Data-Efficient Mapping of Copolymerization Curves and Reactivity Ratios via Gradient Flow Polymerization, Inline...

Copolymerization curves and reactivity-ratio estimates are traditionally derived from sparse offline data collected across many separate experiments. Here, we propose a data-efficient workflow for apparent finite-conversion reactivity-ratio estimation that combines gradient continuous-flow free-radical copolymerization, inline ATR-FTIR spectroscopy, and sparse linear regression to reconstruct dense concentration and composition trajectories from minimal offline quantitation. Using styrene/methyl methacrylate as a benchmark system, we performed a time-programmed zigzag sweep of the styrene/MMA feed ratio at constant total flow and acquired more than 700 inline ATR-FTIR spectra in a single ∼4 h run at each of 70, 80, and 90 °C. Only ten offline UHPLC anchor points per run were used to calibrate Lasso models for monomer concentration prediction, enabling high-density reconstruction of the reference monomer composition f ₁(t), total conversion X _tot(t), and cumulative consumption-based copolymer-composition proxy F _1,cum(t). The reconstructed finite-conversion trajectories were analyzed using a Skeist-type finite-conversion terminal-model formulation, in which the Mayo-Lewis equation was used as the instantaneous composition relation. The maximum X _tot values spanned 0.36-0.62, and the full-data IR-assisted estimates spanned r ₁ = 0.33-0.48 and r ₂ = 0.39-0.44, remaining within the broad range of benchmark literature values for this system. Residual-based Monte Carlo sampling showed that pointwise concentration-prediction uncertainty led to compact r ₁-r ₂ ensembles, whereas block-removal and UHPLC-only subset analyses revealed sensitivity to sparse-anchor selection, most notably at 70 °C. Because the workflow uses finite-conversion trajectory data, aligned reference/reacting concentration differences, and this F _1,cum proxy, the reported values are interpreted as apparent finite-conversion estimates under the present flow and analysis workflow. This approach provides a practical route to dense, uncertainty-aware copolymerization-trajectory mapping from minimal offline measurements and enables unified comparison of apparent finite-conversion reactivity-ratio estimates across temperatures.