Two silica suspensions were tested. One was prepared from dry FS powder. This powder was purchased from Aldrich Chemical (Milwaukee, WI, USA). It has a mean particle size of 0.2–0.3 µm (aggregate) and a Brunauer–Emmett–Teller (BET) surface area of 220 m²/g. This powder was dispersed into ultrapure water (Millipore Milli-Q system, resistivity = 18.2 MΩ cm) to prepare the suspension. The other CS suspension, GEO-40 Sol-1030Na, was obtained from GEO40.com (Auckland, New Zealand) in suspension form. This CS was recovered from geothermal fluids. GEO-40 was selected for water-flow-blocking column tests and sandbox tests because the CS recovered from geothermal fluids has a comparatively low carbon footprint. This suspension contains 30.6 wt% silica particles. Based on the technical data provided by the vendor, the measured particle size is 7.0–10.0 nm with average of 8.0 nm. The surface area ranged from 272 to 388 m²/g with the areage at 341 m²/g. The pH of the suspension is 9.8.

Sodium chloride (NaCl) was used as the destabilizer for the silica suspensions to initiate the gelation process. Amaranth dye was applied to color the silica suspension for the purpose of visually tracking the distribution of the suspension in a porous medium after injection. Potassium bromide (KBr) was selected as the chemical tracer in the sandbox tests.

Accusand (Covia Corp, Ottawa, MN) with 99.8% SiO₂ was the porous medium used in the batch, column, and sandbox experiments. Hanford Site SGW (Emerson et al., 2017) was prepared to saturate the porous media in the batch, column, and sandbox tests.

The viscosities of both FS and Geo-40 silica suspensions increased when the silica concentrations were increased. The FS suspension has a much higher viscosity compared to that of the Geo-40 suspension at comparable concentrations. As illustrated in Figure 3, the viscosities of FS at a concentration of 8 wt% were 506.0, 264.0, and 78.8 mPa·s at shear rates of 10, 22, and 100 s⁻¹, respectively (Figure 3A). In contrast, the viscosity of the Geo-40 suspension at 10 wt% was 2.2 mPa·s at all tested shear rates (Figure 3B). Because the viscosity increases with the silica concentration, it is certain that if the Geo-40 concentration was 8 wt%, the viscosity would be lower than 2.2 mPa·s, which is only about 4.3% of the FS suspension viscosity at a shear rate of 100 s⁻¹, implying that the Geo-40 suspension favors subsurface injection.

The FS suspension showed shear thinning even at a 1 wt% concentration, while the Geo-40 nanosized silica suspension did not exhibit shear thinning. Further discussion of shear thinning is presented in the next section.

The FS suspension shows shear thinning at the tested concentrations (one to eight wt%), while the Geo-40 nanosized silica suspensions did not show a shear-dependent viscosity, even at 30 wt% (Figure 4). Similar shear thinning behavior was reported by Amiri et al. (2009), Yang et al. (2016, 2017), and Zhong et al. (2018) for FS suspensions when the concentration was at and higher than 0.7 wt%. This shear thinning characteristic reveals that the aggregates formed by particle interactions are broken when the shear rate increases (Amiri et al., 2009). For the nanosized silica suspension, such particle–particle interaction does not exist, as indicated by its Newtonian fluid viscosity behavior. It was reported that the nanosized silica particle suspensions showed Newtonian viscosity behavior for particle concentrations up to 41 wt% with shear rates ranging from 1 to 200 s⁻¹ (Metin et al., 2011).

Even though the Geo-40 CS suspensions do not show shear thinning, the viscosity is lower compared to the FS CS at the same concentrations at all tested shear rates (Figure 4). Therefore, Geo-40 has better injectability and a higher penetration potential. Further, because Geo-40 CS is recovered from geothermal fluids, it has a consequently lower carbon footprint than the FS CS. Thus, Geo-40 CS was selected for the gelation and water-flow-blocking study in the column and sandbox tests of this study.

