Block copolymer self-assembly is an alternative approach to photolithography to produce nanopatterns with features and periodicities below 10 nm in size. Termed directed self-assembly, or DSA, the polymers contain sufficient chemical information to undergo nanoscale phase segregation and form a nanopatterned template on a surface, which can be used as a template for further processing. The majority of DSA-based nanopatterning relies on solvent vapor annealing due to its fast processing times compared to thermal annealing, and to not contribute to the thermal budget. Finding a suitable and reproducible combination of materials and processing parameters compatible with the conventional inline manufacturing process on semiconductors is complicated by challenges of small variations in film thickness, and other variables related to the solvent vapor pressure and temperature. Much of the research on DSA has been carried out in static, non-standardized annealing systems that lack in-situ monitoring of film thickness and dewetting. In this work, a computer-interfaced solvent vapor flow annealing system is used to precisely control the process parameters and conditions of DSA using the model system of the block copolymer, PS-<i>b</i>-PDMS (43k-<i>b</i>-8.5k). Small libraries of films of incrementally differing thicknesses and annealing parameters were obtained and machine learning approaches were then applied to analyze the resulting SEM micrographs of the final silica dot patterns. The machine learning approach led to much faster optimization of reproducible parameters and conditions that led to the desired final morphology of the DSA patterns.

Optimization of block copolymer self-assembly using machine learning approaches

ACS Spring Meeting 2021

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.