A General Tensor Prediction Framework Based on Graph Neural Networks.

Graph neural networks (GNNs) have been shown to be extremely flexible and accurate in predicting the physical properties of molecules and crystals. However, traditional invariant GNNs are not compatible with directional properties, which currently limits their usage to the prediction of only invariant scalar properties. To address this issue, here we propose a general framework, i.e., an edge-based tensor prediction graph neural network, in which a tensor is expressed as the linear combination of the local spatial components projected on the edge directions of clusters with varying sizes. This tensor decomposition is rotationally equivariant and exactly satisfies the symmetry of the local structures. The accuracy and universality of our new framework are demonstrated by the successful prediction of various tensor properties from first to third order. The framework proposed in this work will enable GNNs to step into the broad field of prediction of directional properties.

Theoretical study on the magnetic properties of cathode materials in the lithium-ion battery.

The layered LiMO2 (M = Co, Ni, and Mn) materials are commonly used as the cathode materials in the lithium-ion battery due to the distinctive layer structure for lithium extraction and insertion. Although their electrochemical properties have been extensively studied, the structural and magnetic properties of LiNiO2 are still under considerable debate, and the magnetic properties of monoclinic LiMnO2 are seldom reported. In this work, a detailed study of LiNiO2, LiMnO2, and a half-doped material LiNi0.5Mn0.5O2 is performed via both first-principles calculations and Monte Carlo simulations based on the effective spin Hamiltonian model. Through considering different structures, it is verified that a structure with a zigzag-type pattern is the most stable one of LiNiO2. Moreover, in order to figure out the magnetic properties, the spin exchange interactions are calculated, and then magnetic ground states are predicted in these three systems. The results show that LiNiO2 forms a spiral order that is caused by the competition from both the short-range and long-range spin exchange interactions, whereas the magnetic ground state of LiMnO2 is collinearly antiferromagnetic due to its nearest and next-nearest neighbor antiferromagnetic spin exchange interactions. However, LiNi0.5Mn0.5O2 is collinearly ferrimagnetic because of the ferromagnetic nearest neighbor Ni-Ni and Mn-Mn exchange interactions. Our work demonstrates the competition between the different exchange interactions in these cathode materials, which may be relevant to the performance of the lithium-ion battery.

Abnormally weak intervalley electron scattering in MoS2 monolayer: insights from the matching between electron and phonon bands.

It is known that carrier mobility in layered semiconductors generally increases from two-dimensions (2D) to three-dimensions due to fewer scattering channels resulting from decreased densities of electron and phonon states. In this work, we find an abnormal decrease of electron mobility from monolayer to bulk MoS2. By carefully analyzing the scattering mechanisms, we can attribute such abnormality to the stronger intravalley scattering in the monolayer but weaker intervalley scattering caused by few intervalley scattering channels and weaker corresponding electron-phonon couplings compared to the bulk case. We show that it is the matching between the electronic band structure and phonon spectrum rather than their densities of electronic and phonon states that determines scattering channels. We propose, for the first time, the phonon-energy-resolved matching function to identify the intra- and inter-valley scattering channels. Furthermore, we show that multiple valleys do not necessarily lead to strong intervalley scattering if: (1) the scattering channels, which can be explicitly captured by the distribution of the matching function, are few due to the small matching between the corresponding electron and phonon bands; and/or (2) the multiple valleys are far apart in the reciprocal space and composed of out-of-plane orbitals so that the corresponding electron-phonon coupling strengths are weak. Consequently, the searching scope of high-mobility 2D materials can be reasonably enlarged using the matching function as useful guidance with the help of band edge orbital analysis.

Crystal structure prediction by combining graph network and optimization algorithm.

Crystal structure prediction is a long-standing challenge in condensed matter and chemical science. Here we report a machine-learning approach for crystal structure prediction, in which a graph network (GN) is employed to establish a correlation model between the crystal structure and formation enthalpies at the given database, and an optimization algorithm (OA) is used to accelerate the search for crystal structure with lowest formation enthalpy. The framework of the utilized approach (a database + a GN model + an optimization algorithm) is flexible. We implemented two benchmark databases, i.e., the open quantum materials database (OQMD) and Matbench (MatB), and three OAs, i.e., random searching (RAS), particle-swarm optimization (PSO) and Bayesian optimization (BO), that can predict crystal structures at a given number of atoms in a periodic cell. The comparative studies show that the GN model trained on MatB combined with BO, i.e., GN(MatB)-BO, exhibit the best performance for predicting crystal structures of 29 typical compounds with a computational cost three orders of magnitude less than that required for conventional approaches screening structures through density functional theory calculation. The flexible framework in combination with a materials database, a graph network, and an optimization algorithm may open new avenues for data-driven crystal structural predictions.

