High-throughput Calculations Research Articles

2C-Ternary Content Addressable Memory in Memcapacitor Crossbar Array with NAND Flash Structure.

As one of the data-intensive in-memory computing hardware, ternary content addressable memory (TCAM) stands out for its efficient in-memory-searching capability, enabling high-throughput and low-latency computing. However, TCAMs, especially those based on resistive non-volatile memories, face challenges in limited resistance ratio (Rhigh/Rlow) that deteriorate sensing margin and energy efficiency. Addressing these issues, a TCAM cell composed of two memcapacitors (2C-TCAM) based on NAND flash array structure is proposed. The 2C-TCAM utilizes the memcapacitors coupled with the mature technology of flash cells for reliable operation and high-density (8F2) array configuration. Thanks to the capacitive readout of memcapacitors, the 2C-TCAM achieves near-zero static power consumption and minimizes IR drop effect. Consequently, highly parallel and reliable search functionality can be obtained even in large arrays while preserving the sensing margin. Electrical characteristics and operation schemes of the proposed 2C-TCAM cell are validated through fabrication and measurements, and array operations are experimentally demonstrated using a 24 × 48 memcapacitor crossbar array with sensing circuits. Additionally, the system-level performance of the 2C-TCAM array is analyzed, considering the device programming accuracy. Search times of 47ps and energy consumption of 11.7 fJ per bit are achieved by scaling down the device cell area to 1 µm2.

Screening heat-resistant energetic molecules via deep learning and high-throughput computation

Screening heat-resistant energetic molecules via deep learning and high-throughput computation

Accelerated Design of Dual-Metal-Site Catalysts via Machine-Learning Prediction.

Dual-metal site catalysts (DMSCs) supported on nitrogen-doped graphene have shown great potential in heterogeneous catalysis due to their unique properties and enhanced efficiency. However, the precise control and stabilization of metal dimers, particularly in oxygen activation reactions, present significant challenges in practical applications. In this study, we integrate high-throughput density functional theory calculations with machine learning techniques to predict and optimize the catalytic properties of DMSCs. Transfer learning is employed to enhance the model's generalization capability, successfully predicting catalytic performance across new metal combinations. Additionally, the application of the SISSO method enables the derivation of interpretable symbolic regression models, revealing critical correlations between electronic structure features and catalytic efficiency. This approach not only advances the understanding of dual-metal site catalysis but also provides a novel framework for the systematic design and optimization of highly efficient catalysts, with broad applicability in catalytic science.

Machine learning Hubbard parameters with equivariant neural networks

Density-functional theory with extended Hubbard functionals (DFT + U + V) provides a robust framework to accurately describe complex materials containing transition-metal or rare-earth elements. It does so by mitigating self-interaction errors inherent to semi-local functionals which are particularly pronounced in systems with partially-filled d and f electronic states. However, achieving accuracy in this approach hinges upon the accurate determination of the on-site U and inter-site V Hubbard parameters. In practice, these are obtained either by semi-empirical tuning, requiring prior knowledge, or, more correctly, by using predictive but expensive first-principles calculations. Here, we present a machine learning model based on equivariant neural networks which uses atomic occupation matrices as descriptors, directly capturing the electronic structure, local chemical environment, and oxidation states of the system at hand. We target here the prediction of Hubbard parameters computed self-consistently with iterative linear-response calculations, as implemented in density-functional perturbation theory (DFPT), and structural relaxations. Remarkably, when trained on data from 12 materials spanning various crystal structures and compositions, our model achieves mean absolute relative errors of 3% and 5% for Hubbard U and V parameters, respectively. By circumventing computationally expensive DFT or DFPT self-consistent protocols, our model significantly expedites the prediction of Hubbard parameters with negligible computational overhead, while approaching the accuracy of DFPT. Moreover, owing to its robust transferability, the model facilitates accelerated materials discovery and design via high-throughput calculations, with relevance for various technological applications.

Two-Dimensional Transition Metal Dichalcogenides: A Theory and Simulation Perspective.

