Search NASA⌕ Search

SEARCH · Search NASA

Results for “optimized energy density functionals”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,135 records · Page 63

Correlation functions from a unified variational principle: Trial Lie groups

Time-dependent expectation values and correlation functions for many-body quantum systems are evaluated by means of a unified variational principle. It optimizes a generating functional depending on sources associated with the observables of interest. It is built by imposing through Lagrange multipliers constraints that account for the initial state (at equilibrium or off equilibrium) and for the backward Heisenberg evolution of the observables. The trial objects are respectively akin to a density operator and to an operator involving the observables of interest and the sources. We work out here the case where trial spaces constitute Lie groups. This choice reduces the original degrees of freedom to those of the underlying Lie algebra, consisting of simple observables; the resulting objects are labeled by the indices of a basis of this algebra. Explicit results are obtained by expanding in powers of the sources. Zeroth and first orders provide thermodynamic quantities and expectation values in the form of mean-field approximations, with dynamical equations having a classical Lie–Poisson structure. At second order, the variational expression for two-time correlation functions separates–as does its exact counterpart–the approximate dynamics of the observables from the approximate correlations in the initial state. Two building blocks are involved: (i) a commutation matrix which stems from the structure constants of the Lie algebra; and (ii) the second-derivative matrix of a free-energy function. The diagonalization of both matrices, required for practical calculations, is worked out, in a way analogous to the standard RPA. The ensuing structure of the variational formulae is the same as for a system of non-interacting bosons (or of harmonic oscillators) plus, at non-zero temperature, classical Gaussian variables. This property is explained by mapping the original Lie algebra onto a simpler Lie algebra. The results, valid for any trial Lie group, fulfill consistency properties and encompass several special cases: linear responses, static and time-dependent fluctuations, zero- and high-temperature limits, static and dynamic stability of small deviations.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Creating Favorable Pt/Co Interfaces via a Two‐Step Approach for Constructing Highly Durable PtCo Intermetallic Fuel Cell Catalysts

Structurally ordered PtCo intermetallics are one of the most promising oxygen-reduction catalysts in proton exchange membrane fuel cells (PEMFCs) due to their intrinsically improved catalytic activity and stability relative to PtCo solid-solution alloys. However, increasing the heating temperature to achieve a desirable high degree of ordering results in severe particle agglomeration and low mass activity and stability. Herein, a two-step synthesis approach is developed to create an L1 2 -Pt 3 Co intermetallic structure with an increased ordering degree and well-dispersed ultrafine particles. The first step of the synthesis yields ultrafine Pt nanoparticles that are well-dispersed on the ZIF-8-derived carbon support. The second adsorption step enables us to fine-tune the Pt and Co interfaces, assisted by optimal amino acids, to establish a favorable Co-rich environment around fine Pt nanoparticles, facilitating Co diffusion into the Pt crystalline under mild thermal conditions (<800 °C). In conclusion, this two-step ordered L1 2 -Pt 3 Co catalyst is systematically evaluated using membrane electrode assemblies under heavy-duty vehicle (HDV) conditions and demonstrated exceptional performance and durability, retaining 1.35 A cm -2 only a 7% loss in current density at 0.7 V after an extensive accelerated stress test of 150,000 voltage cycles.

30 DIRECT ENERGY CONVERSION↗

Methane Adsorption on Heteroatom-Modified Maquettes of Porous Carbon Surfaces

Experimental and theoretical studies disagree on the energetics of methane adsorption on carbon materials. However, this information is critical for the rational design and optimization of the structure and composition of adsorbents for natural gas storage. The delicate nature of dispersion interactions, polarization of both the adsorbent and the adsorbate, interplay between H-bonding and tetrel bonding, and induced dipole/Coulomb interactions inherent to methane physisorption require computational treatment at the highest possible level of theory. In this study, we employed the smallest reasonable computational model, a maquette of porous carbon surfaces with a central site for substitution and methane binding. The most accurate predictions of methane adsorption energetics were achieved by electron-correlated molecular orbital theory CCSD(T) and hybrid density functional theory MN15 calculations employing a saturated, all-electron basis set. Here, the characteristic geometry of methane adsorption on a carbon surface (“lander approach”) arises due to bonding interactions of the adsorbent π-system with the proximal H–C bonds of methane, in addition to tetrel bonding between the antibonding orbital of the distal C–H bond and the central atom of the maquette (C, B, or N). The polarization of the electron density, structural deformations, and the comprehensive energetic analysis clearly indicate a ∼3 kJ mol –1 preference for methane binding on the N-substituted maquette. The B-substituted maquette showed a comparable or lower binding energy than the unsubstituted, pure C model, depending on the level of theory employed. The calculated thermodynamic results indicate a strategy for incorporating electron-enriched substitutions (e.g., N) into carbon materials as a way to increase methane storage capacity over electron-deficient (e.g., B) modifications. The thermochemical analysis was revised for establishing a conceptual agreement between the experimental isosteric heat of adsorption and the binding enthalpies from statistical thermodynamics principles.

