Mineral surface area accessibility and sensitivity constraints on carbon mineralization in basaltic aquifers

Estimating mineral reactive surface areas in geologic media remains one of the key challenges limiting the accuracy of reactive transport modeling (RTM) predictions of subsurface processes, particularly those controlling the fate of carbon dioxide (CO2) during geologic storage. Although there have been numerous attempts to combine imaging and experimental techniques to estimate mineral reactive surface area for use in RTM predictions of geologic CO2 storage, these techniques have yet to be adapted to basaltic reservoirs, which have pore structure, mineralogy, and chemical composition that is unique compared to their more often-studied sedimentary counterparts. Here, we address this issue by quantifying fluid-accessible mineral surface areas through image analysis of scanning electron microscope (SEM) backscatter electron images (high-resolution 500 nm/pixel) and Raman spectroscopic mapping of a basaltic rock sample from the Eastern Snake River Plain, Idaho. To evaluate whether the determined pore fluid-accessible mineral surface area accurately reflects reactive surface area, a micro-continuum scale RTM was developed and compared with a high-temperature, high-pressure flow-through CO2 mineralization experiment conducted on the characterized basalt. Importantly, simulations employing the image-derived pore fluid-accessible mineral surface areas match the experimental effluent chemistry well within uncertainties. These mineral surface areas were then used to parametrize a field-scale model representative of the Cascadia basin, Northeastern Pacific, to evaluate impacts of surface area variations on mineral carbonation. Simulations were carried out using variations in image-derived surface areas that cover one to two orders of magnitude increase and decrease in surface area, analogous to previously reported magnitudes of difference between total and reactive surface areas. Carbonation efficiency in terms of CO2 volume mineralized over the simulated period was tracked and compared. Simulations with surface area increased and decreased by two orders of magnitude show basalt carbonation efficiency that is three times faster and six times slower, respectively, than predictions with image-derived mineral surface area. These sensitivity analyses demonstrate that accurate quantification of mineral surface area is crucial for efforts to predict CO2 mineralization, and that efforts such as those employed here can dramatically reduce the uncertainty of field-scale predictions of basalt carbonation.