On the crystalline structure of orthorhombic SrRuO3: A benchmark study of DFT functionals

Abstract

By investigating the crystalline structure of ground-state orthorhombic SrRuO 3 , we present a benchmark study of some of the most popular density functional theory (DFT) approaches from the local density approximation (LDA), generalized-gradient approximation (GGA), and hybrid functional families. Recent experimental success in stabilizing tetragonal and monoclinic phases of SrRuO 3 at room temperature sheds a new light on the ability to accurately describe geometry of this material by applying first-principles calculations. Therefore, our work is aimed to analyse the performance of different DFT functionals and provide some recommendations for future research of SrRuO 3 . A comparison of the obtained results to the low-temperature experimental data indicates that revised GGAs for solids are the best choice for the lattice constants and volume due to their nice accuracy and low computational cost. However, when tilting and rotation angles appear on the scene, a combination of the revised GGAs with the hybrid scheme becomes the most preferable option. It is important to note that a worse performance of LDA functional is somewhat compensated by its realistic reproduction of electronic and magnetic structure of SrRuO 3 , making it a strong competitor if the physical features are also taken into account.

The crystalline structure of SrRuO3: Application of hybrid scheme to the density functionals revised for solids

“…A somewhat different behaviour of the SOGGA functional in comparison to those of PBEsol and WC at 20% of HF exchange allows the decrease of MARE 2 to 0.59% showing that the range of a = 0.15-0.2 seems to be the most appropriate choice for the tilting and rotation angles. This observation is perfectly consistent with the result of mB1WC [4] -a combination of WC exchange, PBE correlation, and 16% of HF mixing -which is a bit lower (0.84%) compared to the MARE 2 value of WC at 15% of HF exchange (0.95%). But amounts of E X HF larger than 20% should not lead to a further improvement though, at least for PBEsol and WC approximations.…”

“…2 (b) where tilting and rotation angles of RuO 6 octahedra are taken into account. One can note that here at least ∼ 15% of HF mixing is necessary for WC and slightly less for PBEsol and SOGGA in order to get below a so-called satisfactory threshold of MAREs set to 1% in our previous paper [4]. The higher MARE 2 values compared to the errors of lattice constants and volume may be explained by the fact that variations in tilting and rotation angles involve very subtle energy changes which are much more harder to deal with.…”

“…A recent observation that pairwise differences between the results of modern solid-state codes, based on density functional theory (DFT) approaches, are comparable to those between different high-precision experiments [3] sheds a new light on the predictive potential of the first-principles simulations. In our previous paper [4], we have carefully benchmarked a bunch of DFT functionals -including local density approximation, generalized-gradient approximations (GGAs), and hybrids -in order to identify the ones that are the best at reproducing the crystalline structure of ground-state orthorhombic SrRuO 3 . The importance of such calculations has recently grown to a new level due to the experimental breakthrough in stabilizing tetragonal and monoclinic phases of SrRuO 3 at room temperature -the lack of the precision while determining the exact space group symmetry for these stabilized systems paves the way to exploit the predictive power of DFT simulations.…”

The Crystalline Structure of Tensile Strained SrRuO3: A First-Principles Investigation

“…We have chosen the PBEsol exchangecorrelation functional for the calculations because of its good performance in describing the crystalline structure of SrRuO 3 . 21 Besides, among revised generalized-gradient approximations for solids, this functional seems to be a more common choice for the plane-wave calculations with open-source pseudopotential libraries, making it easier to reproduce our findings. In general, we believe that the bulk-like nature of our modeling should not be an obstacle to realistically simulate thin-film behavior, since the thickness of the considered SrRuO 3 structure, ranging from 20 to 30 nm, 4 is far from the ultrathin limit.…”

“…On the whole, precise identification of the tilt system is a very important point of any thin-film investigation because it was experimentally shown that physical properties of strained SrRuO 3 vary according to the octahedral rotation pattern. 11,13,14 Therefore, by having at our disposal reliable tools for ab initio modeling, thoroughly discussed in a previous paper, 21 we are willing to give a broader perspective on the crystalline as well as electronic structure of tensile strained SrRuO 3 .…”