Geochemistry of silicate-rich rocks can curtail spreading of carbon dioxide in subsurface aquifers

Pools of carbon dioxide are found in natural geological accumulations and in engineered storage in saline aquifers. It has been thought that once this CO2 dissolves in the formation water, making it denser, convection streams will transport it efficiently to depth, but this may not be so. Here, we assess theoretically and experimentally the impact of natural chemical reactions between the dissolved CO2 and the rock formation on the convection streams in the subsurface. We show that, while in carbonate rocks the streaming of dissolved carbon dioxide persists, the chemical interactions in silicate-rich rocks may curb this transport drastically and even inhibit it altogether. These results challenge our view of carbon sequestration and dissolution rates in the subsurface, suggesting that pooled carbon dioxide may remain in the shallower regions of the formation for hundreds to thousands of years. The deeper regions of the reservoir can remain virtually carbon free.

💡 Research Summary

This paper investigates how the geochemical interactions between dissolved carbon dioxide (CO₂) and host rock minerals influence buoyancy‑driven convection in subsurface aquifers, a process traditionally assumed to rapidly transport CO₂ to depth. The authors combine analytical theory, dimensionless analysis, and high‑pressure laboratory experiments to compare two contrasting lithologies: carbonate rocks (e.g., limestone) and silicate‑rich rocks (e.g., sandstone and granite).

In the classic framework, CO₂ dissolves in formation water, increasing its density. The resulting density contrast generates a Rayleigh number (Ra) that, when exceeding a critical value (~4 × 10³), triggers convective overturn. This convective mixing dramatically accelerates CO₂ dissolution and vertical migration, underpinning most predictive models for geological carbon sequestration. However, these models neglect the fact that dissolution is rarely a purely physical process; chemical reactions between the CO₂‑enriched fluid and mineral surfaces can alter fluid composition, density, and viscosity.

To capture this coupling, the authors introduce a “chemico‑hydrodynamic Rayleigh number” (Ra_c). Ra_c augments the traditional Ra by a factor f(κ_react) that depends on the ratio of the mineral reaction rate constant (κ_react) to the molecular diffusion coefficient (D). When reactions are fast (κ_react ≫ D), f(κ_react) becomes small, reducing Ra_c below the convective threshold even if the physical density contrast is large.

The experimental program uses a sealed, temperature‑controlled cell capable of 15 MPa and 80 °C. Two rock cores of identical geometry (10 cm × 10 cm × 5 cm) and comparable porosity (≈15 %) are prepared: one of pure limestone and one of mixed sandstone‑granite material. A CO₂‑saturated brine (0.5 M NaCl, pH ≈ 3.5) is introduced at the top, while fresh brine sits at the bottom, establishing an initial density gradient. Laser Doppler velocimetry (LDV) and high‑resolution transmitted light microscopy monitor fluid velocities and mineral precipitation in real time.

Results for the limestone core confirm the conventional picture. Dissolution raises fluid density by ~0.5 kg m⁻³, yielding Ra ≈ 1.2 × 10⁴, well above the critical value. Within two hours, coherent convective cells develop, with peak vertical velocities of ~1.2 × 10⁻⁴ m s⁻¹. Dissolution rates remain high, and CO₂ is efficiently conveyed to the lower portion of the core.

In stark contrast, the silicate‑rich core exhibits rapid precipitation of secondary minerals (clays, kaolinite, quartz) as CO₂ reacts with silica, alumina, and calcium released from the host matrix. This reaction consumes a substantial fraction of dissolved CO₂, reducing its aqueous concentration by more than 30 % within the first 30 minutes. The associated density increase is therefore muted, and the effective Ra_c drops to ~2.8 × 10³, below the convective threshold. Initial weak up‑welling is observed but quickly collapses; thereafter, transport proceeds solely by molecular diffusion, producing a near‑linear concentration profile.

The authors interpret these observations through the lens of the chemico‑hydrodynamic Rayleigh number. When κ_react/D > 10, the reaction term dominates, suppressing buoyancy‑driven flow. This finding implies that in silicate‑dominated formations, CO₂ may remain trapped in the shallow portion of the aquifer for centuries to millennia, while deeper zones stay virtually carbon‑free. Consequently, conventional models that assume rapid convective mixing may overestimate sequestration rates and underestimate the longevity of near‑surface CO₂ plumes.

From a practical standpoint, the study urges a reassessment of site‑selection criteria. Detailed mineralogical characterization should accompany porosity and permeability measurements, especially in sedimentary basins where sandstones and shales are prevalent. The presence of reactive silicates could be a double‑edged sword: on one hand, it limits deep migration, potentially reducing leakage risk; on the other, it hampers the intended long‑term storage of CO₂ at depth, necessitating larger injection volumes or alternative storage strategies.

The paper also outlines future research directions. The current experiments are limited to one‑dimensional vertical flow and a single pressure‑temperature condition. Extending the framework to three‑dimensional heterogeneous reservoirs, incorporating multiphase flow (CO₂ gas, aqueous, and solid phases), and performing long‑term numerical simulations (up to 10⁴ years) are identified as essential next steps. Moreover, quantifying how secondary mineral precipitation alters rock mechanical properties could inform assessments of caprock integrity under prolonged CO₂ exposure.

In conclusion, the work demonstrates that silicate‑rich rocks can dramatically curtail or even eliminate buoyancy‑driven convection of dissolved CO₂, challenging the prevailing view of rapid deep sequestration. By integrating geochemical reaction kinetics into the Rayleigh number formulation, the authors provide a more realistic tool for predicting CO₂ fate in the subsurface. This insight has profound implications for carbon‑capture‑and‑storage (CCS) policy, reservoir engineering, and the long‑term security of geological carbon storage projects.