Two-dimensional transition metal dichalcogenides (2D TMDs) are a promising class of functional materials for fundamental physics explorations and applications in next-generation electronics, catalysis, quantum technologies, and energy-related fields. Theory and simulations have played a pivotal role in recent advancements, from understanding physical properties and discovering new materials to elucidating synthesis processes and designing novel devices. The key has been developments in ab initio theory, deep learning, molecular dynamics, high-throughput computations, and multiscale methods. This review focuses on how theory and simulations have contributed to recent progress in 2D TMDs research, particularly in understanding properties of twisted moiré-based TMDs, predicting exotic quantum phases in TMD monolayers and heterostructures, understanding nucleation and growth processes in TMD synthesis, and comprehending electron transport and characteristics of different contacts in potential devices based on TMD heterostructures. The notable achievements provided by theory and simulations are highlighted, along with the challenges that need to be addressed. Although 2D TMDs have demonstrated potential and prototype devices have been created, we conclude by highlighting research areas that demand the most attention and how theory and simulation might address them and aid in attaining the true potential of 2D TMDs toward commercial device realizations.

Machine learning models for easily obtainable descriptors of the electrocatalytic properties of Ag-Pd-Ir nanoalloys toward the formate oxidation reaction.

Direct formate fuel cells (DFFCs) have received increasing attention due to their environmentally benign and highly safe characteristics. However, the absence of highly active electrocatalysts for the formate oxidation reaction (FOR) restricts their widespread application. Currently, the design of FOR catalysts, which relies on experimental trial-and-error and high-throughput DFT calculations, is costly and time-consuming. In this study, based on a DFT dataset of FOR overpotentials for 137 Ag-Pd-Ir nanoalloy catalysts, six machine learning (ML) models were trained, where the K-nearest neighbors (KNN) model demonstrated the best performance, with an R2 value of 0.94, an MAE value of 0.041 V and an RMSE value of 0.050 V. Using the KNN model, six optimal catalysts with an overpotential of 0.48 V were screened from 310 candidate catalysts, with an MAE value as low as 0.004 V compared to the DFT results, proving the accuracy of the ML model. This work provides a novel strategy to accelerate the design of high-performance catalysts.

Data-Driven Discovery of High-Performance Heterobilayer Transition Metal Dichalcogenide-Based Sliding Ferroelectrics.

The development of efficient sliding ferroelectric (FE) materials is crucial for advancing next-generation low-power nanodevices. Currently, most efforts focus on homobilayer two-dimensional materials, except for the experimentally reported heterobilayer sliding FE, MoS2/WS2. Here, we first screened 870 transition metal dichalcogenide (TMD) bilayer heterostructures derived from experimentally characterized monolayer TMDs and systematically investigated their sliding ferroelectric behavior across various stacking configurations using high-throughput calculations. On the basis of the generated data, we developed an efficient descriptor, named the amplitude of Allen electronegativity difference (Δχm), for identifying van der Waals heterobilayers with sliding FE properties. Finally, 16 semiconducting TMD heterobilayers are identified as exhibiting interlayer sliding FE alongside low switching barriers (<21 meV/f.u.), with 10 outperforming the experimental MoS2/WS2 system, showing the largest out-of-plane polarization (OPP) values up to 10 times higher than MoS2/WS2. These materials exhibit favorable band gaps (0.60-1.80 eV) using the HSE06 method, making them suitable for sliding FE applications. Our findings reveal that polarization switching in these heterobilayers is strongly influenced by the interplay of stacking patterns, material electronegativity, charge transfer, and electronic structures. This study provides a robust framework for designing novel sliding ferroelectric materials and offers a theoretical basis for future experimental research.

Design of High-Temperature Superconducting Ternary Hydride NaY3H20 at Moderate Pressure via Introducing Hydrogen Vacancies.

Superconducting hydrides exhibiting a high critical temperature (Tc) under extreme pressures have garnered significant interest. However, the extremely high pressures required for their stability have limited their practical applications. The current challenge is to identify high-Tc superconducting hydrides that can be stabilized at lower or even ambient pressures. Here, we propose a strategy for designing high-Tc superconducting hydrides at low pressures by introducing defects into the hydrogen frameworks of clathrate hydrides. We present a type of hydrogen-vacancy structural type AB3H20 derived from type-I clathrate hydrides and identified a stable NaY3H20 through high-throughput calculations. Further calculations show that NaY3H20 is thermodynamically stable above 133 GPa and dynamically stable down to 20 GPa, with a predicted high Tc of approximately 115 K. It significantly reduces the pressure required for stability compared to that of type-I clathrate hydrides with high Tc. Our results provide a foundation for further exploration of high-Tc superconducting hydrides at lower pressures or even ambient conditions.