Enhancing Hole Density and Suppressing Recombination Centers through Illumination in Kesterite Thin Film Solar Cells.

Enhancing carrier density and increasing carrier lifetime are critical for the good performance of thin film solar cells. We apply illumination during the growth of kesterite Cu2ZnSnS4 (CZTS) to enhance hole density and suppress defects of nonradiative electron-hole recombination centers simultaneously. To examine the effect of the injected carriers generated by illumination, we first extend the scheme of detailed balance equations relating free carriers and defects beyond thermal equilibrium conditions by developing an extended Fermi level (EF') to characterize a homogeneous semiconductor with non-equilibrium carriers. On the basis of this scheme, we find that illumination can promote the formation of carrier-providing defects and suppress the formation of carrier-compensating defects. Then, we demonstrate that applying proper illumination during the growth of CZTS will help achieve a higher hole density and simultaneously suppress the formation of the SnZn antisite significantly, which are beneficial for the performance of CZTS solar cells.

Semiconductor-to-metal transition from monolayer to bilayer blue phosphorous induced by extremely strong interlayer coupling: a first-principles study.

Monolayer blue phosphorous has a large band gap of 2.76 eV but counterintuitively the most stable bilayer blue phosphorous has a negative band gap of -0.51 eV. Such a large band gap reduction from just monolayer to bilayer has not been revealed before, the underlying mechanism behind which is important for understanding interlayer interactions. In this work, we reveal the origin of the semiconductor-to-metal transition using first-principles calculations and tight-binding models. We find that the interlayer interactions are extremely strong, which can be attributed to the short layer distance and strong π-like atomic orbital couplings. Therefore, the upshift of the valence band maximum (VBM) from monolayer to bilayer blue-P is so large that the VBM in the bilayer gets higher than the conduction band minimum, leading to a negative band gap and an energy gain. Besides, the interlayer atomic misplacements weaken the couplings of out-of-plane orbitals. Therefore, the energy gain due to the semiconductor-to-metal transition is larger than the energy cost due to interlayer repulsions, thus stabilizing the metallic phase. The large band gap reduction with layer number increasing is expected to exist in other similar layered systems.

Effective lifetime of non-equilibrium carriers in semiconductors from non-adiabatic molecular dynamics simulations.

The lifetimes of non-equilibrium charge carriers in semiconductors calculated using non-adiabatic molecular dynamics often differ from experimental results by orders of magnitude. By revisiting the definition of carrier lifetime, we report a systematic procedure for calculating the effective carrier lifetime in semiconductor crystals under realistic conditions. The consideration of all recombination mechanisms and the use of appropriate carrier and defect densities are crucial to bridging the gap between modeling and measurements. Our calculated effective carrier lifetime of CH3NH3PbI3 agrees with experiments, and is limited by band-to-band radiative recombination and Shockley-Read-Hall defect-assisted non-radiative recombination, whereas the band-to-band non-radiative recombination is found to be negligible. The procedure is further validated by application to the compound semiconductors CdTe and GaAs, and thus can be applied in carrier lifetime simulations in other material systems.

Unusual interlayer coupling in layered Cu-based ternary chalcogenides CuMCh2 (M = Sb, Bi; Ch = S, Se).

