Claudia Draxl

Reproducibility in density functional theory calculations of solids

A comparison of DFT methods Density functional theory (DFT) is now routinely used for simulating material properties. Many software packages are available, which makes it challenging to know which are the best to use for a specific calculation. Lejaeghere et al. compared the calculated values for the equation of states for 71 elemental crystals from 15 different widely used DFT codes employing 40 different potentials (see the Perspective by Skylaris). Although there were variations in the calculated values, most recent codes and methods converged toward a single value, with errors comparable to those of experiment. Science, this issue p. 10.1126/science.aad3000; see also p. 1394 A survey of recent density functional theory methods shows a convergence to more accurate property calculations. [Also see Perspective by Skylaris] INTRODUCTION The reproducibility of results is one of the underlying principles of science. An observation can only be accepted by the scientific community when it can be confirmed by independent studies. However, reproducibility does not come easily. Recent works have painfully exposed cases where previous conclusions were not upheld. The scrutiny of the scientific community has also turned to research involving computer programs, finding that reproducibility depends more strongly on implementation than commonly thought. These problems are especially relevant for property predictions of crystals and molecules, which hinge on precise computer implementations of the governing equation of quantum physics. RATIONALE This work focuses on density functional theory (DFT), a particularly popular quantum method for both academic and industrial applications. More than 15,000 DFT papers are published each year, and DFT is now increasingly used in an automated fashion to build large databases or apply multiscale techniques with limited human supervision. Therefore, the reproducibility of DFT results underlies the scientific credibility of a substantial fraction of current work in the natural and engineering sciences. A plethora of DFT computer codes are available, many of them differing considerably in their details of implementation, and each yielding a certain “precision” relative to other codes. How is one to decide for more than a few simple cases which code predicts the correct result, and which does not? We devised a procedure to assess the precision of DFT methods and used this to demonstrate reproducibility among many of the most widely used DFT codes. The essential part of this assessment is a pairwise comparison of a wide range of methods with respect to their predictions of the equations of state of the elemental crystals. This effort required the combined expertise of a large group of code developers and expert users. RESULTS We calculated equation-of-state data for four classes of DFT implementations, totaling 40 methods. Most codes agree very well, with pairwise differences that are comparable to those between different high-precision experiments. Even in the case of pseudization approaches, which largely depend on the atomic potentials used, a similar precision can be obtained as when using the full potential. The remaining deviations are due to subtle effects, such as specific numerical implementations or the treatment of relativistic terms. CONCLUSION Our work demonstrates that the precision of DFT implementations can be determined, even in the absence of one absolute reference code. Although this was not the case 5 to 10 years ago, most of the commonly used codes and methods are now found to predict essentially identical results. The established precision of DFT codes not only ensures the reproducibility of DFT predictions but also puts several past and future developments on a firmer footing. Any newly developed methodology can now be tested against the benchmark to verify whether it reaches the same level of precision. New DFT applications can be shown to have used a sufficiently precise method. Moreover, high-precision DFT calculations are essential for developing improvements to DFT methodology, such as new density functionals, which may further increase the predictive power of the simulations. Recent DFT methods yield reproducible results. Whereas older DFT implementations predict different values (red darts), codes have now evolved to mutual agreement (green darts). The scoreboard illustrates the good pairwise agreement of four classes of DFT implementations (horizontal direction) with all-electron results (vertical direction). Each number reflects the average difference between the equations of state for a given pair of methods, with the green-to-red color scheme showing the range from the best to the poorest agreement. The widespread popularity of density functional theory has given rise to an extensive range of dedicated codes for predicting molecular and crystalline properties. However, each code implements the formalism in a different way, raising questions about the reproducibility of such predictions. We report the results of a community-wide effort that compared 15 solid-state codes, using 40 different potentials or basis set types, to assess the quality of the Perdew-Burke-Ernzerhof equations of state for 71 elemental crystals. We conclude that predictions from recent codes and pseudopotentials agree very well, with pairwise differences that are comparable to those between different high-precision experiments. Older methods, however, have less precise agreement. Our benchmark provides a framework for users and developers to document the precision of new applications and methodological improvements.

Big Data of Materials Science: Critical Role of the Descriptor

Statistical learning of materials properties or functions so far starts with a largely silent, nonchallenged step: the choice of the set of descriptive parameters (termed descriptor). However, when the scientific connection between the descriptor and the actuating mechanisms is unclear, the causality of the learned descriptor-property relation is uncertain. Thus, a trustful prediction of new promising materials, identification of anomalies, and scientific advancement are doubtful. We analyze this issue and define requirements for a suitable descriptor. For a classic example, the energy difference of zinc blende or wurtzite and rocksalt semiconductors, we demonstrate how a meaningful descriptor can be found systematically.

