Digitalization of microbial electrosynthesis: Understanding operational variables for ethanol production optimization

Microbial electrosynthesis (MES) is a bioelectrochemical technology devoted to carbon dioxide (CO 2 ) conversion into value-added organic products. Traditionally, it has been addressed through trial-and-error approaches, with limited experimental data. This study introduces a data-driven framework to advance the MES digitalisation through systematic analysis of operational variables influencing CO 2 conversion. CO 2 conversion to acetic acid and ethanol was investigated in MES cells integrated within an automated bench-scale platform. It enabled real-time control of key operational variables, i.e. dissolved CO 2 , pH, and pressure, while monitoring electrical conductivity, cell voltage and temperature. The ethanol production rate was improved by 35% (14.80 g m −2 d −1 ), obtained controlling dissolved CO 2 at 100–300 mg L −1 , pH at 4.5–4.8 and pressure at 1.8 bar. Process data were analysed through pattern discovery (based on both operation and production variables) and unsupervised clustering (based solely on operational variables). The latter allowed to identify distinct operational phases, which matched the real production stages, solely based on operation variables. Principal component analysis revealed that acetic acid production was primarily dependent on dissolved CO 2 , with secondary contributions from pH and electrical conductivity. Electrical conductivity emerged as an identifying signal for ethanol production. The proposed data-based methodology elucidates the key variable interactions and supports phase-specific monitoring and control strategies for MES optimization in the absence of time-consuming, punctual product measurements. • Real-time operation control improved ethanol production rate by 35% in MES cells. • CO 2 limitation and pH acidification through fine control triggered solventogenesis. • SCFCM from operational data enabled process partition in product-selective phases. • EC signal dynamics are linked to ethanol selectivity in MES.