The art and science of translucent color organic solar cells

The artistic and scientific perspectives of the translucent color organic solar cells (OSCs), made with the emerging narrowband nonfullerene acceptors are explored. The translucent color OSCs, comprising a Fabry–Pérot microcavity optical coupling layer, have a power conversion efficiency of >15% and a maximum transparency of >20% for the three primary colors. The performance−color relationship of the translucent color OSCs is analyzed using a combination of high-throughput optical computing and experimental optimization, allowing light with desired color to pass through, while absorbing enough light to generate electricity. Replication of Piet Mondrian’s artwork “Composition C (1920)” is demonstrated using a 10 × 10 cm2-sized translucent OSC module with a wide palette of colors and hues. The outcome of the work offers an opportunity for translucent color OSCs to function as both esthetic art and power generating windowpanes for use in our homes, offices, and even greenhouses.

Application of machine learning in polyimide structure design and property regulation

Polyimide (PI) is widely used in modern industry due to its excellent properties. Its synthesis methods and property research have significantly progressed. However, the design and regulation of PI structures through traditional technologies are slow and expensive, which make it difficult to meet the practical demand of modern materials. With the rapid development of high-throughput computing and data-driven technology, machine learning (ML) has become an important method for exploring new materials. Data-driven ML is envisaged as a decisive enabler for new PIs discovery. This paper first introduces the basic workflow and common algorithms of ML. Secondly, applications of ML in material properties prediction, assisting computational simulation technologies and inverse design for desired structures are reviewed. Finally, we discuss the main challenges and possible solutions of ML in PI research.

High-throughput computation and machine learning screening of van der Waals heterostructures for Z-scheme photocatalysis

High-throughput computation and machine learning screening of van der Waals heterostructures for Z-scheme photocatalysis

Screening corrosion-resistant genetic elements of Sn-0.7Cu solders by high-throughput calculations

Screening corrosion-resistant genetic elements of Sn-0.7Cu solders by high-throughput calculations

High-throughput calculations and machine learning modeling of 17O NMR in non-magnetic oxides.

The only NMR-active oxygen isotope, oxygen-17 (17O), serves as a sensitive probe due to its large chemical shift range, the electric field gradient at the oxygen site, and the quadrupolar interaction. Consequently, 17O solid-state NMR offers unique insights into local structures and finds significant applications in the studies of disorder, reactivity, and host-guest chemistry. Despite recent advances in sensitivity enhancement, isotopic labeling, and NMR crystallography, the application of 17O solid-state NMR is still hindered by low natural abundance, costly enrichment, and challenges in handling spectrum signals. Density functional theory calculations and machine learning techniques offer an alternative approach to mapping the local crystal structures to NMR parameters. However, the lack of high-quality data remains a challenge, despite the establishment of some datasets. In this study, we implement and execute a high-throughput workflow combining AiiDA and CASTEP to evaluate the NMR parameters. Focusing on non-magnetic oxides, we have chosen over 7100 binary, ternary, and quaternary compounds from the Materials Project database and performed calculations. Furthermore, using various descriptors for the local crystalline environments, we model the 17O NMR parameters using machine learning techniques, further enhancing our ability to predict and understand 17O NMR parameters in oxide crystals.

Theoretical prediction of efficient Cu-based dual-atom alloy catalysts for electrocatalytic nitrate reduction to ammonia via high-throughput first-principles calculations

The dual-atom sites in Cu-based dual-atom alloy not only exhibit high activity for electrocatalytic nitrate (NO3−) reduction reaction (eNO3RR) to produce valuable ammonia (NH3), but also can effectively inhibit the formation of by-products.

An Area-Efficient VLSI Architecture for High-Throughput Computation of the 2-D DWT

An Area-Efficient VLSI Architecture for High-Throughput Computation of the 2-D DWT

Optical materials discovery and design with federated databases and machine learning.

