Harnessing Semi-Supervised Machine Learning to Automatically Predict Bioactivities of Per- and Polyfluoroalkyl Substances (PFASs).

Many per- and polyfluoroalkyl substances (PFASs) pose significant health hazards due to their bioactive and persistent bioaccumulative properties. However, assessing the bioactivities of PFASs is both time-consuming and costly due to the sheer number and expense of in vivo and in vitro biological experiments. To this end, we harnessed new unsupervised/semi-supervised machine learning models to automatically predict bioactivities of PFASs in various human biological targets, including enzymes, genes, proteins, and cell lines. Our semi-supervised metric learning models were used to predict the bioactivity of PFASs found in the recent Organisation of Economic Co-operation and Development (OECD) report list, which contains 4730 PFASs used in a broad range of industries and consumers. Our work provides the first semi-supervised machine learning study of structure-activity relationships for predicting possible bioactivities in a variety of PFAS species.

FLUID-GPT (Fast Learning to Understand and Investigate Dynamics with a Generative Pre-Trained Transformer): Efficient Predictions of Particle Trajectories and Erosion.

The deleterious impact of erosion due to high-velocity particle impingement adversely affects a variety of engineering and industrial systems, resulting in irreversible mechanical wear of materials/components. Brute force computational fluid dynamics (CFD) calculations are commonly used to predict surface erosion by directly solving the Navier-Stokes equations for fluid and particle dynamics; however, these numerical approaches often require significant computational resources. In contrast, recent data-driven approaches using machine learning (ML) have shown immense promise for more efficient and accurate predictions to sidestep computationally demanding CFD calculations. To this end, we have developed FLUID-GPT (Fast Learning to Understand and Investigate Dynamics with a Generative Pre-Trained Transformer), a new hybrid ML architecture for accurately predicting particle trajectories and erosion on an industrial-scale steam header geometry. Our FLUID-GPT approach utilizes a Generative Pre-Trained Transformer 2 (GPT-2) with a convolutional neural network (CNN) for the first time to predict surface erosion using only information from five initial conditions: particle size, main-inlet speed, main-inlet pressure, subinlet speed, and subinlet pressure. Compared to the bidirectional long- and short-term memory (BiLSTM) ML techniques used in previous work, our FLUID-GPT model is much more accurate (a 54% decrease in the mean squared error) and efficient (70% less training time). Our work demonstrates that FLUID-GPT is an accurate and efficient ML approach for predicting time-series trajectories and their subsequent spatial erosion patterns in these complex dynamic systems.

Implementation of real-time TDDFT for periodic systems in the open-source PySCF software package.

We present a new implementation of real-time time-dependent density functional theory (RT-TDDFT) for calculating excited-state dynamics of periodic systems in the open-source Python-based PySCF software package. Our implementation uses Gaussian basis functions in a velocity gauge formalism and can be applied to periodic surfaces, condensed-phase, and molecular systems. As representative benchmark applications, we present optical absorption calculations of various molecular and bulk systems and a real-time simulation of field-induced dynamics of a (ZnO)4 molecular cluster on a periodic graphene sheet. We present representative calculations on optical response of solids to infinitesimal external fields as well as real-time charge-transfer dynamics induced by strong pulsed laser fields. Due to the widespread use of the Python language, our RT-TDDFT implementation can be easily modified and provides a new capability in the PySCF code for real-time excited-state calculations of chemical and material systems.

Photo-induced degradation of PFASs: Excited-state mechanisms from real-time time-dependent density functional theory.

Per- and polyfluoroalkyl substances (PFASs) are hazardous, carcinogenic, and bioaccumulative contaminants found in drinking water sources. To mitigate and remove these persistent pollutants, recent experimental efforts have focused on photo-induced processes to accelerate their degradation; however, the mechanistic details of these promising degradation processes remain unclear. To shed crucial insight on these electronic-excited state processes, we present the first study of photo-induced degradation of explicitly-solvated PFASs using excited-state, real-time time-dependent density functional theory (RT-TDDFT) calculations. Furthermore, our large-scale RT-TDDFT calculations show that these photo-induced excitations can be highly selective by enabling a charge-transfer process that only dissociates the CF bond while keeping the surrounding water molecules intact. Collectively, the RT-TDDFT techniques used in this work (1) enable a new capability for probing photo-induced mechanisms that cannot be gleaned from conventional ground-state DFT calculations and (2) provide a rationale for understanding ongoing experiments that are actively exploring photo-induced degradation of PFASs and other environmental contaminants.

Harnessing Plasma Environments for Ammonia Catalysis: Mechanistic Insights from Experiments and Large-Scale Ab Initio Molecular Dynamics.

By combining experimental measurements with ab initio molecular dynamics simulations, we provide the first microscopic description of the interaction between metal surfaces and a low-temperature nitrogen-hydrogen plasma. Our study focuses on the dissociation of hydrogen and nitrogen as the main activation route. We find that ammonia forms via an Eley-Rideal mechanism where atomic nitrogen abstracts hydrogen from the catalyst surface to form ammonia on an extremely short time scale (a few picoseconds). On copper, ammonia formation occurs via the interaction between plasma-produced atomic nitrogen and the H-terminated surface. On platinum, however, we find that surface saturation with NH groups is necessary for ammonia production to occur. Regardless of the metal surface, the reaction is limited by the mass transport of atomic nitrogen, consistent with the weak dependence on catalyst material that we observe and has been reported by several other groups. This study represents a significant step toward achieving a mechanistic, microscopic-scale understanding of catalytic processes activated in low-temperature plasma environments.

Indirect but Efficient: Laser-Excited Electrons Can Drive Ultrafast Polarization Switching in Ferroelectric Materials.

To enhance the efficiency of next-generation ferroelectric (FE) electronic devices, new techniques for controlling ferroelectric polarization switching are required. While most prior studies have attempted to induce polarization switching via the excitation of phonons, these experimental techniques required intricate and expensive terahertz sources and have not been completely successful. Here, we propose a new mechanism for rapidly and efficiently switching the FE polarization via laser-tuning of the underlying dynamical potential energy surface. Using time-dependent density functional calculations, we observe an ultrafast switching of the FE polarization in BaTiO3 within 200 fs. A laser pulse can induce a charge density redistribution that reduces the original FE charge order. This excitation results in both desirable and highly directional ionic forces that are always opposite to the original FE displacements. Our new mechanism enables the reversible switching of the FE polarization with optical pulses that can be produced from existing 800 nm experimental laser sources.

An MM and QM Study of Biomimetic Catalysis of Diels-Alder Reactions Using Cyclodextrins.

We performed a computational investigation of the mechanism by which cyclodextrins (CDs) catalyze Diels-Alder reactions between 9-anthracenemethanol and N-cyclohexylmaleimide. Hydrogen bonds (Hbonds) between N-cyclohexylmaleimide and the hydroxyl groups of cyclodextrins were suggested to play an important role in this catalytic process. However, our free energy calculations and molecular dynamics simulations showed that these Hbonds are not stable, and quantum mechanical calculations suggested that the reaction is not promoted by these Hbonds. The binding of 9-anthracenemethanol and N-cyclohexylmaleimide to cyclodextrins was the key to the catalytic process. Cyclodextrins act as a container to hold the two reactants in the cavity, pre-organize them for the reactions, and thus reduce the entropy penalty to the activation free energy. Dimethyl-ß-CD was a better catalyst for this specific reaction than ß-CD because of its stronger van der Waals interaction with the pre-organized reactants and its better performance in reducing the activation energy. This computational work sheds light on the mechanism of the catalytic reaction by cyclodextrins and introduces new perspectives of supramolecular catalysis.