Phase Behavior of Aqueous Biphasic Systems with Choline Alkanoate Ionic Liquids and Phosphate Solutions: The Influence of pH

Aqueous biphasic systems (ABS) composed of the choline alkanoate ionic liquids (ILs) choline acetate [Cho][OAc], choline propanoate [Cho][Pro], choline butyrate [Cho][But], and choline hexanoate [Cho][Hex], mixed with K3PO4 solutions at pH 7.2 and 14.5, were prepared and their phase diagrams were compared. The ability to form ABS with alkaline K3PO4 solutions decreased in the order [Cho][OAc] ≈ [Cho][Pro] > [Cho][But] > [Cho][Hex], while with neutral K3PO4 solutions, [Cho][OAc] could not form an ABS, and the other three ILs performed similarly. All of the biphasic regions of the ABS decreased with the increase in pH. 1H-NMR data indicated anion exchange between phases in ABS at neutral pH. The ABS at neutral pH were evaluated to extract the triazine herbicides simazine, cyanazine, and atrazine, and the ABS formed by [Cho][Pro] and the pH 7.2 K3PO4 solution has shown extraction recoveries higher than 90%.


Introduction
Aqueous biphasic systems (ABS) are, as the name implies, two-phase systems mainly composed of water. The aqueous nature of the two phases, without using volatile organic solvents, suggests that these systems could be more environmentally friendly alternatives for traditional liquid-liquid extraction methods. ABS have been reported for extractions of organic compounds [1][2][3][4][5][6] and biological macromolecules [7][8][9][10]. Based on their components, these systems can be classified as polymer/salt, polymer/polymer, ionic liquid (IL)/polymer, IL/salt, and IL/surfactant. The ABS based on polymers generate systems with high viscosity, which can be a drawback in extraction processes. Therefore, ABS with salts and ILs are preferable choices and by their nature have more tunable properties [11].
Ionic liquids, defined as organic salts with melting points below 100 • C [12], are salts that can be designed to have tunable properties for specific applications, including sample pretreatment and analyte extraction [13]. At present, most of the ABS with salts and ILs focus on systems with imidazolium-based ILs and inorganic salts [14,15]; however, some of the imidazolium-based ILs can have negative impacts on the environment and organisms [16,17]. Therefore, it is very important to design ABS based on biocompatible, environmentally friendly ILs.
Choline-based ILs are biodegradable, present low toxicity, are readily available, and have a low cost [18][19][20][21][22]. However, due to their high affinity for water, these ILs are usually studied in combination with polyethylene glycol (PEG) or polypropylene glycol (PPG)

Results and Discussion
Using the procedure described in the experimental section for the determination of the phase diagrams, a screening step was initially conducted to select the proper phosphate salt able to generate ABS with choline alkanoate ILs. Various aqueous phosphate solutions (Table 1) were added to the aqueous solutions of the ILs to form cloudy or clear solutions. The results indicated that the choline alkanoate ILs could not be salted out by sodiumbased salts, irrespective of the phosphate speciation induced by pH changes. On the other hand, K 3 [OAc] was previously reported to form ABS in combination with K 3 PO 4 under alkaline conditions (pH > 12) [28], resulting in a phase diagram and binodal curves similar to the ones reported in this work.
