arXiv subjects

Alexander Khrabry

Publications and source records attributed to Alexander Khrabry.

12 recordsLinked to original sources

arXiv2025

Accelerating kinetic plasma simulations with machine learning generated initial conditions

Computer aided engineering of multi-time-scale plasma systems which exhibit a quasi-steady state solution are challenging due to the large number of time steps required to reach convergence. Machine learning techniques combined with traditional first-principles simulations and high-performance computing offer many interesting pathways towards resolving this challenge. We consider acceleration of kinetic plasma simulations via machine learning generated initial conditions. The approach is demonstrated through modeling of capacitively coupled plasma discharges relevant to the microelectronics industry. Three models are trained on simulations across a parameter space of device driving frequency and operating pressure. The models incorporate elements of a multi-layer perceptron, principal component analysis, and convolutional neural networks to predict the final time-averaged profiles of ion-density and velocity distribution functions. These data-driven initial condition generators (ICGs) provide a mean speedup of 17.1x in convergence time, when measured using an offline procedure, or a 4.4x speedup with an online procedure, with convolutional neural networks leading to the best performance. The paper also outlines a workflow for continuous data-driven model improvement and simulation speedup, with the aim of generating sufficient data for full device digital twins.

physics.plasm-ph↗

arXiv2025

Hierarchical-embedding autoencoder with a predictor (HEAP) as efficient architecture for learning long-term evolution of complex multi-scale physical systems

We propose a novel efficient architecture for learning long-term evolution in complex multi-scale physical systems which is based on the idea of separation of scales. Structures of various scales that dynamically emerge in the system interact with each other only locally. Structures of similar scale can interact directly when they are in contact and indirectly when they are parts of larger structures that interact directly. This enables modeling a multi-scale system in an efficient way, where interactions between small-scale features that are apart from each other do not need to be modeled. The hierarchical fully-convolutional autoencoder transforms the state of a physical system not just into a single embedding layer, as it is done conventionally, but into a series of embedding layers which encode structures of various scales preserving spatial information at a corresponding resolution level. Shallower layers embed smaller structures on a finer grid, while deeper layers embed larger structures on a coarser grid. The predictor advances all embedding layers in sync. Interactions between features of various scales are modeled using a combination of convolutional operators. We compare the performance of our model to variations of a conventional ResNet architecture in application to the Hasegawa-Wakatani turbulence. A multifold improvement in long-term prediction accuracy was observed for crucial statistical characteristics of this system.

cs.AI↗

arXiv2024

Size-dependent second-order-like phase transitions in Fe nanocluster melting from low-temperature structural isomerization

In this work, the melting phase transitions of $Fe_{n}$ nanoclusters with $10 \leq n \leq 100$ atoms are investigated using classical many-body molecular dynamics simulations. For many cluster sizes, surface melting occurs at much lower temperatures than core melting. Surface and core melting points, and energetic melting points (temperatures of maximum heat capacity, $C_v$) are calculated for all cluster sizes. Melting properties are found to be strong functions of cluster structure. Cluster sizes with closed-shell structures always have first-order-like phase transitions. Almost one-third of cluster sizes in the analyzed range exhibit second-order-like phase transitions due to the presence of multiple structural configurations close in energy. 1-shell clusters with one to a few more atoms than a neighboring closed-shell structure have very low surface melting points and very high energetic melting points compared to their closed-shell counterparts. In clusters above 50 atoms with certain core structures, melting of the surface before the core was observed.

physics.atm-clus↗

arXiv2024

Gibbs free energies of Fe clusters can be approximated by Tolman correction to accurately model cluster nucleation and growth

Accurate Gibbs free energies of Fe clusters are required for predictive modeling of Fe cluster growth during condensation of a cooling vapor. We present a straightforward method of calculating free energies of cluster formation using the data provided by molecular dynamics (MD) simulations. We apply this method to calculate free energies of Fe clusters having from 2 to 100 atoms. The free energies are verified by comparing to an MD-simulated equilibrium cluster size distribution in a sub-saturated vapor. We show that these free energies differ significantly from those obtained with a commonly used spherical cluster approximation - which relies on a surface tension coefficient of a flat surface. The spherical cluster approximation can be improved by using a cluster size-dependent Tolman correction for the surface tension. The values for the Tolman length and effective surface tension were derived, which differ from the commonly used experimentally measured surface tension based on the potential energy. This improved approximation does not account for geometric magic number effects responsible for spikes and troughs in densities of neighbor cluster sizes. Nonetheless, it allows to model cluster formation from a cooling vapor and accurately reproduce the condensation timeline, overall shape of the cluster size distribution, average cluster size, and the distribution width. Using a constant surface tension coefficient resulted in distorted condensation dynamics and inaccurate cluster size distributions. The analytical expression for cluster nucleation rate from classical nucleation theory (CNT) was updated to account for the size-dependence of cluster surface tension.