ElaStic: A tool for calculating second-order elastic constants from first principles

a b s t r a c t Elastic properties play a key role in materials science and technology. The elastic tensors at any order are defined by the Taylor expansion of the elastic energy or stress in terms of the applied strain. In this paper, we present ElaStic, a tool that is able to calculate the full second-order elastic stiffness tensor for any crystal structure from ab initio total-energy and/or stress calculations. This tool also provides the elastic compliances tensor and applies the Voigt and Reuss averaging procedure in order to obtain an evalua- tion of the bulk, shear, and Young moduli as well as the Poisson ratio of poly-crystalline samples. In a first step, the space-group is determined. Then, a set of deformation matrices is selected, and the corre- sponding structure files are produced. In a next step, total-energy or stress calculations for each deformed structure are performed by a chosen density-functional theory code. The computed energies/stresses are fitted as polynomial functions of the applied strain in order to get derivatives at zero strain. The knowl- edge of these derivatives allows for the determination of all independent components of the elastic tensor. In this context, the accuracy of the elastic constants critically depends on the polynomial fit. Therefore, we carefully study how the order of the polynomial fit and the deformation range influence the numer- ical derivatives, and we propose a new approach to obtain the most reliable results. We have applied ElaStic to representative materials for each crystal system, using total energies and stresses calculated with the full-potential all-electron codes exciting and WIEN2k as well as the pseudo-potential code

exciting: a full-potential all-electron package implementing density-functional theory and many-body perturbation theory

Linearized augmented planewave methods are known as the most precise numerical schemes for solving the Kohn-Sham equations of density-functional theory (DFT). In this review, we describe how this method is realized in the all-electron full-potential computer package, exciting. We emphasize the variety of different related basis sets, subsumed as (linearized) augmented planewave plus local orbital methods, discussing their pros and cons and we show that extremely high accuracy (microhartrees) can be achieved if the basis is chosen carefully. As the name of the code suggests, exciting is not restricted to ground-state calculations, but has a major focus on excited-state properties. It includes time-dependent DFT in the linear-response regime with various static and dynamical exchange-correlation kernels. These are preferably used to compute optical and electron-loss spectra for metals, molecules and semiconductors with weak electron-hole interactions. exciting makes use of many-body perturbation theory for charged and neutral excitations. To obtain the quasi-particle band structure, the GW approach is implemented in the single-shot approximation, known as G(0)W(0). Optical absorption spectra for valence and core excitations are handled by the solution of the Bethe-Salpeter equation, which allows for the description of strongly bound excitons. Besides these aspects concerning methodology, we demonstrate the broad range of possible applications by prototypical examples, comprising elastic properties, phonons, thermal-expansion coefficients, dielectric tensors and loss functions, magneto-optical Kerr effect, core-level spectra and more.

Epitaxial growth of π-stacked perfluoropentacene on graphene-coated quartz.

Chemical-vapor-deposited large-area graphene is employed as the coating of transparent substrates for the growth of the prototypical organic n-type semiconductor perfluoropentacene (PFP). The graphene coating is found to cause face-on growth of PFP in a yet unknown substrate-mediated polymorph, which is solved by combining grazing-incidence X-ray diffraction with theoretical structure modeling. In contrast to the otherwise common herringbone arrangement of PFP in single crystals and “standing” films, we report a π-stacked arrangement of coplanar molecules in “flat-lying” films, which exhibit an exceedingly low π-stacking distance of only 3.07 Å, giving rise to significant electronic band dispersion along the π-stacking direction, as evidenced by ultraviolet photoelectron spectroscopy. Our study underlines the high potential of graphene for use as a transparent electrode in (opto-)electronic applications, where optimized vertical transport through flat-lying conjugated organic molecules is desired.

