Accurate and Transferable Machine Learning Potential for Molecular Dynamics Simulation of Sodium Silicate Glasses.

An accurate and transferable machine learning (ML) potential for the simulation of binary sodium silicate glasses over a wide range of compositions (from 0 to 50% Na2O) was developed. The potential energy surface is approximated by the sum of atomic energy contributions mapped by a neural network algorithm from the local geometry comprising information on atomic distances and angles with neighboring atoms using the DeePMD code [Wang, H. Comput. Phys. Commun. 2018, 228, 178-184]. Our model was trained on a large data set of total energies and atomic forces computed at the density functional theory level on structures extracted from classical molecular dynamics (MD) simulations performed at several temperatures from 300 to 3000 K. This allows for the generation of a robust and transferable ML potential applicable over the full compositional range of glass formability at different temperatures that outperforms the empirical potentials available in the literature in reproducing structures and properties such as bond angle distribution, total distribution functions, and vibrational density of state. The generality of the approach enables the future training of a potential with other or more elements allowing for simulations of structures, properties, and behavior of ternary and multicomponent oxide glasses with nearly ab initio accuracy at a fraction of the computational cost.

Synthetic Data Generation for Automatic Segmentation of X-ray Computed Tomography Reconstructions of Complex Microstructures.

The greatest challenge when using deep convolutional neural networks (DCNNs) for automatic segmentation of microstructural X-ray computed tomography (XCT) data is the acquisition of sufficient and relevant data to train the working network. Traditionally, these have been attained by manually annotating a few slices for 2D DCNNs. However, complex multiphase microstructures would presumably be better segmented with 3D networks. However, manual segmentation labeling for 3D problems is prohibitive. In this work, we introduce a method for generating synthetic XCT data for a challenging six-phase Al-Si alloy composite reinforced with ceramic fibers and particles. Moreover, we propose certain data augmentations (brightness, contrast, noise, and blur), a special in-house designed deep convolutional neural network (Triple UNet), and a multi-view forwarding strategy to promote generalized learning from synthetic data and therefore achieve successful segmentations. We obtain an overall Dice score of 0.77. Lastly, we prove the detrimental effects of artifacts in the XCT data on achieving accurate segmentations when synthetic data are employed for training the DCNNs. The methods presented in this work are applicable to other materials and imaging techniques as well. Successful segmentation coupled with neural networks trained with synthetic data will accelerate scientific output.

Parity Violation Energy of Biomolecules - V: Protein Metal Centers.

The parity-violation difference between mirror images of chiral metal centers found in naturally occurring proteins and enzymes is computed at the Dirac-Hartree-Fock level, for both equilibrium and transition state configurations. The systems, selected on the likelihood of yielding high parity violation energies based on atomic mass and coordination geometry, are extracted from: type I Blue Copper Protein active site, Zn and Cd Carbon Anhydrase, Ni Acetyl-Coenzyme-A Synthase, and Mo based CO-Dehydrogenase. Our values provide an approximate upper limit to possible parity-violation effects in biological systems based on static effects.

Deep learning techniques for detecting preneoplastic and neoplastic lesions in human colorectal histological images.

Trained pathologists base colorectal cancer identification on the visual interpretation of microscope images. However, image labeling is not always straightforward and this repetitive task is prone to mistakes due to human distraction. Significant efforts are underway to develop informative tools to assist pathologists and decrease the burden and frequency of errors. The present study proposes a deep learning approach to recognize four different stages of cancerous tissue development, including normal mucosa, early preneoplastic lesion, adenoma and cancer. A dataset of human colon tissue images collected and labeled over a 10-year period by a team of pathologists was partitioned into three sets. These were used to train, validate and test the neural network, comprising several convolutional and a few linear layers. The approach used in the present study is 'direct'; it labels raw images and bypasses the segmentation step. An overall accuracy of >95% was achieved, with the majority of mislabeling referring to a near category. Tests on an external dataset with a different resolution yielded accuracies >80%. The present study demonstrated that the neural network, when properly trained, can provide fast, accurate and reproducible labeling for colon cancer images, with the potential to significantly improve the quality and speed of medical diagnoses.

Role of solvent-anion charge transfer in oxidative degradation of battery electrolytes.

Electrochemical stability windows of electrolytes largely determine the limitations of operating regimes of lithium-ion batteries, but the degradation mechanisms are difficult to characterize and poorly understood. Using computational quantum chemistry to investigate the oxidative decomposition that govern voltage stability of multi-component organic electrolytes, we find that electrolyte decomposition is a process involving the solvent and the salt anion and requires explicit treatment of their coupling. We find that the ionization potential of the solvent-anion system is often lower than that of the isolated solvent or the anion. This mutual weakening effect is explained by the formation of the anion-solvent charge-transfer complex, which we study for 16 anion-solvent combinations. This understanding of the oxidation mechanism allows the formulation of a simple predictive model that explains experimentally observed trends in the onset voltages of degradation of electrolytes near the cathode. This model opens opportunities for rapid rational design of stable electrolytes for high-energy batteries.