The viscosity of the Geo-40 sol with 1,500 mg/L Na⁺ increased with time (Figure 5). As the gelation process proceeded, the viscosity of the suspension increased. When the viscosity reached a certain value, the suspension showed shear thinning behavior, revealing that the SiO₂ particle clusters must reach a certain dimension during the gelation process for the shear thinning behavior to occur.

The shear thinning non-Newtonian flow curves can be fitted with a power-law equation (Zhong et al., 2013): ɳ ṙ = K ṙ − n (1) where ɳ ṙ is the viscosity, ṙ is the shear rate, K is the consistency index (i.e., the solution viscosity at 1.0 s⁻¹ shear rate), and n is the flow behavior index. When n is larger, the shear thinning nature of the solution is more pronounced. This fitting was applied to the flow curves for samples at 19h and longer in the gelation process (Figure 5). It was revealved that when the gelation time was longer, the shear thinning behavior was more profound.

Cross-linking among the silica particles, e.g., the formation of siloxane bonds, gives rise to the viscosity of the suspensions (Iler, 1979; Brinker, 1994; Yossapol, 2002; Sögaard, et al., 2018; Pagano et al., 2022). The cross-linking formed particle clusters that grow with time (Iler, 1979). As the clusters become larger, they do not move randomly anymore in the suspension but will form agglomerates, thus increasing the viscosity of the suspension. This agglomerate microstructure will break when the shear rate is increased in the flow, and the viscosity decreases as the particles are free to flow, resulting in shear thinning.

Adding more Na⁺ ions to the CS system shortens the gelation time. When the CS concentration was high (e.g., 30.0%), the gelation time could be as short as a few minutes when the added Na⁺ concentration was 10,000 mg/L (Figure 6). As the Na⁺ concentration was reduced to 1,000 mg/L, the gelation time increased to 40 h. The time would be even longer when the Na⁺ concentration was further reduced (Figure 6).

Higher Na ion concentration induced shorter gelation time for nano silica particle suspensions was also reported by Ghaffari et al. (2022). For the water-borne CS suspension, there is a “double layer” of water and electrolytes, which suspend the particles, with a thickness of about 2 nm surrounding the silica particles (Roberts and Bergna, 2006). This “layering structure” around the particles is also referred as electrical double layer (Amiri et al., 2009). Increasing the Na⁺ concentration in the silica suspensions diminishes the thickness of the double layer, thus allowing the silica particles to approach more closely to each other. At a closer distance, the van der Waals attractions force exceeds the electrostatic repulsion between particles which induces connections among the particles. Formation of siloxane bonds (Si–O–Si) between silica particles occur in the system and eventually resulting in gelation (Iler, 1979; Brinker, 1994; Yossapol, 2002; Sögaard, et al., 2018; Pagano et al., 2022). Controlling the gelation time of the silica suspensions is an important step in the subsurface applications of silica suspension gelation to provide sufficient time for the handling and injection of the silica suspension before gelation.

Oscillatory measurements reveal the microstructures of the CS suspensions. Figure 7 presents the loss and storage moduli, G″ and G’, of the suspensions over time. The measurements were started once the suspension was prepared and continued until the gelation time was passed. Figures 7A–C compare the influence of the Na⁺ concentration, and Figures 7C–F show the impact of the silica particle concentration.

For suspensions with 30% CS (Figures 7A–C), G″ was greater than G′ in the beginning stage, indicating that the suspensions were viscoelastic liquids. In this stage, the fluid behaved more liquid-like. As time increased, both G″ and G′ increased and eventually reached a point at which G′ was greater than G″, indicating that the material was more solid-like. When the Na⁺ concentration was higher, the rate of increase in G′ was faster, and the time needed for the G′ curve to cross the G″ curve was shorter.

When the CS concentration was reduced from 30% to 10% (Figures 7C–F), the rate of increase in G′ gradually reduced when the Na⁺ concentration remained constant at 10,000 mg/L. For the suspension with 10% CS and 10,000 mg/L Na⁺, G′ and G″ did not increase within the tested time (Figure 7F).

