Michael Probst
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Michael Probst has a total of 211 co-authors. Here are the top co-authors by number of shared publications:
- Stefan Ernest Huber
University of Graz
15 shared publications Jumras Limtrakul
School of Energy Science and Engineering, Vidyasirimedhi Institute of Science and Technology, Rayong 2120, Thailand
11 shared publications- Thana Maihom
Division of Chemistry, Department of Physical and Material Sciences, Faculty of Liberal Arts and Science, Kasetsart University, Kamphaeng Saen Campus, Nakhon Pathom 73140, Thailand
9 shared publications - José Romero
Universidade Nova de Lisboa
7 shared publications - Paulo Limao-Vieira
Departamento de Física, Universidade Federal do Paraná, Caixa Postal 19044, Curitiba 81531-980, PR, Brazil
7 shared publications - Alexander Kaiser
Institut für Ionenphysik und Angewandte Physik, Universität Innsbruck, Technikerstr. 25, A-6020 Innsbruck, Austria
6 shared publications
Latest Publications
- 5 February 2026
- Published by Royal Society of Chemistry (RSC) in Physical Chemistry Chemical Physics
- Vol. 28 (9) , 6069-6076
Computational investigation of acetylene hydrogenation to ethylene over transition metal–exchanged chabazite zeolites: mechanistic insights and descriptor-based predictions
DFT calculations reveal selective acetylene hydrogenation to ethylene over transition metals confined in CHA zeolite. SISSO identifies compact nonlinear combinations of DFT-derived descriptors that accurately predict overall activation barriers.
- 5 February 2026
- Published by Royal Society of Chemistry (RSC) in Physical Chemistry Chemical Physics
- Vol. 28 (9) , 6069-6076
Computational investigation of acetylene hydrogenation to ethylene over transition metal–exchanged chabazite zeolites: mechanistic insights and descriptor-based predictions
DFT calculations reveal selective acetylene hydrogenation to ethylene over transition metals confined in CHA zeolite. SISSO identifies compact nonlinear combinations of DFT-derived descriptors that accurately predict overall activation barriers.
- Open Access
- 7 January 2026
- Published by American Chemical Society (ACS) in ChemRxiv
A Semi-Empirical Orbital-Free DFT Approach for Modeling Partial Charges and Interaction Energies
Accurate functionals for the kinetic energy and exchange-correlation potential energy remain a critical challenge within Orbital-Free Density Functional Theory (OFDFT). The Thomas-Fermi model, one of the earliest and most significant approaches in this domain, provides a computationally efficient framework for modeling very large systems where Kohn-Sham DFT (KSDFT) is computationally prohibitive. In this work, we extend our previous OFDFT methodology [J. Phys. Chem. A (2024), 128, 6, 1163–1172 and J. Chem. Phys. 160 (2024) 235101] by introducing several key enhancements: a more rigorous theoretical formulation that solely relies on variational calculus, fictitious atomic kinetic energy terms that use the Thomas-Fermi and von Weizsäcker functionals, and empirically fitted coefficients that only depend on the atomic numbers of the pair of interacting atoms. The parameters of the model are optimized by fitting to reference data generated using KSDFT calculations on the B3LYP/aug-cc-pVTZ level of theory, allowing us to evaluate the ability of our model to capture complex chemical environments.
- 7 January 2026
- Published by American Chemical Society (ACS) in ChemRxiv
Open AccessA Semi-Empirical Orbital-Free DFT Approach for Modeling Partial Charges and Interaction Energies
Accurate functionals for the kinetic energy and exchange-correlation potential energy remain a critical challenge within Orbital-Free Density Functional Theory (OFDFT). The Thomas-Fermi model, one of the earliest and most significant approaches in this domain, provides a computationally efficient framework for modeling very large systems where Kohn-Sham DFT (KSDFT) is computationally prohibitive. In this work, we extend our previous OFDFT methodology [J. Phys. Chem. A (2024), 128, 6, 1163–1172 and J. Chem. Phys. 160 (2024) 235101] by introducing several key enhancements: a more rigorous theoretical formulation that solely relies on variational calculus, fictitious atomic kinetic energy terms that use the Thomas-Fermi and von Weizsäcker functionals, and empirically fitted coefficients that only depend on the atomic numbers of the pair of interacting atoms. The parameters of the model are optimized by fitting to reference data generated using KSDFT calculations on the B3LYP/aug-cc-pVTZ level of theory, allowing us to evaluate the ability of our model to capture complex chemical environments.
