Point defects in semiconductors and insulators play a pivotal role in electronic, photovoltaic and optical devices. Recently, point defects in solids have emerged as main actors in quantum technology hardware elements, where the tight control of their location is of utmost importance [1]. With today's fabrication techniques, point defects can be deterministically placed into the host material, even those that do not form in nature. The implantation and irradiation techniques revolutionized materials science and industrial sectors relying on it. However, irradiation techniques often introduce mobile defects beside the target quantum hardware defects that can combine with each other and form stable complex configurations. Understanding the atomistic processes behind the formation and the stability of the target defect and parasitic complexes paves the way to realize reliable devices. Furthermore, the urgent quest to find novel solutions, e.g., seeking point defects for given functionalities, for the emerging technologies calls for atomistic simulation techniques that have high predictive power and can guide the research and development directions [2,3]. Recent developments in search algorithms and the increasing computation power have directed the computational discovery of target point defects to the application of machine learning techniques. The vast majority of point defects are complexes that may exist in various configurations, and current machine learning algorithms are conditioned to find the energetically most stable defects configuration for the given applications. It is a common assumption that the most stable defect complexes will form, i.e., the thermal equilibrium of the defect complex configurations can be eventually reached.

 

Ádám Gali’s group has demonstrated by means of accurate atomistic simulations that the kinetics of the formation of defect complex configurations can prohibit to reach the thermal equilibrium [4]. In particular, this effect has been exemplified on a key defect complex in silicon which can be a basic unit for future quantum communication and quantum computation devices. The results imply that the machine learning techniques conditioned to search for thermal equilibrium defect complexes would overlook a key candidate for quantum technology applications. This study may turn the attention to the importance of kinetics in the simulation of implanted or irradiated semiconductors for defect engineering.