The changes in G′ and G″ over time indicate a phase transformation inside a viscoelastic material. When G’ = G″, the viscoelastic material is at a phase transition point, often called the sol–gel transition point. This is the gel point of the material. The observed gel point times were 80, 34, and 14 min for the suspensions with 30 wt% Geo-40 and 6,000, 8,000, and 10,000 mg/L Na⁺, respectively. In addition, the gel point times were >200, 105, 50, and 14 min for suspensions with 10,000 mg/L Na⁺ at 10%, 20%, 25%, and 30% silica concentrations, respectively. It is clearly demonstrated that the increases in both the Na⁺ concentration and silica concentration shortened the gelation time.

The complex viscosities of the 20, 25, and 30 wt% suspensions quickly increased after gelation started, while the 10 wt% suspension did not show gelation (Figure 8A). The viscosity increased with the increase in the SiO₂ concentration. A higher silica concentration led to a stronger gel. The viscosity of the gel continuously increased with time, up to 300 min when the tests were terminated, revealing that the silica network in the gel was becoming stronger over time.

The Na⁺ concentration not only influenced the gelation time but also changed the final strength of the gel, as indicated by the complex viscosity at test completion (300 min) (Figure 8B). The gel strength was also still increasing at 300 min. The rate of increase in the viscosity was higher at the beginning of gelation when the silica concentration was higher and when the Na⁺ concentration was higher. After the initial gelation stage, the rate of increase in the viscosity was lower when the silica concentration was higher and when the Na⁺ concentration was higher, as indicated by the slopes of the curves.

Amiri et al. (2009, 2011) reported that increases in the salinity and silica particle concentration in the suspension resulted in a silica network with greater strength, producing a stronger gel. The changes in the salinity and silica concentration can be used to achieve a desirable gel strength for specific applications.

The changes in the storage and loss moduli (G′, G″) of the silica suspensions over time revealed dynamic gelation processes. The results clearly showed that the gelation process started earlier with a higher Na⁺ concentration (Figures S–1). When the silica suspension and Na⁺ mixtures were freshly prepared, all mixtures behaved more like liquids, as indicated by the larger G″ compared to G’ (Figures S–1A). After aging for 65 and 100 min (Figures S–1B,C), the mixture with 6,000 mg/L Na⁺ still behaved more like a liquid, while suspensions with 8,000 and 10,000 mg/L Na⁺ behaved more like solids with G’ > G”. At 165 min (Figures S–1D), all mixtures showed a higher G′ compared to G”, behaving more like a solid. With further aging to 240 min (Figures S–1E), G′ for the suspensions with 6,000 and 8,000 mg/L Na⁺ further increased, while G′ for the 10,000 mg/L Na⁺ suspension did not show a further increase. This observation is consistent with the change in the rate of increase in G’ in Figure 6.

The complex viscosities of the silica suspensions and Na⁺ mixtures clearly increased with time, especially at low frequencies, indicating a gelling process (Figures S–2). At aging times of 65 and 100 min, the complex viscosity was higher for the gel with a higher Na⁺ concentration (Figures S–2B,C); thus, the gel is more rigid. However, at an aging time of 165 min, there were small differences in the complex viscosities for all three gels (Figures S–2D), indicating that they had similar stiffnesses.

When the silica suspension was injected into the pore spaces in packed sand, it also gelled and grouted the sand pack, as indicated by the sand pack remaining at the bottom end of the vial when the vial was turned upside down (Figure 9). When a silica suspension was injected into a dry sand pack, the whole sand pack was grouted (Figure 9, Vial A). When the silica suspension was injected into the bottom section of an SGW-saturated sand pack, only the bottom section was grouted, where the silica suspension had displaced the SGW and the upper-section sand pack was still loose and separated from the grouted sand pack (Figure 9, Vial B).

