Interpolating Nonadiabatic Molecular Dynamics Hamiltonian with Artificial Neural Networks

in Journal of Physical Chemistry Letters (2022)

Fine-Structure Constant Connects Electronic Polarizability and Geometric van-der-Waals Radius of Atoms

in Journal of Physical Chemistry Letters (2021)

Accurate Many-Body Repulsive Potentials for Density-Functional Tight Binding from Deep Tensor Neural Networks

in Journal of Physical Chemistry Letters (2020), 11(16), 68356843

We combine density-functional tight binding (DFTB) with deep tensor neural networks (DTNN) to maximize the strengths of both approaches in predicting structural, energetic, and vibrational molecular ... [more ▼]

We combine density-functional tight binding (DFTB) with deep tensor neural networks (DTNN) to maximize the strengths of both approaches in predicting structural, energetic, and vibrational molecular properties. The DTNN is used to construct a nonlinear model for the localized many-body interatomic repulsive energy, which so far has been treated in an atom-pairwise manner in DFTB. Substantially improving upon standard DFTB and DTNN, the resulting DFTB-NNrep model yields accurate predictions of atomization and isomerization energies, equilibrium geometries, vibrational frequencies, and dihedral rotation profiles for a large variety of organic molecules compared to the hybrid DFT-PBE0 functional. Our results highlight the potential of combining semiempirical electronic-structure methods with physically motivated machine learning approaches for predicting localized many-body interactions. We conclude by discussing future advancements of the DFTB-NNrep approach that could enable chemically accurate electronic-structure calculations for systems with tens of thousands of atoms. [less ▲]

First-Principles Study of Ferroelastic Twins in Halide Perovskites

in JOURNAL OF PHYSICAL CHEMISTRY LETTERS (2019), 10(6), 1416-1421

Modeling Nonreactive Molecule−Surface Systems on Experimentally Relevant Time and Length Scales: Dynamics and Conductance of Polyfluorene on Au(111)

in Journal of Physical Chemistry Letters (2018), 9

We propose a computationally efficient strategy to accurately model nonreactive molecule−surface interactions that adapts density functional theory calculations with the Tkatchenko−Scheffler scheme for ... [more ▼]

We propose a computationally efficient strategy to accurately model nonreactive molecule−surface interactions that adapts density functional theory calculations with the Tkatchenko−Scheffler scheme for van der Waals interactions into a simple classical force field. The resulting force field requires just two adjustable parameters per atom type that are needed to capture short-range and polarization interactions. The developed strategy allows for classical molecular dynamics simulation of molecules on surfaces with the accuracy of highlevel electronic structure methods but for system sizes (103 to 107 atoms) and timescales (picoseconds to microseconds) that go well beyond what can be achieved with first-principles methods. Parameters for H, sp2 C, and O on Au(111) are developed and employed to atomistically model experiments that measure the conductance of a single polyfluorene on Au(111) as a continuous function of its length. The simulations qualitatively capture both the gross and fine features of the observed conductance decay during initial junction elongation and lead to a revised atomistic understanding of the experiment. [less ▲]

Machine learning predictions of molecular properties: Accurate many-body potentials and nonlocality in chemical space

in Journal of Physical Chemistry Letters (2015), 6(12), 2326-2331

Simultaneously accurate and efficient prediction of molecular properties throughout chemical compound space is a critical ingredient toward rational compound design in chemical and pharmaceutical ... [more ▼]

Simultaneously accurate and efficient prediction of molecular properties throughout chemical compound space is a critical ingredient toward rational compound design in chemical and pharmaceutical industries. Aiming toward this goal, we develop and apply a systematic hierarchy of efficient empirical methods to estimate atomization and total energies of molecules. These methods range from a simple sum over atoms, to addition of bond energies, to pairwise interatomic force fields, reaching to the more sophisticated machine learning approaches that are capable of describing collective interactions between many atoms or bonds. In the case of equilibrium molecular geometries, even simple pairwise force fields demonstrate prediction accuracy comparable to benchmark energies calculated using density functional theory with hybrid exchange-correlation functionals; however, accounting for the collective many-body interactions proves to be essential for approaching the "holy grail" of chemical accuracy of 1 kcal/mol for both equilibrium and out-of-equilibrium geometries. This remarkable accuracy is achieved by a vectorized representation of molecules (so-called Bag of Bonds model) that exhibits strong nonlocality in chemical space. In addition, the same representation allows us to predict accurate electronic properties of molecules, such as their polarizability and molecular frontier orbital energies. © 2015 American Chemical Society. [less ▲]

Size effects in the interface level alignment of dye-sensitized TiO 2 clusters

in Journal of Physical Chemistry Letters (2014), 5(14), 2395-2401

The efficiency of dye-sensitized solar cells (DSCs) depends critically on the electronic structure of the interfaces in the active region. We employ recently developed dispersion-inclusive density ... [more ▼]

The efficiency of dye-sensitized solar cells (DSCs) depends critically on the electronic structure of the interfaces in the active region. We employ recently developed dispersion-inclusive density functional theory (DFT) and GW methods to study the electronic structure of TiO2 clusters sensitized with catechol molecules. We show that the energy level alignment at the dye-TiO2 interface is the result of an intricate interplay of quantum size effects and dynamic screening effects and that it may be manipulated by nanostructuring and functionalizing the TiO2. We demonstrate that the energy difference between the catechol LUMO and the TiO2 LUMO, which is associated with the injection loss in DSCs, may be reduced significantly by reducing the dimensions of nanostructured TiO2 and by functionalizing the TiO2 with wide-gap moieties, which contribute additional screening but do not interact strongly with the frontier orbitals of the TiO2 and the dye. Precise control of the electronic structure may be achieved via "interface engineering" in functional nanostructures. © 2014 American Chemical Society. [less ▲]

Toward Low-Temperature Dehydrogenation Catalysis: Isophorone Adsorbed on Pd(111)

in Journal of Physical Chemistry Letters (2012), 3(5), 582-586

Adsorbate geometry and reaction dynamics play essential roles in catalytic processes at surfaces. Here we present a theoretical and experimental study for a model functional organic/metal interface ... [more ▼]

Adsorbate geometry and reaction dynamics play essential roles in catalytic processes at surfaces. Here we present a theoretical and experimental study for a model functional organic/metal interface: isophorone (C9H14O) adsorbed on the Pd(111) surface. Density functional theory calculations with the Perdew-Burke-Ernzerhoff (PBE) functional including van der Waals (vdW) interactions, in combination with infrared spectroscopy and temperature-programmed desorption (TPD) experiments reveal the reaction pathway between the weakly chemisorbed reactant (C9H14O) and the strongly chemisorbed product (C9H10O), which occurs by the cleavage of four C-H bonds below 250 K. Analysis of the TPD spectrum is consistent with the relatively small magnitude of the activation barrier derived from PBE+vdW calculations, demonstrating the feasibility of low-temperature dehydrogenation. [less ▲]