Accelerating VASP electronic structure calculations using graphic processing units

Giant Capacitive Energy Storage in High-Entropy Lead-Free Ceramics with Temperature Self-Check

Considering the large demand for electricity in the era of artificial intelligence and big data, there is an urgent need to explore novel energy storage media with higher energy density and intelligent temperature self-check functions. High-entropy (HE) ceramic capacitors are of great significance because of their excellent energy storage efficiency and high power density (PD). However, the contradiction between configurational entropy and polarization in traditional HE systems greatly restrains the increase in energy storage density. Herein, the contradiction is effectively solved by regulating the octahedral tilt and cationic displacement in ABO3-type perovskite HE ceramics, i.e., (1-x)[0.6(Bi0.47Na0.47Yb0.03Tm0.01)TiO3-0.4(Ba0.5Sr0.5)TiO3]-xSr(Zr0.5Hf0.5)O3 (BNYTT-BST-xSZH). Combining the tape-casting process and cold isostatic pressing, the optimal BNYTT-BST-0.06SZH ceramic displays a large recoverable energy storage density (10.46 J cm-3) at 685 kV cm-1 and a high PD (332.88 MW cm-3). More importantly, due to Tm/Yb codoping, abnormal fluorescent negative thermal expansion and excellent real-time temperature sensing are developed, thus the application of fault detection and warning in high-voltage transmission line systems is conceptualized. This study provides an effective strategy for enhancing the polarization of energy-storing HE ceramics and offers a promising material for overcoming the problems of insufficient capacitor density and thermal runaway in terminal communication.

Volatile and Nonvolatile Programmable Iontronic Memristor with Lithium Imbued TiOx for Neuromorphic Computing Applications

We demonstrate a lithium (Li) imbued TiOx iontronic device that exhibits synapse-like short-term plasticity behavior without requiring a forming process beforehand or a compliance current during switching. A solid-state electrolyte lithium phosphorus oxynitride (LiPON) behaves as the ion source, and the embedding and releasing of Li ions inside the cathodic like TiOx renders volatile conductance responses from the device and offers a natural platform for hardware simulating neuron functionalities. Besides, these devices possess high uniformity and great endurance as no conductive filaments are present. Different short-term pulse-based phenomena, including paired pulse facilitation, post-tetanic potentiation, and spike rate-dependent plasticity, were observed with self-relaxation characteristics. Based on the voltage excitation period, the time scale of the volatile memory can be tuned. Temperature measurement reveals the ion displacement-induced conductance channels become frozen below 220 K. In addition, the volatile analog devices can be configured into nonvolatile memory units with multibit storage capabilities after an electroforming process. Therefore, on the same platform, we can configure volatile units as nonlinear dynamic reservoirs for performing neuromorphic training and the nonvolatile units as the weight storage layer. We proceed to use voice recognition as an example with the tunable time constant relationship and obtain 94.4% accuracy with a minimal training data set. Thus, this iontronic platform can effectively process and update temporal information for reservoir and neuromorphic computing paradigms.

High-throughput investigation of stability and Li diffusion of doped solid electrolytes via neural network potential without configurational knowledge

Solid electrolytes hold substantial promise as vital components of all-solid-state batteries. Enhancing their performance necessitates simultaneous improvements in their stability and lithium conductivity. These properties can be calculated using first-principles simulations, provided that the crystal structure of the material and the diffusion pathway through the material are known. However, solid electrolytes typically incorporate dopants to enhance their properties, necessitating the optimization of the dopant configuration for the simulations. Yet, performing such calculations via the first-principles approach is so costly that existing approaches usually rely on predetermined dopant configurations informed by existing knowledge or are limited to systems doped with only a few atoms. The proposed method enables the optimization of the dopant configuration with the support of neural network potential (NNP). Our approach entails the use of molecular dynamics to analyze the diffusion after the optimization of the dopant configuration. The application of our approach to Li10 MP2 S12 - x Ox (M = Ge, Si, or Sn) reproduce the experimental results well. Furthermore, analysis of the lithium diffusion pathways suggests that the activation energy of diffusion undergoes a percolation transition. This study demonstrates the effectiveness of NNPs in the systematic exploration of solid electrolytes.

Interplay between Catalyst Corrosion and Homogeneous Reactive Oxygen Species in Electrochemical Ozone Production

Electrochemical ozone production (EOP), a six-electron water oxidation reaction, offers promising avenues for creating value-added oxidants and disinfectants. However, progress in this field is slowed by a dearth of understanding of fundamental reaction mechanisms. In this work, we combine experimental electrochemistry, spectroscopic detection of reactive oxygen species (ROS), oxygen-anion chemical ionization mass spectrometry, and computational quantum chemistry calculations to determine a plausible reaction mechanism on nickel- and antimony-doped tin oxide (Ni/Sb-SnO2, NATO), one of the most selective EOP catalysts. Antimony doping is shown to increase the conductivity of the catalyst, leading to improved electrochemical performance. Spectroscopic analysis and electrochemical experiments combined with quantum chemistry predictions reveal that hydrogen peroxide (H2O2) is a critical reaction intermediate. We propose that leached Ni4+ cations catalyze hydrogen peroxide into solution phase hydroperoxyl radicals (•OOH); these radicals are subsequently oxidized to ozone. Isotopic product analysis shows that ozone is generated catalytically from water and corrosively from the catalyst oxide lattice without regeneration of lattice oxygens. Further quantum chemistry calculations and thermodynamic analysis suggest that the electrochemical corrosion of tin oxide itself might generate hydrogen peroxide, which is then catalyzed to ozone. The proposed pathways explain both the roles of dopants in NATO and its lack of stability. Our study interrogates the possibility that instability and electrochemical activity are intrinsically linked through the formation of ROS. In doing so, we provide the first mechanism for EOP that is consistent with computational and experimental results and highlight the central challenge of instability as a target for future research efforts.

