The issue of structure prediction for a given composition is a longstanding problem even for the simplest crystalline solids. The problem is further complicated in the case of an amorphous material, as the potential energy landscape that describes the glassy region exhibits a large number of minima of varying depths. We circumvented the problem by using an evolutionary algorithm to find a stable structure for a realistic, composition provided by experiments and subjecting it to ab-initio simulated annealing to create disorder. 

In this work we present the results of Density Functional Theory calculations. After characterizing the structural and electronic properties of our material, we addressed the issue of ionic conductivity by calculating the defect formation energies of neutral and charged point defects with respect to two different Li reservoirs: a typical anode, metallic lithium, and a typical cathode, lithium cobalt oxide (LCO). 

We found that charged defects dominate over neutral ones, as expected of an ionic conductor. Li⁺ interstitials dominate by far over vacancies, both when the lithium reservoir is metallic lithium and LCO. The formation of charged interstitials when LiPON is brought into contact with LCO has already been observed by our experimental partners. Furthermore, at the interface with metallic lithium the formation of neutral interstitials is a competitive process that results in the chemical reduction of LiPON and the disruption of the network. The occurrence of side reactions at the interface between LiPON and lithium has been observed and quantified by means of X-ray photoemission spectroscopy by our experimental partners. Following up on this result, we proceeded to investigate the structural and electronic properties of the interface between LiPON and metallic lithium with a special focus on the occurrence of side reactions. We conclude by showing preliminary studies of the interface between LiPON and selected LCO terminations suggested by available experimental results.