The Block Copolymer\nPhase Behavior Database

The Block Copolymer Database (BCDB) is a platform that\nallows users\nto search, submit, visualize, benchmark, and download experimental\nphase measurements and their associated characterization information\nfor di- and multiblock copolymers. To the best of our knowledge, there\nis no widely accepted data model for publishing experimental and simulation\ndata on block copolymer self-assembly. This proposed data schema with\ntraceable information can accommodate any number of blocks and at\nthe time of publication contains over 5400 block copolymer total melt\nphase measurements mined from the literature and manually curated\nand simulation data points of the phase diagram generated from self-consistent\nfield theory that can rapidly be augmented. This database can be accessed\nvia the Community Resource for Innovation in Polymer Technology (CRIPT)\nweb application and the Materials Data Facility. The chemical structure\nof the polymer is encoded in BigSMILES, an extension of the Simplified\nMolecular-Input Line-Entry System (SMILES) into the macromolecular\ndomain, and the user can search repeat units and functional groups\nusing the SMARTS search syntax (SMILES Arbitrary Target Specification).\nThe user can also query characterization and phase information using\nStructured Query Language (SQL) and download custom sets of block\ncopolymer data to train machine learning models. Finally, a protocol\nis presented in which GPT-4, an AI-powered large language model, can\nbe used to rapidly screen and identify block copolymer papers from\nthe literature using only the abstract text and determine whether\nthey have BCDB data, allowing the database to grow as the number of\npublished papers on the World Wide Web increases. The F1 score for\nthis model is 0.74. This platform is an important step in making polymer\ndata more accessible to the broader community.