CO2 Sequestration in Feldspar-Rich Sandstone: The Importance of Saturation State and Fluid Composition

Abstract:

To investigate CO₂ Capture, Utilization, and Storage (CCUS) in sandstones, we performed 150°C flow-through experiments on K-feldspar rich cores from the Eau Claire Formation. We observed mass transfer processes operating on a range of scales by measuring sample porosity, permeability, and surface area and fluid and solid chemistry. The close agreement between our measured K-feldspar dissolution rates and literature values demonstrate that feldspar-fluid interaction can be confidently modeled using our chosen rate parameters and characterization methods, but other hydrogeochemical processes evolve somewhat less predictably. Specifically, a sandstone core through which CO₂-rich deionized water was recycled for 52 days decreased in porosity and permeability and increased in surface area while an Al hydroxide mineral, such as boehmite, precipitated within its pore space. However, two samples subjected to ~3 day single-pass experiments run with CO₂-rich NaCl brines generally increased in porosity and surface area and decreased in permeability as K-feldspar was converted to a phase with kaolinite-like stoichiometry. Notably, boehmite remained two orders of magnitude less supersaturated than kaolinite in all samples from all experiments, and we therefore hypothesize that the identity of the secondary phase is related to the presence of NaCl in solution. Regardless of their identity, long-lived secondary phase supersaturations and low measured Al molalities suggest that Al is precipitated at approximately the same rate at which it is released to solution through feldspar hydrolysis. Finally, we note that, although core permeability measurably decreased in all three experiments, the magnitude of these changes is unlikely to impact CO₂ injectivity over a CCUS reservoir's lifetime. The observations produced here are critical to the application of reactive transport models to CCUS systems, yet more work will be required in order to improve model predictions.