Analytical Methods and Critical Analyses Supporting Thermodynamically Consistent Characterizations of the Marine CO\u3cem\u3e2\u3c/em\u3e System

Chemical equilibria describing the unique behavior of gaseous and ionic forms of dissolved carbon dioxide (CO2) in seawater comprise what is known as the marine CO2 (or carbonate) system. Observations of the marine CO2 system with high degrees of accuracy, reproducibility, spatial coverage, and temporal resolution are critical for evaluating natural cycles of carbon within the Earth system, as well as chemical and biological responses to anthropogenic CO2 emissions. One component of the CO2 system is the carbonate ion (CO2−3), a dissolved ion that is produced when carbonic acid (H2CO30) dissociates both its hydrogen ions. The carbonate ion is an important buffer against seawater pH changes and is vital for marine organisms that build shells and/or skeletons out of calcium carbonate. Concentrations of carbonate, [CO2−3], are typically inferred from measurements of other CO2 system variables followed by calculations using established thermodynamic relationships. This dissertation advances and evaluates a method for direct measurement of [CO2−3]. The method is based on observations of seawater absorbance in the ultraviolet spectrum after addition of dissolved lead (Pb2+). The absorbance is caused by lead carbonate and lead chloride species, and its magnitude is a function of the carbonate ion concentration. In this dissertation, an instrument-dependent artifact in [CO2−3] measurements is identified and corrected, improving differences between measured and calculated [CO2−3] from –2.78 ± 2.9 µmol kg–1 to –0.03 ± 1.9 µmol kg–1 for the datasets examined in chapter two (where μ ± σ represents the mean, μ, and one standard deviation, σ). An algorithm is introduced to convert [CO2−3] measured at laboratory conditions to [CO32−] at in situ ocean conditions. Aragonite saturation states appropriate to in situ conditions are determined from laboratory-measured [CO2−3] using this algorithm. The resulting saturation states show very good agreement with saturation states calculated from laboratory-measured pH and total dissolved inorganic carbon; differences are centered around zero with a standard deviation of ±0.031. The [CO2−3] measurement method is then extended to an extensive range of salinity (20 to 40) and temperature (3 to 40 °C). With the new, temperature-dependent algorithm, differences between measured and calculated [CO2−3] for the datasets examined in chapter two are 0.02 ± 2.0 μmol kg−1. This result shows almost no degradation compared to the algorithm optimized for 25 °C. Further, the extended algorithm allows for benchtop measurements of [CO2−3] to be performed without temperature control, and opens up the potential for in situ measurements of [CO2−3]. Estimated measurement imprecisions and systematic uncertainties are then used to evaluate combined uncertainties in CO2 system variables that are determined via calculations involving [CO2−3]. An open-source code for CO2 system error propagations was modified for this purpose and made publicly available. Notably, pairing [CO2−3] with total alkalinity (AT, or TA) or total dissolved inorganic carbon (CT, or DIC) can lead to well-constrained characterizations of the CO2 system. One of the most frequently measured CO2 system variables is AT. The typical method for AT measurement is titration with a strong acid of known concentration, along with measurements of initial volume or mass, amount of acid added, and pH (either by electrode or spectrophotometer). The acid amount and pH measurements are made either in a stepwise manner (typical) or after a single acid addition (less common). These measurements are followed by some form of curve fitting or the application of a single equation to determine AT. Within this general framework, different methods of AT measurement are used in the marine chemistry community, and different lab groups introduce minor tweaks to a general method to arrive at their own specific protocols. Importantly, any and all proton-binding species that are active over the pH range of an AT titration will contribute to measured AT. Numerical simulations detailed in this dissertation show that proton-binding organics introduce differences between AT values determined by different commonly employed titration methods. These differences can exceed 50% of the total organic concentration. Further, proton-binding organics can cause incorrect deductions of carbonate alkalinity from AT, which propagate to calculations of other CO2 system variables. Each of these effects is modulated by the proton-binding affinity (or dissociation constant) of the dissolved organic matter, the method used to determine AT, and the carbonate chemistry of the titrated sample. The demonstration of differences between AT determined by different titration methods (given the likely omnipresence of organic proton acceptors) has implications for CO2 system thermodynamic consistency analyses. Experimental determinations of CO2 system variables (especially AT and CT) are aided by certified reference materials (CRMs), which are used to verify numerical accuracy, temporal reproducibility, and inter-laboratory consistency of measurements. CRMs are prepared from natural seawater in large, uniform batches, and are certified for AT and CT. CRMs are irradiated with ultraviolet light to reduce organic contamination and are poisoned with mercuric chloride to suppress biological activity. However, CRMs are not ensured to be free of dissolved organics. This dissertation details the results of experiments that indicate excess alkalinity in CRMs, likely due to the presence of proton-binding organic molecular structures. The experiments were initially designed to investigate the total boron to salinity ratio in seawater; however, inconsistent results between CRM batches and natural seawater from the Gulf of Mexico indicated that isolation of the borate alkalinity component from the unexpected excess alkalinity component would be impossible. Instead, the excess alkalinity component of CRM batches is described and evaluated in the context of experimental uncertainties. Then, possible effects of the excess alkalinity on CRM-based evaluations of AT measurement consistency and implications for seawater acid–base chemistry are discussed. The amount of excess alkalinity detected in CRMs has the potential to bias evaluations of AT measurement consistency by up to about 5 μmol kg−1, depending on a few factors including AT measurement method and the nature of the excess alkalinity contributor. The existence of excess alkalinity in CRMs implies that, in certain ocean regions, total alkalinity is not an exclusive function of inorganic chemical species, which has implications for evaluations of thermodynamic consistency between measured and calculated CO2 system variables. Determinations of excess alkalinity are highly influenced by the investigator’s choice of total boron to salinity ratio, the two most commonly used values of which differ by about 4%. This dissertation significantly advances an analytical method for direct determinations of a fifth measurable CO2 system variable. The method is uniquely suited for in situ application, which is critically important as the marine biogeochemical community looks more toward autonomous in situ sensors for ocean monitoring capabilities. This dissertation also offers a critical analysis of the effects of dissolved proton-binding organic molecules on total alkalinity measurements. “Organic alkalinity” resulting from these proton-binding molecules is one of the most likely explanations for the confounding nature of inconsistencies between CO2 system measurements and calculations, and the work detailed here provides insight that will be helpful in solving that important issue. Finally, this dissertation describes a novel detection of excess alkalinity in CO2 system reference materials. The implications for quality-control efforts using those reference materials and for the acid–base chemistry of natural seawater are discussed