Ab Initio Prediction of Vapor Pressure for Diverse Atomic Layer Deposition Precursors.

Understanding the saturated vapor pressure (Pvap) is vital for evaluating atomic layer deposition (ALD) precursors, as it directly influences the ALD temperature window and, by extension, the processability of compounds. The early estimation of vapor pressure ranges is crucial during the initial stages of novel precursor design, reducing the reliance on empirical synthesis or experimentation. However, predicting vapor pressure through computer simulations is often impeded by the scarcity of suitable empirical force fields for molecular dynamics simulations. This challenge is further compounded by the diverse chemical substances and the introduction of new elements into modern ALD processes, necessitating robust force fields that can accommodate metals, organics, and halides. In response, this study introduces a novel approach utilizing a quantum mechanically derived force field for the prediction of vapor pressure across a wide spectrum of potential ALD precursors. This approach enables the creation of system-specific force fields through parametrization based on ab initio calculations for a single molecule. We develop a comprehensive workflow to simulate both liquid and gaseous equilibrium phases, allowing the calculation of vapor pressure across a wide temperature range. Our methodology has been validated with a diverse set of ALD precursors, demonstrating its robustness in predicting Pvap at specified temperatures. The approach yields a Pearson's correlation coefficient (R2) greater than 0.9 on a logarithmic scale and a root-mean-squared deviation in self-solvation-free energies as low as 1.3 kcal mol-1. This innovative workflow, which does not require any prior experimental data, marks a significant advancement in the computer-aided design of novel ALD precursors, paving the way for accelerating developments in technology.

Computational Discovery of TTF Molecules with Deep Generative Models.

We present a computational workflow based on quantum chemical calculations and generative models based on deep neural networks for the discovery of novel materials. We apply the developed workflow to search for molecules suitable for the fusion of triplet-triplet excitations (triplet-triplet fusion, TTF) in blue OLED devices. By applying generative machine learning models, we have been able to pinpoint the most promising regions of the chemical space for further exploration. Another neural network based on graph convolutions was trained to predict excitation energies; with this network, we estimate the alignment of energy levels and filter molecules before running time-consuming quantum chemical calculations. We present a comprehensive computational evaluation of several generative models, choosing a modification of the Junction Tree VAE (JT-VAE) as the best one in this application. The proposed approach can be useful for computer-aided design of materials with energy level alignment favorable for efficient energy transfer, triplet harvesting, and exciton fusion processes, which are crucial for the development of the next generation OLED materials.

pyEFP: Automatic decomposition of the complex molecular systems into rigid polarizable fragments.

We present an open source tool able to describe intermolecular electrostatic interactions within the framework of the effective fragment potential (EFP) method. Complex molecular structure is subdivided into compact rigid fragments and parameters of their interactions are obtained from ab initio calculations. Automatic procedure allows for searching of these parameters into the existing database and merge new fragments into it. A set of standard fragments useful for the studies of organic semiconductors is also provided. Input files both for purely EFP and hybrid QM/MM calculations can be generated. The program is written in python and freely available on GitHub: https://github.com/ale-odinokov/pyEFP © 2017 Wiley Periodicals, Inc.

First principles crystal engineering of nonlinear optical materials. I. Prototypical case of urea.

The crystalline materials with nonlinear optical (NLO) properties are critically important for several technological applications, including nanophotonic and second harmonic generation devices. Urea is often considered to be a standard NLO material, due to the combination of non-centrosymmetric crystal packing and capacity for intramolecular charge transfer. Various approaches to crystal engineering of non-centrosymmetric molecular materials were reported in the literature. Here we propose using global lattice energy minimization to predict the crystal packing from the first principles. We developed a methodology that includes the following: (1) parameter derivation for polarizable force field AMOEBA; (2) local minimizations of crystal structures with these parameters, combined with the evolutionary algorithm for a global minimum search, implemented in program USPEX; (3) filtering out duplicate polymorphs produced; (4) reoptimization and final ranking based on density functional theory (DFT) with many-body dispersion (MBD) correction; and (5) prediction of the second-order susceptibility tensor by finite field approach. This methodology was applied to predict virtual urea polymorphs. After filtering based on packing similarity, only two distinct packing modes were predicted: one experimental and one hypothetical. DFT + MBD ranking established non-centrosymmetric crystal packing as the global minimum, in agreement with the experiment. Finite field approach was used to predict nonlinear susceptibility, and H-bonding was found to account for a 2.5-fold increase in molecular hyperpolarizability to the bulk value.

Structural Degradation and Swelling of Lipid Bilayer under the Action of Benzene.

Benzene and other nonpolar organic solvents can accumulate in the lipid bilayer of cellular membranes. Their effect on the membrane structure and fluidity determines their toxic properties and antibiotic action of the organic solvents on the bacteria. We performed molecular dynamics simulations of the interaction of benzene with the dimyristoylphosphatidylcholine (DMPC) bilayer. An increase in the membrane surface area and fluidity was clearly detected. Changes in the acyl chain ordering, tilt angle, and overall bilayer thickness were, however, much less marked. The dependence of all computed quantities on the benzene content showed two regimes separated by the solubility limit of benzene in water. When the amount of benzene exceeded this point, a layer of almost pure benzene started to grow between the membrane leaflets. This process corresponds to the nucleation of a new phase and provides a molecular mechanism for the mechanical rupture of the bilayer under the action of nonpolar compounds.

