arXiv subjects

Arslan Mazitov

Publications and source records attributed to Arslan Mazitov.

17 recordsLinked to original sources

arXiv2026

High-quality, high-information datasets for universal atomistic machine learning

The quality, consistency, and information content of training data is often what determines the practical value of machine-learning models for atomistic simulations. Yet, many widely used electronic-structure databases are assembled having materials screening as primary goal rather than robust force-field learning, are limited in their scope to a specific class of chemical compounds, and/or employ inconsistent DFT functionals and settings. Here we introduce MAD-1.6, a highly curated dataset designed explicitly for training broadly applicable atomistic models across the periodic table at high levels of theory. MAD-1.6 extends the MAD dataset with targeted enrichment strategies that improve the coverage of chemical space to 102 elements while keeping the total number of configurations compact. All structures are computed with a single, standardized all-electron DFT workflow using the r$^2$SCAN meta-GGA functional and consistent convergence settings, ensuring uniformity across chemically heterogeneous systems. The dataset encompasses molecules, clusters, bulk crystals, surfaces, and low-dimensional structures, and its quality and consistency are further enhanced by outlier removal using uncertainty quantification. We demonstrate the high accuracy that can be achieved with the proposed dataset by training PET-MAD-1.6, a generally applicable r$^2$SCAN interatomic potential that covers 102 elements in the periodic table and achieves exceptional levels of benchmark accuracy and stability in challenging simulation protocols.

cond-mat.mtrl-sci↗

arXiv2026

Pushing the limits of unconstrained machine-learned interatomic potentials

Machine-learned interatomic potentials (MLIPs) are increasingly used to replace computationally demanding electronic-structure calculations to model matter at the atomic scale. The most commonly used model architectures are constrained to fulfill a number of physical laws exactly, from geometric symmetries to energy conservation. Evidence is mounting that relaxing some of these constraints can be beneficial to the efficiency and (somewhat surprisingly) accuracy of MLIPs, even though care should be taken to avoid qualitative failures associated with the breaking of physical symmetries. Given the recent trend of scaling up models to larger numbers of parameters and training samples, a very important question is how unconstrained MLIPs behave in this limit. Here we investigate this issue, showing that -- when trained on large datasets -- unconstrained models can be superior in accuracy and speed when compared to physically constrained models. We assess these models both in terms of benchmark accuracy and in terms of usability in practical scenarios, focusing on static simulation workflows such as geometry optimization and lattice dynamics. We conclude that accurate unconstrained models can be applied with confidence, especially since simple inference-time modifications can be used to recover observables that are consistent with the relevant physical symmetries.

physics.chem-ph↗

arXiv2026

A universal machine learning model for the electronic density of states

In the last few years several ``universal'' interatomic potentials have appeared, using machine-learning approaches to predict energy and forces of atomic configurations with arbitrary composition and structure, with an accuracy often comparable with that of the electronic-structure calculations they are trained on. Here we demonstrate that these generally-applicable models can also be built to predict explicitly the electronic structure of materials and molecules. We focus on the electronic density of states (DOS), and develop PET-MAD-DOS, a rotationally unconstrained transformer model built on the Point Edge Transformer (PET) architecture, and trained on the Massive Atomistic Diversity (MAD) dataset. We demonstrate our model's predictive abilities on samples from diverse external datasets, showing also that the DOS can be further manipulated to obtain accurate band gap predictions. A fast evaluation of the DOS is especially useful in combination with molecular simulations probing matter in finite-temperature thermodynamic conditions. To assess the accuracy of PET-MAD-DOS in this context, we evaluate the ensemble-averaged DOS and the electronic heat capacity of three technologically relevant systems: lithium thiophosphate (LPS), gallium arsenide (GaAs), and a high entropy alloy (HEA). By comparing with bespoke models, trained exclusively on system-specific datasets, we show that our universal model achieves semi-quantitative agreement for all these tasks. Furthermore, we demonstrate that fine-tuning can be performed using a small fraction of the bespoke data, yielding models that are comparable to, and sometimes better than, fully-trained bespoke models.