03 NATURAL GAS↗

Development and Validation of a SiC Based 50 kW Grid-Connected PV Inverter: Preprint

In this digest a 50kW string PV inverter formed using all SiC semiconductor devices is presented. The inverter design includes additively manufactured power block, symmetrical Y-core inductors for ac side filter, and advanced inverter functionality for grid support. This inverter utilizes the conventional three-phase Voltage Source Inverter topology and optimizes the design for SiC based devices. This paper presents the full functional validation of the designed inverter along with quantification of the efficiency and power density. The performance of the inverter for high dc-bus voltage without multi-level topology to exploit SiC advantages is also presented in this digest. The full paper will include details of inverter design, optimization and further experimental validation.

14 SOLAR ENERGY↗

The Critical Stack Pressure to Alter Void Generation at Li/Solid-Electrolyte Interfaces during Stripping

The lithium stripping process generates vacancies, which may accumulate as voids and lead to uneven current distribution and dendrite growth in the following plating cycles. A stack pressure is typically required during stripping, but how to optimize the stack pressure is not clear. In this work, extremely lithiophilic Li/Li 2 O and lithiophobic Li/LiF interfaces were used to reveal the combining effect of interface interaction and stack pressure induced lithium creep on the stripping critical current density (CCD). A multiscale simulation scheme with Density Functional Theory (DFT), kinetic Monte Carlo (KMC) simulations, and an analytical model was developed. The analytical model predicted lithiophobic interfaces require a higher stack pressure than lithiophilic interfaces to reach the same CCD. The KMC simulations also showed higher stack pressure is needed at lithiophobic interfaces to accelerate Li vacancy diffusion into the bulk and maintain a flat surface. This stack pressure needs to be high enough to alter the Li forward-and-backward hopping barriers at the interface. This multiscale simulation scheme illustrates the importance to include the chemical-mechanical effects during Li stripping morphology evolution. It can be used to design ideal interlayer coating materials to maintain a flat Li surface during cycling.

25 ENERGY STORAGE↗

Combinatorial Insights into Doping Control and Transport Properties of Zinc Tin Nitride

ZnSnN 2 is an Earth-abundant semiconductor analogous to the III–nitrides with potential as a solar absorber due to its direct bandgap, steep absorption onset, and disorder-driven bandgap tunability. Despite these desirable properties, discrepancies in the fundamental bandgap and degenerate n-type carrier density have been prevalent issues in the limited amount of literature available on this material. We we use a combinatorial RF co-sputtering approach, we explored a growth-temperature-composition space for Zn 1+x Sn 1-x N 2 over the ranges 35–340 °C and 0.30–0.75 Zn/(Zn + Sn). In this way, we identified an optimal set of deposition parameters for obtaining as-deposited films with wurtzite crystal structure and carrier density as low as 1.8 × 10 18 cm -3 . Films grown at 230 °C with Zn/(Zn + Sn) = 0.60 were found to have the largest grain size overall (70 nm diameter on average) while also exhibiting low carrier density (3 × 10 18 cm -3 ) and high mobility (8.3 cm 2 V -1 s -1 ). Using this approach, we establish the direct bandgap of cation-disordered ZnSnN 2 at 1.0 eV. Moreover, we report tunable carrier density as a function of cation composition, in which lower carrier density is observed for higher Zn content. Consequently, this relationship manifests as a Burstein–Moss shift widening the apparent bandgap as cation composition moves away from Zn-rich. Collectively, these findings provide important insight into the fundamental properties of the Zn–Sn–N material system and highlight the potential to utilize ZnSnN 2 for photovoltaics.