The binodal curves for the seven possible ABS were then constructed by cloud point titration of 80 wt% aqueous solutions of [Cho][OAc], [Cho][Pro], [Cho][But], and [Cho][Hex] mixed with aqueous K 3 PO 4 solutions at pH 14.5 or 7.2. The alkaline phosphate solution was prepared by directly adding the appropriate amount of K 3 PO 4 into deionized (DI) water, while the neutral phosphate solution was formed by mixing the appropriate amount of K 3 PO 4 and H 3 PO 4 with DI water. The phosphate solution was then dripped into the IL solution to form cloudy solutions, and water was then added to clarify the solutions. The procedure was repeated and the mass fractions of IL and K 3 PO 4 were recorded to establish the phase diagrams. The binodal curves at room temperature are presented in Figure 1 and the binodal data are provided in the Supplementary Materials (SM). The phase forming ability with the K 3  [Hex] were more easily salted out and all exhibited essentially the same ability to form two phases. If the phase diagram is represented using the phosphate anion (instead of K 3 PO 4 ) as the horizontal axis, the phase-forming ability of the selected ionic liquids follows the same pattern, although the effect of the pH is weaker ( Figure S3 The binodal curves at room temperature are presented in Figure 1  [Hex] were more easily salted out and all exhibited essentially the same ability to form two phases. If the phase diagram is represented using the phosphate anion (instead of K3PO4) as the horizontal axis, the phase-forming ability of the selected ionic liquids follows the same pattern, although the effect of the pH is weaker ( Figure S3 Figure S2. It has typically been reported that salt anions have the greatest effect on a salt's ability to salt out or be salted out [47,48], and the results here suggest that some protonation of the IL anions occurs at the higher pH, since the ILs would be considered the kosmotropic (salted out) salts. Prior reports focusing on the protonation of the kosmotropic (salting out) salts (citrate buffer solutions, pH 5-8) indicate that it is easier to salt out imidazoliumor quaternary ammonium-based ILs by increasing pH of the aqueous solution due to the degree of protonation of the citrate ions at different pH [49,50]. The previous literature also indicates that ABS form more easily under alkaline conditions rather than acidic or neutral conditions [51,52]. At alkali conditions, we observe that the PO 4  anion are needed to form ABS (i.e., a decrease in salting out ability) at higher pH.
This discrepancy might be due to the higher complexity of our systems, with protic salts salting out other protic salts. The speciation of the phosphate salt changes from a H 2 PO 4 − /HPO 4 2− system at pH 7.2 to PO 4 3− at pH 14.5 [53]. In addition, the hydroxide group present in the choline cation is protonated at pH 7.2 but deprotonated (i.e., presents a negative charge) at pH 14.5 (pKa 13.9) [54], while the anions of the ILs are weak organic carboxylic acids that, at both neutral and alkaline pH, are in their anionic forms (pKa values: 4.75-4.90) [55].
To further study the effect of solution pH on the speciation of IL anions, we conducted  It has typically been reported that salt anions have the greatest effect on a salt's ability to salt out or be salted out [47,48], and the results here suggest that some protonation of the IL anions occurs at the higher pH, since the ILs would be considered the kosmotropic (salted out) salts. Prior reports focusing on the protonation of the kosmotropic (salting out) salts (citrate buffer solutions, pH 5-8) indicate that it is easier to salt out imidazolium-or quaternary ammonium-based ILs by increasing pH of the aqueous solution due to the degree of protonation of the citrate ions at different pH [49,50]. The previous literature also indicates that ABS form more easily under alkaline conditions rather than acidic or neutral conditions [51,52]. At alkali conditions, we observe that the PO4 3− anion is able to salt out all the [Cho]-based ILs, except for [Cho] [Oct]. However, larger concentrations of the PO4 3− anion are needed to form ABS (i.e., a decrease in salting out ability) at higher pH.
This discrepancy might be due to the higher complexity of our systems, with protic salts salting out other protic salts. The speciation of the phosphate salt changes from a H2PO4 − /HPO4 2− system at pH 7.2 to PO4 3− at pH 14.5 [53]. In addition, the hydroxide group present in the choline cation is protonated at pH 7.2 but deprotonated (i.e., presents a negative charge) at pH 14.5 (pKa 13.9) [54], while the anions of the ILs are weak organic carboxylic acids that, at both neutral and alkaline pH, are in their anionic forms (pKa values: 4.75-4.90) [55].
To further study the effect of solution pH on the speciation of IL anions, we con-  Water has the largest influence on the hydrogen atoms in the cation residing on C atoms directly bonded to the N (H3, H2). At 40 wt% water (60 wt% IL, ~7.5 H2O:IL mol ratio), these atoms experience a negative deviation of approximately −0.2, while the H4 hydrogen atoms in the anion on the C bonded to the carboxylate group exhibit the largest Water has the largest influence on the hydrogen atoms in the cation residing on C atoms directly bonded to the N (H3, H2). At 40 wt% water (60 wt% IL,~7.5 H 2 O:IL mol ratio), these atoms experience a negative deviation of approximately −0.2, while the H4 hydrogen atoms in the anion on the C bonded to the carboxylate group exhibit the largest positive deviation of approximately 0.05. The H atoms on C atoms further away from the charge bearing groups are much less affected with essentially no or very little change in chemical shift. Interestingly, as the water concentration increases, all of the chemical shift deviations become more positive until 70 wt% (i.e., 30 wt% IL,~64.5 H 2 O:IL mol ratio), where they turn more negative, eventually ending up nearly the same as they were with only 40% water.  Figure S4). After phase separation, the pH of each phase (the top, IL-rich phase and the bottom, K 3 PO 4 -rich phase) was measured (Table S3 in  positive deviation of approximately 0.05. The H atoms on C atoms further away from the charge bearing groups are much less affected with essentially no or very little change in chemical shift. Interestingly, as the water concentration increases, all of the chemical shift deviations become more positive until 70 wt% (i.e., 30 wt% IL, ~64.5 H2O:IL mol ratio), where they turn more negative, eventually ending up nearly the same as they were with only 40% water.
