Deep learning workflow for the inverse design of molecules with specific optoelectronic properties.

The inverse design of novel molecules with a desirable optoelectronic property requires consideration of the vast chemical spaces associated with varying chemical composition and molecular size. First principles-based property predictions have become increasingly helpful for assisting the selection of promising candidate chemical species for subsequent experimental validation. However, a brute-force computational screening of the entire chemical space is decidedly impossible. To alleviate the computational burden and accelerate rational molecular design, we here present an iterative deep learning workflow that combines (i) the density-functional tight-binding method for dynamic generation of property training data, (ii) a graph convolutional neural network surrogate model for rapid and reliable predictions of chemical and physical properties, and (iii) a masked language model. As proof of principle, we employ our workflow in the iterative generation of novel molecules with a target energy gap between the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO).

Two excited-state datasets for quantum chemical UV-vis spectra of organic molecules.

We present two open-source datasets that provide time-dependent density-functional tight-binding (TD-DFTB) electronic excitation spectra of organic molecules. These datasets represent predictions of UV-vis absorption spectra performed on optimized geometries of the molecules in their electronic ground state. The GDB-9-Ex dataset contains a subset of 96,766 organic molecules from the original open-source GDB-9 dataset. The ORNL_AISD-Ex dataset consists of 10,502,904 organic molecules that contain between 5 and 71 non-hydrogen atoms. The data reveals the close correlation between the magnitude of the gaps between the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO), and the excitation energy of the lowest singlet excited state energies quantitatively. The chemical variability of the large number of molecules was examined with a topological fingerprint estimation based on extended-connectivity fingerprints (ECFPs) followed by uniform manifold approximation and projection (UMAP) for dimension reduction. Both datasets were generated using the DFTB+ software on the "Andes" cluster of the Oak Ridge Leadership Computing Facility (OLCF).

High-pressure and temperature neural network reactive force field for energetic materials.

Reactive force fields for molecular dynamics have enabled a wide range of studies in numerous material classes. These force fields are computationally inexpensive compared with electronic structure calculations and allow for simulations of millions of atoms. However, the accuracy of traditional force fields is limited by their functional forms, preventing continual refinement and improvement. Therefore, we develop a neural network-based reactive interatomic potential for the prediction of the mechanical, thermal, and chemical responses of energetic materials at extreme conditions. The training set is expanded in an automatic iterative approach and consists of various CHNO materials and their reactions under ambient and shock-loading conditions. This new potential shows improved accuracy over the current state-of-the-art force fields for a wide range of properties such as detonation performance, decomposition product formation, and vibrational spectra under ambient and shock-loading conditions.

A Surface-Oxide-Rich Activation Layer (SOAL) on Ni2 Mo3 N for a Rapid and Durable Oxygen Evolution Reaction.

The oxygen evolution reaction (OER) is key to renewable energy technologies such as water electrolysis and metal-air batteries. However, the multiple steps associated with proton-coupled electron transfer result in sluggish OER kinetics and catalysts are required. Here we demonstrate that a novel nitride, Ni2 Mo3 N, is a highly active OER catalyst that outperforms the benchmark material RuO2 . Ni2 Mo3 N exhibits a current density of 10 mA cm-2 at a nominal overpotential of 270 mV in 0.1 m KOH with outstanding catalytic cyclability and durability. Structural characterization and computational studies reveal that the excellent activity stems from the formation of a surface-oxide-rich activation layer (SOAL). Secondary Mo atoms on the surface act as electron pumps that stabilize oxygen-containing species and facilitate the continuity of the reactions. This discovery will stimulate the further development of ternary nitrides with oxide surface layers as efficient OER catalysts for electrochemical energy devices.

First principles study on hydrogen doping induced metal-to-insulator transition in rare earth nickelates RNiO3 (R = Pr, Nd, Sm, Eu, Gd, Tb, Dy, Yb).

Rare earth nickelates (RNiO3), consisting of a series of correlated transition metal oxides, have received increasing attention due to their sharp metal-to-insulator transition (MIT). Previous reports focused on understanding the origin and modulation of thermally driven MIT by strain effects, cation doping, or external electric field. Recently, it was reported that isothermal chemical doping of hydrogen can induce MIT and increase resistivity by ∼8 orders of magnitude, which opens up the possibility of utilizing these oxides to develop advanced electronic and sensing devices. In this study, we applied first principles methods to study geometric and electronic structures of MIT driven by hydrogen doping in a series of rare earth nickelates RNiO3 (R = Pr, Nd, Sm, Eu, Gd, Tb, Dy, Yb). Hybrid functional HSE06 calculations predict that all oxides under study exhibit sharp MIT, opening up an ∼3 eV band gap after hydrogen doping, with band gap values slightly increasing from Pr to Yb. We find that the R site elements play a key role in determining hydrogen adsorption energies and hydrogen migration barriers, which controls how difficult it would be for the hydrogen atoms to migrate inside the oxides. Detailed information on geometries, electronic structures, migration barriers and adsorption energies of hydrogen provides guidance for further optimizing these materials for future experiments and applications.