physics.chem-ph↗

arXiv2025

Giant exciton binding energy in bulk CrCl3

Van der Waals (vdW) materials, with their unique combination of electronic, optical, and magnetic properties, are emerging as promising platforms for exploring excitonic phenomena. Thus far, the choice of materials with exceptional excitonic response has been limited to two-dimensional (2D) configurations of vdW materials. At the same time, large interlayer distance and the possibility to create a variety of heterostructures offers an opportunity to control the dielectric screening in van der Waals heterostructures and van der Waals 3D materials, thus engineering the excitonic properties. Here, we reveal that bulk vdW crystal CrCl3 answers this quest with a record exciton binding energy of 1.64 eV owing to a delicate interplay of quasi-2D electronic confinement and local magnetic correlations. We also suggest that the non-local magnetic correlations play an important role in the temperature dependence of photoluminescence intensity. Furthermore, we observe colossal binding energies in vdW crystals NbOCl2 (0.66 eV) and MoCl3 (0.35 eV) and formulate a universal exciton binding energy dependence on bandgap for 2D and 3D vdW materials. Hence, our findings establish a fundamental link between the layered structure of vdW materials and their excitonic properties.

cond-mat.mtrl-sci↗

arXiv2025

Metatensor and metatomic: foundational libraries for interoperable atomistic machine learning

Incorporation of machine learning (ML) techniques into atomic-scale modeling has proven to be an extremely effective strategy to improve the accuracy and reduce the computational cost of simulations. It also entails conceptual and practical challenges, as it involves combining very different mathematical foundations, as well as software ecosystems that are very well developed in their own right, but do not share many commonalities. To address these issues and facilitate the adoption of ML in atomistic simulations, we introduce two dedicated software libraries. The first one, metatensor, provides multi-platform and multi-language storage and manipulation of arrays with many potentially sparse indices, designed from the ground up for atomistic ML applications. By combining the actual values with metadata that describes their nature and that facilitates the handling of geometric information and gradients with respect to the atomic positions, metatensor provides a common framework to enable data sharing between ML software -- typically written in Python -- and established atomistic modeling tools -- typically written in Fortran, C or C++. The second library, metatomic, provides an interface to store an atomistic ML model and metadata about this model in a portable way, facilitating the implementation, training and distribution of models, and their use across different simulation packages. We showcase a growing ecosystem of tools, including low-level libraries, training utilities, and interfaces with existing software packages that demonstrate the effectiveness of metatensor and metatomic in bridging the gap between traditional simulation software and modern ML frameworks.

physics.chem-ph↗

arXiv2025

Giant optical anisotropy and visible-frequency epsilon-near-zero in hyperbolic van der Waals MoOCl2

The realization of extreme optical anisotropy is foundational to nanoscale light manipulation. Van der Waals (vdW) crystal MoOCl2 has emerged as a promising candidate for this quest, hosting hyperbolic plasmon polaritons in the visible and near-infrared wavelengths. However, the fundamental anisotropic dielectric tensor governing this behavior has remained elusive. Here, we resolve this problem by providing the first experimental determination of the full dielectric tensor of hyperbolic vdW MoOCl2. Via spectroscopic ellipsometry, Mueller matrix, and reflectance measurements, we quantify the material's optical duality: a metallic optical response (ε_1 < 0) along the crystallographic a-axis and a dielectric response (ε_1 > 0) along the orthogonal directions. This dichotomy drives an epsilon-near-zero (ENZ) condition at \approx 512 nm and results in giant in-plane birefringence of δn \approx 2.2 for MoOCl2. As a result, our work provides the critical missing experimental parameters for MoOCl2, establishing it as a benchmark hyperbolic and ENZ material.