Interlayer interactions play important roles in manipulating the electronic properties of layered semiconductors. One common mechanism is that the valence band maximum (VBM) and the conduction band minimum (CBM) in one layer couple to the VBM and CBM in another layer, respectively, resulting in the decrease of the band gap from the monolayer to the bilayer. Here we report an unusual interlayer coupling mechanism in layered Cu-based ternary chalcogenides CuMCh2 (M = Sb, Bi; Ch = S, Se) that the CBM in one layer strongly couples to the VBM in the other layer, leading to the band gap increase from the monolayer to the bilayer. Such an unusual interlayer interaction arises from the entangling between the electronic structures and the structures of CuMCh2 in which the cations M and anions Ch are alternatively arranged at the outmost part of each layer. Consequently, the M atom at the bottom of the upper layer is very close to the Ch atom at the top of the bottom layer, so that the orbitals of the M atom which dominate the CBM can strongly couple to the orbitals of the Ch atom which dominate the VBM, as demonstrated by the orbital hopping integrals obtained from the Wannier function analysis. The exceptional case of the unusual interlayer interaction revealed in this work enriches the diversity of the interlayer interactions in layered materials and is expected to exist in similar layered systems in which cations and anions are alternatively arranged at the outmost part of each layer.

Hyperdynamics simulations with ab initio forces.

By applying the locally optimal rotation method to deal with the lowest eigenvalue of a Hessian matrix, we have efficiently incorporated the hyperdynamics method into the ab initio scheme. In the present method, we only need to calculate the first derivative of the potential and several more force calls in each molecular dynamics (MD) step, which makes hyperdynamics simulation applicable in ab initio MD simulations. With this implementation, we are able to simulate defect diffusion in silicon with boost factors up to 105. We utilized both direct MD and the hyperdynamics method to investigate diffusion of lithium atoms and silicon vacancies in silicon. We identified the complex diffusion process. The obtained diffusion coefficients of Li atoms and Si vacancies are in good agreement with the direct MD results.

Dimensionality-Inhibited Chemical Doping in Two-Dimensional Semiconductors: The Phosphorene and MoS2 from Charge-Correction Method.

Despite the great appeal of two-dimensional semiconductors for electronics and optoelectronics, to achieve the required charge carrier concentrations by means of chemical doping remains a challenge due to large defect ionization energies (IEs). Here, by decomposing the defect IEs into three parts based on ionization process, we propose a conceptual picture that the large defect IEs are caused by two effects of reduced dimensionality. While the quantum confinement effect makes the neutral single-electron point defect levels deep, the reduced screening effect leads to high energy cost for the electronic relaxation. The first-principles calculations for black phosphorus and MoS2 do demonstrate the general trend. Using BP monolayer either embedded into dielectric continuum or encapsulated between two hBN layers, we demonstrate the feasibility of increasing the screening to reduce the defect IEs. Our analysis is expected to help achieve effective carrier doping and open ways toward more extensive applications of 2D semiconductors.

More Se Vacancies in Sb2 Se3 under Se-Rich Conditions: An Abnormal Behavior Induced by Defect-Correlation in Compensated Compound Semiconductors.

It was believed that the Se-rich synthesis condition can suppress the formation of deep-level donor defect VSe (selenium vacancy) in Sb2 Se3 and is thus critical for fabricating high-efficiency Sb2 Se3 solar cells. However, here it is shown that by first-principles calculations the density of VSe increases unexpectedly to 1016 cm-3 when the Se chemical potential increases, so Se-rich condition promotes rather than suppresses the formation of VSe . Therefore, high density of VSe is thermodynamically inevitable, no matter under Se-poor or Se-rich conditions. This abnormal behavior can be explained by a physical concept "defect-correlation", i.e., when donor and acceptor defects compensate each other, all defects become correlated with each other due to the formation energy dependence on Fermi level which is determined by densities of all ionized defects. In quasi-1D Sb2 Se3 , there are many defects and the complicated defect-correlation can give rise to abnormal behaviors, e.g., lowering Fermi level and thus decreasing the formation energy of ionized donor VSe 2+ in Se-rich Sb2 Se3 . Such behavior exists also in Sb2 S3 . It explains the recent experiments that the extremely Se-rich condition causes the efficiency drop of Sb2 Se3 solar cells, and demonstrates that the common chemical intuition and defect engineering strategies may be invalid in compensated semiconductors.

Fully Boron-Sheet-Based Field Effect Transistors from First-Principles: Inverse Design of Semiconducting Boron Sheets.