Organic/Inorganic Hybrid Materials: Challenges for ab Initio Methodology

CONSPECTUS: Organic/inorganic hybrid structures are most exciting since one can expect new properties that are absent in either of their building blocks. They open new perspectives toward the design and tailoring of materials with desired features and functions. Prerequisite for real progress is, however, the in-depth understanding of what happens on the atomic and electronic scale. In this respect, hybrid materials pose a challenge for electronic-structure theory. Methods that proved useful for describing one side may not be applicable for the other one, and they are likely to fail for the interfaces. In this Account, we address the question to what extent we can quantitatively describe hybrid materials and where we even miss a qualitative description. We note that we are dealing with extended systems and thus adopt a solid-state approach. Therefore, density-functional theory (DFT) and many-body perturbation theory (MBPT), the GW approach for charged and the Bethe-Salpeter equation for neutral excitations, are our methods of choice. We give a brief summary of the used methodology, focusing on those aspects where problems can be expected when materials of different character meet at an interface. These issues are then taken up when discussing hybrid materials. We argue when and why, for example, standard DFT may fall short when it comes to the electronic structure of organic/metal interfaces or where the framework of MBPT can or must take over. Selected examples of organic/inorganic interfaces, structural properties, electronic bands, optical excitation spectra, and charge-transport properties as obtained from DFT and MBPT highlight which properties can be reliably computed for such materials. The crucial role of van der Waals forces is shown for sexiphenyl films, where the subtle interplay between intermolecular and molecule-substrate interactions is decisive for growth and morphologies. With a PTCDA monolayer on metal surfaces we discuss the performance of DFT in terms of interfacial electronic structure. We face the problem of a so far hidden variable, namely, electron-vibrational coupling, regarding level alignment at interfaces between organic and inorganic semiconductors. Poly(para-phenylene) adsorbed on graphene and encapsulated in carbon nanotubes represent case studies to demonstrate the impact of polarization effects and exciton delocalization in optoelectronic excitations, respectively. Polaron-induced band narrowing and its consequences for charge transport in organic crystals is exemplified for the HOMO bandwidth in naphthalene crystals. On the basis of these prototypical systems, we discuss what is missing to reach predictive power on a quantitative level for organic/inorganic hybrid materials and, thus, open a perspective toward the computational discovery of new materials for optoelectronic applications.

Confined Pyrolysis within Metal–Organic Frameworks To Form Uniform Ru3 Clusters for Efficient Oxidation of Alcohols

Here we report a novel approach to synthesize atomically dispersed uniform clusters via a cage-separated precursor preselection and pyrolysis strategy. To illustrate this strategy, well-defined Ru3(CO)12 was separated as a precursor by suitable molecular-scale cages of zeolitic imidazolate frameworks (ZIFs). After thermal treatment under confinement in the cages, uniform Ru3 clusters stabilized by nitrogen species (Ru3/CN) were obtained. Importantly, we found that Ru3/CN exhibits excellent catalytic activity (100% conversion), high chemoselectivity (100% for 2-aminobenzaldehyde), and significantly high turnover frequency (TOF) for oxidation of 2-aminobenzyl alcohol. The TOF of Ru3/CN (4320 h-1) is about 23 times higher than that of small-sized (ca. 2.5 nm) Ru particles (TOF = 184 h-1). This striking difference is attributed to a disparity in the interaction between Ru species and adsorbed reactants.

Computing Equilibrium Shapes of Wurtzite Crystals: The Example of GaN.

Crystal morphologies are important for the design and functionality of devices based on low-dimensional nanomaterials. The equilibrium crystal shape (ECS) is a key quantity in this context. It is determined by surface energies, which are hard to access experimentally but can generally be well predicted by first-principles methods. Unfortunately, this is not necessarily so for polar and semipolar surfaces of wurtzite crystals. By extending the concept of Wulff construction, we demonstrate that ECSs can nevertheless be obtained for this class of materials. For the example of GaN, we identify different crystal shapes depending on the chemical potential, shedding light on experimentally observed GaN nanostructures.

Learning physical descriptors for materials science by compressed sensing

The availability of big data in materials science offers new routes for analyzing materials properties and functions and achieving scientific understanding. Finding structure in these data that is not directly visible by standard tools and exploitation of the scientific information requires new and dedicated methodology based on approaches from statistical learning, compressed sensing, and other recent methods from applied mathematics, computer science, statistics, signal processing, and information science. In this paper, we explain and demonstrate a compressed-sensing based methodology for feature selection, specifically for discovering physical descriptors, i.e., physical parameters that describe the material and its properties of interest, and associated equations that explicitly and quantitatively describe those relevant properties. As showcase application and proof of concept, we describe how to build a physical model for the quantitative prediction of the crystal structure of binary compound semiconductors.

Band renormalization of a polymer physisorbed on graphene investigated by many-body perturbation theory

Many-body perturbation theory at the G0W0 level is employed to study the electronic properties of poly(para-phenylene) (PPP) on graphene. Analysis of the charge density and the electrostatic potential shows that the polymer-surface interaction gives rise to the formation of only weak surface dipoles with no charge transfer between the polymer and the surface. In the local-density approximation (LDA) of density-functional theory, the band structure of the combined system appears as a superposition of the eigenstates of its constituents. Consequently, the LDA band gap of PPP remains unchanged upon adsorption onto graphene. G0W0 calculations, however, renormalize the electronic levels of the weakly physisorbed polymer. Thereby, its band gap is considerably reduced compared to that of the isolated PPP chain. This effect can be understood in terms of image charges induced in the graphene layer, which allows us to explain the quasi-particle gap of PPP versus polymer-graphene distance by applying a classical image-potential model. For distances below 4.5 ˚ A, however, deviations from this simple classical model arise, which we qualitatively explain by taking into account the polarizablity of the adsorbate. For a quantitative description with predictive power, however, we emphasize the need for an accurate ab initio description of the electronic structure for weakly coupled systems at equilibrium bonding distances.