Point-defect optical transitions and thermal ionization energies from quantum Monte Carlo methods: Application to theF-center defect in MgO

Abstract

We present an approach to calculation of point defect optical and thermal ionization energies based on the highly accurate quantum Monte Carlo methods. The use of an inherently many-body theory that directly treats electron correlation offers many improvements over the typically-employed density functional theory Kohn-Sham description. In particular, the use of quantum Monte Carlo methods can help overcome the band gap problem and obviate the need for ad-hoc corrections. We demonstrate our approach to the calculation of the optical and thermal ionization energies of the F-center defect in magnesium oxide, and obtain excellent agreement with experimental and/or other high-accuracy computational results.From electronics to optoelectronics to photovoltaics, point defects influence and even dominate the properties of semiconducting materials [1][2][3][4][5][6]. Quantitative descriptions of the effect of point defects on electronic, optical, and transport properties is critical to enabling point-defect engineering for materials design. However, accurate prediction of point-defect energetics, thermal ionization energies, and optical transition energies from first principles remains a challenge. Currently, the most widely-used approach based on conventional density functional theory (DFT) suffers from poor descriptions of band gaps that render difficult the accurate description of mid-gap defect states [5,[7][8][9]]. Here we demonstrate that, by contrast, an inherently many-body approach based on quantum Monte Carlo (QMC) methods [10,11] can eliminate these problems and enable high-accuracy calculations of point defect optical and thermal ionization energies. Our computed optical transition energies are in excellent agreement with experimental and/or other highaccuracy computational results for the same system [12], and demonstrate that QMC can obtain quantitatively accurate descriptions.

Structural stability and defect energetics of ZnO from diffusion quantum Monte Carlo

“…3.26 (16). Close agreement between DMC quasi-particle gap and experimental band gap have also been reported for MgO, 22 TiO 2 , 18 FeO, 101 MnO, 102 Based on the calculated cohesive energy of Zn and ZnO, the overall accuracy of our DMC calculations is ~0.3 eV. These errors, which are likely fixed-node errors, mostly cancel out for energy differences between similar systems.…”

“…Similar approaches have been employed previously to study the oxygen vacancy in MgO with DMC 22 and various ionic defects in ZnO with hybrid functionals. 92,99 We do not expect any qualitative difference due to this approximation for .…”

“…DMC has been previously applied to study ionic defects in various materials and the results compared with local and semi-local DFT approximations. [22][23][24] For instance, formation energies evaluated with DMC are higher (by 0.3 -1 eV) than GGA for neutral oxygen vacancy in MgO 22 and self-interstitial in aluminum 23 and silicon. 24 The main error of GGA for in MgO was attributed to the overly delocalized description in GGA, which results in a small and underestimated band gap for MgO.…”

“…24 The main error of GGA for in MgO was attributed to the overly delocalized description in GGA, which results in a small and underestimated band gap for MgO. 22 As a result of the small, the deep doubly occupied level is located too close to the VBM, resulting in low formation energy. 22 A similar argument can explain the fact that LDA and GGA yield lower formation energies than DMC for in ZnO (Figure 4).…”

“…For instance, DMC has recently been applied to study different properties of transition-metal oxides, 10,[16][17][18] MgSiO 3 perovskite, 19 boron-nitride, 20 and point defects in MgO, 21,22 Al, 23 Si [24][25][26] and diamond. 27 Other QMC methods have also recently been applied to study transition-metal oxides.…”