14 SOLAR ENERGY↗

Quantum-accurate spectral neighbor analysis potential models for Ni-Mo binary alloys and fcc metals

In recent years, efficient interatomic potentials approaching the accuracy of density functional theory (DFT) calculations have been developed using rigorous atomic descriptors satisfying strict invariances, for example, for translation, rotation, permutation of homonuclear atoms, among others. In this work, we generalize the spectral neighbor analysis potential (SNAP) model to bcc-fcc binary alloy systems. We demonstrate that machine-learned SNAP models can yield significant improvements even over the well-established high-performing embedded atom method (EAM) and modified EAM potentials for fcc Cu and Ni. We also report on the development of a SNAP model for the fcc Ni-bcc Mo binary system by machine learning a carefully constructed large computed data set of elemental and intermetallic compounds. We demonstrate that this binary Ni-Mo SNAP model can achieve excellent agreement with experiments in the prediction of a Ni-Mo phase diagram as well as near-DFT accuracy in the prediction of many key properties, such as elastic constants, formation energies, melting points, etc., across the entire binary composition range. In contrast, the existing Ni-Mo EAM has significant errors in the prediction of the phase diagram and completely fails in binary compounds. Lastly, this paper provides a systematic model development process for multicomponent alloy systems, including an efficient procedure to optimize the hyperparameters in the model fitting, and paves the way for long-time large-scale simulations of such systems.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Moltensaltpropnet

MoltenSaltPropnet is a physics-informed machine learning framework that aims to predict the thermophysical properties of molten fluoride and chloride salt mixtures, which are crucial for the design and safety of Generation IV molten salt reactors. The code processes data from the Molten-Salt Thermal Properties Database (MSTDB-TP) and the Janz compendium, converting critically evaluated correlations into fast, differentiable surrogate models for density, viscosity, thermal conductivity, and heat capacity across 448 distinct salt systems. The implementation consists of several key components: 1. Data Curation: The code parses and cleans the raw data, normalizing elemental mole fractions and extracting relevant regression coefficients for various thermophysical properties. 2. Feature Engineering: It generates fixed-length numerical descriptors that encapsulate the composition and temperature, incorporating polynomial interaction terms and dimensionality-reduction techniques to optimize model performance. 3. Coefficient Learning: Four different machine learning architectures are employed: a deep residual network (ResNet), a Kolmogorov–Arnold network (KAN), a sparsity-inducing neural network (SNN), and classical regression models. Each model learns to predict coefficients that define the temperature-dependent correlations for the thermophysical properties. 4. Property Reconstruction: The predicted coefficients are used to compute temperature-dependent property values, ensuring positivity and monotonic trends through a composite loss function that enforces physical constraints. 5. User Interface: An open-source web application enables users to filter the database, train task-specific models, and visualize the results, allowing for rapid exploration of candidate salt mixtures. MoltenSaltPropnet bridges the gap between limited experimental data and high-fidelity reactor simulations, providing a powerful tool for researchers in the field of molten salt reactors and advanced nuclear energy systems.

Retamales, Mauricio Eduardo Tano [Idaho National L↗

A universal model for nanoporous carbon supercapacitors

Supercapacitors based on nanoporous carbon materials, commonly called electric double-layer capacitors (EDLCs), are emerging as a novel type of energy-storage device with the potential to substitute batteries in applications that require high power densities. Nanoporous carbon supercapacitors are generally viewed as a parallel-plate capacitor since supercapacitors store energy by charge separation in an electric double layer formed at the electrode/electrolyte interface. The EDLC model has been used to characterize the energy storage of supercapacitors for decades. We comment in this chapter on the shortcomings of the EDLC model when applied to nanoporous carbon supercapacitors. In response to the latest experimental breakthrough in nanoporous carbon supercapacitors, we have proposed a heuristic model that takes pore curvature into account as a replacement for the EDLC model. When the pore size is in the mesopore regime (2 50 nm), electrolyte counterions enter mesoporous carbons and approach the pore wall to form an electric double-cylinder capacitor (EDCC); in the micropore regime (< 2 nm), solvated/desolvated counterions line up along the pore axis to form an electric wire-in-cylinder capacitor (EWCC). In the macropore regime (> 50 nm), where pores are large enough so that pore curvature is no longer significant, the EDCC model can be reduced to the EDLC model. With the backing of experimental data and quantum density functional theory calculations, we have shown that the EDCC/EWCC model is universal for carbon supercapacitors with diverse carbon materials and electrolytes. The strengths and limitations of this new model are discussed. The new model allows the supercapacitor properties to be correlated with pore size, specific surface area, Debye length, electrolyte concentration, dielectric constant, and solute ion size, and may lend support to the systematic optimization of the properties of carbon supercapacitors through experiments.