NMR studies were also conducted on the separated phases of the  (Figure 3), overlaid on the binodal curves for these two ABS (tie-line data included in Supplementary Materials Figure S4). After phase separation, the pH of each phase (the top, IL-rich phase and the bottom, K3PO4-rich phase) was measured (Table S3 in the Supplementary Materials): When the neutral pH solution was used to generate the ABS, the resulting phases were weakly alkaline (pH 8.08-8.47), with a slightly higher pH on the top phase (8. 35-8.47) than on the bottom phase (8.08-8.20), possibly indicating ion exchange among the phases. On the other hand, the two phases of the ABS formed by alkaline K3PO4 solution remained strongly alkaline (pH 13.50-14.42), with a slightly lower pH on the top phase (13.50-13.81) than on the bottom phase (14.20-14.42). In addition, the NMR spectra of the top and bottom phases of the ABS were measured (shown in the Supplementary Materials).   In the top phases of the two ABS ( Figure 4, red symbols), a downfield chemical shift deviation (negative Δδ values in comparison to pure IL) was observed for all of the groups of protons (H1 to H5) at low IL concentrations. This negative deviation decreased with the increase in IL concentration, and reached a plateau at 60 wt% IL. Comparing the deviations of the protons of the ILs in the ABS formed with alkaline solutions (Figure 4, red, open symbols) to those at neutral pH ( Figure 4, red, filled symbols) indicates that the main differences are based on IL concentration rather than other effects due to the pH. The highest deviation was observed for proton H3 and H2; both protons are located in the cation. These two protons seem to interact with the water molecules at low IL concentrations, interactions that seem to decrease at higher concentrations of IL, possibly due to the formation of oligomeric ions, which is a characteristic of protic ILs.
An analysis of the protons in the bottom phases of the two [Cho][Pro]/K3PO4 ABS ( Figure 4, blue symbols) indicates a downfield chemical shift deviation, i.e., the same behavior observed in the top phase but with more negative Δδ values. As described for the top phases, the shifts decrease with the increase of IL concentration, and this decrease is linear except for the H1 protons in the pH 7.2 ABS. Also, as previously described for the protons in the top phase, the H2 and H3 protons suffer the highest negative shift. Different from what was observed in the top phases, pH has a greater influence on the proton deviation, with much larger negative deviations in the system at pH 14.5 ( Figure  4, blue, open symbols) than those observed in the protons of the system at pH 7.2 ( Figure  4, blue, filled symbols). This difference is possibly due to the favorable interaction between the ions of the IL and water in the ABS at high pH, where the hydroxide of the choline alkyl chain is also in its negative form, forming strong ionic hydrogen-bonded complexes with water.
The  In the top phases of the two ABS ( Figure 4, red symbols), a downfield chemical shift deviation (negative ∆δ values in comparison to pure IL) was observed for all of the groups of protons (H1 to H5) at low IL concentrations. This negative deviation decreased with the increase in IL concentration, and reached a plateau at 60 wt% IL. Comparing the deviations of the protons of the ILs in the ABS formed with alkaline solutions (Figure 4, red, open symbols) to those at neutral pH (Figure 4, red, filled symbols) indicates that the main differences are based on IL concentration rather than other effects due to the pH. The highest deviation was observed for proton H3 and H2; both protons are located in the cation. These two protons seem to interact with the water molecules at low IL concentrations, interactions that seem to decrease at higher concentrations of IL, possibly due to the formation of oligomeric ions, which is a characteristic of protic ILs.