Specific Interactions of Neutral Side Chains of an Adsorbed Protein with the Surface of α-Quartz and Silica Gel.

Many key features of the protein adsorption on the silica surfaces still remain unraveled. One of the open questions is the interaction of nonpolar side chains with siloxane cavities. Here, we use nonequilibrium molecular dynamics simulations for the detailed investigation of the binding of several hydrophobic and amphiphilic protein side chains with silica surface. These interactions were found to be a possible driving force for protein adsorption. The free energy gain was larger for the disordered surface of amorphous silica gel as compared to α-quartz, but the impact depended on the type of amino acid. The dependence was analyzed from the structural point of view. For every amino acid an enthalpy-entropy compensation behavior was observed. These results confirm a hypothesis of an essential role of hydrophobic interactions in protein unfolding and irreversible adsorption on the silica surface.

Preferential solvation of spherical ions in binary DMSO/benzene mixtures.

We consider a new qualitative approach for treating theoretically the solvation of single-atomic ionic solutes in binary mixtures of polar and nonpolar aprotic solvents. It is based on the implicit continuum electrostatic model of the solvent mixture involving distance-dependent dielectric permittivity epsilon(R) (where R is the distance from the ion) and local concentrations C(1)(R) and C(2)(R) of the solvent ingredients. For a given R, the condition for local thermodynamic equilibrium provides the transcendental equation for explicitly establishing the permittivity and concentration profiles. Computations performed with real Cl(-) and model Cl(+) ions as solutes in benzene/DMSO mixtures are compared with the molecular dynamics simulations of the same systems. A significant discrepancy of molecular and continuum results is revealed for the concentration profiles in the close vicinity of the ion boundary, although the general trends are similar. The continuum methodology cannot account for the formation of rigid solvent structures around ions, which is most significant for the case of Cl(+). Such defect, however, proves to become of less importance in calculations of the solvation free energy, which are quite satisfactory for Cl(-) ion. Free energy calculations for Cl(+) are less successful in the range of low DMSO concentration.

Advanced dielectric continuum model of preferential solvation.

A continuum model for solvation effects in binary solvent mixtures is formulated in terms of the density functional theory. The presence of two variables, namely, the dimensionless solvent composition y and the dimensionless total solvent density z, is an essential feature of binary systems. Their coupling, hidden in the structure of the local dielectric permittivity function, is postulated at the phenomenological level. Local equilibrium conditions are derived by a variation in the free energy functional expressed in terms of the composition and density variables. They appear as a pair of coupled equations defining y and z as spatial distributions. We consider the simplest spherically symmetric case of the Born-type ion immersed in the benzene/dimethylsulfoxide (DMSO) solvent mixture. The profiles of y(R) and z(R) along the radius R, which measures the distance from the ion center, are found in molecular dynamics (MD) simulations. It is shown that for a given solute ion z(R) does not depend significantly on the composition variable y. A simplified solution is then obtained by inserting z(R), found in the MD simulation for the pure DMSO, in the single equation which defines y(R). In this way composition dependences of the main solvation effects are investigated. The local density augmentation appears as a peak of z(R) at the ion boundary. It is responsible for the fine solvation effects missing when the ordinary solvation theories, in which z=1, are applied. These phenomena, studied for negative ions, reproduce consistently the simulation results. For positive ions the simulation shows that z>>1 (z=5-6 at the maximum of the z peak), which means that an extremely dense solvation shell is formed. In such a situation the continuum description fails to be valid within a consistent parametrization.

Molecular simulation of solvent-induced stokes shift in absorption/emission spectra of organic chromophores.

The values of steady-state solvatochromic Stokes shifts (SS) in absorption/emission electronic spectra of organic chromophores are studied theoretically in the framework of the Hush-Marcus model. Charge distributions for chromophore solutes in their S0 and S1 states are found by means of conventional quantum-chemical methods combined with the continuum PCM approach for treating solvation effects. The solvent reorganization energies, which are expected to correlate with the solvent-induced part of 1/2 SS, are found in a molecular dynamics (MD) simulation which invokes a novel method for separation of the inertial piece of the electrostatic response (Vener, et al. J. Phys. Chem. B 2006, 110, 14950). Computations, performed in two solvents (acetonitrile and benzene), consider three organic dyes: coumarin 153 as a benchmark system and two other chromophores, for which experimental spectra are also reported. The results are found to be in reasonable agreement with the experiment. A consistent treatment of nonlinear effect in the solvent response, promoted by the polarizability of solutes and contributing to the solvent reorganization energies (Ingrosso, et al. J. Phys. Chem. B 2005, 109, 3553), improves the results of computations.