42 ENGINEERING↗

Vibrational spectroscopic, structural and nonlinear optical activity studies on 2-amino-3-chloro-5-trifluoromethyl pyridine: A DFT approach

The conformational analysis was carried out for 2-amino-3-chloro-5-trifluoromethylpyridine using potential energy surface (PES) scan and the most stable optimized conformer was predicted. The theoretical vibrational frequencies were calculated for the optimized geometry using DFT/B3LYP cc-pVQZ basis set by Gaussian 09 Program. The vibrational frequencies were assigned on the basis of potential energy distribution calculation using VEDA 4.0 program package. The Mulliken atomic charge values were calculated. In the Frontier molecular orbitals analysis, the molecular reactivity, kinetic stability, intermolecular charge transfer studies and the calculation of ionization energy, electron affinity, global hardness, chemical potential, electrophilicity index and softness of the molecule were carried out. The nonlinear optical (NLO) activity was studied and the first order hyperpolarizability value was computed, which was 3.48 times greater than the urea. The natural bond orbital analysis was also performed to confirm the NLO activity of the molecule. Hence, the ACTP molecule is a promising candidate for NLO materials.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

A critical comparison of CH∙∙∙π versus π∙∙∙π interactions in the benzene dimer: obtaining benchmarks at the CCSD(T) level and assessing the accuracy of lower scaling methods

In this study, we have established CCSD(T)/CBS (Complete Basis Set) limits for 3 stationary points on the benzene dimer potential energy surface, corresponding to the π∙∙∙π (parallel displaced or PD(C 2h ), minimum) and CH∙∙∙π (T-shaped or T(C 2v ), transition state) and tilted T-shaped (or TT(Cs), minimum) bonding scenarios considering both the structure and binding energy. The CCSD(T)/CBS binding energies are -2.65 ± 0.02 (PD), -2.74 ± 0.03 (T), and -2.83 ± 0.01 kcal mol -1 (TT). To this end, the CH∙∙∙π is ~0.2 kcal mol -1 stronger than the π∙∙∙π interaction, whereas the tilting of the CH donating benzene molecule with respect to the other benzene is worth 0.1 kcal mol -1 . As previously discussed in the literature, the MP2 level of theory does not provide a close match for either the energy or structure, yet the SCS-MP2 yields structures in excellent agreement with respect to the CCSD(T) result. It is found that the SCS-MI-MP2 also gives optimized structures very close to SCS-MP2 (within ~0.01 Å of the benchmark). Despite the closer match in structure, the spin-biased MP2 methods (SCS-, SCS-MI-, and SOS-MP2) incorrectly predict the relative stabilities of the isomers. That said, none of the spin biased MP2 methods offers a good compromise between energy and structure for the systems examined. Finally, the CCSD(T)/CBS benchmarks were used to assess the performance of 13 DFT functionals selected from different rungs of Jacob's ladder. Several functionals such as TPSS-D3, B3LYP-D3, B97-D, B97-D3, and B2PLYP-D3 provided a good description of the binding energies for both CH∙∙∙π and π∙∙∙π interactions, yielding values within 6% of the CCSD(T)/CBS benchmark values. Unlike the MP2 methods, these functionals correctly predict the relative stability of the PD(C 2h ) and T(C 2v ) dimers. Further, we find that there is no systematic improvement as Jacob's ladder is ascended (increased complexity of functional). The best functionals that result in a good compromise between structure and energy accuracy are B97-D3 and B2PLYP-D3 for both the CH∙∙∙π and π∙∙∙π interaction. Despite the impressive performance of these functionals, a challenge that remains is ensuring the transferability of these density functionals in accurately describing the interaction between dimers of larger aromatic molecules, the latter requiring high-level benchmarks for these systems.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Band-gap and band-edge engineering of multicomponent garnet scintillators from first principles