Complementary probes for the electrochemical interface

The functions of electrochemical energy conversion and storage devices rely on the dynamic junction between a solid and a fluid: the electrochemical interface (EI). Many experimental techniques have been developed to probe the EI, but they provide only a partial picture. Building a full mechanistic understanding requires combining multiple probes, either successively or simultaneously. However, such combinations lead to important technical and theoretical challenges. In this Review, we focus on complementary optoelectronic probes and modelling to address the EI across different timescales and spatial scales - including mapping surface reconstruction, reactants and reaction modulators during operation. We discuss how combining these probes can facilitate a predictive design of the EI when closely integrated with theory.

Poisoning of Pt/γ-Al2O3 Aqueous Phase Reforming Catalysts by Ketone and Diketone-Derived Surface Species

Strong adsorption of ketone and diketone byproducts and their fragmentation products during the aqueous phase reforming of biomass derived oxygenates is believed to be responsible for the deactivation of supported Pt catalysts. This study involves a combined experimental and theoretical approach to demonstrate the interactions of several model di/ketone poisons with Pt/γ-Al2O3 catalysts. Particular di/ketones were selected to reveal the effects of hydroxyl groups (acetone, hydroxyacetone), conjugation with C=C bonds (mesityl oxide), intramolecular distance between carbonyls in diketones (2,3-butanedione, 2,4-pentanedione), and length of terminal alkyl chains (3,4-hexanedione). The formation of adsorbed carbon monoxide (1900-2100 cm-1) as a decarbonylation product was probed using infrared spectroscopy and to calculate the extent of poisoning during subsequent methanol dehydrogenation based on the reduction of the ν(C≡O) band integral relative to experiments in which only methanol was dosed. Small Pt particles appeared less active in decarbonylation and were perhaps poisoned by strongly adsorbed di/ketones on undercoordinated metal sites and bulky conjugated species formed on the γ-Al2O3 support from aldol self-condensation. Larger Pt particles were more resistant to di/ketone poisoning due to higher decarbonylation activity yet still fell short of the expected yield of adsorbed CO from subsequent methanol activity. Vibrational spectra acquired using inelastic neutron scattering showed evidence for strongly binding methyl and acyl groups resulting from di/ketone decarbonylation on a Pt sponge at 250 °C. Adsorption energies and molecular configurations were obtained for di/ketones on a Pt(111) slab using density functional theory, revealing potential descriptors for predicting decarbonylation activity on highly coordinated metal sites. Calculated reaction energies suggest it is energetically favorable to reform surface methyl groups into adsorbed CO and H. However, the rate of this surface reaction is limited by a high activation barrier indicating that either improved APR catalyst designs or regeneration procedures may be necessary.

Distributed Multi-GPU Ab Initio Density Matrix Renormalization Group Algorithm with Applications to the P-Cluster of Nitrogenase

The presence of many degenerate d/f orbitals makes polynuclear transition-metal compounds, such as iron-sulfur clusters in nitrogenase, challenging for state-of-the-art quantum chemistry methods. To address this challenge, we present the first distributed multi-graphics processing unit (GPU) ab initio density matrix renormalization group (DMRG) algorithm suitable for modern high-performance computing (HPC) infrastructures. The central idea is to parallelize the most computationally intensive part─the multiplication of O(K2) operators with a trial wave function, where K is the number of spatial orbitals, by combining operator parallelism for distributing the workload with a batched algorithm for performing contractions on GPU. With this new implementation, we are able to reach an unprecedentedly large bond dimension D = 14,000 on 48 GPUs (NVIDIA A100 80 GB SXM) for an active space model (114 electrons in 73 active orbitals) of the P-cluster, which is nearly 3 times larger than the bond dimensions reported in previous DMRG calculations for the same system using only central processing units (CPUs).

Bound-State Breaking and the Importance of Thermal Exchange-Correlation Effects in Warm Dense Hydrogen

Hydrogen at extreme temperatures and pressures is of key relevance for cutting-edge technological applications, with inertial confinement fusion research being a prime example. In addition, it is ubiquitous throughout our universe and naturally occurs in a variety of astrophysical objects. In the present work, we present exact ab initio path integral Monte Carlo (PIMC) results for the electronic density of warm dense hydrogen along a line of constant degeneracy across a broad range of densities. Using the well-known concept of reduced density gradients, we develop a new framework to identify the breaking of bound states due to pressure ionization in bulk hydrogen. Moreover, we use our PIMC results as a reference to rigorously assess the accuracy of a variety of exchange-correlation (XC) functionals in density functional theory calculations for different density regions. Here, a key finding is the importance of thermal XC effects for the accurate description of density gradients in high-energy-density systems. Our exact PIMC test set is freely available online and can be used to guide the development of new methodologies for the simulation of warm dense matter and beyond.