physics.optics↗

arXiv2025

Giant optical anisotropy in CrSBr from giant exciton oscillator strength

The interplay between dimensionality and electronic correlations in van der Waals (vdW) materials offers a powerful toolkit for engineering light-matter interactions at the nanoscale. Excitons, bound electron-hole pairs, are central to this endeavor, yet maximizing their oscillator strength, which dictates the interaction cross-section, remains a challenge. Conventional wisdom suggests a trade-off, where the observable oscillator strength often decreases in strongly bound systems due to population dynamics. Here, we unveil a colossal oscillator strength associated with the quasi-one-dimensional (quasi-1D) excitons in the layered magnetic semiconductor CrSBr, which fundamentally defies this established scaling law. Through comprehensive optical characterization and ab initio calculations, we establish that this anomalous enhancement originates directly from the reduced dimensionality, which enforces an increased electron-hole wavefunction overlap. Moreover, we find a close connection between fundamental exciton and local spin fluctuations that contribute to the opening of the gap in the electronic spectrum. The resulting optical anisotropy shows a giant in-plane birefringence (Delta_n = 1.45) and profoundly anisotropic waveguiding, which we directly visualize using nano-optical imaging. Leveraging this extreme response, we realize a true zero-order quarter-wave plate with an unprecedented wavelength-to-thickness ratio (lambda/t) exceeding 3.4, surpassing the limits of current miniaturization technologies, including state-of-the-art metasurfaces. Our findings underscore the profound impact of dimensionality engineering in magnetic vdW materials for realizing novel regimes of light-matter coupling and developing next-generation ultracompact photonic architectures.

cond-mat.mtrl-sci↗

arXiv2025

PET-MAD, a lightweight universal interatomic potential for advanced materials modeling

Machine-learning interatomic potentials (MLIPs) have greatly extended the reach of atomic-scale simulations, offering the accuracy of first-principles calculations at a fraction of the cost. Leveraging large quantum mechanical databases and expressive architectures, recent ''universal'' models deliver qualitative accuracy across the periodic table but are often biased toward low-energy configurations. We introduce PET-MAD, a generally applicable MLIP trained on a dataset combining stable inorganic and organic solids, systematically modified to enhance atomic diversity. Using a moderate but highly-consistent level of electronic-structure theory, we assess PET-MAD's accuracy on established benchmarks and advanced simulations of six materials. Despite the small training set and lightweight architecture, PET-MAD is competitive with state-of-the-art MLIPs for inorganic solids, while also being reliable for molecules, organic materials, and surfaces. It is stable and fast, enabling the near-quantitative study of thermal and quantum mechanical fluctuations, functional properties, and phase transitions out of the box. It can be efficiently fine-tuned to deliver full quantum mechanical accuracy with a minimal number of targeted calculations.

cond-mat.mtrl-sci↗

arXiv2025

Massive Atomic Diversity: a compact universal dataset for atomistic machine learning

The development of machine-learning models for atomic-scale simulations has benefited tremendously from the large databases of materials and molecular properties computed in the past two decades using electronic-structure calculations. More recently, these databases have made it possible to train universal models that aim at making accurate predictions for arbitrary atomic geometries and compositions. The construction of many of these databases was however in itself aimed at materials discovery, and therefore targeted primarily to sample stable, or at least plausible, structures and to make the most accurate predictions for each compound - e.g. adjusting the calculation details to the material at hand. Here we introduce a dataset designed specifically to train machine learning models that can provide reasonable predictions for arbitrary structures, and that therefore follows a different philosophy. Starting from relatively small sets of stable structures, the dataset is built to contain massive atomic diversity (MAD) by aggressively distorting these configurations, with near-complete disregard for the stability of the resulting configurations. The electronic structure details, on the other hand, are chosen to maximize consistency rather than to obtain the most accurate prediction for a given structure, or to minimize computational effort. The MAD dataset we present here, despite containing fewer than 100k structures, has already been shown to enable training universal interatomic potentials that are competitive with models trained on traditional datasets with two to three orders of magnitude more structures. We describe in detail the philosophy and details of the construction of the MAD dataset. We also introduce a low-dimensional structural latent space that allows us to compare it with other popular datasets and that can be used as a general-purpose materials cartography tool.

cond-mat.mtrl-sci↗

arXiv2024

Tunable Nanostructuring for van der Waals Materials

Van der Waals (vdW) materials are becoming increasingly popular in scientific and industrial applications because of their unique mixture of record electronic, optical, and mechanical properties. However, nanostructuring of vdW materials is still in its infancy and strongly depends on the specific vdW crystal. As a result, the universal self-assembled technology of vdW materials nanostructuring opens vast technological prospects. This work demonstrates an express and universal synthesis method of vdW nanoparticles with well-defined geometry using femtosecond laser ablation and fragmentation. The disarming simplicity of the technique allows us to create nanoparticles from over 50 vdW precursor materials covering transition metal chalcogenides, MXenes, and other vdW materials. Obtained nanoparticles manifest perfectly defined crystalline structures and diverse shapes, from nanospheres to nanocubes and nanotetrahedrons. Thus, our work provides a new paradigm for vdW nanostructuring with a vast potential of tunability for size, shape, and materials specific to the particular application.

