Anion source. Two methods were used to load the anions: Via A, from acids, or via B, involving the corresponding ammonium salt (Scheme 1 and Table 1).

The AER (OH⁻ form) was packed in a column and treated with an aqueous or hydromethanolic solution of the acid or ammonium salt. The loading effectiveness was then checked by passing a methanolic solution of [bmim]I through the AER column loaded with the target anion and the halide ion to another anion exchange proceeded in quantitative yield.

Thus, following via A, the resin was charged with organic oxoanions derived from carboxilate (R-CO₂⁻), including chiral (S)-lactate, sulfonate (MeSO₃⁻) or phosphate (Bu₂PO₄⁻), together with inorganic anions such as Cl⁻, NO₃⁻ or ClO₄⁻, by treatment with the corresponding 1% aqueous acidic solutions. When the loading was performed with the aqueous solution of CF₃SO₃H, HF, H₃PO₄ or H₂SO₄, the polymeric matrix was partially denaturalized by overheating. For this reason, anions such as CF₃SO₃⁻, F⁻, H₂PO₄⁻ or HSO₄⁻ were loaded in the resin using aqueous solutions of their ammonium salts (via B). In order to confirm the efficiency of the method, both procedures were used to load AcO⁻, Cl⁻, PF₆⁻ or BF₄⁻ anions, and identical results were obtained. A few attempts to load anions from their corresponding Na⁺, K⁺ or Li⁺ salt showed, however, that the replacement of OH⁻ in the AER was incomplete, as evidenced by an observed mixture of anions in the checking, and this was not further studied.

Solvent selection. We extended our studies to the loading of hydrophobic anions, and explored alternative solvents and solvent mixtures. Benzoic acid was selected to prepare the AER (BzO⁻ from) and then a methanolic solution of [bmim]I was used to check the iodide-to-benzoate anion switch. The resin was first packed in a column and generously washed with the solvent, which was used afterwards to load the benzoate anion. Pure solvents such as distilled CH₃OH, CH₃CN, THF and CH₂Cl₂ were assayed, but only CH₃OH provided the optimal loading. Then, several solvent mixtures containing CH₃CN or THF with H₂O or CH₃OH were applied. Among the successful loading solvent mixtures that provided the AER in the BzO⁻ form, those with the lowest proportions of water or methanol were CH₃CN:H₂O (9:1), CH₃CN:CH₃OH (9.5:0.5), THF:H₂O (1:1) or THF:CH₃OH (4:1) (Scheme 1 and Table 1).

These results indicated that a non-aqueous mixture can be used to incorporate lipophylic anions, although the presence of a protic solvent was necessary for the OH⁻ replacement in the AER. Once the suitable solvent conditions were found, acetonitrile solvent mixtures were used to load representative hydrophobic anions: The anti-inflammatory acid ibuprofen to explore via A and ammonium tetraphenylborate to explore via B.

In order to check the loading effectiveness, a methanolic solution of [bmim]I was passed through the AER (Ibu⁻ form) or AER (Ph₄B⁻ form) and the pure [bmim][Ibu] [34] or [bmim][Ph₄B] [35] was obtained (see later). These results confirmed that lipophylic anions replace the OH⁻ anion in resin when using the appropriate solvent and the corresponding AER (A¯ form) obtained can then be used for the halide-to-anion switch.

Loading and exchange ability. The anion amount that the AER can load and the amount of halide that can then be exchanged were examined. Thus, 2.5 g (~3 cm³) of commercial wet A-26 (OH form) was treated with a 1% NH₄AcO aqueous solution until the pH value of the eluates indicated that loading was complete. Thus, 14.54 mmol of AcO⁻ was loaded with a maximum loading of 5.8 mmol of AcO⁻ per 1 g of this AER. In this context, the synthesis and characterization of resin-supported organotrifluoroborates have recently been reported and the loading was quantified by a UV/Vis spectroscopic analysis [36].

