The Department of Chemistry and Physics continues its Seminar series. Details for this week’s talk are given below.
Cameron Brown, PhD
Senior Applications Research Chemist
Machine learning assisted design and synthesis of industrial polyester resins for bisphenol-A-Non-Intent (BPA-NI) coating applications
In food contact coating applications, the need to feed a growing world population with safer, healthier packaging material choices is the primary driver for growing innovation in BPA-NI technology. Originated from the initial success of 2,2,4,4-tetramethyl-1,3-cyclobutanediol (TMCD) in BPA-free thermoplastics, a new family of TMCD-based polyester resins has emerged as a promising candidate to lead the conversion from BPA to BPA-NI in packaging coatings due to their superior hydrolytic stability, corrosion resistance and high temperature resistance, in combination with good solubility, compatibility and lower viscosity. To deliver first-to-market solutions that can meet or exceed the performance of BPA benchmarks, Eastman must innovate with complex design to meet significant performance and numerous technical challenges while going up against aggressive timelines. From the molecular and structural perspective, polyester properties can be tuned by taking advantage of the enormous design space including complex monomer selections, compositions, specific end-group and molecular weight distributions, and polymer architectures. However, such a complex design space makes developing structure-process-property relationships and deciding where to focus limited laboratory resources challenging. Owing to the continued development of machine learning, we effectively free up laboratory resources for only the most promising candidates through initial digital exploration of this vast chemical design space. Here, we detail our strategies for construction of machine learning models that accelerate the planning and execution of specialty polyester resin syntheses for BPA-NI food contact coating applications using a combination of theoretical inputs, properties from historical datasets, and molecular descriptors.
Keenan E. Dungey, PhD
Department Head, Chemistry & Physics