Realistic finite temperature simulations of matter are a formidable challenge for first principles methods. Long simulation times and large length scales are required, demanding years of compute time. Here we present an on-the-fly machine learning scheme that generates force fields automatically during molecular dynamics simulations. This opens up the required time and length scales, while retaining the distinctive chemical precision of first principles methods and minimizing the need for human intervention. The method is widely applicable to multi-element complex systems. We demonstrate its predictive power on the entropy driven phase transitions of hybrid perovskites, which have never been accurately described in simulations. Using machine learned potentials, isothermal-isobaric simulations give direct insight into the underlying microscopic mechanisms. Finally, we relate the phase transition temperatures of different perovskites to the radii of the involved species, and we determine the order of the transitions in Landau theory.
Two seemingly similar crystal structures of the low-temperature (~100 K) MAPbX$_3$ (X=I,Br,Cl) perovskites, but with different relative Methylammonium (MA) ordering, have appeared as representatives of this orthorhombic phase. Distinguishing them by X-ray diffraction experiments is difficult and conventional first-principles based molecular-dynamics approaches are often too computationally intensive to be feasible. Therefore, to determine the thermodynamically stable structure, we use a recently introduced on-the-fly Machine-Learning Force Field method, which reduces the computation time from …
Using the dynamical matrix of a crystal obtained from ab initio calculations, we have evaluated for the first time the strength of the dynamic flexoelectric effect and found it comparable to that of the static bulk flexoelectric effect, in agreement with earlier order-of-magnitude estimates. We also proposed a method of evaluation of these effects directly from the simulated phonon spectra. This method can also be applied to the analysis of the experimental phonon spectra, being currently the only one enabling experimental characterization of the static bulk flexoelectric effect.
Frequency-dependent and temperature-dependent dielectric measurements are performed on double perovskite Tb$_2$NiMnO$_6$. The real ($ε_1$) and imaginary ($ε_2$) parts of dielectric permittivity show three plateaus suggesting dielectric relaxation originating from bulk, grain boundaries and the sample-electrode interfaces respectively. The temperature and frequency variation of $ε_1$ and $ε_2$ are successfully simulated by a $RC$ circuit model. The complex plane of impedance, $Z'$-$Z"$, is simulated using a series network with a resistor $R$ and a constant phase element. Through the analysis of…
A combination of quantum mechanics calculations with machine learning (ML) techniques can lead to a paradigm shift in our ability to predict materials properties from first principles. Here we show that on-the-fly training of an interatomic potential described through moment tensors provides the same accuracy as state-of-the-art {\it ab inito} molecular dynamics in predicting high-temperature elastic properties of materials with two orders of magnitude less computational effort. Using the technique, we investigate high-temperature bcc phase of titanium and predict very weak, Elinvar, temperatu…
Inorganic-organic perovskites semiconductors are becoming increasingly interesting due to their remarkable optical properties, such as a high photoluminescence quantum yield and the possibility to show optical gain in a broad range of wavelengths. We have here simulated microcavities that embed MaPBI3 and 2D hybrid organic inorganic perovskite semiconductors by sandwiching such active layers between periodic, aperiodic and disordered photonic structures. We have carefully considered the refractive index dispersion of all the materials employed, such silicon dioxide, titanium dioxide and the pe…
New approach on interpretation and processing of Raman spectra of complex perovskites is suggested. Raman spectra of $PbMg_{1/3}Nb_{2/3}O_{3}$ and $% PbSc_{1/2}Ta_{1/2}O_{3}$ are successfully described on the basis of the phonon-confinement model and validity of the method is demonstrated on good agreement of experimentally obtained and computer simulated spectra of another complex perovskite, $BaMg_{1/3}Ta_{2/3}O_{3}$. The study showed that the size of nanoscale regions is a kind of an inborn characteristic of the sample and is not temperature dependent. For $PbSc_{1/2}Ta_{1/2}O_{3}$, two spe…
The phase-transition sequence of a ferroelectric perovskite such as BaTiO_3 can be simulated by computing the statistical mechanics of a first-principles derived effective Hamiltonian [Zhong, Vanderbilt and Rabe, Phys. Rev. Lett. 73, 1861 (1994)]. Within this method, the effect of an external pressure (in general, of any external field) can be studied by considering the appropriate "enthalpy" instead of the effective Hamiltonian itself. The legitimacy of this approach relies on two critical assumptions that, to the best of our knowledge, have not been adequately discussed in the literature to …
