doped: Python toolkit for robust and repeatable charged defect supercell calculations

Abstract: Defects are a universal feature of crystalline solids, dictating the key properties and performance of many functional materials. Given their crucial importance yet inherent difficulty in measuring experimentally, computational methods (such as DFT and ML/classical force-fields) are widely used to predict defect behaviour at the atomic level and the resultant impact on macroscopic properties. Here we report doped, a Python package for the generation, pre-/post-processing, and analysis of defect supercell calculations. doped has been built to implement the defect simulation workflow in an efficient and user-friendly -- yet powerful and fully-flexible -- manner, with the goal of providing a robust general-purpose platform for conducting reproducible calculations of solid-state defect properties.

• The paper introduces doped, which optimizes supercell generation to maximize defect image distances and reduces finite-size errors by an average of 36%.
• It implements an advanced charge-state estimation algorithm that integrates oxidation state probabilities with the host's electronic structure for improved accuracy.
• The toolkit automates competing phase selection and symmetry handling, ensuring efficient and reproducible charged defect simulations in high-throughput environments.

A Review of "Doped: Python Toolkit for Robust and Repeatable Charged Defect Supercell Calculations"

The paper by Kavanagh et al. introduces "doped," a Python toolkit designed to streamline and enhance the computational modeling of charged defects in crystalline materials. This package is positioned to address the complexities inherent in defect modeling workflows, facilitating the pre-processing, simulation, and analysis stages. This toolkit promises to cater to the requirements of both experienced researchers and newcomers in the domain, offering robust, reproducible, and flexible computational capabilities.

The "doped" toolkit represents a significant advancement in the computational assessment of defect properties due to several key features. Among these, its supercell generation method stands out. The toolkit optimizes the choice of simulation supercells to maximize the minimum distance between periodic images of a defect, thus effectively minimizing finite-size errors while maintaining computational efficiency. This capability is particularly highlighted by the superior minimum image distances compared to existing methods like those used by the pymatgen and ASE libraries, achieving mean improvements of 36% over a test set ranging from 2-50 unit cells.

Charge-state estimation is another innovative feature of the toolkit. Defect charge states are estimated using an algorithm that incorporates oxidation state probabilities and the electronic state of the host crystal, reducing the frequency of false positives and negatives in charge-state calculations. This estimation process, detailed in the paper, enhances the predictive accuracy of defect simulations compared to prior methods.

An efficient competing phase selection mechanism is implemented within "doped," allowing users to determine elemental chemical potentials with greater precision. It automates the phase selection process by querying the Materials Project database, focusing only on phases that potentially affect chemical potential limits. This selection method reduces unnecessary calculations and enhances the simulation's efficiency.

The toolkit simplifies the previously labor-intensive process of symmetry and degeneracy handling. It automates the determination of symmetry groups for defects and computes degeneracy factors without requiring extensive user input. Such automation is crucial given the large impact these factors have on defect concentration predictions.

Additionally, "doped" ensures high levels of reproducibility through the storage of input parameters and calculation results in JSON files, supporting easy sharing and consistency in computational research. Its compatibility with high-throughput computational environments like atomate2 and AiiDA further broadens its applicability, making it suitable for large-scale materials research.

Theoretical implications of "doped" include enhanced accuracy and efficiency in defect simulation workflows, potentially leading to more insightful predictions about material properties. From a practical standpoint, the toolkit facilitates the exploration of defects in novel materials, aiding in the development of functional materials with optimized properties.

Looking forward, one can anticipate extensions of "doped" to incorporate additional computational methods and support for a broader range of DFT codes. Given the modular nature of its Python-based framework, continued development to include machine learning models for charge state estimation can be envisaged, further refining its capabilities.

In summary, "doped" addresses critical challenges in defect simulations and offers a comprehensive suite of tools that bolster both the accuracy and efficiency of computational methods in materials science. The implications of such a toolkit are vast, offering the potential for significant advancements in the field of computational materials discovery and defect engineering.