A Low-Cost Automated Platform for Fast and Accurate pH Control via Physics-Informed Active Learning

Abstract

The precise adjustment of pH in complex buffered systems represents a critical process in chemical synthesis and biopharmaceutical development. However, the intricate multi-buffer equilibria pose significant challenges for conventional methods, leading to modeling difficulties and low optimization efficiency. We have developed a low-cost automated titration platform and established a hybrid data-driven modeling framework that achieves target pH values within remarkably few experimental iterations. Validation across diverse buffer systems, including phosphate, acetate, citrate, and ammonium buffers, demonstrates substantial efficiency improvements over purely data-driven methods, with rapid convergence to target pH achieved in minimal experimental iterations. Beyond its scientific contributions, this work also offers important pedagogical value by offering a low-cost, transparent, and modular platform that allows students and early-stage researchers to engage hands-on with automated experimentation, chemical equilibria, and machine learning.