Active sites engineering on FeNi alloy/Cr3C2 heterostructure for superior oxygen evolution activity

Exploring high active electrocatalysts for oxygen evolution reaction (OER) is of great significance for a sustainable hydrogen economy. The development of non-precious transition metals, with sufficient active sites and ample intrinsic activity, remains a challenge. Herein, a new type of FeNi-Cr3C2 heterostructure anchored on carbon sheets (FeNi-Cr3C2@C) was reported, which can effectively catalyze OER with swift kinetics and outstanding intrinsic activity. The introduced Cr3C2 phase not only serves as a support material but also effectively suppresses the thermal coarsening of FeNi alloy nanoparticle. The FeNi-Cr3C2@C displays a robust OER activity with a low overpotential of 283 mV at the current density of 10 mA cm-2, a high turnover frequency value of 1.69 s-1 at the overpotential of 300 mV (10 times higher than that of FeNi@C) and good stability in alkaline media. Density functional theory calculations (DFT) calculations show that Cr3C2 can facilitate the generation of electron-rich region at the Ni site in FeNi alloys as an active site, exhibiting an optimized adsorption behavior toward oxygen intermediate species with regard to decreased thermodynamic energy barriers. Our work opens up a promising path to modulate the electrocatalytic active sites using inexpensive and durable Cr3C2 for electrochemical catalytic reactions.

Mechanistic Study on the Possibility of Converting Dissociated Oxygen into Formic Acid on χ-Fe5C2(510) for Resource Recovery in Fischer-Tropsch Synthesis

During Fischer-Tropsch synthesis, O atoms are dissociated on the surface of Fe-based catalysts. However, most of the dissociated O would be removed as H2O or CO2, which results in a low atom economy. Hence, a comprehensive study of the O removal pathway as formic acid has been investigated using the combination of density functional theory (DFT) and kinetic Monte Carlo (kMC) to improve the economics of Fischer-Tropsch synthesis on Fe-based catalysts. The results show that the optimal pathway for the removal of dissociated O as formic acid is the OH pathway, of which the effective barrier energy (0.936 eV) is close to that of the CO activation pathway (0.730 eV), meaning that the removal of dissociated O as formic acid is possible. The main factor in an inability to form formic acid is the competition between the formic acid formation pathway and other oxygenated compound formation pathways (H2O, CO2, methanol-formaldehyde); the details are as follows: 1. If the CO is hydrogenated first, then the subsequent reaction would be impossible due to its high effective Gibbs barrier energy. 2. If CO reacts first with O to become CO2, it is difficult for it to be hydrogenated further to become HCOOH because of the low adsorption energy of CO2. 3. When the CO + OH pathway is considered, OH would react easily with H atoms to form H2O due to the hydrogen coverage effect. Finally, the removal of dissociated O to formic acid is proposed via improving the catalyst to increase the CO2 adsorption energy or CO coverage.

Distributed memory, GPU accelerated Fock construction for hybrid, Gaussian basis density functional theory

With the growing reliance of modern supercomputers on accelerator-based architecture such a graphics processing units (GPUs), the development and optimization of electronic structure methods to exploit these massively parallel resources has become a recent priority. While significant strides have been made in the development GPU accelerated, distributed memory algorithms for many modern electronic structure methods, the primary focus of GPU development for Gaussian basis atomic orbital methods has been for shared memory systems with only a handful of examples pursing massive parallelism. In the present work, we present a set of distributed memory algorithms for the evaluation of the Coulomb and exact exchange matrices for hybrid Kohn-Sham DFT with Gaussian basis sets via direct density-fitted (DF-J-Engine) and seminumerical (sn-K) methods, respectively. The absolute performance and strong scalability of the developed methods are demonstrated on systems ranging from a few hundred to over one thousand atoms using up to 128 NVIDIA A100 GPUs on the Perlmutter supercomputer.

Dynamic Precision Approach for Accelerating Large-Scale Eigenvalue Solvers in Electronic Structure Calculations on Graphics Processing Units

Single precision (SP) arithmetic can be greatly accelerated as compared to double precision (DP) arithmetic on graphics processing units (GPUs). However, the use of SP in the whole process of electronic structure calculations is inappropriate for the required accuracy. We propose a 3-fold dynamic precision approach for accelerated calculations but still with the accuracy of DP. Here, SP, DP, and mixed precision are dynamically switched during an iterative diagonalization process. We applied this approach to the locally optimal block preconditioned conjugate gradient method to accelerate a large-scale eigenvalue solver for the Kohn-Sham equation. We determined a proper threshold for switching each precision scheme by examining the convergence pattern on the eigenvalue solver only with the kinetic energy operator of the Kohn-Sham Hamiltonian. As a result, we achieved up to 8.53× and 6.60× speedups for band structure and self-consistent field calculations, respectively, for test systems under various boundary conditions on NVIDIA GPUs.