High-performance two-dimensional (2D) field effect transistors (FETs) have a broad application prospect in future electronic devices. The lack of an ideal material system, however, hinders the breakthrough of 2D FETs. Recently, phase engineering offers a promising solution, but it requires both semiconducting and metallic phases of materials. Here we suggest borophenes as ideal systems for 2D FETs by theoretically searching semiconducting phases. Using multiobjective differential optimization algorithms implemented in the IM2ODE package and the first-principles calculations, we have successfully identified 16 new semiconducting borophenes. Among them, the B12-1 borophene is the most stable semiconducting phase, whose total energy is lower than any other known semiconducting borophenes. By considering not only the band alignments but also the lattice matches between semiconducting and metallic borophenes, we then have theoretically proposed several device models of fully boron-sheet-based 2D FETs. Our work provides beneficial ideas and attempts for discovering novel borophene-based 2D FETs.

Au60-: The Smallest Gold Cluster with the High-Symmetry Icosahedral Core Au13.

Among coinage metal nanoclusters with 55 atoms, only Ag55- and Cu55- are the geometric magic-number clusters, as both exhibit icosahedral symmetry. Au55-, however, exhibits much lower symmetry due largely to the strong relativistic bonding effect. In this study, we collect a much larger population (>10,000 isomers) of low-energy isomers of Au55- to Au60- by using the combined density-functional theory and basin-hopping global optimization method. We also include the spin-orbit effect in the density-functional theory computation to achieve simulated photoelectron spectra in quantitative fashion. Remarkably, we uncover that the Au13 core with the highest icosahedral ( Ih) symmetry emerges at the size of Au60-. Stability analysis suggests that Au57- with 58 valence electrons, an electronic magic number, is the relatively more stable cluster in the size range considered. Overall, in this size range we reveal a compromise between the trend toward having a perfect icosahedral 13-atom core and the strong relativistic bonding effect.

Chemical-to-Electricity Carbon: Water Device.

The ability to release, as electrical energy, potential energy stored at the water:carbon interface is attractive, since water is abundant and available. However, many previous reports of such energy converters rely on either flowing water or specially designed ionic aqueous solutions. These requirements restrict practical application, particularly in environments with quiescent water. Here, a carbon-based chemical-to-electricity device that transfers the chemical energy to electrical form when coming into contact with quiescent deionized water is reported. The device is built using carbon nanotube yarns, oxygen content of which is modulated using oxygen plasma-treatment. When immersed in water, the device discharges electricity with a power density that exceeds 700 mW m-2 , one order of magnitude higher than the best previously published result. X-ray absorption and density functional theory studies support a mechanism of operation that relies on the polarization of sp2 hybridized carbon atoms. The devices are incorporated into a flexible fabric for powering personal electronic devices.

Topological Dirac States beyond π-Orbitals for Silicene on SiC(0001) Surface.

The discovery of intriguing properties related to the Dirac states in graphene has spurred huge interest in exploring its two-dimensional group-IV counterparts, such as silicene, germanene, and stanene. However, these materials have to be obtained via synthesizing on substrates with strong interfacial interactions, which usually destroy their intrinsic π(pz)-orbital Dirac states. Here we report a theoretical study on the existence of Dirac states arising from the px,y orbitals instead of pz orbitals in silicene on 4H-SiC(0001), which survive in spite of the strong interfacial interactions. We also show that the exchange field together with the spin-orbital coupling give rise to a detectable band gap of 1.3 meV. Berry curvature calculations demonstrate the nontrivial topological nature of such Dirac states with a Chern number C = 2, presenting the potential of realizing quantum anomalous Hall effect for silicene on SiC(0001). Finally, we construct a minimal effective model to capture the low-energy physics of this system. This finding is expected to be also applicable to germanene and stanene and imply great application potentials in nanoelectronics.

Thermal conductivity of disordered two-dimensional binary alloys.

Using non-equilibrium molecular dynamics simulations, we have studied the effect of disorder on the thermal conductivity of two-dimensional (2D) C1-xNx alloys. We find that the thermal conductivity not only depends on the substitution concentration of nitrogen, but also strongly depends on the disorder distribution. A general linear relationship is revealed between the thermal conductivity and the participation ratio of phonons in 2D alloys. Localization mode analysis further indicates that the thermal conductivity variation in the ordered alloys can be attributed to the number of inequivalent atoms. As for the disordered alloys, we find that the thermal conductivity variation can be described by a simple linear formula with the disorder degree and the substitution concentration. The present study suggests some general guidance for phonon manipulation and thermal engineering in low dimensional alloys.