This work is aimed to study mixed-cation (hybrid methyl ammonium plus inorganic cations) halide perovskites using first principles Density Functional Theory (DFT) formalism in order to find their potential applications in perovskite based photovoltaics. This is an addition to the two previous studies performed by the author on high-throughput screening of hybrid halide and inorganic halide perovskites. This work involves mixing certain amounts of the inorganic cation to the organic site (methyl ammonium) and study the electronic structure of the resultant. 63 perovskites have been simulated in…
Metal halide perovskites (MHPs) are nowadays one of the most studied semiconductors due to their exceptional performance as active layers in solar cells. Although MHPs are excellent solid-state semiconductors, they are also ionic compounds, where ion migration plays a decisive role in their formation, their photovoltaic performance and their long-term stability. Given the above-mentioned complexity, molecular dynamics simulations based on classical force fields are especially suited to study MHP properties, such as lattice dynamics and ion migration. In particular, the possibility to model mix…
The accurate prediction of solid-solid structural phase transitions at finite temperature is a challenging task, since the dynamics is so slow that direct simulations of the phase transitions by first-principles (FP) methods are typically not possible. Here, we study the $α$-$β$ phase transition of Zr at ambient pressure by means of on-the-fly machine-learned force fields. These are automatically generated during FP molecular dynamics (MD) simulations without the need of human intervention, while retaining almost FP accuracy. Our MD simulations successfully reproduce the first-order displacive…
In this study, the optoelectronic properties of a monolithically integrated series-connected tandem solar cell are simulated. Following the large success of hybrid organic-inorganic perovskites, which have recently demonstrated large efficiencies with low production costs, we examine the possibility of using the same perovskites as absorbers in a tandem solar cell. The cell consists in a methylammonium mixed bromide-iodide lead perovskite, CH3NH3PbI3(1-x)Br3x (0 < x < 1), top sub-cell and a single-crystalline silicon bottom sub-cell. A Si-based tunnel junction connects the two sub-cells. Numer…
Ferroelectric and multiferroic materials form an important class of functional materials. Over the last twenty years, first-principles-based effective Hamiltonian approaches have been successfully developed to simulate these materials. In recent years, effective Hamiltonian approaches were further combined with molecular dynamics methods to investigate terahertz dynamical properties of various perovskites. With this combination, a variety of ferroelectric and multiferroic materials, including BaTiO_{3}, Ba(Sr,Ti)O_{3}, Pb(Zr,Ti)O_{3}, BiFeO_{3}, and SrTiO_{3} bulks and films have been simulate…
Perovskite insulator SrTiO3 is expected to be applied to the next generation of electronic and photonic devices as high-k capacitors and photocatalysts. However, reproducible growth of highly insulating stoichiometric SrTiO3 films remains challenging due to the difficulty of the precise stoichiometry control in perovskite oxide films. Here, to grow stoichiometric SrTiO3 thin films by fine-tuning multiple growth conditions, we developed a new Bayesian optimization (BO)-based machine learning method that encourages the exploration of the search space by varying the prior mean to get out of subop…
Machine-learning screening and first-principles lattice dynamics resolve the most stable structure of NaCa(BH$_4$)$_3$ from a data-driven candidate search
The Pennsylvania State University（宾夕法尼亚州立大学）；The American University in Cairo（开罗美国大学）
A multi-fidelity surrogate model for highly nonlinear multiscale problems is proposed. It is based on the introduction of two different surrogate models and an adaptive on-the-fly switching. The two concurrent surrogates are built incrementally starting from a moderate set of evaluations of the full order model. Therefore, a reduced order model (ROM) is generated. Using a hybrid ROM-preconditioned FE solver, additional effective stress-strain data is simulated while the number of samples is kept to a moderate level by using a dedicated and physics-guided sampling technique. Machine learning (M…
Crystal structure prediction is a central problem of theoretical crystallography and materials science, which until mid-2000s was considered intractable. Several methods, based on either energy landscape exploration$^{1,2}$ or, more commonly, global optimization$^{3-8}$, largely solved this problem and enabled fully non-empirical computational materials discovery$^{9,10}$. A major shortcoming is that, to avoid expensive calculations of the entropy, crystal structure prediction was done at zero Kelvin and searched for the global minimum of the enthalpy, rather than free energy. As a consequence…
Predicting challenging phase transitions with Bayesian active learning
Universität Bremen（不来梅大学）；École Polytechnique Fédérale de Lausanne（洛桑联邦理工学院）；Univ Rennes, ENSCR, CNRS, Institut des Sciences Chimiques de Rennes, UMR 6226（雷恩大学、国立高等化学学院、法国国家科学研究中心、雷恩化学科学研究所）；Sapienza Università di Roma（罗马第一大学）