Increasing the Strain Resistance of Si/SiO2 Interfaces for Flexible Electronics

Understanding the changes that occur in the micro-mechanical properties of semiconductor materials is of utmost importance for the design of new flexible electronic devices, especially to control the properties of newly designed materials. In this work, we present the design, fabrication, and application of a novel tensile-testing device coupled to FTIR measurements that enables in situ atomic investigations of samples under uniaxial tensile load. The device allows for mechanical studies of rectangular samples with dimensions of 30 mm × 10 mm × 0.5 mm. By recording the alternation in dipole moments, the investigation of fracture mechanisms becomes feasible. Our results show that thermally treated SiO₂ on silicon wafers has a higher strain resistance and breaking force than the SiO₂ native oxide. The FTIR spectra of the samples during the unloading step indicate that for the native oxide sample, the fracture happened following the propagation of cracks from the surface into the silicon wafer. On the contrary, for the thermally treated samples, the crack growth starts from the deepest region of the oxide and propagates along the interface due to the change in the interface properties and redistribution of the applied stress. Finally, density functional theory calculations of model surfaces were conducted in order to unravel the differences in optic and electronic properties of the interfaces with and without applied stress.

Differences in solvation thermodynamics of oxygenates at Pt/Al2O3 perimeter versus Pt(111) terrace sites

A prominent role of water in aqueous-phase heterogeneous catalysis is to modify free energies; however, intuition about how is based largely on pure metal surfaces or even homogeneous solutions. Using multiscale modeling with explicit liquid water molecules, we show that the influence of water on the free energies of adsorbates at metal/support interfaces is different than that on pure metal surfaces. We specifically compute free energies of solvation for methanol and its constituents on a Pt/Al2O3 catalyst and compare the results to analogous values calculated on a pure Pt catalyst. We find that the more hydrophilic Pt/Al2O3 interface leads to smaller (more positive) free energies of solvation due to an increased entropy penalty resulting from the additional work necessary to disrupt the interfacial water structure and accommodate the interfacial species. The results will be of interest in other fields, including adsorption and proteins.

Large Scale Quantum Chemistry with Tensor Processing Units

We demonstrate the use of Googles cloud-based Tensor Processing Units (TPUs) to accelerate and scale up conventional (cubic-scaling) density functional theory (DFT) calculations. Utilizing 512 TPU cores, we accomplish the largest such DFT computation to date, with 247848 orbitals, corresponding to a cluster of 10327 water molecules with 103270 electrons, all treated explicitly. Our work thus paves the way toward accessible and systematic use of conventional DFT, free of any system-specific constraints, at unprecedented scales.

Quantum Perturbation Theory Using Tensor Cores and a Deep Neural Network

Time-independent quantum response calculations are performed using Tensor cores. This is achieved by mapping density matrix perturbation theory onto the computational structure of a deep neural network. The main computational cost of each deep layer is dominated by tensor contractions, i.e., dense matrix-matrix multiplications, in mixed-precision arithmetics, which achieves close to peak performance. Quantum response calculations are demonstrated and analyzed using self-consistent charge density-functional tight-binding theory as well as coupled-perturbed Hartree-Fock theory. For linear response calculations, a novel parameter-free convergence criterion is presented that is well-suited for numerically noisy low-precision floating point operations and we demonstrate a peak performance of almost 200 Tflops using the Tensor cores of two Nvidia A100 GPUs.

Towards universal neural network potential for material discovery applicable to arbitrary combination of 45 elements

Computational material discovery is under intense study owing to its ability to explore the vast space of chemical systems. Neural network potentials (NNPs) have been shown to be particularly effective in conducting atomistic simulations for such purposes. However, existing NNPs are generally designed for narrow target materials, making them unsuitable for broader applications in material discovery. Here we report a development of universal NNP called PreFerred Potential (PFP), which is able to handle any combination of 45 elements. Particular emphasis is placed on the datasets, which include a diverse set of virtual structures used to attain the universality. We demonstrated the applicability of PFP in selected domains: lithium diffusion in LiFeSO4F, molecular adsorption in metal-organic frameworks, an order-disorder transition of Cu-Au alloys, and material discovery for a Fischer-Tropsch catalyst. They showcase the power of PFP, and this technology provides a highly useful tool for material discovery.

The effect of oxygen impurities on the stability and structural properties of vacancy-ordered and -disordered ZrC x

Theoretical calculations predict several long-range ordered sub-stoichiometric zirconium carbide phases to be stable at low temperature, rather than a random (disordered solution) distribution of vacancies. However, experimental synthesis of vacancy-ordered phases is extremely challenging and not all predicted phases have been experimentally observed. It has been hypothesised that the inevitable oxygen contamination in experimental samples may affect the vacancy ordering. In this work, the stability and structural properties of the vacancy-ordered and vacancy-disordered phases are investigated as a function of oxygen defect concentration using first-principles calculations. The observed trends are explained in terms of changes to the local bonding in the presence of varying amounts of oxygen and vacancies. It is found that the relative stability of the ordered phases (compared to the disordered phase at the same composition) decreases as oxygen concentration increases, and some vacancy-ordered phases are destabilised by the level of oxygen impurities found in experimental samples. This suggests that oxygen contamination is a contributing factor to the challenge of synthesising ordered zirconium carbides, and gives insight that may assist fabrication in the future. The volume of all ZrC x (x ≤ 1) phases was found to decrease with increasing oxygen concentration, which can be attributed to the different ionocovalent nature of the C-Zr and O-Zr bonds. The volume of the vacancy-ordered phases within the expected oxygen solubility limit is greater than the disordered phase of the same composition, which is explained in terms of the relative bond strengths surrounding different vacancy distributions.