- Open Access
- 17 September 2025
- Published by Royal Society of Chemistry (RSC) in RSC Advances
- Vol. 15 (41) , 34274-34281
Molecular dynamics simulations of the sputtering of boron and boron oxide surfaces
We employ classical molecular dynamics (MD) simulations to study processes governed by particle–surface interactions. The interatomic potential energy functions are described by a neural network potential (NNP) trained on an extensive set of density functional theory (DFT) calculations in a semi-iterative fashion. Potential construction and simulation set up follow the Behler–Parrinello approach. As a specific example we investigate sputtering, reflection, and adsorption phenomena occurring on boron and boron oxide surfaces under the impact of deuterium atoms, systems that reflect recent developments in materials science. Besides the frequent use of boron surfaces as an oxygen-gathering material in technical applications, boron-based compounds will be used in future fusion devices. Understanding their interaction with energetic plasma particles is essential, yet their stability and sputtering behavior at the atomic level have remained largely unexplored. From our simulations, we analyzed sputtering yields, adsorption, and reflection events on both boron and boron oxide surfaces. The production simulations included about 750 atoms and covered a range of incident energies and impact angles. Increasing the deuterium impact energy generally leads to an increase in sputtering yields but with distinct energy-dependent trends. The sputtering yield of boron from boron oxide surfaces remains significantly lower than from surfaces of pristine boron. We use an analytical approach to estimate an effective surface binding energy. The work presented here, especially the various energy- and angle dependent rates, can be used to create a parametric model of the B and B2O3 surfaces under the impact of hot particles.
- 17 September 2025
- Published by Royal Society of Chemistry (RSC) in RSC Advances
- Vol. 15 (41) , 34274-34281
Open AccessMolecular dynamics simulations of the sputtering of boron and boron oxide surfaces
We employ classical molecular dynamics (MD) simulations to study processes governed by particle–surface interactions. The interatomic potential energy functions are described by a neural network potential (NNP) trained on an extensive set of density functional theory (DFT) calculations in a semi-iterative fashion. Potential construction and simulation set up follow the Behler–Parrinello approach. As a specific example we investigate sputtering, reflection, and adsorption phenomena occurring on boron and boron oxide surfaces under the impact of deuterium atoms, systems that reflect recent developments in materials science. Besides the frequent use of boron surfaces as an oxygen-gathering material in technical applications, boron-based compounds will be used in future fusion devices. Understanding their interaction with energetic plasma particles is essential, yet their stability and sputtering behavior at the atomic level have remained largely unexplored. From our simulations, we analyzed sputtering yields, adsorption, and reflection events on both boron and boron oxide surfaces. The production simulations included about 750 atoms and covered a range of incident energies and impact angles. Increasing the deuterium impact energy generally leads to an increase in sputtering yields but with distinct energy-dependent trends. The sputtering yield of boron from boron oxide surfaces remains significantly lower than from surfaces of pristine boron. We use an analytical approach to estimate an effective surface binding energy. The work presented here, especially the various energy- and angle dependent rates, can be used to create a parametric model of the B and B2O3 surfaces under the impact of hot particles.
- Open Access
- 3 September 2025
- Published by American Chemical Society (ACS) in ACS Omega
- Vol. 10 (36) , 41890-41898
Glyphosate Adsorption on Metal–Organic Framework M3(BTC)2: Mechanistic Insights and Descriptors for the Adsorption Energy from Density Functional Calculations
Adsorption-based removal has emerged as an effective strategy for mitigating the harmful impact of glyphosate. Here, the adsorption of glyphosate over the coordinatively unsaturated metal–organic framework (MOF) M3(BTC)2 (M = V, Cr, Mn, Fe, Co, Ni, Cu, Zn) was investigated by means of density functional theory (DFT) calculations. Among the studied MOFs, Zn3(BTC)2 and Ni3(BTC)2 show the highest adsorption energy. The Lewis acidity of the MOFs, as determined by ammonia adsorption, is found to significantly influence glyphosate adsorption. Decomposing the adsorption free energy reveals that the adsorption is dominated by the bond energy. A moderately strong linear correlation between the adsorption energies of glyphosate and ammonia suggests that ammonia adsorption can serve as a descriptor for predicting the glyphosate adsorption energy. To enhance the predictive accuracy, we employed the sure independence screening and sparsifying operator (SISSO) method, which explores a large pool of mathematical combinations of DFT descriptors to identify the most relevant ones. The resulting models enable accurate and transferable prediction of glyphosate adsorption energy.
