ALKALINITY TITRATION IN SEAWATER - HOW ACCURATELY CAN THE DATA BE FITTED BY AN EQUILIBRIUM-MODEL

Displaying 'calculated minus observed' data for precise titrations of seawater with strong acid permits direct evaluation of important parameters and detection of systematic errors. At least two data sets from the GEOSECS (Geochemical Ocean Sections) program fit an equilibrium model (which includes carbonate, borate, sulfate, silicate, fluoride, and phosphate) within the most stringent experimental error, less than 2-mu-mol kg-1. The effect of various parameters on the fit of calculated to observed values depends strongly on pH. Although standard potential E0, total alkalinity A(t), total carbonate C(t), and first acidity constant of carbon dioxide pK1 are nearly independent, and can be determined for each data set, other parameters are strongly correlated. Within such groups, alt but one parameter must be determined from data other than the titration curve. Adding an acid-base pair to the theoretical model (e.g. C(x)=20-mu-mol kg-1, pK(x)=6.2) produces a deviation approaching 20-mu-mol kg-1 at constant C(t); however, adjustment of C(t) by about - 18-mu-mol kg-1 to produce a good fit leaves only +/- 1.5-mu-mol kg-1 residual deviation from the reference values. Thus, at current standards of precision, an unidentified weak acid cannot be distinguished from carbonate purely on the basis of the titration curve shape. There are few full sets of numerical data published, and most show larger systematic errors (3-12-mu-mol l-1) than the above; one well-defined source is experiments performed in unsealed vessels. Total carbonate can be explicitly obtained as a function of pH by a rearrangement of the titration curve equation; this can reveal a systematic decrease in C(t) in the pH range 5-6, as a result of CO2 gas loss from the titration vessel. Attempts to compensate for this by adjustment of A(t), C(t), or pK1 produce deviations which mimic those produced by an additional acid-base pair. Changing from the free H+ scale (for which [HSO4-] and [HF] are explicit terms in the alkalinity) to the seawater scale (SWS) (where those terms are part of a constant factor multiplying [H+]) requires modification of the titration curve equation as well as adjustment of acidity constants. Even with this change, however, omission of pH-dependent terms in [HSO4-] and [HF] produces small systematic errors at low pH. Shifts in liquid junction potential also introduce small systematic errors, but are significant only at pH < 3. High-pH errors due to response of the glass electrode to Na+ as well as H+ can be adequately compensated to pH 9.5 by a linear selectivity expression.