obscure. Here, we explore the hyperparameter space by developing the hyperparameter tuning approach for the DNN model for RC and investigate how the parameter set affects the RC quality. The DNN model is built to predict the committor along the RC from various collective variables by minimizing the cross-entropy function; the hyperparameters are automatically determined using the Bayesian optimization method. The approach is applied to study the isomerization of alanine dipeptide in vacuum and in water, and the features that characterize the RC are extracted using the explainable AI (XAI) tools. The results show that the DNN models with diverse structures can describe the RC with similar accuracy, and furthermore, the features analyzed by XAI are highly similar. This indicates that the hyperparameter space is multimodal. The electrostatic potential from the solvent to the hydrogen H18 plays an important role in the RC in water. The current study shows that the structure of the DNN models can be rather flexible, while the suitably optimized models share the same features; therefore, a common mechanism from the RC can be extracted.

The pesticide contamination in the environment has become a global concern. So far, pesticide adsorption from waste solution is one of the most economic strategies for pesticide removal. Carbon-based nanomaterials were reported to be potential pesticide sorbents. To date, nanosized graphene oxide (GO) has been discovered. Its nanosize, which is comparable to pesticide sizes, is attractive enough to explore its performance to be the pesticide sorbent. Thus, herein, the adsorption mechanisms of a single pesticide on GO were studied by comparing 6-pesticide systems. Three types of common pesticides (cyfluthrin (CFT) (pyrethroid), ivermectin (IVM) (avermectin), and diazinon (DZ) (organophosphate)) were used as pesticide models. All pesticides rapidly adhere to GO at the graphene-like region. The π-π and π-alkyl interactions contribute most to pesticide adhesion. The adsorption of CFT and DZ is led by the π-π stacking, whereas bulky IVM uses the π-alkyl forces. Having more pesticides results in self-clustering. Pesticides pile up and avoid lying on the oxygenated area. IVM is the most favorable for GO and shows tight self-packing via dispersion force and hydrogen bonding. Overall, this work displays the encouraging ability of nanosized GO to effectively absorb all pesticides which will benefit future applications in pest control.

Chiral ferrocenyl aminoalcohols were enantioselectively synthesized via the rhodium‐catalyzed asymmetric aminative ring‐opening of 1,4‐epoxy‐1,4‐dihydronaphthalenylferrocenes. A planar chiral phosphine‐olefin ligand consisting of a cyclopentadienyl (Cp) manganese scaffold provided excellent catalytic performance, furnishing the ring‐opening products in high yield and enantioselctivity (up to 99 : 1 enantiomeric ratio (e.r.)). The enantioselectivity could be further improved by changing the steric congestion in the cyclopentadienyl unit of the ferrocenyl fragment of the starting 1,4‐epoxy‐1,4‐dihydronaphthalenylferrocene. Density functional theory (DFT) calculations confirmed the high enantioselective induction in the ring‐opening reaction, estimating an e.r. of 99 : 1, which is in excellent agreement with the experimental results. The application of the thus obtained chiral ferrocenyl aminoalcohols as ligands in asymmetric reactions was also investigated.

Chiral ferrocenyl aminoalcohols were enantioselectively synthesized via the rhodium‐catalyzed asymmetric aminative ring‐opening of 1,4‐epoxy‐1,4‐dihydronaphthalenylferrocenes. A planar chiral phosphine‐olefin ligand consisting of a cyclopentadienyl (Cp) manganese scaffold provided excellent catalytic performance, furnishing the ring‐opening products in high yield and enantioselctivity (up to 99 : 1 enantiomeric ratio (e.r.)). The enantioselectivity could be further improved by changing the steric congestion in the cyclopentadienyl unit of the ferrocenyl fragment of the starting 1,4‐epoxy‐1,4‐dihydronaphthalenylferrocene. Density functional theory (DFT) calculations confirmed the high enantioselective induction in the ring‐opening reaction, estimating an e.r. of 99 : 1, which is in excellent agreement with the experimental results. The application of the thus obtained chiral ferrocenyl aminoalcohols as ligands in asymmetric reactions was also investigated.

This study demonstrates the rhodium-catalyzed divergent synthesis of 5-, 5,6-, and 8-membered silacyclic compounds, such as benzosilole, silolosiline, and disilocine, from (2-hydrosilylaryl)acetylenes. To achieve high product selectivity, intra- and intermolecular reactions are controlled by tuning the mono- and bidentate phosphine ligands on the rhodium catalyst precursor. Deuterium labeling experiments and computational studies reveal that the annulation reaction proceeds via hydride transfer to the alkyne moiety.

We have developed a gold‐catalyzed cascade reaction of aryldiynes bearing a hydrosilyl group to afford a variety of unexplored 5H‐benzo[b]indeno[2,1‐d]silines. The reaction system is applicable to the synthesis of bidirectionally π‐extended silacycles from tetra(alkynyl)aryl compounds. Computational studies suggest that 5H‐benzo[b]indeno[2,1‐d]silines are formed via the insertion of a vinyl carbocation intermediate into the Si−H bond.

The regioselective radical CH trifluoromethylation of aromatic compounds have been shown to proceed in good yield and high regioselectivity when cyclodextrin (CD) is present. Yet, the reaction mechanism and the role of CD during the reaction have remained obscure. To this end, here we performed density functional theory (DFT) calculations to the conformations obtained by semiempirical quantum mechanical molecular dynamics calculations to reveal the reaction mechanism and the role of CD in controlling regioselectivity. The results show that metal salt increases the yield but do not affect the regioselectivity, which we further confirmed by an experiment. In contrast, multiple CD‐substrate complex conformations and reaction pathways were obtained, and CD was shown to contribute to improving the regioselectivity by stabilizing the intermediate state via encapsulation. The present study indicates that CDs can increase the regioselectivity by stabilizing the intermediate and product states while only marginally affecting the transition state.

Novel triarylmethanols and triarylmethyl cations with diphenylamino groups were designed and synthesized. The triarylmethyl cations exhibited near‐infrared (NIR) emission in toluene solution, solid film state, and polystyrene (PS) film. The emission intensities of the triarylmethyl cations were strongest in the PS film. The relative quantum yield of the PS film increased as the number of diphenylamino groups increased. In addition, triarylmethanols exhibit dual emission in their aggregated state owing to the intermolecular interactions between triphenyl amine moieties and/or hydroxy groups. These findings are interesting because NIR and dual emissions are not observed in the dimethylamino group‐substituted triarylmethyl cation, which is referred to as the “malachite green,” and the corresponding triarylmethanol, respectively.

The wood-rot basidiomycete Ganoderma colossus has been chemically investigated. Comparative analyses of the natural fruiting body, artificially cultivated fruiting bodies, and mycelial cultures resulted in the isolation, in total, of 13 new highly modified lanostanes, ganocolossusins A–H (1–8) and ganodermalactones T–X (9–13), together with 23 known compounds (14–36). There were significant overlaps of the same compounds among the three different states of the fungal materials. Ganocolossusin D (4) displayed the most potent antimalarial activity against Plasmodium falciparum K1 (multi-drug-resistant strain) with an IC₅₀ value of 2.4 μM, while it was noncytotoxic to Vero cells at 50 μg/mL.