Complex doping schemes in R 3 Al 5 O 12 (where R is the rare-earth element) garnet compounds have recently led to pronounced improvements in scintillator performance. Specifically, by admixing lutetium and yttrium aluminate garnets with gallium and gadolinium, the band gap is altered in a manner that facilitates the removal of deleterious electron trapping associated with cation antisite defects. Here, we expand upon this initial work to systematically investigate the effect of substitutional admixing on the energy levels of band edges. Density-functional theory and hybrid density-functional theory (HDFT) are used to survey potential admixing candidates that modify either the conduction-band minimum (CBM) or valence-band maximum (VBM). We consider two sets of compositions based on Lu 3 B 5 O 12 where B is Al, Ga, In, As, and Sb, and R 3 Al 5 O 12 , where R is Lu, Gd, Dy, and Er. We find that admixing with various R cations does not appreciably affect the band gap or band edges. In contrast, substituting Al with cations of dissimilar ionic radii has a profound impact on the band structure. We further show that certain dopants can be used to selectively modify only the CBM or the VBM. Specifically, Ga and In decrease the band gap by lowering the CBM, while As and Sb decrease the band gap by raising the VBM, the relative change in band gap is quantitatively validated by HDFT. These results demonstrate a powerful approach to quickly screen the impact of dopants on the electronic structure of scintillator compounds, identifying those dopants which alter the band edges in very specific ways to eliminate both electron and hole traps responsible for performance limitations. Furthermore, this approach should be broadly applicable for the optimization of electronic and optical performance for a wide range of compounds by tuning the VBM and CBM.

36 MATERIALS SCIENCE↗

An efficient virus–induced gene silencing (VIGS) system for gene functional studies in Miscanthus

Virus-induced gene silencing (VIGS) is a powerful tool for transient gene functional analysis in plants, especially for monocot species (e.g., grasses) that are recalcitrant to transformation. Despite various VIGS systems that have been developed in different plant species, none was previously available for the bioenergy crop Miscanthus. Here, we report the establishment of an efficient and robust VIGS system mediated by Tobacco Rattle Virus (TRV) in Miscanthus. We first investigated the impact of various factors that may affect gene silencing efficiency using the Miscanthus sinensis Phytoene Desaturase (MsPDS) gene as a visual indicator of photobleaching. Then, we optimized the TRV-elicited VIGS procedure using an orthogonal experimental design with four factors (sprout size, Agrobacterium concentration, vacuum infiltration time, and co-incubation time) each at three levels. The following led to the highest silencing efficiency (~76%): inoculation of germinating seedlings (1.0–2.0 mm), Agrobacterium tumefaciens culture grown to optical density at 600 nm (OD 600 ) of 0.4, vacuum infiltration for 90 min, and co-incubation for 5 h. The VIGS system established was applicable for both M. sinensis and M. lutarioriparius, with comparable gene silencing efficiency. We verified the efficacy of the VIGS system via the functional characterization of the role of a MYB transcription factor, MsMYB112, in salt stress tolerance. Expression of MsMYB112 was successfully knocked down using the VIGS system, and this led to compromised salt tolerance in the silenced Miscanthus plants. The TRV-based VIGS system established may, therefore, substantially facilitate functional genomic studies in Miscanthus.

60 APPLIED LIFE SCIENCES↗

Modeling Water Uptake and Pt Utilization in High Surface Area Carbon

Local transport losses limit high-current density performance of fuel-cells, especially at low Pt-loadings. Optimizing catalyst-layer structure and material properties to mitigate these losses is critical for wide-scale fuel-cell commercialization. Carbon supports form the primary porous structure of catalyst layers in which Pt particles are deposited and ionomer films are distributed. Unlike non-porous carbon supports, high surface area carbon support is unique in its relative humidity dependent Pt utilization and performance due to presence of Pt particles in the micropores interior of carbon particles. We model water uptake and Pt utilization in catalyst layers with high-surface-area carbon support. Interactions between functional groups on the carbon pore walls and water play a vital role in controlling water uptake from water vapor by catalyst layers. Ionomer is shown to influence proton access in micropores, even though it may not deposit into the micropores.

08 HYDROGEN↗

Potential high-performance magnet materials: Co- and Al-alloyed Sm 2 Fe 17

Sm 2 Fe 17 has long been known as a potential high-performance magnet whose deficiencies—planar anisotropy and lower-than-optimal T c —can be remedied by nitrogen addition, but which presents synthesis difficulties. In this work, we apply first-principles calculations to search for alternative low-cost, high-performance permanent magnets in this family, by exploring simultaneous Fe and Al substitution. Specifically, the goal is to improve properties of Sm 2 Fe 14 Al 3 easy-plane magnet at the stoichiometric composition. Density functional theory calculations were executed for three series of compounds, i.e., Sm 2 (Fe 1-x Co x ) 14 Al 3 , Sm 2 (Fe 1-x Co x ) 15 Al 2 , and Sm 2 (Fe 1-x Cox) 16 Al. We find that substitution of Fe with 12–18 of Co in % Sm 2 Fe 14 Al 3 modifies the magnetic anisotropy type from easy plane to easy axis with a substantial anisotropy of 7.1 MJ/m 3 . We also demonstrate that the largest part of magnetic anisotropy is introduced by 4f Sm atom electrons. Thus the rotation of magnetic moment orientation from $\langle$$1\bar10$$\rangle$ to $\langle$111$\rangle$ is followed by an increase of the occupied 4f state number and, as a result, the orbital part of the magnetic moment of one of the Sm atoms. This increase of the occupied 4f state number at an energy ~ -4.3 eV results in a significant reduction of band structure energy. The substitution of Fe by Co does not significantly reduce the magnetization of the compound and keeps it slightly above 1 T. This combination of magnetic anisotropy and magnetization makes the compound a promising candidate for a permanent magnet.

