RANDOM STRUCTURE SEARCHING RESULTS ---------------------------------- The folders contain the search results. The following python dictionary is used to map the structure from its job_id to the names mentioned in the paper: { 2926: "A", 1718: "B", 908: "C", 4351: "D", 8402: "E", 2987: "F", 2402: "G", 6314: "H", 7889: "I", 395 : 'S1', 8378: 'S2', 2719: 'S3', 6482: 'S4', 1451: 'S5', 7881: 'S6', } These keys are ids of the run, and they refer to that of in run11. The other runs included are performed with a range of randomization zone size and number of bulk layers. The files are in the SHELX format compatible to the `cryan` analysis program of the Ab Inito Random Structure Searching package (https://www.mtg.msm.cam.ac.uk/Codes/AIRSS). Python code used to wrap the `buildcell` program in AIRSS can be found at https://pypi.org/project/airsspy/. The file name for each structure is in the form of: STO_CEO_75_Ti_1-180319-134616-9.res Where the first part 'STO_CEO_75_Ti_1-180319-134616' is the seed name (for generation) and '9' is the job_id which is unique for each run. The name 'Ti_1' means the interface is terminated with a TiO2 plane and includes one extra layer in the randomization zone. The seed used for the search is stored in the cell file format in each folder.