Soultz-sous-Forêts geothermal water was synthesized using various chemicals to set the concentration of major cations, i.e., Na⁺, Ca²⁺, K⁺, Li⁺, Mg²⁺, Sr²⁺, and Ba²⁺, for a chemical composition similar to that of actual geothermal fluid (Scheiber et al., 2012). The main chemical properties of synthetic geothermal water are shown in Table 1.

After synthesizing geothermal water, solutions with different concentrations of dissolved SiO₂ were added to observe the lithium extraction efficiency. The content of Si ions is measured and converted into SiO₂ content that can be converted to a soluble form of silica in the geothermal water. In this study, 0, 150, and 350 mg/L of dissolved SiO₂ solutions were put to the geothermal water, and they were classified as samples A, B, and C in this study. As shown in the previous studies by authors, high TDS concentration of a solution might inhibit the efficient lithium recovery during solvent extraction process (Jang et al., 2017) and solvent extraction was applied in a 50× diluted shale gas produced water of which the TDS concentration was up to 157,000 mg/L. Soultz-sous-Forêts geothermal water has a high TDS value, i.e., approximately 100,000 mg/L (Scheiber et al., 2012), therefore, three different dilution rates for the synthetic geothermal fluid were tested. The results were provided in Supplementary Material. When the original geothermal fluid was tested, most of the divalent ions showed a low removal efficiency (<40%). For In the 25× diluted geothermal water, Sr²⁺ and Ba²⁺ showed > 90% removal efficiency and Mg²⁺ showed 84% removal efficiency. When the solvent extraction method was applied to the 50× and diluted geothermal water, all divalent ions showed significantly higher removal efficiency than for the 25× diluted geothermal water, especially, Sr²⁺ and Ba²⁺ showed removal efficiencies greater than 97%. In both 25× and 50× diluted geothermal water, Ca²⁺ showed the highest removal efficiency (close to 100%) and Li⁺ showed removal efficiency about 69–70%. Based on the results, it was concluded that the 50× diluted geothermal water was optimal to be tested for the lithium recovery from geothermal fluid.

In all stages of the solvent extraction, D2EHPA (97%, Sigma Aldrich, St. Louis, MO, United States) was used as an extractant, and kerosene (Sigma-Aldrich, St. Louis, MO, United States) was used as a diluent to control the molarity of the organic phases. During the solvent extraction process, the ratio of the organic phase to the aqueous phase was 1:1, and the two phases were mixed using a shaking incubator (SH-BSI16R, Samheung Instrument, Korea). The shaking speed and temperature were set to 150 rpm and 25°C, respectively, for 30 min. After mixing the organic and aqueous phases, the two-phase solution was equilibrated until the two phases were separated in a separating funnel. The aqueous solution was reused for repetitive extractions up to four times to improve the efficiency of the removal of the divalent cations and fresh organic solvents (1.0 M D2EHPA) were applied in every repetitive extraction. The process of the second stage was similar to the previous one but TBP (Tributyl phosphate, C₁₂H₂₇O₄P, 98.5% purity, Daejung Co., Siheung, Korea) was used for lithium recovery in the mixture of D2EHPA and kerosene as an additive. Fresh organic solvent (1.5 M D2EHPA + 0.3 M TBP) was used in every repetition and the aqueous solution was reused for repetitive extractions.

After every extraction cycle, 10 ml of the aqueous phase was withdrawn and filtered with a 0.45 μm polytetrafluoroethylene (PTFE) filter (Millipore, Germany) to remove particulates. The permeate was then analyzed cations and anions by inductively coupled plasma optical emission spectrometry (ICP-OES, Optima 8300, Perkin Elmer, United States) and ion chromatography (IC, Dionex ICS-1100, Thermo Scientific, United States). The pH level was measured using a multimeter (Orion Star A329, Thermo Fisher Scientific, United States). The organic solution was analyzed by Fourier transform infrared spectroscopy (FT-IR spectra, Nicolet 6700, Thermo Fisher Scientific, Waltham, MA, United States) to analyze the changes in the absorbances of the major functional groups.

According to Hano et al. (1992), the chemical reaction of D2EHPA (HR) and metal ion (Mⁿ⁺) can be described as Eq. 1 (x: solvation number of complex) M n + + ( n + X ) 2 ( HR ) 2 ↔ M R n ( HR ) x + n H + (1)

A chelate (metal-extractant complex) can be formed during the interaction of deprotonated D2EHPA and metal ions, and the metal ions move from the aqueous phase to the organic phase (Lee et al., 2011). A proper amount of TBP would act as a synergistic reagent in the mixture of D2EHPA, TBP, and metals to increase the extraction efficiency of metal ions, and this reaction can be described as Eq. 2 (Hano et al., 1992; Amani et al., 2017). M R X ( HR ) + TBP ↔ M R ( x − 1 ) HR ( TBP ) + HR (2)

One molecule of D2EHPA from the D2EHPA-metal complex can be replaced with one molecule of TBP to react with another metal ion to increase the extraction efficiency.