Achieving High Aqueous Energy Storage via Hydrogen-Generation Passivation.

A new design strategy for polyimides/carbon nanotube networks is reported, aiming to passivate the hydrogen-evolution mechanism on the molecular structures of electrodes, thus substantially boosting their aqueous energy-storage capabilities. The intrinsic sluggish hydrogen-evolution activity of polyimides is further passivated via Li(+) association during battery charging, leading to a much wider voltage window and exceptional energy-storage capability.

Photoelectrochemical Conversion from Graphitic C3N4 Quantum Dot Decorated Semiconductor Nanowires.

Despite the recent progress of developing graphitic carbon nitride (g-C3N4) as a metal-free photocatalyst, the synthesis of nanostructured g-C3N4 has still remained a complicated and time-consuming approach from its bulk powder, which substantially limits its photoelectrochemical (PEC) applications as well as the potential to form composites with other semiconductors. Different from the labor-intensive methods used before, such as exfoliation or assistant templates, herein, we developed a facile method to synthesize graphitic C3N4 quantum dots (g-CNQDs) directly grown on TiO2 nanowire arrays via a one-step quasi-chemical vapor deposition (CVD) process in a homemade system. The as-synthesized g-CNQDs uniformly covered over the surface of TiO2 nanowires and exhibited attractive photoluminescence (PL) properties. In addition, compared to pristine TiO2, the heterojunction of g-CNQD-decorated TiO2 nanowires showed a substantially enhanced PEC photocurrent density of 3.40 mA/cm(2) at 0 V of applied potential vs Ag/AgCl under simulated solar light (300 mW/cm(2)) and excellent stability with ∼82% of the photocurrent retained after over 10 h of continuous testing, attributed to the quantum and sensitization effects of g-CNQDs. Density functional theory calculations were further carried out to illustrate the synergistic effect of TiO2 and g-CNQD. Our method suggests that a variety of g-CNQD-based composites with other semiconductor nanowires can be synthesized for energy applications.

Three-dimensional WS2 nanosheet networks for H2O2 produced for cell signaling.

Hydrogen peroxide (H2O2) is an important molecular messenger for cellular signal transduction. The capability of direct probing of H2O2 in complex biological systems can offer potential for elucidating its manifold roles in living systems. Here we report the fabrication of three-dimensional (3D) WS2 nanosheet networks with flower-like morphologies on a variety of conducting substrates. The semiconducting WS2 nanosheets with largely exposed edge sites on flexible carbon fibers enable abundant catalytically active sites, excellent charge transfer, and high permeability to chemicals and biomaterials. Thus, the 3D WS2-based nano-bio-interface exhibits a wide detection range, high sensitivity and rapid response time for H2O2, and is capable of visualizing endogenous H2O2 produced in living RAW 264.7 macrophage cells and neurons. First-principles calculations further demonstrate that the enhanced sensitivity of probing H2O2 is attributed to the efficient and spontaneous H2O2 adsorption on WS2 nanosheet edge sites. The combined features of 3D WS2 nanosheet networks suggest attractive new opportunities for exploring the physiological roles of reactive oxygen species like H2O2 in living systems.

Incorporation of well-dispersed sub-5-nm graphitic pencil nanodots into ordered mesoporous frameworks.

Over the past few decades the direct assembly of optical nanomaterials into ordered mesoporous frameworks has proved to be a considerable challenge. Here we propose the incorporation of ultrasmall (sub-5-nm) graphitic pencil nanodots into ordered mesoporous frameworks for the fabrication of optoelectronic materials. The nanodots, which were prepared from typical commercial graphite pencils by an electrochemical tailoring process, combine properties such as uniform size (∼3 nm), excellent dispersibility and high photoconversion efficiency (∼27%). These nanodots were incorporated into a variety of ordered mesoporous frameworks (TiO2, silica, carbon and silica-carbon materials) by co-assembly, driven by hydrogen bonding, with the frameworks' precursors. The resulting materials showed a high degree of ordering, and a sharp increase in their optical performance (for example, photocurrent density). We envisage that the large-scale synthesis of ultrasmall carbon nanodots and their incorporation into ordered mesoporous frameworks may facilitate the preparation of materials with a variety of optical properties.