physics.app-ph↗

arXiv2024

Substrate-aware computational design of two-dimensional materials

Two-dimensional materials have attracted considerable attention due to their remarkable electronic, mechanical and optical properties, making them prime candidates for next-generation electronic and optoelectronic applications. Despite their widespread use in combination with substrates in practical applications, including the fabrication process and final device assembly, computational studies often neglect the effects of substrate interactions for simplicity. This study presents a novel method for predicting the atomic structure of 2D materials on arbitrary substrates by combining an evolutionary algorithm, a lattice-matching technique, an automated machine learning interatomic potentials training protocol, and the ab initio thermodynamics approach for predicting the possible conditions of experimental synthesis of the predicted 2D structures. Using the Mo-S system on a c-cut sapphire substrate as a case study, we reveal several new stable and metastable structures, including previously known 1H-MoS2 and newly found Pmma Mo3S2, P-1 Mo2S, P21m Mo5S3, and P4mm Mo4S, where the Mo4S structure is specifically stabilized by interaction with the substrate. Electronic band structure calculations of Mo3S2, Mo2S, Mo5S3, and Mo4S showed their metallic behavior, while phonon properties calculations indicated their dynamic stability and substrate-induced modulation of the phonon density of states. Finally, we use the ab initio thermodynamics approach to predict the synthesis conditions of the discovered structures in the parameter space of the commonly used CVD technique. These results provide insights into computational substrate engineering, allowing one to study the substrate effect on the thermodynamic and dynamical stability of 2D materials and to modulate their electronic and phonon properties for their future applications, as well as to provide guide maps for their experimental synthesis.

cond-mat.mtrl-sci↗

arXiv2024

Surface segregation in high-entropy alloys from alchemical machine learning

High-entropy alloys (HEAs), containing several metallic elements in near-equimolar proportions, have long been of interest for their unique mechanical properties. More recently, they have emerged as a promising platform for the development of novel heterogeneous catalysts, because of the large design space, and the synergistic effects between their components. In this work we use a machine-learning potential that can model simultaneously up to 25 transition metals to study the tendency of different elements to segregate at the surface of a HEA. We use as a starting point a potential that was previously developed using exclusively crystalline bulk phases, and show that, thanks to the physically-inspired functional form of the model, adding a much smaller number of defective configurations makes it capable of describing surface phenomena. We then present several computational studies of surface segregation, including both a simulation of a 25-element alloy, that provides a rough estimate of the relative surface propensity of the various elements, and targeted studies of CoCrFeMnNi and IrFeCoNiCu, which provide further validation of the model, and insights to guide the modeling and design of alloys for heterogeneous catalysis.

cond-mat.mtrl-sci↗

arXiv2023

Analysis of stability and near-equilibrium dynamics of self-assembled Casimir cavities

Vacuum fluctuations are a fundamental and irremovable property of a quantized electromagnetic field. These fluctuations are the cause of the Casimir effect -- mutual attraction of two electrically neutral metallic plates in vacuum in the absence of any other interactions. For most geometries and materials, Casimir effect is strictly attractive, leading to the only stable equilibrium configuration with merged plates. Recent observation showed, however, that this unavoidable vacuum-induced attraction can be mitigated by the presence of electrostatic repulsion produced by the formation of double electric layers, and a stable equilibrium between two charged metallic plates in a solution of an organic salt can be reached. Here, we study theoretically in details equilibrium configurations and their dynamical behavior in the system of two parallel metallic films coupled by the Casimir and electrostatic interactions. We analyze the effect of various parameters of the system -- such as the salt concentartion and temperature -- on the equilibrium cavity thicknesses, inspect resonant properties of the resulting an-harmonic optomechanical system near equilibrium, and examine its stochastic dynamics under the influence of thermal fluctuations of the environment.

