Quantifying Salt Precipitation During CO2 Injection: How Flow Rate, Temperature, and Phase State Control Near-Wellbore Crystallization

Abstract

Salt precipitation near injection wells can reduce permeability, induce excess pressure buildup, and reduce injectivity within days to weeks of CO2 injection, yet the pore-scale mechanisms coupling multiphase flow, evaporation, and crystallization warrant further detailed quantification across variable phase states and flow regimes. We present high-resolution microfluidic experiments that systematically quantify the dynamics of halite crystallization during CO2-driven brine evaporation across liquid, gaseous, and supercritical phases (50-80 bar, 20--60 C, Pe = 50--1440). Crystallization kinetics are controlled by transport, with the Avrami rate constant (K) increasing by two orders of magnitude with the Peclet number and exhibiting the dependence of the temperature of Arrhenius (Ea = 58.6 kJ/mol. Supercritical CO2 achieves superior displacement efficiency (residual saturation 0.22-0.36, fractal dimension D = 1.79-1.82) and the fastest evaporation (Sherwood numbers 2-3x higher than the liquid phase), reducing the nucleation time from 57 min (20C, liquid) to <1 min (40-60 C, gas/supercritical). The final fractions of crystal increase 10-fold from liquid (0.008) to gas-phase conditions (0.08--0.12), confirming that convective transport and phase state dominate over diffusion-limited mechanisms. Despite probabilistic nucleation, final crystal distributions are spatially rather uniform with no systematic inlet-outlet bias. These quantitative relationships between dimensionless parameters (Pe, Sh), kinetic constants (K, Ea) and phase-dependent displacement patterns provide critical benchmarks for validating pore-scale models and predicting near-wellbore permeability impairment in geological storage of saline and hypersaline CO2.

Figures (7)

• Figure 1: Experimental configuration and chip characterization. (a) Schematic diagram of the high-pressure high-temperature microfluidic system showing CO$_2$ delivery via syringe pump, temperature-controlled enclosure, microscopic imaging setup, and backpressure regulation. (b) Microscopic image of the microfluidic chip with inlet channels and porous network. The dashed rectangle indicates the field of view used for data acquisition. The inset shows a magnified view of the pore-grain structure. (c) Pore size distribution (left plot) and spatial porosity variation along the chip length (right plot).
• Figure 2: Image acquisition and processing workflow for a representative experiment (100 mL/min CO$_2$ flow, 80 bar, 40$^\circ$C). Upper panel: Raw microscopic images showing (left to right) initial brine saturation, post-breakthrough residual brine pools, first crystal formation, and complete dryout. The arrows identify glass grains, brine, crystals, and CO$_2$-filled pores. Lower panel: Processed images obtained by subtracting the saturated reference frame from each time-series image, enhancing contrast between phases. The inset shows a magnified view of the pore-scale crystal distribution across two time stages. Note the minimal contrast between brine and glass in raw images, necessitating image subtraction for accurate segmentation.
• Figure 3: CO$_2$ breakthrough dynamics and displacement efficiency analysis. (a) Time-lapse contrast images showing CO$_2$ invasion at 100 mL/min, 80 bar, 40$^\circ$C. It shows the tree-like branching structure and progressive brine isolation. (b) Binary mask of CO$_2$-occupied pore space extracted from panel (a). (c) Post-breakthrough brine saturation versus Reynolds number for gaseous (green), liquid (blue), and supercritical (red) CO$_2$ phase states. Supercritical conditions yield the lowest mean saturation and the smallest variability. (d) Representative box-counting analysis for fractal dimension determination. The slope of the log--log plot yields $D$ = 1.78 for this example. (e) Fractal dimension of CO$_2$ invasion patterns versus Reynolds number. Higher $D$ correlates with more connected flow networks and lower residual saturation, with supercritical CO$_2$ approaching the dry-chip limit ($D$ = 1.85).
• Figure 4: Comparative evaporation and halite precipitation dynamics for gaseous (50 bar, 20$^\circ$C), liquid (65 bar, 20$^\circ$C), and supercritical (80 bar, 40$^\circ$C) CO$_2$ at 1000 mL/min. First column: First nucleation events (white arrows mark crystals). Note earlier nucleation for liquid/supercritical phases (1 min) versus gas (20 min). Second column: Final crystal distributions after complete dryout. Third column: Spatiotemporal evaporation maps (blue: early; red: late). The supercritical and liquid phases show predominantly early evaporation, whereas the gas phase exhibits more sustained evaporation. Fourth column: Crystal growth maps (red: early; blue: late). Insets show single-pore crystal evolution. (a) Equivalent diameter distributions for final crystals. Gaseous CO$_2$ produces larger, more heterogeneous crystals. (b,c) Normalized center-of-mass trajectories in flow-parallel and flow-perpendicular directions. Center of mass convergence to 0.5 indicates spatially uniform final distributions despite stochastic nucleation.
• Figure 5: Temporal evolution of brine saturation and crystallization for gaseous, liquid, and supercritical CO$_2$ (conditions as in Figure \ref{['fig:TimeEvolutionThreePhases']}). Upper panel: Brine saturation $S_w(t)$ (blue) and exponential decay model fits (Equation \ref{['eq:evaporationModel']}, red). Gaseous CO$_2$ exhibits more sustained evaporation, while liquid and supercritical phases show rapid early-stage drying. Lower panel: Crystal fraction $X_c(t)$ (blue) with Avrami model fits ($n$ = 3, Equation \ref{['eq:avrami']}, red) and detected crystal count (orange, right axis). Early-stage crystallization dominates for all phases. Non-monotonic crystal count reflects coalescence and detection limitations.