Bethe–Salpeter equation for absorption and scattering spectroscopy: implementation in the exciting code
Christian Vorwerk,Benjamin Aurich,Caterina Cocchi,Claudia Draxl
DOI: https://doi.org/10.1088/2516-1075/ab3123
2019-08-21
Electronic Structure
Abstract:The Bethe–Salpeter equation for the electron–hole correlation function is the state-of-the-art formalism for optical and core spectroscopy in condensed matter. Solutions of this equation yield the full dielectric response, including both the absorption and the inelastic scattering spectra. Here, we present an efficient implementation within the all-electron full-potential code <kbd>exciting</kbd>, which employs the linearized augmented plane-wave (L)APW+LO basis set. Being an all-electron code, <kbd>exciting</kbd> allows the calculation of optical and core excitations on the same footing. The implementation fully includes the effects of finite momentum transfer which may occur in inelastic x-ray spectroscopy and electron energy-loss spectroscopy. Our implementation does not require the application of the Tamm–Dancoff approximation that is commonly employed in the determination of absorption spectra in condensed matter. The interface with parallel linear-algebra libraries enables the calculation for complex systems. The capability of our implementation to compute, analyze, and interpret the results of different spectroscopic techniques is demonstrated by selected examples of prototypical inorganic and organic semiconductors and insulators.The Bethe–Salpeter equation (BSE) is the state-of-the-art method to describe light absorption in crystalline materials [<a class="cite hastip" href="#estab3123bib001">1</a>–<a class="cite hastip" href="#estab3123bib007">7</a>] and molecular systems in their condensed phase [<a class="cite hastip" href="#estab3123bib008">8</a>–<a class="cite hastip" href="#estab3123bib019">19</a>]. This approach enables the calculation of the dynamical polarizability including the effects of the electron–hole interaction and thus yields insight into energy, strength, and character of (bound) excitonic states. The application of the BSE formalism to treat transitions from core electrons also gives access to x-ray absorption and inelastic x-ray scattering (IXS) spectra [<a class="cite hastip" href="#estab3123bib020">20</a>–<a class="cite hastip" href="#estab3123bib034">34</a>].BSE implementations typically focus on the optical limit, where the momentum transferred from the photons to the electronic system can be neglected. While this approximation is justified for optical absorption spectra, in both IXS and electron-energy loss spectroscopy (EELS) the influence of the photon momentum loss needs to be accounted for. BSE calculations at finite momentum transfer have been performed only in a limited number of works [<a class="cite hastip" href="#estab3123bib035">35</a>–<a class="cite hastip" href="#estab3123bib040">40</a>], which nonetheless demonstrate their relevance for accessing and complementing scattering spectroscopy experiments. Furthermore, currently available BSE implementations for solids are typically limited to the Tamm–Dancoff approximation (TDA), where the coupling between excitations and de-excitations is neglected. While the TDA is reasonable for calculations related to optical absorption measurements of conventional inorganic semiconductors, the spectroscopic characterization of other material classes requires an extension of this approach, as, for instance, discussed in the context of molecular systems [<a class="cite hastip" href="#estab3123bib041">41</a>–<a class="cite hastip" href="#estab3123bib044">44</a>] and nanostructures [<a class="cite hastip" href="#estab3123bib041">41</a>, <a class="cite hastip" href="#estab3123bib045">45</a>]. Pseudo-potential implementations of the BSE formalism beyond the TDA have been reported recently [<a class="cite hastip" href="#estab3123bib046">46</a>, <a class="cite hastip" href="#estab3123bib047">47</a>].All-electron full-potential methods, in particular those employing the LAPW+LO basis set, treat core and valence electrons on the same footing, thus enabling a reliable access to both valence and core excitations [<a class="cite hastip" href="#estab3123bib026">26</a>–<a class="cite hastip" href="#estab3123bib034">34</a>]. In this paper, we present the comprehensive implementation of the BSE formalism in the all-electron full-potential code <tt>exciting</tt>, with a focus on recent developments that comprise the construction and solution of the full BSE without the application of the TDA, the solution of the BSE beyond the optical limit, and the unified description of core and valence excitations. After reviewing the theoretical background for optical and core-level excitations, we describe the general structure of the code and address how we solve computational challenges within our implementation. We demonstrate the functionalities of the new developments with selected examples, ranging from optical to core spectroscopy and spanning a broad set of materials, including bulk semiconductors and insulators as well as organic crystalline structures.In linear response theory, both absorption<p>-Abstract Truncated-</p>