Machine-learning methods for structure prediction of multi-component perovskites

@article{Singraber2019,

title = {Parallel Multistream Training of High-Dimensional Neural Network Potentials},

author = {Andreas Singraber and Tobias Morawietz and Jörg Behler and Christoph Dellago},

doi = {10.1021/acs.jctc.8b01092},

year  = {2019},

date = {2019-04-17},

journal = {Journal of Chemical Theory and Computation},

volume = {15},

number = {5},

pages = {3075--3092},

publisher = {American Chemical Society (ACS)},

abstract = {Over the past years high-dimensional neural network potentials (HDNNPs), fitted to accurately reproduce ab initio potential energy surfaces, have become a powerful tool in chemistry, physics and materials science. Here, we focus on the training of the neural networks that lies at the heart of the HDNNP method. We present an efficient approach for optimizing the weight parameters of the neural network via multistream Kalman filtering, using potential energies and forces as reference data. In this procedure, the choice of the free parameters of the Kalman filter can have a significant impact on the fit quality. Carrying out a large parameter study, we determine optimal settings and demonstrate how to optimize training results of HDNNPs. Moreover, we illustrate our HDNNP training approach by revisiting previously presented fits for water and developing a new potential for copper sulfide. This material, accessible in computer simulations so far only via first-principles methods, forms a particularly complex solid structure at low temperatures and undergoes a phase transition to a superionic state upon heating. Analyzing MD simulations carried out with the Cu_{2}S HDNNP, we confirm that the underlying ab initio reference method indeed reproduces this behavior.},

keywords = {P12, pre-TACO},

pubstate = {published},

tppubtype = {article}

}

@article{Schlegel2019,

title = {How water flips at charged titanium dioxide: an SFG-study on the water–TiO_{2} interface},

author = {Simon J Schlegel and Saman Hosseinpour and Maximilian Gebhard and Anjana Devi and Mischa Bonn and Ellen H. G. Backus},

doi = {10.1039/c9cp01131e},

year  = {2019},

date = {2019-04-05},

urldate = {2019-04-05},

journal = {Physical Chemistry Chemical Physics},

volume = {21},

number = {17},

pages = {8956--8964},

publisher = {Royal Society of Chemistry (RSC)},

abstract = {Photocatalytic splitting of water into hydrogen and oxygen by utilizing sunlight and a photocatalyst is a promising way of generating clean energy. Here, we report a molecular-level study on heavy water (D_{2}O) interacting with TiO_{2} as a model photocatalyst. We employed the surface specific technique Sum-Frequency-Generation (SFG) spectroscopy to determine the nature of the hydrogen bonding environment and the orientation of interfacial water molecules using their OD-stretch vibrations as reporters. By examining solutions with various pD-values, we observe an intensity-minimum at around pD 5, corresponding to the balance of protonation and deprotonation of TiO_{2} (point of zero charge). The majority of water molecules’ deuterium atoms point away from the interface when the pD is below 5, and point towards the surface when the pD is higher than 5, with strong hydrogen bonds towards the surface.},

keywords = {P11, pre-TACO},

pubstate = {published},

tppubtype = {article}

}

@article{Diebold2019,

title = {Preface: Surface Science of functional oxides},

author = {Ulrike Diebold and Günther Rupprechter},

doi = {10.1016/j.susc.2018.11.017},

year  = {2019},

date = {2019-03-01},

journal = {Surface Science},

volume = {681},

pages = {A1},

publisher = {Elsevier BV},

keywords = {P02, P08, pre-TACO},

pubstate = {published},

tppubtype = {article}

}

@article{Singraber2019a,

title = {Library-Based LAMMPS Implementation of High-Dimensional Neural Network Potentials},

author = {Andreas Singraber and Jörg Behler and Christoph Dellago},

doi = {10.1021/acs.jctc.8b00770},

year  = {2019},

date = {2019-01-24},

journal = {Journal of Chemical Theory and Computation},

volume = {15},

number = {3},

pages = {1827--1840},

publisher = {American Chemical Society (ACS)},

abstract = {Neural networks and other machine learning approaches have been successfully used to accurately represent atomic interaction potentials derived from computationally demanding electronic structure calculations. Due to their low computational cost, such representations open the possibility for large scale reactive molecular dynamics simulations of processes with bonding situations that cannot be described accurately with traditional empirical force fields. Here, we present a library of functions developed for the implementation of neural network potentials. Written in C++, this library incorporates several strategies resulting in a very high efficiency of neural network potential-energy and force evaluations. Based on this library, we have developed an implementation of the neural network potential within the molecular dynamics package LAMMPS and demonstrate its performance using liquid water as a test system.},

keywords = {P12, pre-TACO},

pubstate = {published},

tppubtype = {article}

}

@article{Reticcioli2019,

title = {Interplay between Adsorbates and Polarons: CO on Rutile TiO_{2}(110)},

author = {Michele Reticcioli and Igor Sokolović and Michael Schmid and Ulrike Diebold and Martin Setvin and Cesare Franchini},