The extraction efficiency of each metal ion is calculated by Eq. 3 (Jafari et al., 2018): Extraction efficiency  ( % )   = [ C ] i n ,   a q − [ C ] f i n ,   a q [ C ] i n ,   a q × 100 (3)

In Eq. 3, the concentration of a specific metal ion in the aqueous phase before solvent extraction is [C]_in,aq, and the concentration of a specific metal ion in the aqueous phase after solvent extraction is [C]_fin,aq.

Distribution ratio (D_m) is the ratio of the concentration of the metal ion in the organic phase to the concentration of the metal ion in the aqueous phase and is calculated by Eq. 4 (Ganji et al., 2016). D m = [ M R n ( H R ) x ] [ M n + ]   = c o n c e n t r a t i o n   o f   s p e c i f i c   m e t a l   i o n   s p e c i e s   i n   t h e   o r g a n i c   p h a s e   c o n c e n t r a t i o n   o f   s p e c i f i c   m e t a l   i o n   s p e c i e s   i n   t h e   a q u e o u s   p h a s e (4)

The separation factor of a specific metal ion (M₁) over another metal ion (M₂) in the extraction was calculated by the distribution ratio (D_m), as described by Eq. 5 (Jang et al., 2017). S M 2 M 1 = D M 1 D M 2 =   ( M 1   i o n   c o n c e n t r a t i o n   i n   t h e   o r g a n i c   p h a s e   /   M 1   i o n   c o n c e n t r a t i o n   i n   t h e   a q u e o u s   p h a s e ) ( M 2   i o n   c o n c e n t r a t i o n   i n   t h e   o r g a n i c   p h a s e   /   M 2   i o n   c o n c e n t r a t i o n   i n   t h e   a q u e o u s   p h a s e ) (5)

The influence of different dissolved SiO₂ concentrations (0, 150, 350 mg/L) in geothermal water was observed during the solvent extraction process of metal ions. Figure 1 shows the extraction efficiency of cations (Ca²⁺, Sr²⁺, Mg²⁺, Li⁺, Na⁺) in geothermal fluid, with different dissolved SiO₂ concentrations for four extraction processes. Ba²⁺ was not drawn in the figure due to its lower concentration (<0.4 mg/L) in the initial fluid after 50× dilution. The pH of the aqueous solution was measured after every repetition extraction process, and the range of the pH values was 1.8–2.1 due to the production of hydrogen ions, as described in Eq. 1.

The removal efficiencies of Ca²⁺, Mg²⁺, and Sr²⁺ were greater than 90% after four repetitions in all three samples. The removal rates of SiO₂ were also calculated after each step and the values were significantly low. For sample A and C, the removal rates of silicate ions from the aqueous side were 0% and, for sample B, the removal rate was measured as about 7% after 4 repetitions. D2EHPA affinity for cation extraction has been reported to be in the order of Ca²⁺ > Sr²⁺ > Mg²⁺ > Li⁺ > Na⁺ (Hano et al., 1992; Jang et al., 2017), and all the cations tendency satisfied in three samples. For example, as D2EHPA shows the highest affinity for Ca²⁺ compared to other divalent ions (Hano et al., 1992; Jang et al., 2017), the removal efficiency showed the highest values in all conditions as shown in Figure 1.

The removal efficiency of cations including lithium increased as the SiO₂ concentration increased. Especially, the lithium removal efficiency in solution A at the first extraction was 55.5%, and the efficiency increased to 66.8 % and 77.9% in solutions B and C, respectively. To observe the effect of dissolved SiO₂ in the solution, distribution ratio of the cations were calculated for the first (1-1) extraction (Table 2). The D_Na values in all samples indicate that the extraction of Na⁺ was not significant during the first solvent extraction. The distribution ratio of Ca²⁺ showed a decrease while the SiO₂ concentration increases but the other divalent ions (Sr²⁺ and Mg²⁺) and Li⁺ showed higher distribution ratio in higher SiO₂ concentration solution.

After four repetitive stages of extraction, more cations were extracted from the aqueous phase and the distribution ratio of cations increased. The separation factors of three divalent ions over lithium after four stages were calculated as shown in Table 3. Ca²⁺ had the highest selectivity factors in the geothermal water without SiO₂, and the selectivity increased when the SiO₂ concentrations increased in the solution. However, Sr²⁺ and Mg²⁺ showed slight decreased separation factors with the increase of SiO₂ concentrations.