Translational dependence of the geometry of metallic mono- and bilayers optimized on semi-ionic supports: the cases of Pd on γ-Al2O3(110), monoclinic ZrO2(001), and rutile TiO2(001)

Geometries and cohesion energies while shifting (by 1 Å)/optimizing the Pd24 monolayer (2 × 2 unit cells) at γ-Al2O3(110) along the OX axis.

Inq, a Modern GPU-Accelerated Computational Framework for (Time-Dependent) Density Functional Theory

We present inq, a new implementation of density functional theory (DFT) and time-dependent DFT (TDDFT) written from scratch to work on graphic processing units (GPUs). Besides GPU support, inq makes use of modern code design features and takes advantage of newly available hardware. By designing the code around algorithms, rather than against specific implementations and numerical libraries, we aim to provide a concise and modular code. The result is a fairly complete DFT/TDDFT implementation in roughly 12 000 lines of open-source C++ code representing a modular platform for community-driven application development on emerging high-performance computing architectures.

Hole-Pinned Defect Clusters for a Large Dielectric Constant up to GHz in Zinc and Niobium Codoped Rutile SnO2

High permittivity materials for a gigahertz (GHz) communication technology have been actively sought for some time. Unfortunately, in most materials, the dielectric constant starts to drop as frequencies increase through the megahertz (MHz) range. In this work, we report a large dielectric constant of ∼800 observed in defect-mediated rutile SnO₂ ceramics, which is nearly frequency and temperature independent over the frequency range of 1 mHz to 35 GHz and temperature range of 50-450 K. Experimental and theoretical investigations demonstrate that the origin of the high dielectric constant can be attributed to the formation of locally well-defined Zn²⁺-Nb⁴⁺ defect clusters, which create hole-pinned defect dipoles. We believe that this work provides a promising strategy to advance dipole polarization theory and opens up a direction for the design and development of high frequency, broadband dielectric materials for use in future communication technology.

Quantum-Based Molecular Dynamics Simulations Using Tensor Cores

Tensor cores, along with tensor processing units, represent a new form of hardware acceleration specifically designed for deep neural network calculations in artificial intelligence applications. Tensor cores provide extraordinary computational speed and energy efficiency but with the caveat that they were designed for tensor contractions (matrix-matrix multiplications) using only low-precision floating-point operations. Despite this perceived limitation, we demonstrate how tensor cores can be applied with high efficiency to the challenging and numerically sensitive problem of quantum-based Born-Oppenheimer molecular dynamics, which requires highly accurate electronic structure optimizations and conservative force evaluations. The interatomic forces are calculated on-the-fly from an electronic structure that is obtained from a generalized deep neural network, where the computational structure naturally takes advantage of the exceptional processing power of the tensor cores and allows for high performance in excess of 100 Tflops on a single Nvidia A100 GPU. Stable molecular dynamics trajectories are generated using the framework of extended Lagrangian Born-Oppenheimer molecular dynamics, which combines computational efficiency with long-term stability, even when using approximate charge relaxations and force evaluations that are limited in accuracy by the numerically noisy conditions caused by the low-precision tensor core floating-point operations. A canonical ensemble simulation scheme is also presented, where the additional numerical noise in the calculated forces is absorbed into a Langevin-like dynamics.

Stability and structural properties of vacancy-ordered and -disordered ZrC x

Vacancy-ordered superstructural phases of zirconium carbide have been intermittently observed at low temperatures for over 50 years. However, little is known about these ordered phases as they have proven to be challenging to fabricate experimentally, although theoretical predictions suggest that they should be significantly more stable than the more-observed vacancy-disordered solid solution ZrC x (x ≤ 1) phase at low temperatures. The stability and structural properties of the vacancy-ordered and vacancy-disordered phases are investigated using first-principles calculations. The stability of the ordered superstructural phases is related to the driving force from the relative instability of certain vacancy configurations, which are preferred or avoided in ordered structures. The trend of the vacancy ordering and the underlying mechanisms of the relative instability are explained in terms of the geometry of the crystal structures and the electronic charge distribution and atomic bonding features.

An evaluation of solvent effects and ethanol oxidation

Understanding liquid-metal interfaces in catalysis is important, as the liquid can speed up surface reactions, increase the selectivity of products, and open up new favorable reaction pathways. In this work we modeled using density functional theory various steps in ethanol oxidation/decomposition over Rh(111). We considered implicit (continuum), explicit, and hybrid (implicit combined with explicit) solvation approaches, as well as two solvents, water and ethanol. We focused on modeling adsorption steps, as well as C-C/C-H bond scission and C-O bond formation reactions. Implicit solvation had very little effect on adsorption and reaction free energies. However, using the explicit and hybrid models, some free energies changed significantly. Furthermore, ethanol solvent had a more considerable impact than water solvent. We observed that preferred reaction pathways for C-C scission changed depending on the solvation model and solvent choice (ethanol or water). We also applied the bond-additivity solvation method to calculate heats of adsorption. Heats of adsorption and reaction using the bond-additivity model followed the same trends as the other solvation models, but were ∼1.1 eV more endothermic. Our work highlights how different solvation approaches can influence analysis of the oxidation/decomposition of organic surface species.