With the same amount of CS suspension injected into a set of sand packs saturated with SGW and different gelling times applied to each sand pack, the influence of the gelling time on the grouted sand pack volume was revealed (Figure 10). With a longer time allowed for gelation, a larger volume of the sand pack was grouted. In these gelation tests, a silica suspension was injected into the bottom of the sand pack to displace SGW. At the interface between the silica suspension and the SGW, the silica concentration and Na⁺ concentration decreased. These decreases would increase the time needed for the gelation of the silica suspension at the interface. Therefore, a longer gelation time resulted in a larger grouted sand pack volume (Figure 10). Qualitatively, by touching the grouted sand packs, it is obvious that these sand packs became more rigid over time.

When a silica suspension is injected into a sand pack for gelation, the gelation time and gel strength may be affected by the sand since the sand grains have similar functional group as the silica particles in the suspension. The gelation times of a silica suspensions in systems with and without sand were compared and no significant differences were observed, revealing that the presence of sand did not significantly affect the gelation time. The strength of the silica gel form in conditions with and without sand was not compared as it was not in the scope of this study.

The pressure drops over the sand-packed and SGW-saturated column during water injection and flow through the column are presented in Figure 11. Before the injection of a silica suspension, the pressure drop during water injection at a flow rate of 1.5 mL/min was stabilized at 0.1 psig (0.69 kPa) (Figure 11). After injection of the CS suspension and gelation in the column, the water injection pressure increased quickly once water was injected into the column inflow end at a rate of 1.5 mL/min. No effluent flow was measured. It was certain that the silica gel completely blocked the water flow through the column.

Water pumping was stopped when the pressure reached 20, 25, 35, 45, and 50 psig (137.88, 172.55, 241.29, 310.23, 344.7 kPa) for the injections performed at times of 1, 2, 3, 4, and 15 days after gelation, respectively (Figure 11). The final injection pressure was held constant for at least 15 min for each of these injections, and no column effluent was observed.

After day 15, 45 psig (310.23 kPa) injection pressure was maintained at the column inflow end for 120 days. At this pressure, no water breakthrough was observed at the column effluent, revealing that the gel in the sand pack could sustain the tested pressure within the testing duration.

During all the injecitons for these pressure tests, no effluent was observed, indicating that the silica gel was able to sustain a injection pressure up to 45 psig (310.23 kPa) and the gel is not permeable to water, resulted in a complete water flow blocking.

Prior to the gelation of the pore space with silica gel, the chemical tracer KBr injected into the sandbox annulus above the “Gel zone” (Figure 2) was transported through the zone to the annulus section below the gel zone, as indicated by the increasing Br⁻ concentration over time (Figure 12A). However, transport of the tracer through this zone was completely blocked when silica gel formed and filled up the pore spaces of the gel zone in the annulus, as no Br⁻ was detected below the treated zone (Figure 12B).

The pressure drop through the sandbox annulus during water injection at 5.0 mL/min was 0.1 psig (0.69 kPa) prior to the silica gel treatment. With an 11 cm thick section of silica gel formed in the annulus, the injection pressure increased to 7.7 psig (53.08 kPa) at the same flow rate. When water pumping stopped, the pressure decreased slowly, indicating that water was slowly leaking through the gel zone. Based on the results from a separate column pressure test (30 cm thick silica gel), which showed no water flow (or pressure leaking) at 50 psig (344.70 kPa), the observed pressure leakage through the sandbox annulus likely indicates that there was incomplete annular sealing between the outside of the well due to the roughness and irregularities in the screen material used in the test setup.

The electrical conductivity measured for sand saturated with SGW was 2.12 mS/m. In comparison, the value for a sand pack saturated with DIW was 0.73 mS/m (Figure 13). The conductivity for the sand pack with the silica suspension containing 10,000 mg/L Na⁺ was 369.50 mS/m, more than two orders of magnitude higher than that of the sand pack with SGW. After gelation and with the aging of the gel, the electrical conductivity showed minimal changes (Figure 13). The significant difference in conductivity between the sand pack with SGW and the silica suspension and gel suggests that there is a sufficient electrical response to image the distribution of the injected silica suspension with the ERT geophysical approach, which could be evaluated in future testing.