A 50 mM methanolic solution of [bbim]Br was passed through the packed column and aliquots were collected periodically and examined by ¹H-NMR. The related integration of signals corresponding to the anion and imidazolium cation indicated that the exchange process was quantitative up to nearly 14.54 mmol of ionic liquid, suggesting that the Br⁻ exchange could take place as long as there was enough AcO⁻ anion (Scheme 2).

Additionally, it should also be considered that the AER used in the exchange can be recycled by treatment with 10% NaOH aqueous solution, and the recovered AER (OH⁻ form) can be re-utilized for a new anion loading. In the present study, the chosen resin was Amberlyst A-26, given that it allows the use of aqueous mixtures and non-aqueous solvents, but other similar strongly basic anion exchange resins can be used instead.

Having achieved the loading of several anions in the AER, we examined their efficiency in the counterion exchange in imidazolium-based ILs, including [bmim]I or Br, [bbim]I or Br or [mmim]I as well as [bm₂im]Br. Thus, a methanolic solution of IL was passed through a column packed with the AER (A⁻ form) previously prepared, and the solvent was removed from the collected eluates. Following this simple method, in almost all cases I⁻ or Br⁻≥ 95% halide-for-anion swapping was obtained except for the hydrophobic anions Ph₄B⁻ and Ibu⁻, which gave for example, from [bmim]I in 65% and 95% yield, respectively (Table 2 and Scheme 3).

Moreover, no evidence of N-heterocyclic carbenes (NHCs) and/or dealkylation by-product formation was observed despite the basic environment, e.g., anion basicity [13,37,38]. The purity of the ionic liquids obtained by this method was qualitatively determined using ¹H-NMR spectra, and/or ESI(−)-MS experiments, and according to the silver chromate test, most analyses indicated low halide contents (<20 ppm for I⁻ or <13 ppm for Br⁻). Further quantification of possible halide impurity was restricted by instrumental limitation [32].

Although the halide exchange occurred with lipophylic anions such as Ph₄B⁻, when the process was carried out in methanol the yield of the recovered compound decreased to 65%, due to the change of solubility of the new ion pair, which caused their partial retention in the resin. Hence, organic solvents such as CH₃CN or CH₂Cl₂ or CH₃CN:CH₂Cl₂ solvent mixtures were then selected to perform the halide switch, the treatment of [bmim]I with AER (BzO⁻ form) being used to check the process. The results indicated that the exchange was successful in both aprotic organic solvents, while the use of pure CH₂Cl₂ as a solvent in our usual exchange procedure was discarded due to experimental difficulties, after testing several combinations, the mixture with the highest proportion of dichloromethane that was workable was found to be CH₃CN:CH₂Cl₂ (3:7). This enabled a quantitative iodide-for-benzoate swap and afforded the possibility for those exchanges of hydrophobic ionic species.

Accordingly, the preparation of [bmim][Ph₄B] or [bbim][Ph₄B] from their corresponding iodide salts using the AER (Ph₄B^¯ form) in CH₃OH provided the corresponding ion pair in 65% and 45% yield, respectively. The yield increased to 95% and 100% when CH₃CN was used, confirming that less polar solvents in the exchange process substantially improved the recovery of the less hydrophilic ion pair (Scheme 4). Similarly, [bm₂im]Br was directly studied in CH₃CN and the exchange of Ph₄B⁻ and Ibu⁻ anions proceeded in 91% and 96% yields, respectively (Table 2).

Hydrophobic salts such as hexylmethylimidazolium chloride [hmim]Cl or decylmethylimidazolium chloride [dmim]Cl were used to swap the chloride for the ibuprofenate anion. A solution of the corresponding ionic liquid in CH₃CN was passed through the AER (Ibu⁻ form) affording the anion exchange in ≤95% yields. A more lipophylic solvent was then used and quantitative results were obtained with the dipolar nonhydroxylic organic solvent mixture CH₃CN:CH₂Cl₂ (3:7) (Scheme 4 and Table 3).