An analysis of the protons in the bottom phases of the two [Cho][Pro]/K 3 PO 4 ABS ( Figure 4, blue symbols) indicates a downfield chemical shift deviation, i.e., the same behavior observed in the top phase but with more negative ∆δ values. As described for the top phases, the shifts decrease with the increase of IL concentration, and this decrease is linear except for the H1 protons in the pH 7.2 ABS. Also, as previously described for the protons in the top phase, the H2 and H3 protons suffer the highest negative shift. Different from what was observed in the top phases, pH has a greater influence on the proton deviation, with much larger negative deviations in the system at pH 14.5 (Figure 4, blue, open symbols) than those observed in the protons of the system at pH 7.2 (Figure 4, blue, filled symbols). This difference is possibly due to the favorable interaction between the ions of the IL and water in the ABS at high pH, where the hydroxide of the choline alkyl chain is also in its negative form, forming strong ionic hydrogen-bonded complexes with water.
The  Figure 5). A downfield chemical shift deviation (negative ∆δ values in comparison to pure IL) was observed for all protons (H1 to H5) in the two phases, with the largest deviations observed for proton H3 and H2, as observed in the [Pro]-based ABS. In the top phases of the two ABS ( Figure 5, red symbols), this negative deviation slightly decreased with the increase in IL concentration. The deviations were slightly larger for the protons of the ILs in ABS formed with neutral pH (Figure 5, red filled symbols) in comparison to those observed in alkaline solutions ( Figure 5, red open symbols). The protons in the bottom phases of the [Cho][But]/K 3 PO 4 ABS ( Figure 5, blue symbols) also showed more negative ∆δ values than those in top phases. Also, as previously described for the protons in the top phase, the H2 and H3 protons suffer the highest negative shift. Different from what was observed in the top phases, pH did not have an influence on the proton deviation; it seemed to be more influenced by the IL concentration than by the pH of the solution (pH 14.5, Figure 5, blue open symbols, or pH 7.2, Figure 5, blue filled symbols). solution and the IL concentration.
The NMR data were also used to calculate the chemical shift deviations of the protons of [Cho][But] in the top and bottom phases of the [Cho][But]/K3PO4 ABS at pH 7.2 and 14.5 ( Figure 5). A downfield chemical shift deviation (negative Δδ values in comparison to pure IL) was observed for all protons (H1 to H5) in the two phases, with the largest deviations observed for proton H3 and H2, as observed in the [Pro]-based ABS. In the top phases of the two ABS ( Figure 5, red symbols), this negative deviation slightly decreased with the increase in IL concentration. The deviations were slightly larger for the protons of the ILs in ABS formed with neutral pH ( Figure 5, red filled symbols) in comparison to those observed in alkaline solutions ( Figure 5, red open symbols). The protons in the bottom phases of the [Cho][But]/K3PO4 ABS ( Figure 5, blue symbols) also showed more negative Δδ values than those in top phases. Also, as previously described for the protons in the top phase, the H2 and H3 protons suffer the highest negative shift. Different from what was observed in the top phases, pH did not have an influence on the proton deviation; it seemed to be more influenced by the IL concentration than by the pH of the solution (pH 14.5, Figure 5, blue open symbols, or pH 7.2, Figure 5 Table 3.

Extraction of Triazine Herbicides Using Choline Alkanoate/K3PO4 (pH 7.5) ABS
Triazine-based herbicides such as simazine, cyanazine, and atrazine are reported to be unstable in strong acid or alkaline solutions [56] In addition to the stability of the analytes and the extraction efficiency, the viscosity of the system was another parameter to consider, since it could interfere with the mass transfer of the analytes. Hence, the viscosity of the aqueous solutions with different IL

Extraction of Triazine Herbicides Using Choline Alkanoate/K 3 PO 4 (pH 7.5) ABS
Triazine-based herbicides such as simazine, cyanazine, and atrazine are reported to be unstable in strong acid or alkaline solutions [56] In addition to the stability of the analytes and the extraction efficiency, the viscosity of the system was another parameter to consider, since it could interfere with the mass transfer of the analytes. Hence, the viscosity of the aqueous solutions with different IL concentrations was determined (Table S5, Supplementary Materials). Although, as expected, the viscosity of the solutions increased with increasing concentrations of the IL, the viscosity values were still relatively low, even at relatively high concentrations (e.g., at over 50% IL, viscosities were only 10−12 cP). Given these results, mass transfer of the analytes during extraction should not be significantly influenced by viscosity, and fast phase equilibrium is expected, which is beneficial for the partition of the analytes between the two phases.