On the simulation of vibrationally resolved electronic spectra of medium-size molecules: the case of styryl substituted BODIPYs.

BODIPY dyes are used in a variety of applications because of their peculiar spectroscopic and photo-physical properties that vary depending on the stereochemistry of the functional groups attached to the boron-dipyrromethene core structure. In this work, we have applied several computational methods, adapted for semi-rigid molecules based on the Franck-Condon principle, for the study of the optical properties of BODIPY systems and for the understanding of the influence of functional groups on their spectroscopic features. We have analyzed the electronic spectra of two styryl substituted BODIPY molecules of technological interest, properly taking into account the vibronic contribution. For comparison with recently recorded experimental data in methanol, the vibrationally resolved electronic spectra of these systems were computed using both Time-Independent (TI) and Time-Dependent (TD) formalisms. The first step toward the analysis of optical properties of the styryl modified BODIPYs was a benchmark of several density functionals, to select the most appropriate one. We have found that all benchmarked functionals provide good results in terms of band shape but some of them show strong discrepancies in terms of band position. Beyond the issue of the electronic structure calculation method, different levels of sophistication can be adopted for the calculation of vibronic transitions. In this study, the effect of mode couplings and the influence of the Herzberg-Teller terms on the theoretical spectra has been investigated. It has been found that all levels of theory considered give reproducible results for the investigated systems: band positions and shapes are similar at all levels and little improvements have been found in terms of band shape with the inclusion of Herzberg-Teller effect. Inclusion of temperature effects proved to be challenging due to the important impact of large amplitude motions. Better agreement can be achieved by adopting a suitable set of coordinates coupled with a reduced-dimensionality scheme.

Factors affecting cyclic durability of all-solid-state lithium batteries using poly(ethylene oxide)-based polymer electrolytes and recommendations to achieve improved performance.

A detailed experimental analysis of the factors affecting cyclic durability of all-solid-state lithium batteries using poly(ethylene oxide)-based polymer electrolytes was published in EES by Nakayama et al. We use quantum mechanics to interpret these results, identifying processes involved in the degradation of rechargeable lithium batteries based on polyethylene oxide (PEO) polymer electrolyte with LiTFSI. We consider that ionization of the electrolyte near the cathode at the end of the recharge step is probably responsible for this degradation. We find that an electron is likely removed from PEO next to a TFSI anion, triggering a sequence of steps leading to neutralization of a TFSI anion and anchoring of another TFSI to the PEO. This decreases the polymer conductivity near the cathode, making it easier to ionize additional PEO and leading to complete degradation of the battery. We refer to this as the Cathode Overpotential Driven Ionization of the Solvent (CODIS) model. We suggest possible ways to confirm experimentally our interpretation and propose modifications to suppress or reduce electrolyte degradation.

Computational Study of the Stability of Nanotube Fragments.

Extremely short (<1 nm) fragments of zig-zag carbon nanotubes are studied with ab-initio techniques to determine their geometric and electronic structure as well as their magnetic susceptibility. It is found that for lengths of a few carbon-carbon bonds, each fragment can be viewed as composed of crowns, that is, zig-zag rings of carbon atoms along the circumference of the tube. In this case, two kinds of electronic structures are found, depending on whether the number of carbon atoms in each crown is even or odd. Systems comprising three or more crowns either have a high spin ground state or involve a charge transfer across the length of the fragment. Conjugation changes qualitatively when the length of the fragment approaches and surpasses its girth. Indications regarding the predicted chemical stability and electronic response are provided and interpreted in terms of current densities induced within each crown by a magnetic field along the tube axis.

Predicting autoxidation stability of ether- and amide-based electrolyte solvents for Li-air batteries.

Finding suitable solvents remains one of the most elusive challenges in rechargeable, nonaqueous Li-air battery technology. Although ether and amides are identified as stable classes of aprotic solvents against nucleophilic attack by superoxide, many of them are prone to autoxidation under oxygen atmosphere. In this work, we use density functional theory calculations coupled with an implicit solvent model to investigate the autoxidative stability of ether- and N,N-dialkylamide-based solvents. The change in the activation free energy for the C-H bond cleavage by O(2) is consistent with the extent of peroxide production for each class of solvent. Conversely, the thermodynamic stability alone is not sufficient to account for the observed variation in solvent reactivity toward O(2). A detailed understanding of the factors influencing the autoxidative stability provides several strategies for designing molecules with enhanced air/O(2) stability, comparable or superior to that of structurally related hydrocarbons. The mechanism of superoxide-mediated oxidation of hydroperoxides derived from ethers and amides is presented. The degradation mechanism accounts for the primary decomposition products (esters and carboxylates) observed in the Li-air battery with ether-based electrolytes. The identification of solvents having resistance to autoxidation is critical for the development of rechargeable Li-air batteries with long cycle life.

Parity-violation energy of biomolecules-IV: protein secondary structure.

