Characterization and quantification of numerical errors in threshold displacement energy calculated by molecular dynamics in bcc-Fe

[Display omitted] The accumulation of radiation defects in nuclear fusion/fission materials often causes adverse changes in material properties. The threshold displacement energy (TDE) is one of the fundamental quantities used to evaluate the number of formed radiation defects. In the present study, to understand the error sources in TDE calculation using the molecular dynamics method and to find an optimal strategy for accurate and precise calculation, we analyze how the calculated TDE is affected by the calculation settings, including the system shape and size, the recoil energy increment in searching the threshold value, the sampling numbers of recoil directions and recoil event time samples, and the potential model. Bcc-Fe is used as a test material, and three potential models are employed for comparison. As a result, the calculation errors caused by these settings are successfully classified into accuracy error and precision error and then quantified. The type and trend of the error caused by each setting are not qualitatively different between the tested potential models, although they are quantitatively different. This potential model dependence can often be reasonably explained from the recoil direction dependence of the TDE and the defect formation mechanisms. Finally, using the equations derived to quantify the accuracy and precision of the calculated TDE and the calculation cost, we suggest some strategies to minimize the error with a given limited computational resource.