- 3 September 2025
- Published by American Chemical Society (ACS) in ACS Omega
- Vol. 10 (36) , 41890-41898
Open AccessGlyphosate Adsorption on Metal–Organic Framework M3(BTC)2: Mechanistic Insights and Descriptors for the Adsorption Energy from Density Functional Calculations
Adsorption-based removal has emerged as an effective strategy for mitigating the harmful impact of glyphosate. Here, the adsorption of glyphosate over the coordinatively unsaturated metal–organic framework (MOF) M3(BTC)2 (M = V, Cr, Mn, Fe, Co, Ni, Cu, Zn) was investigated by means of density functional theory (DFT) calculations. Among the studied MOFs, Zn3(BTC)2 and Ni3(BTC)2 show the highest adsorption energy. The Lewis acidity of the MOFs, as determined by ammonia adsorption, is found to significantly influence glyphosate adsorption. Decomposing the adsorption free energy reveals that the adsorption is dominated by the bond energy. A moderately strong linear correlation between the adsorption energies of glyphosate and ammonia suggests that ammonia adsorption can serve as a descriptor for predicting the glyphosate adsorption energy. To enhance the predictive accuracy, we employed the sure independence screening and sparsifying operator (SISSO) method, which explores a large pool of mathematical combinations of DFT descriptors to identify the most relevant ones. The resulting models enable accurate and transferable prediction of glyphosate adsorption energy.
- Open Access
- 7 April 2025
- Published by AIP Publishing in APL Machine Learning
- Vol. 3 (2)
Accurate classification of materials with elEmBERT: Element embeddings for chemical benchmarks
We introduce the elEmBERT model for chemical classification tasks. It is based on deep learning techniques, such as a multilayer encoder architecture. We demonstrate the opportunities offered by our approach on sets of organic, inorganic, and crystalline compounds. In particular, we developed and tested the model using the Materials Project and MoleculeNet benchmarks, which include crystal properties and drug design-related benchmarks. We also conduct an analysis of vector representations of chemical compounds, shedding light on the underlying patterns in structural data. Our model exhibits exceptional predictive capabilities and proves universally applicable to molecular and material datasets. For instance, on the Tox21 dataset, we achieved an average precision of 96%, surpassing the previously best result by 10%.
- 7 April 2025
- Published by AIP Publishing in APL Machine Learning
- Vol. 3 (2)
Open AccessAccurate classification of materials with elEmBERT: Element embeddings for chemical benchmarks
We introduce the elEmBERT model for chemical classification tasks. It is based on deep learning techniques, such as a multilayer encoder architecture. We demonstrate the opportunities offered by our approach on sets of organic, inorganic, and crystalline compounds. In particular, we developed and tested the model using the Materials Project and MoleculeNet benchmarks, which include crystal properties and drug design-related benchmarks. We also conduct an analysis of vector representations of chemical compounds, shedding light on the underlying patterns in structural data. Our model exhibits exceptional predictive capabilities and proves universally applicable to molecular and material datasets. For instance, on the Tox21 dataset, we achieved an average precision of 96%, surpassing the previously best result by 10%.