The ABS were prepared with 10 wt% K 3 PO 4 (pH 7.5) and variable amounts ( Figure 6). After equilibration, the two phases were separated, and the herbicides were quantified using High Performance Liquid Chromatography (HPLC). In all cases, the concentrations of the herbicides that remained in the lower, K 3 PO 4 -rich phases of the ABS after extraction were below the detection limit of the HPLC, indicating high partition coefficients to the upper, IL-rich phases. Therefore, the recoveries of the herbicides extracted by the different ABS were calculated by measuring the concentration of the herbicides in the IL-rich phase (extracting phase) to evaluate the extraction efficiency of the ABS (Table 2).
the viscosity values were still relatively low, even at relatively high concentrations (e.g., at over 50% IL, viscosities were only 10−12 cP). Given these results, mass transfer of the analytes during extraction should not be significantly influenced by viscosity, and fast phase equilibrium is expected, which is beneficial for the partition of the analytes between the two phases.
The ABS were prepared with 10 wt% K3PO4 (pH 7.5) and variable amounts ( Figure 6). After equilibration, the two phases were separated, and the herbicides were quantified using High Performance Liquid Chromatography (HPLC). In all cases, the concentrations of the herbicides that remained in the lower, K3PO4-rich phases of the ABS after extraction were below the detection limit of the HPLC, indicating high partition coefficients to the upper, IL-rich phases. Therefore, the recoveries of the herbicides extracted by the different ABS were calculated by measuring the concentration of the herbicides in the IL-rich phase (extracting phase) to evaluate the extraction efficiency of the ABS (Table 2).    The recoveries of the analytes were higher than 60% in all cases, and in most systems, the recoveries increased with increasing concentration of the ILs. Interestingly, the anion has a big effect on the extractions, and the highest recoveries were achieved in the ABS formed by [ [Pro] and neutral phosphate solution is the preferred system for the extraction of these analytes.
Salt aqueous solutions were prepared at the desired pH (7.2 and 14.5) by mixing appropriate amounts of potassium phosphate tribasic and phosphoric acid in DI water. Neutral phosphate solution (27.52 wt%) was prepared with 9.9590 g K 3 PO 4 and 3.00 g H 3 PO 4 slowly added into 23.2292 g water, stirred until completely dissolved and cooled to room temperature. Alkaline phosphate solution (48.78 wt%) was formed with 24.001 g K 3 PO 4 added into 25.2015 g water and stirred to form a transparent solution. A similar procedure was followed to prepare the aqueous solutions of Na 2 HPO 4 (pH 9.0), NaH 2 PO 4 (pH 3.3), NaH 2 PO 4 /Na 2 HPO 4 (pH 5.5), and NaH 2 PO 4 /Na 2 HPO 4 (pH 7.0). In all cases, the pH of the solutions was measured with a pH meter (Accumet XL 600, Fischer Scientific, Ottawa, ON, Canada).
Stock standard solutions of individual herbicides (1.0 mg/mL) were prepared in methanol. An intermediate mixture of the three herbicides (100 µg/mL in methanol) was prepared by mixing the appropriate amount of the individual stock solutions, which was stored at 4 • C. Standard solutions with lower concentrations (10.0, 5.0, 2.0, 1.0, 0.5, and 0.2 µg/mL) were prepared daily in methanol by serial dilution and used for the calibration curve. A solution of 4.0 µg/mL of the three herbicides was used to study the recoveries and extraction efficiency of the ABS system.

Synthesis and Characterization of ILs
Choline alkanoate ILs were synthesized via neutralization of the base with the appropriate organic acids, following reported methods [42,43,57]. Briefly, 0.1 mol organic acid (glacial acetic acid, propionic acid, butyric acid, hexanoic acid, or octanoic acid) was added dropwise into an aqueous solution of choline hydroxide (0.1 mol). The mixture was stirred continuously using a magnetic stirrer (RZR 2051, Heidolph, Schwabach, Germany) for 12 h at room temperature. The obtained ILs were dried for 6 h under vacuum using a rotary evaporator (R-210, Büchi, Flawil, Switzerland), followed by freeze drying (Freezone 2.5, Labconco, Kansas City, MO, USA) for approximately 4 days.