36 MATERIALS SCIENCE↗

Enhanced nucleation mechanism in ruthenium atomic layer deposition: Exploring surface termination and precursor ligand effects with RuCpEt(CO)2

Miniaturization of microelectronic devices necessitates atomic precision in manufacturing techniques, particularly in the deposition of thin films. Atomic layer deposition (ALD) is recognized for its precision in controlling film thickness and composition on intricate three-dimensional structures. This study focuses on the ALD nucleation and growth mechanisms of ruthenium (Ru), a metal that has significant future implications for microelectronics. Despite its advantages, the deposition of a high surface-free energy material like Ru on a low surface-free energy material such as an oxide often faces challenges of large nucleation delays and non-uniform growth. To address these challenges, we explored the effectiveness of organometallic surface pretreatments using trimethylaluminum (TMA) or diethylzinc (DEZ) to enhance Ru film nucleation and growth. Our study employed a less-studied Ru precursor, cyclopentadienylethyl(dicarbonyl)ruthenium [RuCpEt(CO)2], which demonstrated promising results in terms of reduced nucleation delay and increased film continuity. Ru ALD was performed on silicon substrates with native oxide, using RuCpEt(CO)2 and O2 as coreactants. Our findings reveal that surface pretreatment significantly improves nucleation density and film thickness within the initial 60 ALD cycles, achieving up to a 3.2-fold increase in Ru surface coverage compared to nonpretreated substrates. Supported by density functional theory calculations, we propose that the enhanced nucleation observed with RuCpEt(CO)2 compared to previously-studied Ru(Cp)2 is due to two key mechanisms: the facilitated removal of CO ligands during deposition, which enhances the reactivity of the precursor, and a hydrogen-abstraction reaction involving the ethyl ligand of RuCpEt(CO)2 and the metal-alkyl groups on the surface. This study not only advances our understanding of Ru ALD processes but also highlights the significant impact of precursor chemistry and surface treatments in optimizing ALD for advanced microelectronic applications.

Materials Science↗

Advancing density functional tight-binding method for large organic molecules through equivariant neural networks

Semi-empirical quantum-mechanical (QM) methods have become valuable tools for studying complex (bio)molecular systems due to their balance between computational efficiency and accuracy. A key aspect of these methods is their parameterization, which not only governs the reliability of the results but also provides an opportunity to enhance their overall performance. In our previous work [J. Phys. Chem. Lett., 2021, 11, 16], we advanced the third-order semi-empirical density functional tight-binding (DFTB3) method for computing multiple properties of small molecules by developing the machine learning (ML) potential NN rep to bridge the gap between DFTB3 electronic components and those of the hybrid DFT-PBE0 functional. To overcome the limitations of NN rep , we introduce the EquiDTB framework, which leverages physics-inspired equivariant neural networks (NN) to parameterize scalable and transferable many-body Δ TB potentials, replacing the standard pairwise DFTB repulsive potential. This advancement extends the applicability of our ML-corrected DFTB approach to larger molecules and non-covalent systems (including only C, N, O, and H atoms), going beyond the chemical space represented in the training QM datasets. The enhanced performance of EquiDTB over the standard TB methods is demonstrated by the accurate computation of the atomic forces of S66x8 molecular dimers, as well as their interaction energies. Moreover, EquiDTB can be effectively employed to explore the potential energy surfaces of large and flexible drug-like molecules—for example, to determine the minimum energy path between isomers, analyze structural transitions during dynamical simulations, compute vibrational modes, and investigate energetic rankings. The performance for single molecules slightly decreases when the DFTB electronic energy is reduced to first-order but remains superior to standard TB methods. Our work thus demonstrates that an optimal integration of an equivariant NN with QM datasets can advance the DFTB method while maintaining high efficiency, paving the way for reliable (bio)molecular simulations.

Medrano Sandonas, Leonardo [Technische Universität↗