physics.optics↗

arXiv2023

Exploring van der Waals materials with high anisotropy: geometrical and optical approaches

The emergence of van der Waals (vdW) materials resulted in the discovery of their giant optical, mechanical, and electronic anisotropic properties, immediately enabling countless novel phenomena and applications. Such success inspired an intensive search for the highest possible anisotropic properties among vdW materials. Furthermore, the identification of the most promising among the huge family of vdW materials is a challenging quest requiring innovative approaches. Here, we suggest an easy-to-use method for such a survey based on the crystallographic geometrical perspective of vdW materials followed by their optical characterization. Using our approach, we found As2S3 as a highly anisotropic vdW material. It demonstrates rare giant in-plane optical anisotropy, high refractive index and transparency in the visible range, overcoming the century-long record set by rutile. Given these benefits, As2S3 opens a pathway towards next-generation nanophotonics as demonstrated by an ultrathin true zero-order quarter-waveplate that combines classical and the Fabry-Perot optical phase accumulations. Hence, our approach provides an effective and easy-to-use method to find vdW materials with the utmost anisotropic properties.

physics.optics↗

arXiv2023

A hidden advantage of van der Waals materials for overcoming limitations in photonic integrated circuitry

With the advance of on-chip nanophotonics, there is a high demand for high refractive index, low-loss materials. Currently, this technology is dominated by silicon, but van der Waals (vdW) materials with high refractive index can offer a very advanced alternative. Still, up to now it was not clear if the optical anisotropy perpendicular to the layers might be a hindering factor for the development of vdW nanophotonics. Here, we studied WS2-based waveguides in terms of their optical properties and, particularly, in terms of possible crosstalk distance. Surprisingly, we discovered that the low refractive index in the direction perpendicular to the atomic layers improves the characteristics of such devices, mainly due to expanding the range of parameters at which single-mode propagation can be achieved. Thus, using anisotropic materials offers new opportunities and novel control knobs when designing the nanophotonic devices.

physics.optics↗

arXiv2022

Fast general two- and three-body interatomic potential

We introduce a new class of machine learning interatomic potentials - fast General Two- and Three-body Potential (GTTP), which is as fast as conventional empirical potentials and require computational time that remains constant with increasing fitting flexibility. GTTP does not contain any assumptions about the functional form of two- and three-body interactions. These interactions can be modeled arbitrarily accurately, potentially by thousands of parameters not affecting resulting computational cost. Time complexity is O(1) per every considered pair or triple of atoms. The fitting procedure is reduced to simple linear regression on ab initio calculated energies and forces and leads to effective two- and three-body potential, reproducing quantum many-body interactions as accurately as possible. Our potential can be made continuously differentiable any number of times at the expense of increased computational time. We made a number of performance tests on one-, two- and three-component systems. The flexibility of the introduced approach makes the potential transferable in terms of size and type of atomic systems. We show that trained on randomly generated structures with just 8 atoms in the unit cell, it significantly outperforms common empirical interatomic potentials in the study of large systems, such as grain boundaries in polycrystalline materials.

physics.comp-ph↗

arXiv2021

Topological phase singularities in atomically thin high-refractive-index materials

Atomically thin transition metal dichalcogenides (TMDCs) present a promising platform for numerous photonic applications due to excitonic spectral features, possibility to tune their constants by external gating, doping, or light, and mechanical stability. Utilization of such materials for sensing or optical modulation purposes would require a clever optical design, as by itself the 2D materials can offer only a small optical phase delay - consequence of the atomic thickness. To address this issue, we combine films of 2D semiconductors which exhibit excitonic lines with the Fabry-Perot resonators of the standard commercial SiO$_2$/Si substrate, in order to realize topological phase singularities in reflection. Around these singularities, reflection spectra demonstrate rapid phase changes while the structure behaves as a perfect absorber. Furthermore, we demonstrate that such topological phase singularities are ubiquitous for the entire class of atomically thin TMDCs and other high-refractive-index materials, making it a powerful tool for phase engineering in flat optics. As a practical demonstration, we employ PdSe$_2$ topological phase singularities for a refractive index sensor and demonstrate its superior phase sensitivity compared to typical surface plasmon resonance sensors.

physics.optics↗