Harnessing the Power of Multi-GPU Acceleration into the Quantum Interaction Computational Kernel Program

We report a new multi-GPU capable ab initio Hartree-Fock/density functional theory implementation integrated into the open source QUantum Interaction Computational Kernel (QUICK) program. Details on the load balancing algorithms for electron repulsion integrals and exchange correlation quadrature across multiple GPUs are described. Benchmarking studies carried out on up to four GPU nodes, each containing four NVIDIA V100-SXM2 type GPUs demonstrate that our implementation is capable of achieving excellent load balancing and high parallel efficiency. For representative medium to large size protein/organic molecular systems, the observed parallel efficiencies remained above 82% for the Kohn-Sham matrix formation and above 90% for nuclear gradient calculations. The accelerations on NVIDIA A100, P100, and K80 platforms also have realized parallel efficiencies higher than 68% in all tested cases, paving the way for large-scale ab initio electronic structure calculations with QUICK.

Mixed Precision Fermi-Operator Expansion on Tensor Cores from a Machine Learning Perspective

We present a second-order recursive Fermi-operator expansion scheme using mixed precision floating point operations to perform electronic structure calculations using tensor core units. A performance of over 100 teraFLOPs is achieved for half-precision floating point operations on Nvidia's A100 tensor core units. The second-order recursive Fermi-operator scheme is formulated in terms of a generalized, differentiable deep neural network structure, which solves the quantum mechanical electronic structure problem. We demonstrate how this network can be accelerated by optimizing the weight and bias values to substantially reduce the number of layers required for convergence. We also show how this machine learning approach can be used to optimize the coefficients of the recursive Fermi-operator expansion to accurately represent the fractional occupation numbers of the electronic states at finite temperatures.

Efficient approaches to solutions of partition function for condensed matters

The key problem of statistical physics standing over one hundred years is how to exactly calculate the partition function (or free energy), which severely hinders the theory to be applied to predict the thermodynamic properties of condensed matters. Very recently, we developed a direct integral approach (DIA) to the solutions and achieved ultrahigh computational efficiency and precision. In the present work, the background and the limitations of DIA were examined in details, and another method with the same efficiency was established to overcome the shortage of DIA for condensed system with lower density. The two methods were demonstrated with empirical potentials for solid and liquid cooper, solid argon and C₆₀ molecules by comparing the derived internal energy or pressure with the results of vast molecular dynamics simulations, showing that the precision is about ten times higher than previous methods in a temperature range up to melting point. The ultrahigh efficiency enables the two methods to be performed with ab initio calculations and the experimental equation of state of solid copper up to ∼600 GPa was well reproduced, for the first time, from the partition function via density functional theory implemented.

Examination of permittivity for depolarization field of ferroelectric by ab initio calculation, suggesting hidden mechanisms

Electrostatics of depolarization field Ed in relation to the polarization is studied. In particular, the value of permittivity for Ed (εd) in prototypical situations of ferroelectrics, including Mehta formula, is examined by ab initio calculations. By using spontaneous polarization PS corresponding to accurate experiment ones, we show εd = 1, which suggests that the results of εd ≫ 1 indicate hidden mechanisms; εd = 1 suggests that the effect of Ed is significant to induce intriguing important phenomena overlooked by εd ≫ 1. A bridge between εd = 1 and εd ≫ 1, i.e. the consistency of εd = 1 with conventional results is presented. The exact electrostatic equality of head-to-head-tail-to-tail domains to free-standing ferroelectrics is deduced. Hence, most stoichiometric clean freestanding monodomain ferroelectrics and head-to-head-tail-to-tail domains are shown unstable regardless of size, unless partially metallic. This verifies the previous results in a transparent manner. This conclusion is shown consistent with a recent hyperferroelectric LiBeSb and "freestanding" monolayer ferroelectrics, of which origin is suggested to be adsorbates. In addition, this restriction is suggested to break in externally strained ultrathin ferroelectrics. The macroscopic formulas of Ed are found valid down to a several unit-cells, when electronic and atomic-scale surface effects are unimportant and accurate PS is used.

Nanostructured LiMnO2 with Li3PO4 Integrated at the Atomic Scale for High-Energy Electrode Materials with Reversible Anionic Redox

Nanostructured LiMnO₂ integrated with Li₃PO₄ was successfully synthesized by the mechanical milling route and examined as a new series of positive electrode materials for rechargeable lithium batteries. Although uniform mixing at the atomic scale between LiMnO₂ and Li₃PO₄ was not anticipated because of the noncompatibility of crystal structures for both phases, our study reveals that phosphorus ions with excess lithium ions dissolve into nanosize crystalline LiMnO₂ as first evidenced by elemental mapping using STEM-EELS combined with total X-ray scattering, solid-state NMR spectroscopy, and a theoretical ab initio study. The integrated phase features a low-crystallinity metastable phase with a unique nanostructure; the phosphorus ion located at the tetrahedral site shares faces with adjacent lithium ions at slightly distorted octahedral sites. This phase delivers a large reversible capacity of ∼320 mA h g^-1 as a high-energy positive electrode material in Li cells. The large reversible capacity originated from the contribution from the anionic redox of oxygen coupled with the cationic redox of Mn ions, as evidenced by operando soft XAS spectroscopy, and the superior reversibility of the anionic redox and the suppression of oxygen loss were also found by online electrochemical mass spectroscopy. The improved reversibility of the anionic redox originates from the presence of phosphorus ions associated with the suppression of oxygen dimerization, as supported by a theoretical study. From these results, the mechanistic foundations of nanostructured high-capacity positive electrode materials were established, and further chemical and physical optimization may lead to the development of next-generation electrochemical devices.

