Comparison of CO₂ detection methods tested in shallow groundwater monitoring wells at a geological sequestration site

The geological storage of anthropogenic carbon dioxide (CO2) is one method of reducing the amount of CO2 released into the atmosphere. Monitoring programs typically determine baseline conditions in surface and near-surface environments before, during, and after CO2 injection to evaluate if impacts related to injection have occurred. Because CO2 concentrations in groundwater fluctuate naturally due to complex geochemical and geomicrobiological interactions, a clear understanding of the baseline behavior of CO2 in groundwater near injection sites is important. Numerous ways of measuring aqueous CO2 in the field and lab are currently used, but most methods have significant shortcomings (e.g., are tedious, lengthy, have interferences, or have significant lag time before a result is determined). In this study, we examined the effectiveness of two novel CO2 detection methods and their ability to rapidly detect CO2 in shallow groundwater monitoring wells associated with the Illinois Basin – Decatur Project geological sequestration site. The CarboQC beverage carbonation meter was used to measure the concentration of CO2 in water by monitoring temperature and pressure changes and calculating the PCO2 from the ideal gas law. Additionally, a non-dispersive infrared (NDIR) CO2 sensor enclosed in a gas-permeable, water-impermeable membrane measured CO2 by determining an equilibrium concentration. Results showed that the CarboQC method provided rapid (< 3 min) and repeatable results under field conditions within a measured concentration range of 15 – 125 mg/L CO2. The NDIR sensor results correlated well (r2 = 0.93) with the CarboQC data, but CO2 equilibration required at least 15 minutes, making the method somewhat less desirable under field conditions. In contrast, NDIR-based sensors have a greater potential for long-term deployment. Both systems are adaptable to in-line groundwater sampling methods. Other specific advantages and disadvantages associated with the two approaches, and anomalies associated with specific samples, are discussed in greater detail.