High-throughput electronic band structure calculations: Challenges and tools

Abstract

The article is devoted to the discussion of the high-throughput approach to band structures calculations. We present scientific and computational challenges as well as solutions relying on the developed framework (Automatic Flow, AFLOW/ACONVASP). The key factors of the method are the standardization and the robustness of the procedures. Two scenarios are relevant: 1) independent users generating databases in their own computational systems (off-line approach) and 2) teamed users sharing computational information based on a common ground (on-line approach). Both cases are integrated in the framework: for off-line approaches, the standardization is automatic and fully integrated for the 14 Bravais lattices, the primitive and conventional unit cells, and the coordinates of the high symmetry k-path in the Brillouin zones. For on-line tasks, the framework offers an expandable web interface where the user can prepare and set up calculations following the proposed standard. Few examples of band structures are included. LSDA+U parameters (U, J) are also presented for Nd, Sm, and Eu. Over the past decade computational materials science has undergone a tremendous growth thanks to the availability, power and relatively limited cost of high-performance computational equipment. The highthroughput (HT) method, started from the seminal paper by Xiang et al. for combinatorial discovery of superconductors [1], has become an effective and efficient tool for materials development [2,3,4,5,6] and prediction [7,8,9,10,11,12,13]. Recent examples of computational HT are the Pareto-optimal search for alloys and catalysts [14,15], the data-mining of quantum calculations method leading to the principle-component analysis of the formation energies of many alloys in several configurations [10,11,12,16,17], the highthroughput Kohn-anomalies search in ternary lithiumborides [18,19,20], and the multi-optimization techniques used for the study of high-temperature reactions in multicomponent hydrides [21,22,23].In its practical implementation, HT uses some sort of automatic optimization technique to screen through a library of candidate compounds and to direct further refinements. The library can be a set of alloy prototypes [24,12] or a database of compounds such as the Pauling File [25] or the ICSD Database [26,27]. An important difference between the several-calculations and the HT philosophies is that the former concentrates on the calculation of a particular property, while the latter focuses on the extraction of property correlations which are used to guide the search for systems with ad-hoc characteristics. The power of HT comes with a cost. Due to the enormous amount of information produced, standardization and robustness of the procedures are necessary. This is especially true if one were concomitantly optimizing thermodynamics and electronic structure, which is required, for instance, in catalyst design [28], in accelerated "battery materials" discovery [29], and superconducting materials development [19,20]. Therefore a ratio...

Evaluation of Tavorite-Structured Cathode Materials for Lithium-Ion Batteries Using High-Throughput Computing

“…VASP calculations were managed using the AFLOW/ACONVASP framework. 29 We have compiled a database of nearly 100,000 structural energies calculated using these same parameters, with the following exceptions: In the database all calculations were initialized in a ferromagnetic spin state, the electronic minimization convergence criterion was set to 5×10 -5 eV per atom, and the ionic minimization convergence criterion was set to 5×10 -4 eV per atom. The settings used for the energies in the database are well-suited for high-throughput calculations, but they will typically result in slightly higher calculated energies than the settings used in this paper.…”

Novel high-κ dielectrics for next-generation electronic devices screened by automated ab initio calculations

“…16 On the basis of the results, the k-points corresponding to the valence top and conduction minimum are determined. Because the band gap underestimation is severe in the semilocal functional, we additionally perform a hybrid-functional (HSE06) calculation 17 (without further structural relaxation) and calculate the energy levels at the band-edge k-points identified by GGA (this scheme is called HSE@GGA hereafter).…”

Assessing the Thermoelectric Properties of Sintered Compounds via High-ThroughputAb-InitioCalculations

“…The electronic-structure calculations of the AFLOWLIB repository [36] are performed within the high-throughput framework AFLOW [10,11,[37][38][39], based on the ab-initio VASP package [40,41]. AFLOW automatically scans the ICSD [28,29], produces geometrical input files, sets appropriate parameters, runs the VASP electronic-structure calculations, and upload the results in the online AFLOWLIB database.…”

“…In addition to a small , a large P is another ingredient for efficient materials. While experimental trial-and-error approaches are expensive and time-consuming, theoretical investigations, especially within the high-throughput approach [10,11], could be both cost-and time-effective in pinpointing good candidates.…”

“…[17]). The HT approach has been used to find new materials and trends with data-mining algorithms [10,11,[18][19][20][21]. As an example, Madsen reported an ab-initio search for thermoelectrics containing Zn and Sb [22].…”