- 11 February 2025
- Published by American Chemical Society (ACS) in The Journal of Physical Chemistry C
- Vol. 129 (10) , 5182-5190
Computational Study of the Transfer Hydrogenation of Furfural to Furfuryl Alcohol in the UiO-66 Metal–Organic Frameworks: Mechanistic Insights and Predictive Descriptors for Transition States and Activation Energies
- 11 February 2025
- Published by American Chemical Society (ACS) in The Journal of Physical Chemistry C
- Vol. 129 (10) , 5182-5190
Computational Study of the Transfer Hydrogenation of Furfural to Furfuryl Alcohol in the UiO-66 Metal–Organic Frameworks: Mechanistic Insights and Predictive Descriptors for Transition States and Activation Energies
- Open Access
Modelling the Impact of Argon Atoms on a WO3 Surface by Molecular Dynamics Simulations
Machine learning potential energy functions can drive the atomistic dynamics of molecules, clusters, and condensed phases. They are amongst the first examples that showed how quantum mechanics together with machine learning can predict chemical reactions as well as material properties and even lead to new materials. In this work, we study the behaviour of tungsten trioxide (WO3) surfaces upon particle impact by employing potential energy surfaces represented by neural networks. Besides being omnipresent on tungsten surfaces exposed to air, WO3 plays an important role in nuclear fusion experiments due to the preferred use of tungsten for plasma-facing components. In this instance, the formation of WO3 is caused by the omnipresent traces of oxygen. WO3 becomes a plasma-facing material, but its properties, especially concerning degradation, have hardly been studied. We employ molecular dynamics simulations to investigate sputtering, reflection, and adsorption phenomena occurring on WO3 surfaces irradiated with Argon. The machine-learned potential energy function underlying the MD simulations is modelled using a neural network (NNP) trained from large sets of density functional theory calculations by means of the Behler–Parrinello method. The analysis focuses on sputtering yields for both oxygen and tungsten (W), for various incident energies and impact angles. An increase in Ar incident energy increases the sputtering yield of oxygen, with distinct features observed in different energy ranges. The sputtering yields of tungsten remain exceedingly low, even compared to pristine W surfaces. The ratios between the reflection, adsorption, and retention of the Ar atoms have been analyzed on their dependence of impact energy and incident end angles. We find that the energy spectrum of sputtered oxygen atoms follows a lognormal distribution and offers information about surface binding energies on the WO3 surface.
Open AccessModelling the Impact of Argon Atoms on a WO3 Surface by Molecular Dynamics Simulations
Machine learning potential energy functions can drive the atomistic dynamics of molecules, clusters, and condensed phases. They are amongst the first examples that showed how quantum mechanics together with machine learning can predict chemical reactions as well as material properties and even lead to new materials. In this work, we study the behaviour of tungsten trioxide (WO3) surfaces upon particle impact by employing potential energy surfaces represented by neural networks. Besides being omnipresent on tungsten surfaces exposed to air, WO3 plays an important role in nuclear fusion experiments due to the preferred use of tungsten for plasma-facing components. In this instance, the formation of WO3 is caused by the omnipresent traces of oxygen. WO3 becomes a plasma-facing material, but its properties, especially concerning degradation, have hardly been studied. We employ molecular dynamics simulations to investigate sputtering, reflection, and adsorption phenomena occurring on WO3 surfaces irradiated with Argon. The machine-learned potential energy function underlying the MD simulations is modelled using a neural network (NNP) trained from large sets of density functional theory calculations by means of the Behler–Parrinello method. The analysis focuses on sputtering yields for both oxygen and tungsten (W), for various incident energies and impact angles. An increase in Ar incident energy increases the sputtering yield of oxygen, with distinct features observed in different energy ranges. The sputtering yields of tungsten remain exceedingly low, even compared to pristine W surfaces. The ratios between the reflection, adsorption, and retention of the Ar atoms have been analyzed on their dependence of impact energy and incident end angles. We find that the energy spectrum of sputtered oxygen atoms follows a lognormal distribution and offers information about surface binding energies on the WO3 surface.
- Open Access
- 15 July 2024
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 64 (15) , 5762-5770
Structure to Property: Chemical Element Embeddings for Predicting Electronic Properties of Crystals
We present a new general-purpose machine learning model that is able to predict a variety of crystal properties, including Fermi level energy and band gap, as well as spectral ones such as electronic densities of states. The model is based on atomic representations that enable it to effectively capture complex information about each atom and its surrounding environment in a crystal. The accuracy achieved for band gaps exceeds results previously published. By design, our model is not restricted to the electronic properties discussed here but can be extended to fit diverse chemical descriptors. Its advantages are (a) its low computational requirements, making it an efficient tool for high-throughput screening of materials; and (b) the simplicity and flexibility of its architecture, facilitating implementation and interpretation, especially for researchers in the field of computational chemistry.
- 15 July 2024
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 64 (15) , 5762-5770
Open AccessStructure to Property: Chemical Element Embeddings for Predicting Electronic Properties of Crystals
We present a new general-purpose machine learning model that is able to predict a variety of crystal properties, including Fermi level energy and band gap, as well as spectral ones such as electronic densities of states. The model is based on atomic representations that enable it to effectively capture complex information about each atom and its surrounding environment in a crystal. The accuracy achieved for band gaps exceeds results previously published. By design, our model is not restricted to the electronic properties discussed here but can be extended to fit diverse chemical descriptors. Its advantages are (a) its low computational requirements, making it an efficient tool for high-throughput screening of materials; and (b) the simplicity and flexibility of its architecture, facilitating implementation and interpretation, especially for researchers in the field of computational chemistry.