cond-mat.mtrl-sci↗

arXiv2024

Quantum Chemistry Model of Surface Reactions and Kinetic Model of Diamond Growth: Effects of CH3 Radicals and C2H2 Molecules at Low-Temperatures CVD

The objective of this study is to explore conditions that facilitate a significant reduction in substrate temperature during diamond growth. The typical temperature for this process is around 1200K; we aim to reduce it to a much lower level. To achieve this, we need to understand processes that limit the diamond growth at low temperatures. Therefore, we developed a detailed chemical kinetic model to analyze diamond growth on the (100) surface. This model accounts for variations in substrate temperature and gas composition. Using an ab initio quantum chemistry, we calculated the reaction rates of all major gas phase reactants with the diamond surface, totaling 91 elemental surface reactions. Consistent with previous studies, the model identifies that CH3 is a major precursor of diamond growth, and the contribution from C2H2 to the growth is significantly smaller. However, C2H2 can also contribute to forming a sp2-phase instead of a sp3-phase, and this process becomes dominant below a critical temperature. As a result, C2H2 flux inhibits diamond growth at low temperatures. To quantify this deleterious process, we developed a new mechanism for sp2-phase nucleation on the (100) surface. Similar to the so-called HACA mechanism for soot formation it involves hydrogen abstraction and C2H2 addition. Consequently, optimal low-temperature CVD growth could be realized in a reactor designed to maximize the CH3 radical production, while minimizing the generation of C2H2 and other sp and sp2 hydrocarbons.

physics.chem-ph↗

arXiv2023

Compact and accurate chemical mechanism for methane pyrolysis with PAH growth

A reliable and compact chemical mechanism of gas-phase methane pyrolysis leading to formation of large polycyclic aromatic hydrocarbon (PAH) molecules has been developed. This model is designed for studies of carbon nanostructure synthesis such as carbon black and graphene flakes, including soot growth kinetics. Methane pyrolysis with carbon nanostructure synthesis is a two-stage process where conversion of CH4 to C2H2 precedes the growth of PAH molecules from acetylene. We present a single chemical mechanism that accurately describes both stages. We have constructed a compact and accurate chemical mechanism capable of modeling both stages of methane pyrolysis based on the ABF mechanism which was expanded with most prominent reaction pathways from the mechanism by Tao for small PAH molecules and HACA pathways for larger PAH molecules, up to 37 aromatic rings. The resulting mechanism was validated through comparison to multiple available sets of experimental data. Good agreement with the experimental data for both processes was obtained. Performance of the mechanism was tested for pyrolysis of methane-rich mixtures under long residence times leading to abundant formation of PAH molecules. It is shown that the inclusion of larger PAH species (up to A37) in the chemical mechanism is important for accurate prediction of the fraction of carbon converted to PAH molecules and, correspondingly, residual fraction of acetylene in the mixture.

physics.chem-ph↗

arXiv2021

Boron nitride nanotube precursor formation during high-temperature synthesis: kinetic and thermodynamic modelling

We performed integrated modelling of the chemical pathways of formation for boron nitride nanotube (BNNT) precursors during high-temperature synthesis in a B/N2 mixture. Modelling includes quantum chemistry, quantum-classical molecular dynamics, thermodynamic, and kinetic approaches. It is shown that BN compounds are formed in the interaction of N2 molecules with small boron clusters (N2 molecule fixation) rather than with less reactive liquid boron. We demonstrate that the transformation and consumption of liquid boron proceeds through the evaporation of clusters, Bm with m less than or equal to 5 and their subsequent conversion into BmNn chains. The production of such chains is crucial to the growth of BNNTs because these chains form the building blocks of bigger and longer BN chains and rings, which are themselves the building blocks of fullborenes and BNNTs. Moreover, kinetic modelling revealed that B4N4 and B5N4 species play a major role in the N2 molecule fixation process. The formation of these species via reactions with B4 and B5 clusters is not adequately described under the assumption of thermodynamic equilibrium because the accumulation of both B4N4 and B5N4 depends on the background gas pressure and the gas cooling rate. Long BN chains and rings, which are precursors of the fullborene and BNNT growth, form via self-assembly of component B4N4 and B5N4. Our modelling results (particularly the increased densities of B4N4 and B5N4 species at higher gas pressures) explain the experimentally observed effect of gas pressure on the yield of high-quality BNNTs. The catalytic role of hydrogen was also studied; it is shown that HBNH molecules can be the main precursor of BNNT synthesis in the presence of hydrogen.

