Bradley Lamb, Sri Vishnuvardhan Reddy Akepati, Saroj Upreti, Xiaodan Gu, Boran Ma. “Chemical similarity-informed machine learning for navigating multi-component polymeric material design space.” ChemRxiv (). https://doi.org/10.26434/chemrxiv.15007670/v1

Values presented only in the GISAXS plots and AFM images in the main paper, pp. 8–10, were not transcribed. Machine-learning results were not retained. The main paper refers to supporting information, but no separate supporting information file was found among the files available from the ChemRxiv record

Block-copolymer thin-film dataset
GISAXS, primary-peak fit, 741 samples Domain spacing 20.99–82.87 nm; FWHM 0.00195–0.0076 Å⁻¹. Samples with domain spacing greater than 40 nm had FWHM values below 0.003 Å⁻¹
Main, pp. 8–9
34 kDa PS-b-P2VP blended with 1.2 kDa PS homopolymer
GISAXS, domain spacing as a function of homopolymer fraction Continuously decreased from 25.6 to 21.0 nm. Measurable order persisted up to a homopolymer fraction of 0.6; beyond 0.6 the film became disordered
Atomic force microscopy Long-range order was maintained at homopolymer fractions below 0.5 and was lost at fractions above 0.5
Main, pp. 9–10
34 kDa PS-b-P2VP blended with 42.5 or 58 kDa PS-b-P2VP
GISAXS, domain spacing as a function of blend fraction Increased from 25 nm to 40 nm for the 42.5 kDa blend and to 42.5 nm for the 58 kDa blend; hexagonal-cylinder morphology was preserved across all blends