Next, the AER (A⁻ form) method was extended to other anions. Thus, a methanolic solution of [bmim]Cl was passed through the AER (PF₆⁻ form) packed column and the eluates were analyzed after the solvent was removed. The ¹H-NMR spectrum coincided with that of [bmim][PF₆], which indicated a successful exchange confirmed by the silver chromate test (<6 ppm of Cl⁻). Similarly, a methanolic solution of [bmim][PF₆] was passed through the AER (Cl⁻ form) packed column and the ¹H-NMR spectrum also showed the quantitative exchange (Scheme 5). Thus, a conveniently loaded AER can be used to carry out the swapping from a range of anions other than halides. The process was followed by ¹H-NMR, since the signal corresponding to the C(2)-H of the imidazolium moiety is generally the most influenced by the nature of the anion (see Experimental section); for example, the chemical shift value measured in CDCl₃ (0.02 M) is 9.07 ppm in [bmim][PF₆] while in the same conditions this value is 10.99 ppm in [bmim]Cl.

Regarding other heteroaromatic cationic systems, pyridinium ([bmpy]I) or benzimidazolium (2·I) nuclei were chosen as examples to carry out the anion swap, together with the well known NHC precursor 1,3-dimesitylimidazolium salt (1·Cl) (Figure 1 and Scheme 6). A methanolic solution of [bmpy]I was passed through a column packed with the convenient AER (A⁻ form), and the corresponding pure [bmpy][A] were obtained in ≥ 98% yield, except for the acetate anion, which was recovered in 84% yield. Changing to a more hydrophobic solvent, the iodide-for-acetate swap in acetonitrile proceeded in quantitative yield. In the treatment of [bmpy]I with the AER (A⁻ form), there was no evidence in any case of the formation of decomposition byproducts, despite the basicity of some anions, e.g., acetate (Table 4).

Following the same procedure, a methanolic solution of the new benzimidazolium salt 2·I was used to obtain the corresponding ion pair 2·A, with excellent yields. The iodide exchange of the white solid 2·I (m.p. 150–1 °C) led to oily ion pairs at room temperature or solids with a low melting point (see Experimental section). The new benzimidazolium salts 2·A are related to previously reported benzimidazolium salts with potential use as new materials, e.g., ionic liquid crystals [39] and crystalline metal-containing ILs [40,41,42]. Likewise, the chloride anion in 1,3-dimesitylimidazolium salt 1·Cl can be successfully displaced by a wide range of anions using the AER (A⁻ form). When the swapping took place in methanol, the recovery of 1·A was between 80 to 95%, but in acetonitrile yields were nearly quantitative (Table 4). In all cases the silver chromate test revealed the low chloride content after the exchange (<6 ppm), which confirmed the easy swapping of Cl⁻ anion. These examples demonstrated that the method is also effective with non IL cationic systems, and is a general method for preparing tuneable quaternary heteroaromatic salts. Accordingly, the well-known catalyst precursor 1·Cl [43] was easily transformed in 1·A, and the presence of different counteranions could potentially modulate the formation of organometallic complexes due to their improved solubility and the stabilizing effect of anion participation.

Two examples of quaternary ammonium salts were selected from the API family to confirm the efficiency of the method with this type of ILs. The choline lactate ([Cho][Lact]) [44] was quantitatively prepared from the corresponding [Cho]I using the AER (Lact⁻ form) in methanol. Didecyldimethylammonium bromide ([d₂m₂N]Br) was transformed to the antibacterial-anti-inflammatory didecyldimethylammonium ibuprofenate [d₂m₂N][Ibu] [45].

This hydrophobic ammonium salt required the lipophylic solvent mixture CH₃CN:CH₂Cl₂ (3:7) to afford the quantitatively iodide-to-ibuprofenate switch, since in acetonitrile the yield was only 61% (Scheme 7 and Table 5).

The quaternary heteroaromatic and ammonium ILs prepared taking advantage of the AER (A⁻ form) method in organic solvents were characterized by ¹H-NMR, electrospray ionization mass spectrometry in the negative mode and the halide content was determined by the silver chromate test. When the recovery of the new ion pair [IL][A] was ≤95%, a new assay was performed using a less polar organic solvent, which improved the yield in the range of 95% to 100%. As mentioned above, the use of an anion exchange resin implies the possibility of sorbet contamination [13,46], so, nano-particulates may be an issue to analyze. The analysis of possible nano-particulate contamination was, however, beyond the scope of the present study.