physics.chem-ph↗

arXiv2020

Convenient analytical formula for cluster mean diameter and diameter dispersion after nucleation burst

We propose a new method of estimating the mean diameter and dispersion of clusters formed in a cooling gas, right after the nucleation stage. Using a moment model developed by Friedlander [S.K. Friedlander, Ann. N.Y. Acad. Sci. 354 (1983)], we derive an analytic relationship for both cluster diameter and diameter dispersion as a function of two of the characteristic times of the system - the cooling time and primary constituents collision time. These formulas can be used to predict diameter and dispersion variation with process parameters such as the initial monomer pressure or cooling rate. It is also possible to use them as an input to the coagulation stage, without the need to compute complex cluster generation during the nucleation burst. We compared our results with a nodal code and got excellent agreement.

cond-mat.soft↗

arXiv2019

Determining Gas Composition for Growth of BNNTs Using Thermodynamic Approach

A high-yield production of high-quality boron-nitride nanotubes (BNNTs) was reported recently in several publications. A boron-rich material is evaporated by a laser or plasma in a nitrogen-rich atmosphere to supply precursor gaseous species for nucleation and growth of BNNTs. Either hydrogen was added or pressure was increased in the system to achieve high yield and high purity of the synthesized nanotubes. According to the widely-accepted "root grow" mechanism, upon the gas cooling, boron droplets form first, then they adsorb nitrogen from surrounding gas species, and BNNTs grow on their surfaces. However, what are these precursor species that provide nitrogen for the growth is still an open question. To answer this question, we performed thermodynamic calculations of B-N mixture composition considering broad set of gas species. In enhancement of previous studies, the condensation of boron is now taken into account and is shown to have drastic effect on the gas chemical composition. B2N molecules were identified to be a major source of nitrogen for growth of BNNTs. Presence of B2N molecules in a B-N gas mixture was verified by our spectroscopic measurements during a laser ablation of boron-rich targets in nitrogen. It was shown that the increase of pressure has a quantitative effect on the mixture composition yielding increase of the precursor density. The hydrogen addition might open an additional channel of nitrogen supply to support growth of BNNTs. Nitrogen atoms react with abundant H2 molecules to form NH2 and then NH3 precursor species, instead of just recombining back to inert N2 molecules, as in the no-hydrogen case. In addition, thermodynamics was applied in conjunction with agglomeration theory to predict the size of boron droplets upon growth of BNNTs. Analytical relations for identification of crucial species densities were derived.

physics.chem-ph↗

arXiv2019

Validated Modeling of Atmospheric-Pressure Anodic Arc

We performed self-consistent modeling of the atmospheric-pressure plasmas produced by arc discharges. Special numerical procedure for coupling of the plasma current, emission current, ion current, sheath voltage drop, heat fluxes at plasma-electrode interfaces was developed and implemented into the ANSYS CFX code. Validation of the simulation results is performed through comparison of simulation results with the previously available experimental data obtained by a range of various plasma diagnostics.

physics.plasm-ph↗

arXiv2018

Synthesis of nanoparticles in carbon arc: measurements and modeling

This work studies the region of nanoparticle growth in atmospheric pressure carbon arc. Detection of the nanoparticles is realized via the planar laser induced incandescence (PLII) approach. Measurements revealed large clouds of nanoparticles in the arc periphery, bordering the region with high density of diatomic carbon molecules. Two-dimensional computational fluid dynamic simulations of the arc combined with thermodynamic modeling explain these results due to interplay of the condensation of carbon molecular species and the convection flow pattern. The results have shown that the nanoparticles are formed in the colder, outside regions of the arc and described the parameters necessary for coagulation.

physics.plasm-ph↗

arXiv2017

Effect of polarization forces on atom deposition on a non-spherical nanoparticle. Monte Carlo simulations

Trajectories of a polarizable species (atoms or molecules) in the vicinity of a negatively charged nanoparticle (at a floating potential) are considered. The atoms are pulled into regions of strong electric field by polarization forces. The polarization increases the deposition rate of the atoms and molecules at the nanoparticle. The effect of non-spherical shape of the nanoparticle is investigated by the Monte Carlo method. The shape of the non-spherical nanoparticle is approximated by an ellipsoid. Total deposition rate and its flux density distribution along the nanoparticle surface are calculated. It is shown that the flux density is not uniform along the surface. It is maximal at the nanoparticle tips.

physics.plasm-ph↗