The parity-violation energy difference between enantiomeric forms of the same amino acid sequence, from the amyloid ß-peptide involved in Alzheimer's disease, in both α-helix and ß-sheet configurations, is investigated with ab-initio techniques. To this end, we develop an extension of the N2 computational scheme that selectively includes neighboring amino acids to preserve the relevant H-bonds. In agreement with previous speculations, it is found that the helical α structure is associated with larger parity-violation energy differences than the corresponding ß form. Implications for the evolution of biological homochirality are discussed as well as the relative importance of various effects in determining the parity-violation energy.

Stability of lithium superoxide LiO2 in the gas phase: computational study of dimerization and disproportionation reactions.

Knowledge of the precise molecular mechanisms during the discharge and recharge processes in the lithium-air battery is critical for achieving desired improvements in specific capacity, current density, and cyclability. The initial oxygen reduction product formed in the presence of Li(+) ions is lithium superoxide LiO(2). In this study, we report the computed structures and thermodynamic parameters of LiO(2) dimerization in the gas phase, which enables us to provide a baseline for the reaction free energy profile of the subsequent disproportionation of (LiO(2))(2) to lithium peroxide Li(2)O(2) and O(2). Our calculations identified several low-lying (LiO(2))(2) dimers, with the singlet bipyramidal structure giving IR bands that are consistent with the characteristic IR vibration frequencies of (LiO(2))(2) in the oxygen matrix at T = 15-40 K. The activation barrier for (LiO(2))(2) = Li(2)O(2)+O(2) is 10.9 kcal/mol at the UCCSD(T)/CBS level (T = 298 K), suggesting that in the gas phase LiO(2) and its aggregates could only be observed at low temperatures.

Aqueous-phase quantitative NMR determination of amino acid enantiomer ratio by 13C-NMR using chiral neodymium shift reagent.

A neodymium-(S)-PDTA (PDTA = N,N,N',N'-tetrakis[(hydroxycarbonyl)methyl]-1,2-diaminopropane) complex was found exceptionally useful in the quantitative determination of enantiomer ratios of water-soluble natural amino acids by (13)C-NMR. The method is demonstrated on mixtures of L- and D-enantiomers of various amino acids. The interactions of the chiral shift reagent with the amino acid molecules were rationalized by molecular orbital calculations.

Alkylation of phenol: a mechanistic view.

The current work utilizes the ab initio density functional theory (DFT) to develop a molecular level of the mechanistic understanding on the phenol alkylation in the presence of a cation-exchange resin catalyst, Amberlyst-15. The catalyst is modeled with the benzene sulfonic acid, and the effect of this acid on olefins such as isopropene (i-Pr) and tributene (t-Bu) in a phenol solution mimics the experimental condition. A neutral-pathway mechanism is established to account for early-stage high concentration of the phenolic ether observed in experiments. The mechanism involves an exothermic reaction between olefin and the benzene sulfonic acid to form ester followed by three reaction pathways leading to direct O-alkylation, o-C-alkylation, and p-C-alkylation. Our calculations conclude that O-alkylation to form the phenolic ether is the most energetically favorable in the neutral condition. An ionic rearrangement mechanism describes intramolecular migrations of the alkyl group from the phenolic ether to form C-alkylphenols, while the positively charged protonation significantly lowers transition barriers for these migrations. The ionic rearrangement mechanism accounts for high yields of o-C-alkylphenol and p-C-alkylphenol. Competition between the H atom and the alkyl R group at the substitutive site of the protonated ortho configuration is found to be the determining factor to the ortho/para ratio of C-alkylation products.

Parity violation energy of biomolecules--I: polypeptides.

A computational algorithm is developed to compute the energy of parity-violation (E(pv)) due to weak-nuclear forces acting within chain polymers. The method is applied to estimate the magnitude of E(pv) associated with the folding of alpha-helices in polypeptide chains and, in turn, of typical proteins. Implications to the development of biological homochirality in nature are discussed in terms of a simple evolutionary model to assess the likelihood for energies of the order of the computed E(pv) to have determined the direction of natural homochirality.

Energetics of hydrogen coverage on group VIII transition metal surfaces and a kinetic model for adsorption/desorption.

We determined the binding energy of hydrogen to the closest packed surface for all nine group VIII transition metals as a function of surface coverage using quantum mechanics (density functional theory with the generalized gradient approximation) with periodic boundary conditions. The study provides a systematic comparison of the most stable surfaces of the nine group VIII transition metals, leading to results consistent with available surface science studies. We then use these to develop a simple thermodynamic model useful in estimating the surface coverage under typical heterogeneous catalysis conditions and compare these results to temperature programmed desorption experiments.

Understanding parity violation in molecular systems.

Parity-violation energy (E(PV)) due to weak nuclear interactions between nucleons and electrons in chiral molecular systems provides a fundamental tool to verify our understanding of electronic behavior in complex systems. We used both a relativistic and a nonrelativistic approach to study a number of simple molecules and analyze the corresponding E(PV) in terms of intuitive electrodynamic concepts. We developed a qualitative model to predict the sign of E(PV) and its behavior against selected geometric distortions. Our model provides a valuable tool to screen large sets of molecules and select interesting candidates for more expensive investigations.