doi = {10.1103/physrevlett.122.016805},

year  = {2019},

date = {2019-01-09},

journal = {Physical Review Letters},

volume = {122},

number = {1},

pages = {016805},

publisher = {American Physical Society (APS)},

abstract = {Polaron formation plays a major role in determining the structural, electrical, and chemical properties of ionic crystals. Using a combination of first-principles calculations, scanning tunneling microscopy, and atomic force microscopy, we analyze the interaction of polarons with CO molecules adsorbed on the reduced rutile TiO_{2}(110) surface. Adsorbed CO shows attractive coupling with polarons in the surface layer, and repulsive interaction with polarons in the subsurface layer. As a result, CO adsorption depends on the reduction state of the sample. For slightly reduced surfaces, many adsorption configurations with comparable adsorption energies exist and polarons reside in the subsurface layer. At strongly reduced surfaces, two adsorption configurations dominate: either inside an oxygen vacancy, or at surface Ti_{5c} sites, coupled with a surface polaron. Similar conclusions are predicted for TiO_{2}(110) surfaces containing near-surface Ti interstitials. These results show that polarons are of primary importance for understanding the performance of polar semiconductors and transition metal oxides in catalysis and energy-related applications.},

keywords = {P02, P07, pre-TACO},

pubstate = {published},

tppubtype = {article}

}

@article{Cheng2019,

title = {Ab initio thermodynamics of liquid and solid water},

author = {Bingqing Cheng and Edgar A Engel and Jörg Behler and Christoph Dellago and Michele Ceriotti},

doi = {10.1073/pnas.1815117116},

year  = {2019},

date = {2019-01-04},

urldate = {2019-01-04},

journal = {Proceedings of the National Academy of Sciences},

volume = {116},

number = {4},

pages = {1110--1115},

publisher = {Proceedings of the National Academy of Sciences},

abstract = {A central goal of computational physics and chemistry is to predict material properties by using first-principles methods based on the fundamental laws of quantum mechanics. However, the high computational costs of these methods typically prevent rigorous predictions of macroscopic quantities at finite temperatures, such as heat capacity, density, and chemical potential. Here, we enable such predictions by marrying advanced free-energy methods with data-driven machine-learning interatomic potentials. We show that, for the ubiquitous and technologically essential system of water, a first-principles thermodynamic description not only leads to excellent agreement with experiments, but also reveals the crucial role of nuclear quantum fluctuations in modulating the thermodynamic stabilities of different phases of water.},

keywords = {P12, pre-TACO},

pubstate = {published},

tppubtype = {article}

}

@article{Lukashuk2018,

title = {Operando Insights into CO Oxidation on Cobalt Oxide Catalysts by NAP-XPS, FTIR, and XRD},

author = {Liliana Lukashuk and Nevzat Yigit and Raffael Rameshan and Elisabeth Kolar and Detre Teschner and Michael Hävecker and Axel Knop-Gericke and Robert Schlögl and Karin Föttinger and Günther Rupprechter},

doi = {10.1021/acscatal.8b01237},

year  = {2018},

date = {2018-08-07},

urldate = {2018-08-07},

journal = {ACS Catalysis},

volume = {8},

number = {9},

pages = {8630--8641},

publisher = {American Chemical Society (ACS)},

abstract = {Cobalt oxide Co_{3}O_{4} has recently emerged as promising, noble metal-free catalyst for oxidation reactions but a better understanding of the active catalyst under working conditions is required for further development and potential commercialization. An operando approach has been applied, combining near ambient (atmospheric) pressure X-ray photoelectron spectroscopy (NAP-XPS), Fourier transform infrared spectroscopy (FTIR), or X-ray diffraction (XRD) with simultaneous catalytic tests of CO oxidation on Co_{3}O_{4}, enabling one to monitor surface and bulk states under various reaction conditions (steady-state and dynamic conditions switching between CO and O_{2}). On the basis of the surface-specific chemical information a complex network of different reaction pathways unfolded: Mars-van-Krevelen (MvK), CO dissociation followed by carbon oxidation, and formation of carbonates. A possible Langmuir–Hinshelwood (LH) pathway cannot be excluded because of the good activity when no oxygen vacancies were detected. The combined NAP-XPS/FTIR results are in line with a MvK mechanism above 100 °C, involving the Co^{3+}/Co^{2+} redox couple and oxygen vacancy formation. Under steady state, the Co_{3}O_{4} surface appeared oxidized and the amount of reduced Co^{2+} species at/near the surface remained low up to 200 °C. Only in pure CO, about 15% of surface reduction were detected, suggesting that the active sites are a minority species. The operando spectroscopic studies also revealed additional reaction pathways: CO dissociation followed by carbon reoxidation and carbonate formation and its decomposition. However, due to their thermal stability in various atmospheres, the carbonates are rather spectators and also CO dissociation seems a minor route. This study thus highlights the benefits of combining operando surface sensitive techniques to gain insight into catalytically active surfaces.},