Observation of Molecular Hydrogen Produced from Bridging Hydroxyls on Anatase TiO2(101)

Anatase TiO₂ is used extensively in a wide range of catalytic and photocatalytic processes and is a promising catalyst for hydrogen production. Here, we show that molecular hydrogen was produced from bridging hydroxyls (HO_b) on the (101) surface of single-crystal anatase (TiO₂(101)). This stands in contrast to rutile TiO₂(110), where HO_b pairs react to form H₂O. Electron bombardment at 30 K produced bridging oxygen vacancies in the surface. Deuterated bridging hydroxyls (DO_b) were subsequently formed via dissociation of adsorbed D₂O and confirmed by infrared reflection-absorption spectroscopy. During temperature-programmed desorption (TPD) spectroscopy, D₂ desorption was observed at 520 K. Density functional theory calculations show that both H₂ and H₂O production from HO_b are endothermic at 0 K on TiO₂(101), but H₂ (H₂O) desorption is entropically driven above 230 K (800 K). The calculated activation barrier for H₂ desorption is 1.40 eV, which is similar to the desorption energy obtained from analysis of the D₂ TPD spectra. The H₂ desorption likely proceeds in two steps: H atom diffusion on the surface and then recombination.

Boosting Upconversion Photoluminescence and Multielectrical Properties via Er-Doping-Modulated Vacancy Control in Ba0.85Ca0.15Ti0.9Zr0.1O3

The lead-free 0.5(Ba_0.7Ca_0.3)TiO₃-0.5Ba(Ti_0.8Zr_0.2)O₃ (BCTZ) ceramics with Er doping have shown good upconversion photoluminescence (PL) and desirable optical temperature sensing properties. To bridge a relationship between the structure/intrinsic defects and properties of rare-earth-doped ferroelectrics, we designed and fabricated a series of BCTZ ceramics doped with 1 mol % Er³⁺ by combining the first-principles calculations and experimental measurements. Theoretically, we discovered that Er can occupy both A sites (i.e., replacing Ba or Ca) and B sites (i.e., replacing Ti or Zr) in the BCTZ lattice and highlighted that the Er-doping-induced vacancy concentration decreases for both the oxygen vacancies (V _o) and cation vacancies (V _c). Experimentally, the enhanced PL performance and the dielectric, ferroelectric, and piezoelectric properties of the Er-doped BCTZ ceramics have been observed. Finally, the physical origin of Er-induced property enhancement in BCTZ has been elaborated according to the charge density and chemical bonding analysis. These results open up a path to investigate the effects of site substitution and vacancies on optoelectronic properties of multifunctional rare-earth-doped ferroelectrics.

Amorphous graphene: a constituent part of low density amorphous carbon

In this paper, we provide evidence that low density nano-porous amorphous carbon (a-C) consists of interconnected regions of amorphous graphene (a-G). We include experimental information in producing models, while retaining the power and accuracy of ab initio methods with no biasing assumptions. Our models are highly disordered with predominant sp2 bonding and ring connectivity mainly of sizes 5-8. The structural, dynamical and electronic signatures of our 3-D amorphous graphene are similar to those of monolayer amorphous graphene. We predict an extended x-ray absorption fine structure (EXAFS) signature of amorphous graphene. Electronic density of states calculations for 3-D amorphous graphene reveal similarity to monolayer amorphous graphene and the system is non conducting.

Predicted detonation properties at the Chapman-Jouguet state for proposed energetic materials (MTO and MTO3N) from combined ReaxFF and quantum mechanics reactive dynamics

The development of new energetic materials (EMs) with improved detonation performance but low sensitivity and environmental impact is of considerable importance for applications in civilian and military fields. Often new designs are difficult to synthesize so predictions of performance in advance is most valuable. Examples include MTO (2,4,6-triamino-1,3,5-triazine-1,3,5-trioxide) and MTO3N (2,4,6-trinitro-1,3,5-triazine-1,3,5-trioxide) suggested by Klapötke as candidate EMs but not yet successfully synthesized. We propose and apply to these materials a new approach, RxMD(cQM), in which ReaxFF Reactive Molecular Dynamics (RxMD) is first used to predict the reaction products and thermochemical properties at the Chapman Jouguet (CJ) state for which the system is fully reacted and at chemical equilibrium. Quantum mechanics dynamics (QMD) is then applied to refine the pressure of the ReaxFF predicted CJ state to predict a more accurate final CJ point, leading to a very practical calculation that includes accurate long range vdW interactions needed for accurate pressure. For MTO, this RxMD(cQM) method predicts a detonation pressure of P_CJ = 40.5 GPa and a detonation velocity of D_CJ = 8.8 km s^-1, while for MTO3N it predicts P_CJ = 39.9 GPa and D_CJ = 8.4 km s^-1, making them comparable to HMX (P_CJ = 39.5 GPa, D_CJ = 9.1 km s^-1) and worth synthesizing. This first-principles-based RxMD(cQM) methodology provides an excellent compromise between computational cost and accuracy including the formation of clusters that burn too slowly, providing a practical mean of assessing detonation performances for novel candidate EMs. This RxMD(cQM) method that links first principles atomistic molecular dynamics simulations with macroscopic properties to promote in silico design of new EMs should also be of general applicability to materials synthesis and processing.