A computational study of hydrogen doping induced metal-to-insulator transition in CaFeO3, SrFeO3, BaFeO3 and SmMnO3.

The metal-to-insulator transition (MIT) in rare earth perovskite oxides has drawn significant research interest for decades to unveil the underlying physics and develop novel electronic materials. Recently, chemical doping induced MIT in SmNiO3 has been observed experimentally, with its resistivity changed by eight orders of magnitude. The mechanism of switching from one singly occupied Ni eg orbital to two singly occupied eg orbitals upon doping has been proposed by experimentalists and verified by computation. Here, we tested if this mechanism can be generally applied to other perovskite oxides with non-Ni B site elements. We applied first principles density functional theory (DFT) to study a series of perovskite oxides, CaFeO3, SrFeO3, BaFeO3 and SmMnO3. We investigated the geometry and electronic structures of pure and hydrogen doped oxides. We found that pure CaFeO3, SrFeO3 and BaFeO3 are metallic while pure SmMnO3 has a small band gap of 0.69 eV. Upon hydrogen doping, band gap opening was predicted for all four oxides: HSE06 predicted band gap values of 1.58 eV, 1.40 eV, 1.20 eV and 2.55 eV for H-doped CaFeO3, SrFeO3, BaFeO3 and SmMnO3, respectively. This finding opens up research opportunities for exploring a broader range of materials for MIT to be used in optical and electronic devices.

Molecular engineering of organic-inorganic hybrid perovskites quantum wells.

Semiconductor quantum-well structures and superlattices are key building blocks in modern optoelectronics, but it is difficult to simultaneously realize defect-free epitaxial growth while fine tuning the chemical composition, layer thickness and band structure of each layer to achieve the desired performance. Here we demonstrate the modulation of the electronic structure-and consequently the optical properties-of organic semiconducting building blocks that are incorporated between the layers of perovskites through a facile solution processing step. Self-aggregation of the conjugated organic molecules is suppressed by functionalization with sterically demanding groups and single crystalline organic-perovskite hybrid quantum wells (down to one-unit-cell thick) are obtained. The energy and charge transfers between adjacent organic and inorganic layers are shown to be fast and efficient, owing to the atomically flat interface and ultrasmall interlayer distance of the perovskite materials. The resulting two-dimensional hybrid perovskites are very stable due to protection given by the bulky hydrophobic organic groups.

Highly Stable Lead-Free Perovskite Field-Effect Transistors Incorporating Linear π-Conjugated Organic Ligands.

Sn(II)-based halide perovskite semiconductor materials are promising for a variety of electronics and optoelectronics applications but suffer from poor intrinsic materials stability. Here, we report the synthesis and characterization of a stable Sn (II)-based two-dimensional perovskite featuring a π-conjugated oligothiophene ligand, namely (4Tm)2SnI4, where 4Tm is 2-(3″',4'-dimethyl-[2,2':5',2″:5″,2″'-quaterthiophen]-5-yl)ethan-1-ammonium. The conjugated ligands facilitate formation of micrometer-size large grains, improve charge injections, and stabilize the inorganic perovskite layers. Thin film field-effect transistors based on (4Tm)2SnI4 exhibit enhanced hole mobility up to 2.32 cm2 V-1 s-1 and dramatically improved stability over the previous benchmark material (PEA)2SnI4. Stabilization mechanisms were investigated via single-crystal structure analysis as well as density functional theory calculations. It was found that the large conjugated organic layers not only serve as thick and dense barriers for moisture and oxygen but also increase the crystal formation energy via strong intermolecular interactions. This work demonstrates the great potential of molecular engineering for organic-inorganic hybrid perovskite materials toward applications in high-performance electronics and optoelectronics.

Regulation of the forming process and the set voltage distribution of unipolar resistance switching in spin-coated CoFe2O4 thin films.

We report the preparation of (111) preferentially oriented CoFe2O4 thin films on Pt(111)/TiO2/SiO2/Si substrates using a spin-coating process. The post-annealing conditions and film thickness were varied for cobalt ferrite (CFO) thin films, and Pt/CFO/Pt structures were prepared to investigate the resistance switching behaviors. Our results showed that resistance switching without a forming process is preferred to obtain less fluctuation in the set voltage, which can be regulated directly from the preparation conditions of the CFO thin films. Therefore, instead of thicker film, CFO thin films deposited by two times spin-coating with a thickness about 100 nm gave stable resistance switching with the most stable set voltage. Since the forming process and the large variation in set voltage have been considered as serious obstacles for the practical application of resistance switching for non-volatile memory devices, our results could provide meaningful insights in improving the performance of ferrite material-based resistance switching memory devices.