keywords = {P08, P10, pre-TACO},

pubstate = {published},

tppubtype = {article}

}

@article{Reticcioli2017,

title = {Polaron-Driven Surface Reconstructions},

author = {Michele Reticcioli and Martin Setvin and Xianfeng Hao and Peter Flauger and Georg Kresse and Michael Schmid and Ulrike Diebold and Cesare Franchini},

doi = {10.1103/physrevx.7.031053},

year  = {2017},

date = {2017-09-25},

urldate = {2017-09-25},

journal = {Physical Review X},

volume = {7},

number = {3},

pages = {031053},

publisher = {American Physical Society (APS)},

abstract = {Geometric and electronic surface reconstructions determine the physical and chemical properties of surfaces and, consequently, their functionality in applications. The reconstruction of a surface minimizes its surface free energy in otherwise thermodynamically unstable situations, typically caused by dangling bonds, lattice stress, or a divergent surface potential, and it is achieved by a cooperative modification of the atomic and electronic structure. Here, we combined first-principles calculations and surface techniques (scanning tunneling microscopy, non-contact atomic force microscopy, scanning tunneling spectroscopy) to report that the repulsion between negatively charged polaronic quasiparticles, formed by the interaction between excess electrons and the lattice phonon field, plays a key role in surface reconstructions. As a paradigmatic example, we explain the (1×1) to (1×2) transition in rutile TiO_{2}(110).},

keywords = {P02, P03, P07, pre-TACO},

pubstate = {published},

tppubtype = {article}

}

@article{Legrain2017,

title = {How Chemical Composition Alone Can Predict Vibrational Free Energies and Entropies of Solids},

author = {Fleur Legrain and Jesús Carrete and Ambroise van Roekeghem and Stefano Curtarolo and Natalio Mingo},

doi = {10.1021/acs.chemmater.7b00789},

year  = {2017},

date = {2017-06-22},

journal = {Chemistry of Materials},

volume = {29},

number = {15},

pages = {6220--6227},

publisher = {American Chemical Society (ACS)},

abstract = {Computing vibrational free energies (F_{vib}) and entropies (S_{vib}) has posed a long-standing challenge to the high-throughput ab initio investigation of finite temperature properties of solids. Here, we use machine-learning techniques to efficiently predict F_{vib} and S_{vib} of crystalline compounds in the Inorganic Crystal Structure Database. Using descriptors based simply on the chemical formula and using a training set of only 300 compounds, mean absolute errors of less than 0.04 meV/K/atom (15 meV/atom) are achieved for S_{vib} (F_{vib}), whose values are distributed within a range of 0.9 meV/K/atom (300 meV/atom.) In addition, for training sets containing fewer than 2000 compounds, the chemical formula alone is shown to perform as well as, if not better than, four other more complex descriptors previously used in the literature. The accuracy and simplicity of the approach means that it can be advantageously used for fast screening of chemical reactions at finite temperatures.},

keywords = {P09, pre-TACO},

pubstate = {published},

tppubtype = {article}

}

@article{Lukashuk2016,

title = {Operando XAS and NAP-XPS studies of preferential CO oxidation on Co_{3}O_{4} and CeO_{2}-Co_{3}O_{4 }catalysts},

author = {Liliana Lukashuk and Karin Föttinger and Elisabeth Kolar and Christoph Rameshan and Detre Teschner and Michael Hävecker and Axel Knop-Gericke and Nevzat Yigit and Hao Li and Eamon McDermott and Michael Stöger-Pollach and Günther Rupprechter},

doi = {10.1016/j.jcat.2016.09.002},

year  = {2016},

date = {2016-12-01},

urldate = {2016-12-01},

journal = {Journal of Catalysis},

volume = {344},

pages = {1--15},

publisher = {Elsevier BV},

abstract = {Co_{3}O_{4} is a promising catalyst for removing CO from H_{2} streams via the preferential CO oxidation (PROX). A Mars-van-Krevelen redox mechanism is often suggested but a detailed knowledge especially of the oxidation state of the catalytically active surface under reaction conditions is typically missing. We have thus utilized operando X-ray absorption spectroscopy to examine structure and oxidation state during PROX, and near atmospheric pressure-XPS at low photoelectron kinetic energies and thus high surface sensitivity to monitor surface composition changes. The rather easy surface reduction in pure CO (starting already at ∼100 °C) and the easy reoxidation by O_{2} suggest that molecularly adsorbed CO reacts with lattice oxygen, which is replenished by gas phase O_{2}. Nevertheless, the steady state concentration of oxygen vacancies under reaction conditions is too low even for XPS detection so that both the bulk and surface of Co_{3}O_{4} appear fully oxidized during PROX. Furthermore, the effect of adding CeO_{2} (a less active material) to Co_{3}O_{4} was studied. Promotion of Co_{3}O_{4} with 10 wt% CeO_{2} increases the reduction temperatures in CO and H_{2} and enhances the PROX activity. Since CeO_{2} is a less active material, this can only be explained by a higher activity of the Co-O-Ce interface.},

keywords = {P08, P10, pre-TACO},

pubstate = {published},

tppubtype = {article}

}