- Open Access
- 21 June 2024
- Published by AIP Publishing in The Journal of Chemical Physics
- Vol. 160 (23)
A polarizable valence electron density based force field for high-energy interactions between atoms and molecules
High-accuracy molecular force field models suited for hot gases and plasmas are not as abundant as those geared toward ambient pressure and temperature conditions. Here, we present an improved version of our previous electron-density based force field model that can now account for polarization effects by adjusting the atomic valence electron contributions to match ab initio calculated Mulliken partial charges. Using a slightly modified version of the Hohenberg–Kohn theorem, we also include an improved theoretical formulation of our model when applied to systems with degenerate ground states. We present two variants of our polarizable model, fitted from ab initio reference data calculated at CCSD(T)/cc-pVTZ and CCSD(T)/CEP-31G levels of theory, that both accurately model water dimer interaction energies. Further improvements include the additional interaction components with fictitious non-spherically symmetric, yet atom-centered, electron densities and fitting the exchange and correlation coefficients against analytical expressions. The latter removes all unphysical oscillations that are observed in the previous non-polarizable variant of our force field.
- 21 June 2024
- Published by AIP Publishing in The Journal of Chemical Physics
- Vol. 160 (23)
Open AccessA polarizable valence electron density based force field for high-energy interactions between atoms and molecules
High-accuracy molecular force field models suited for hot gases and plasmas are not as abundant as those geared toward ambient pressure and temperature conditions. Here, we present an improved version of our previous electron-density based force field model that can now account for polarization effects by adjusting the atomic valence electron contributions to match ab initio calculated Mulliken partial charges. Using a slightly modified version of the Hohenberg–Kohn theorem, we also include an improved theoretical formulation of our model when applied to systems with degenerate ground states. We present two variants of our polarizable model, fitted from ab initio reference data calculated at CCSD(T)/cc-pVTZ and CCSD(T)/CEP-31G levels of theory, that both accurately model water dimer interaction energies. Further improvements include the additional interaction components with fictitious non-spherically symmetric, yet atom-centered, electron densities and fitting the exchange and correlation coefficients against analytical expressions. The latter removes all unphysical oscillations that are observed in the previous non-polarizable variant of our force field.
- Open Access
- 27 March 2024
- Published by American Chemical Society (ACS) in ChemRxiv
A Polarizable Valence Electron Density Based Force Field for High-Energy Interactions between Atoms and Molecules
High-accuracy molecular force field models suited for hot gases and plasmas are not as abundant as those geared towards ambient pressure and temperature conditions. Here we present an improved version of our previous electron- density based force field model that can account for polarization effects by adjusting the atomic valence electron contributions to match ab initio calculated Mulliken partial charges. Using a slightly modified version of the Hohenberg-Kohn theorem, we also present an improved theoretical formulation of our model when dealing with systems with degenerate ground states. Preliminary results obtained from this methodology for water dimer calculations using CCSD(T)/cc-pVTZ and CCSD(T)/CEP-31G level of theory are presented using an order ten approximation. Further improvements include the additional interaction components with fictitious non-spherically symmetric, yet atom-centered, electron densities and fitting the exchange and correlation coefficients against analytical expressions. The latter removes all unphysical oscillations that are observed in the previous non-polarizable variant of our force field.
- 27 March 2024
- Published by American Chemical Society (ACS) in ChemRxiv
Open AccessA Polarizable Valence Electron Density Based Force Field for High-Energy Interactions between Atoms and Molecules
High-accuracy molecular force field models suited for hot gases and plasmas are not as abundant as those geared towards ambient pressure and temperature conditions. Here we present an improved version of our previous electron- density based force field model that can account for polarization effects by adjusting the atomic valence electron contributions to match ab initio calculated Mulliken partial charges. Using a slightly modified version of the Hohenberg-Kohn theorem, we also present an improved theoretical formulation of our model when dealing with systems with degenerate ground states. Preliminary results obtained from this methodology for water dimer calculations using CCSD(T)/cc-pVTZ and CCSD(T)/CEP-31G level of theory are presented using an order ten approximation. Further improvements include the additional interaction components with fictitious non-spherically symmetric, yet atom-centered, electron densities and fitting the exchange and correlation coefficients against analytical expressions. The latter removes all unphysical oscillations that are observed in the previous non-polarizable variant of our force field.