Abstract
Data-driven polymer research has experienced a dramatic upswing in recent years owing to the emergence of artificial intelligence (AI) alongside automated laboratory synthesis. However, the chemical complexity of polymers employed in automated synthesis still lacks in terms of defined functionality to meet the need of next-generation high-performance polymer materials. In this work, the automated self-optimization of the reversible addition-fragmentation chain-transfer (RAFT) polymerization of pentafluorophenyl acrylate (PFPA) is presented, a versatile polymer building-block enabling efficient post-polymerization modifications (PPM). The polymerization system consisted of a computer-operated flow reactor with orthogonal analytics comprising an inline benchtop nuclear magnetic resonance (NMR) spectrometer, and an online size exclusion chromatography (SEC). This setup enabled the automatic determination of optimal polymerization conditions by implementation of a multi-objective Bayesian self-optimization algorithm. The obtained poly(PFPA) is precisely modified by amidation taking advantage of the active pentafluorophenyl (PFP) ester. By controlling the feed ratios of solutions containing different amines, their incorporation ratio into the polymer, and therefore its resulting properties, can be tuned and predicted, which is shown using NMR, differential scanning calorimetry (DSC), and infrared (IR) analysis. The described strategy represents a versatile method to synthesize and modify reactive polymers in continuous flow, expanding the range of functional polymer materials accessible by continuous, high-throughput synthesis.