The absorbance of the metal-D2EHPA complex in the organic phase was analyzed using FT-IR to observe the change in the functional group during solvent extraction (Figure 2). The result for D2EHPA before the extraction is shown in red line, and the sample A, B, and C results are shown in Figures 2A–C, respectively. Due to the formation of a metal-D2EHPA complex during the extraction with no SiO₂ ions (sample A), the absorbance of the P=O bond (1034 cm⁻¹) and the P-O-H bond (1230 cm⁻¹) decreased after extraction (Figure 2A). It seems that the metal ion and an electronegative P=O bond from D2EHPA form a bond, and a hydrogen from the P–O–H bond is replaced with a metal ion. Figure 2A shows that the wavenumbers for the absorbance peaks increase as the SiO₂ concentrations increased, which indicates that the existence of the SiO₂ ions interferes with the loading of the metal ion during the solvent extraction.

Although the formation of metal-D2EHPA complex was slightly inhibited by the SiO₂ ions in the aqueous fluid (Figure 2), more metal ions were extracted from the aqueous solution in the existence of the SiO₂. One of the explanation for that could be the silica polymerization. The silanol group (Si-O-H) originated from dissolved SiO₂ in geothermal water can form polymeric, colloidal, and particulate silica due to various conditions, such as pH or the presence of other ions (Park et al., 2020). The polymerization of silicic acid occurs rapidly in neutral or slightly alkaline pH values and is formed slowly below pH 6.5 (Putera et al., 2018; Park et al., 2020). Due to a low pH solution (1.8–2.1) after every repetition extraction process, the polymerization of silicic acid actually occurs probably at a slow rate. With other divalent cations, silica polymerization can be formed because silicic acid is classified as a weak acid in geothermal water (Brown, 2013; Park et al., 2020).

After removing the divalent cations from the geothermal water in the first extraction step, lithium was extracted during four repetitive cycles in the second step (Figure 3). The [C]_in,aq value in Eq. 3 for all data points (2-1, 2-2, 2-3 and 2-4) is the concentration value measured after the stage 1–4 is completed, and the [C]_fin,aq value implies the concentration measured after each stage. Li⁺ had a higher extraction efficiency than the other monovalent cation, Na⁺, for all three samples and the Li⁺ extraction efficiency ranged from 21.9 to 24.8%. The separation factor of Li⁺ over Na⁺ could not be calculated due to the negative extraction efficiency of Na⁺ ions. The separation factor of Li⁺ over Na⁺ could not be calculated due to the negative extraction efficiency of Na⁺ ions. Jang (2016) explained that the effect of TBP addition can dissolve the hydrated complex (HRn ∙ xH₂O) that was not extracted in the previous extraction stage. If the coordination number of the metal ion is not satisfied to load from the aqueous phase to the organic phase during the extraction reaction, the metal ions can be stocked with water molecules with a creation of a hydrated complex (HRn ∙ xH₂O) (Tanaka and Akaiwa, 2009). The Na⁺ showed relatively low extraction efficiency in the first stage due to the formation of a hydrated chelate complex with D2EHPA. TBP molecules have a low affinity with Na⁺ ions and the metal chelate complex (HRn ∙ xH₂O) can be resolved in the solution by the addition of TBP, which might cause an increase of Na⁺ ions concentration (Jang, 2016).

After removing most of the divalent ions in the first step, the overall extraction efficiency of Li was calculated as shown in Figure 4. In the first stage extraction, 71.9–79.50% of the original lithium was lost and 21.9–24.8% of the remaining lithium was recovered during the second stage extraction. Therefore, the overall lithium recovery rate was calculated as 4.5–6.8%. The overall recovery of Li⁺ decreased from 6.8 % to 4.5% as the SiO₂ concentration increased. The difference between the Li⁺ recovery rates depends mainly on the loss of Li⁺ in the first step of the extraction process.

A certain amount of the geothermal fluid is generally lost in EGSs during the circulation of the fluid, and less than 10% of the fluid loss would make the long-term operation of the system possible (Clark et al., 2013; Schill et al., 2017). Therefore, in some cases, the input of new fluid such as river or lake water near the site would be required to operate the system properly. In this study, it was shown that 50× dilution was favorable for the lithium extraction in the geothermal water. If 0.2% of the geothermal water is used for the lithium extraction, the volume of the solution after the extraction will be close to 10% of the original geothermal water volume. If further study shows that the lithium recovery is feasible from the diluted geothermal fluid with the freshwater near the EGSs, the simultaneous lithium recovery and EGS operation without fluid loss might be possible.