Shedding Light on the Protonation States and Location of Protonated N Atoms of Adenine in Metal-Organic Frameworks

We report the syntheses and structures of five metal–organic frameworks (MOFs) based on transition metals (Ni^II, Cu^II, and Zn^II), adenine, and di-, tri-, and tetra-carboxylate ligands. Adenine, with multiple N donor sites, was found to coordinate to the metal centers in different binding modes including bidentate (through N7 and N9, or N3 and N9) and tridentate (through N3, N7, and N9). Systematic investigations of the protonation states of adenine in each MOF structure via X-ray photoelectron spectroscopy revealed that adenine can be selectively protonated through N1, N3, or N7. The positions of H atoms connected to the N atoms were found from the electron density maps, and further supported by the study of C–N–C bond angles compared to the literature reports. DFT calculations were performed to geometrically optimize and energetically assess the structures simulated with different protonation modes. The present study highlights the rich coordination chemistry of adenine and provides a method for the determination of its protonation states and the location of protonated N atoms of adenine within MOFs, a task that would be challenging in complicated adenine-based MOF structures.

The protonation states and positions of hydrogen atoms in five adenine-based metal−organic frameworks were revealed using geometrical studies based on single-crystal XRD data supported by XPS spectra and DFT calculations.

Colossal permittivity behavior and its origin in rutile (Mg1/3Ta2/3)xTi1-xO2

This work investigates the synthesis, chemical composition, defect structures and associated dielectric properties of (Mg²⁺, Ta⁵⁺) co-doped rutile TiO₂ polycrystalline ceramics with nominal compositions of (Mg²⁺_1/3Ta⁵⁺_2/3)_xTi_1−xO₂. Colossal permittivity (>7000) with a low dielectric loss (e.g. 0.002 at 1 kHz) across a broad frequency/temperature range can be achieved at x = 0.5% after careful optimization of process conditions. Both experimental and theoretical evidence indicates such a colossal permittivity and low dielectric loss intrinsically originate from the intragrain polarization that links to the electron-pinned \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\bf{M}}{{\bf{g}}}_{{\bf{T}}{\bf{i}}}^{{\prime}{\prime} }+{{\bf{V}}}_{{\bf{O}}}^{\bullet \bullet }+{\bf{2}}{\bf{T}}{{\bf{a}}}_{{\bf{T}}{\bf{i}}}^{\bullet }+{\bf{2}}{\bf{T}}{{\bf{i}}}_{{\bf{T}}{\bf{i}}}^{\prime}$$\end{document}MgTi′′+VO••+2TaTi•+2TiTi′ defect clusters with a specific configuration, different from the defect cluster form previously reported in tri-/pent-valent ion co-doped rutile TiO₂. This work extends the research on colossal permittivity and defect formation to bi-/penta-valent ion co-doped rutile TiO₂ and elucidates a likely defect cluster model for this system. We therefore believe these results will benefit further development of colossal permittivity materials and advance the understanding of defect chemistry in solids.

Bimetallic Ions Codoped Nanocrystals: Doping Mechanism, Defect Formation, and Associated Structural Transition

Ionic codoping offers a powerful approach for modifying material properties by extending the selection of potential dopant ions. However, it has been a major challenge to introduce certain ions that have hitherto proved difficult to use as dopants (called "difficult-dopants") into crystal structures at high concentrations, especially through wet chemical synthesis. Furthermore, the lack of a fundamental understanding of how codopants are incorporated into host materials, which types of defect structures they form in the equilibrium state, and what roles they play in material performance, has seriously hindered the rational design and development of promising codoped materials. Here we take In³⁺ (difficult-dopants) and Nb⁵⁺ (easy-dopants) codoped anatase TiO₂ nanocrystals as an example and investigate the doping mechanism of these two different types of metal ions, the defect formation, and their associated impacts on high-pressure induced structural transition behaviors. It is experimentally demonstrated that the dual mechanisms of nucleation and diffusion doping are responsible for the synergic incorporation of these two dopants and theoretically evidenced that the defect structures created by the introduced In³⁺, Nb⁵⁺ codopants, their resultant Ti³⁺, and oxygen vacancies are locally composed of both defect clusters and equivalent defect pairs. These formed local defect structures then act as nucleation centers of baddeleyite- and α-PbO₂-like metastable polymorphic phases and induce the abnormal trans-regime structural transition of codoped anatase TiO₂ nanocrystals under high pressure. This work thus suggests an effective strategy to design and synthesize codoped nanocrystals with highly concentrated difficult-dopants. It also unveils the significance of local defect structures on material properties.