To confirm the identity and purity of the synthesized ILs, 500 MHz NMR spectroscopy (AVIIIHD 500, Bruker, Fällanden, Switzerland) and infrared spectroscopy (Attenuated total reflection (ATR) Fourier transform infrared (FTIR) spectrometer, Alpha, Bruker, Billerica, MA, USA) were used (spectra are included in the Supplementary Materials). Viscosity was measured using a VISCOlab 3000 viscometer (Cambridge Viscosity, Inc., Boston, MA, USA) at 25 • C. The water content of the choline ILs was measured using a Karl Fisher Titrator (C20, Mettler Toledo, Greifensee, Switzerland). The chemical properties of the synthesized ILs are shown in Table 3. Bruker, Billerica, MA, USA) were used (spectra are included in the Supplementary Materials). Viscosity was measured using a VISCOlab 3000 viscometer (Cambridge Viscosity, Inc., Boston, MA, USA) at 25 °C. The water content of the choline ILs was measured using a Karl Fisher Titrator (C20, Mettler Toledo, Greifensee, Switzerland). The chemical properties of the synthesized ILs are shown in Table 3. The pure ILs (25.00 g) and 5.00 g water were added into a 100 mL beaker and mixed for 2 min, forming an approximately 80 wt% of the IL solutions. Then, 0.550 g of the 80 wt% IL in water was placed in a test tube and an aqueous solution with a known concentration of phosphate was added dropwise until the mixture became cloudy. A known mass of DI water was then added to make the mixture clear again. This procedure was repeated until enough data was obtained to develop the binodal curve of the ABS. The mass fraction of the phase components was determined by weight quantification of all the components added to the tube within an uncertainty of 0.001 g. The water content of the ILs was considered for the calculation of the compositions of the ABS mixtures. The experimental binodal data is provided in the Supplementary Materials.

Determination of the Tie-Lines
To determine the tie-lines (TLs), mixtures in the biphasic region were prepared in 10 mL glass vials, vigorously stirred, and allowed to achieve equilibrium by separation of the phases for 12 h at 25 °C. After phase separation, both the top (IL-rich solution) and bottom (K3PO4-rich solution) phases were carefully collected. The compositions of the top and bottom phases were determined using the gravimetric method described by Merchuk et al. [58], and the concentrations of the IL components were confirmed using 1 H-NMR. Each TL was determined using the lever rule to calculate the relationship between the top mass phase composition and the overall system composition. The tie-line length (TLL) was calculated using Equation (1): w h e where X and Y are the weight fraction of the IL and salt, respectively; and the subscript T and B indicate the top and bottom phases, respectively. The TLL of the studied systems increased at alkali conditions (Table S3 in the Supplementary Materials).

NMR Analysis
Mixtures in the biphasic regions with mass fractions of 10% neutral and 15% alkaline phosphate salt, and with 35, 40, and 45% ILs, were prepared. Once phase separation was achieved (as described above), the top and bottom phases were separated. The IL aque- The pure ILs (25.00 g) and 5.00 g water were added into a 100 mL beaker and mixed for 2 min, forming an approximately 80 wt% of the IL solutions. Then, 0.550 g of the 80 wt% IL in water was placed in a test tube and an aqueous solution with a known concentration of phosphate was added dropwise until the mixture became cloudy. A known mass of DI water was then added to make the mixture clear again. This procedure was repeated until enough data was obtained to develop the binodal curve of the ABS. The mass fraction of the phase components was determined by weight quantification of all the components added to the tube within an uncertainty of 0.001 g. The water content of the ILs was considered for the calculation of the compositions of the ABS mixtures. The experimental binodal data is provided in the Supplementary Materials.

Determination of the Tie-Lines
To determine the tie-lines (TLs), mixtures in the biphasic region were prepared in 10 mL glass vials, vigorously stirred, and allowed to achieve equilibrium by separation of the phases for 12 h at 25 • C. After phase separation, both the top (IL-rich solution) and bottom (K 3 PO 4 -rich solution) phases were carefully collected. The compositions of the top and bottom phases were determined using the gravimetric method described by Merchuk et al. [58], and the concentrations of the IL components were confirmed using 1 H-NMR. Each TL was determined using the lever rule to calculate the relationship between the top mass phase composition and the overall system composition. The tie-line length (TLL) was calculated using Equation (1): where X and Y are the weight fraction of the IL and salt, respectively; and the subscript T and B indicate the top and bottom phases, respectively. The TLL of the studied systems increased at alkali conditions (Table S3 in the Supplementary Materials).

