Blank variability in coulometric measurements of dissolved inorganic carbon

Marine dissolved inorganic carbon (DIC) is the largest pool of carbon in the Earth surface system, so measurements of DIC are essential for understanding the changing global carbon cycle. The most accurate widely-used method for measuring DIC is coulometric titration. DIC is extracted from a water sample as CO₂ and delivered to a coulometric cell, where it reacts, and an electrical current applied across the cell reverses the reaction. There is also a quasi-continuous background current (termed the ‘blank’). The total integrated current (termed ‘counts’) minus the blank is directly proportional to the amount of CO₂ in the sample. Here, we show how determining the blank on a per-sample basis can reveal changes through time that would not be noticed with the SOP-recommended approach of a single blank determination at the start of each analysis session. If not accounted for, these changes in the blank lead to an apparent drift in the DIC results and reduced accuracy. We show how the per-sample blanks can be best computed and the results applied across an analysis session, an approach which we have implemented in an open source Python package (Koolstof). We quantify the improvement in the reproducibility of DIC measurements when using this approach by application to several different measurement datasets from our laboratory.