Combinatorial and guided screening of materials space with density-functional theory and related approaches has provided a wealth of hypothetical inorganic materials, which are increasingly tabulated in open databases. The OPTIMADE API is a standardised format for representing crystal structures, their measured and computed properties, and the methods for querying and filtering them from remote resources. Currently, the OPTIMADE federation spans over 20 data providers, rendering over 30 million structures accessible in this way, many of which are novel and have only recently been suggested by machine learning-based approaches. In this work, we outline our approach to non-exhaustively screen this dynamic trove of structures for the next-generation of optical materials. By applying MODNet, a neural network-based model for property prediction, within a combined active learning and high-throughput computation framework, we isolate particular structures and chemistries that should be most fruitful for further theoretical calculations and for experimental study as high-refractive-index materials. By making explicit use of automated calculations, federated dataset curation and machine learning, and by releasing these publicly, the workflows presented here can be periodically re-assessed as new databases implement OPTIMADE, and new hypothetical materials are suggested.

Combining machine-learning models with first-principles high-throughput calculations to accelerate the search for promising thermoelectric materials

From high-throughput screening of high-power-factor materials, through first-principles calculation of transport properties, to training machine-learning models for identifying good thermoelectric materials.

Identification of BRCA new prognostic targets and neoantigen candidates from fusion genes

Cancer-associated gene fusions serve as a potential source of highly immunogenic neoantigens. In this study, we identified fusion proteins from fusion genes and extracted fusion peptides to accurately predict Breast cancer (BRCA) neo-antigen candidates by high-throughput artificial intelligence computation. Firstly, Deepsurv was used to evaluate the prognosis of patients, providing a landscape of prognostic fusion genes in BRCA. Next, AGFusion was utilized to generate full-length fusion protein sequences and annotate functional domains. Advanced neural networks and Transformer-based analyses were implemented to predict the binding of fusion peptides to 112 types of HLA, thereby forming a new immunotherapy candidates’ library of BRCA neo-antigens (n = 7791, covering 88.41% of patients). Among them, 15 neo-antigens were validated and factually translated into mass spectrometry data of BRCA patients. Finally, AlphaFold2 was applied to predict the binding sites of these neo-antigens to MHC (HLA) molecules. Notably, we identified a prognostic neoantigen from the TBC1D4–COMMD6 fusion that significantly improves patient prognosis and extensively binds to 16 types of HLA alleles. These highly immunogenic and tumor-specific neoantigens offer emerging targets for personalized cancer immunotherapies and act as prospective predictors for tumor survival prognosis and responses to immune checkpoint therapies.

Convergence Analysis and Predictions for Optimizing Reciprocal Grids: A First-Principles and Machine Learning Study.

When crystalline materials are investigated by performing first-principles density functional theory (DFT) calculations, the reciprocal grid should be fine enough to obtain the converged total energy and electronic structure. Herein, we performed a convergence test of the total energy for the density of reciprocal points to determine fine enough reciprocal grids for high-throughput calculations. Our results show that the nonlinearity of the band structures affects the convergence of the total energy, especially for materials with a finite band gap. We further investigate which physical properties make a finer reciprocal grid necessary based on machine learning (ML) analysis. Our developed models using DFT-based features and elemental properties-based features exhibit R2 of 0.803 and 0.880, respectively. Our ML model quantitatively shows the importance of nonlinearity and band gaps in predicting errors in total energy calculations. Furthermore, our ML model using elemental features can be applied to estimate the appropriate reciprocal grid, facilitating high-throughput calculations.

Electric-Field-Induced Switchable Two-Dimensional Altermagnets.

Altermagnetism, as a recently discovered unconventional antiferromagnetism, allows the lifting of spin degeneracy without net magnetization. The spin splitting of the intrinsic altermagnets is protected by the spin space group symmetry and is therefore difficult to control externally. Here, we propose an extrinsic altermagnet as a complement to the intrinsic altermagnet, whose spin splitting is induced by and can be significantly modulated by the electric field. We then screened intrinsic and extrinsic altermagnets by combining symmetry analysis and high-throughput calculations, identifying 16 intrinsic altermagnets and 24 extrinsic altermagnets. We demonstrate that the spin splitting of extrinsic altermagnets is proportional to the electric field strength, and its sign can be switched by reversing the electric field. Some extrinsic altermagnets exhibit considerable spin splitting, such as CaMnSi with 398 meV under an electric field of 0.1 V/Å. This work provides a realistic material platform for the potential application of 2D altermagnets.