NMR Analysis
Mixtures in the biphasic regions with mass fractions of 10% neutral and 15% alkaline phosphate salt, and with 35, 40, and 45% ILs, were prepared. Once phase separation was achieved (as described above), the top and bottom phases were separated. The IL aqueous solutions were prepared with concentrations in the range of 10-60 wt%. Pure ILs, IL solutions, and the top and bottom phases collected from the ABS were loaded solventless in a flame-sealed capillary, and the 1 H-NMR spectra were collected at 25 • C using d 6 -DMSO (dimethyl sulfoxide) as the external lock. Three sets of 1 H-NMR determinations were performed and the chemical shift deviations of the protons of the ILs, as a function of water and IL concentrations, were determined. The chemical shift deviations of the different mixtures were calculated using Equations (2)-(4): where the subscripts K 3 PO 4 and IL denote the K 3 PO 4 (bottom) or IL (top) phase in the ABS.

Extraction and Determination of the Target Herbicides Using the ABS Formed by ILs and pH 7.5 Phosphate Solutions
First, in 10.0 mL centrifuge tubes, known masses of 80 wt% IL, 24 wt% pH 7.5 phosphate solution (prepared as described in the section above), water, and the solution containing a mixture of the three target herbicides (spiked at 4.0 µg/g), were sequentially added to form the ABS with different mass fractions of IL and K 3 PO 4 . The mixtures were stirred for 2 min by vortex at room temperature and then centrifuged at 4500 revolutions per minute (rpm) for 10 min (Sorvall ST8, Thermo Scientific, Waltham, MA, USA). After centrifugation, the centrifuge tube was left on the bench at room temperature for 2 h to form the ABS. The top phase was primarily comprised of IL, the analytes, and water, while the bottom phase was mainly comprised of phosphate salts and water. A blank sample was also prepared and treated according to the procedure mentioned above without adding the target herbicide solutions. The top and bottom phases were carefully separated using plastic syringes, and the volumes and masses of the top and bottom phases were recorded.
The top phases were filtered with a 0.45 µm syringe filter membrane (VWR International, Ville Mont-Royal, QC, Canada) and injected into an HPLC for the separation and quantification of the herbicides. An Agilent 1260 series HPLC system with a quaternary pump, autosampler, and multiwavelength UV-Vis detector (Agilent Technologies, Santa Clara, CA, USA) was used for the determination of the studied herbicides. HPLC separations were performed using an XDB-C 18 column (Zorbax Eclipse 250 mm × 4.6 mm, 5 µm, Agilent Technologies, Santa Clara, CA, USA) at 25 • C. Gradient elution was performed with water and acetonitrile (ACN) as the mobile phase for the separation of analytes. The analysis started with 30% (v/v) ACN to 50% in 20 min. The column was then washed by increasing the proportion of ACN from 50% to 85% in 5 min and then to 95% in 1 min, held at that composition for 4 min, and then returning to 30% ACN, followed by a reequilibration time of 5 min. The flow rate and the injection volume were 1.0 mL/min and 20 µL, respectively. The detection wavelength was 220 nm.
Recovery (R) of the target compounds was used to evaluate the extraction efficiency of the analytes by the ABS and was determined using Equation (5): where C IL and C 0 are the concentration of the analytes in the IL (top) phase and the initial concentration of the analytes in the spiked sample solution, respectively; and V IL and V 0 are the volumes of the IL phase and the spiked sample solution, respectively.

Conclusions
In this manuscript, we demonstrated the ability of neutral and strong alkaline K 3 PO 4 solutions to salt out choline alkanoate ILs to form ABS. The ability to form ABS with alkaline K 3 [OAc] would not form an ABS, and the other three ILs performed similarly. All biphasic regions of the ABS decreased with increases in pH. 1 H-NMR confirmed an anion exchange between the phases, especially at neutral pH. Further studies are needed to fully understand the factors driving ABS formation using [Cho]-based ILs and inorganic salts at neutral pH, including a complete understanding of the speciation occurring in each phase and the probability of anion exchange with the IL anion.
The ABS at neutral pH were then evaluated for herbicide extraction efficiency, due to the instability of the analytes in acidic or basic solutions. A higher affinity of the analytes (i.e., simazine, cyanazine, and atrazine) to the IL-rich phase was observed, with recoveries higher than 60% in all cases. Overall, the ABS formed with [Cho][Pro] and neutral K 3 PO 4 solution exhibited the highest extraction recovery of the triazine herbicides, with recoveries higher than 90%.