Recapping the results, the AER (A⁻ form) method applied to different examples of quaternary heteroaromatic salts and ionic liquids permitted the halide to be swapped for assorted anions in excellent yields of ≥95% when the appropriate organic solvent or solvent mixture was used. It was confirmed that the AER (A⁻ form) method is efficient with imidazolium-based ILs, improving the currently operative procedures of classical counteranion exchange. Against a large pool of quaternary heteroaromatic and ammonium salts, we limited ourselves to the eleven examples shown in Figure 1 to validate the AER (A⁻ form) method in non-aqueous media.

Ion exchanger resin Amberlyst A-26 (Aldrich, OH⁻ form), [hmim]Cl, [dmim]Cl, [Cho]I and [d₂m₂N]Br together with all acids, ammonium salts, reagents and solvents were purchased from commercial suppliers, unless mentioned otherwise, and used without further purification. All solvents were reagent grade and methanol and THF were distilled prior to use. [bmim]I [32], [bmim]Br [32], [bbim]I [32], [bbim]Br [32], [mmim]I [32], [bm₂im]Br [47], [bmpy]I [48], and 1·Cl [49] were prepared according with the literature. ¹H-NMR and ¹³C-NMR spectra were recorded on a Varian Gemini 300 (300 MHz for ¹H and 75.4 MHz for ¹³C) spectrometer at 298 K. ¹H and ¹³C chemical shifts were referenced with TMS as an internal reference. Mass spectrometric analyses were performed on a LC/MSD-TOF (2006) mass spectrometer with a pumping system HPLC Agilent 1100 from Agilent Technologies at Serveis Científico-Tècnics of universitat de Barcelona. The pH was measured with benchmeter pH1100 (Eutech Instrunments, Nijkerk, The Netherlands), using Hamilton Flushtrode pH electrode for hydroalcoholic solutions.

A glass column (1 cm diameter ) packed with 2.5 g (~3 cm³) of commercial wet strongly basic anion exchange Amberlyst A-26 (OH⁻ form) was washed with water, and the column bed was equilibrated progressively with water-solvent mixtures until reaching the selected solvent media used afterwards for anion loading (~25 mL of each solvent mixture). A 1% acid or ammonium salt solution in the appropriate solvent was passed slowly through the resin until the eluates had the same pH value as the original selected acid solution, and then the resin was washed generously with solvent until constant pH. The process was carried out at room temperature, using gravity as the driving force.

A solution of the imidazolium salt (0.5–0.6 mmol) in 10 mL of the selected solvent was passed slowly through a column packed with ~3 cm³ of Amberlyst A-26 (A^– form), and then washed with 25 mL of solvent. The combined eluates were evaporated, and the residue obtained was dried in a vacuum oven at 60 °C with P₂O₅ and KOH pellets.

The amount of halide contents was determined by a silver chromate test following a similar protocol to that described by Sheldon and co-workers [31]. An aqueous solution (5 mL) of potassium chromate (5% p/v in Milli-Q water, 0.257 M) was added to the sample (5–10 mg). To 1 mL of the resulting dark yellow solution was added a minimum amount of silver nitrate aqueous solution (0.24% p/v in Milli-Q water, 0.014 M). A persistent red suspension of silver chromate would be observed if the sample was free of halide. The minimum measurable amount of silver nitrate aqueous solution was 0.011 mL; consequently, the detection limit is approx. 6 ppm for Cl⁻, 13 ppm for Br⁻ or 20 ppm for I⁻. The halide content was determined at least twice for each sample.

A suspension of 5,6-dimethyl-1H-benzimidazole (1.00 g, 6.84 mmol) and NaH (0.40 g, 16.66 mmol) in dry THF (100 mL) was stirred under argon atmosphere at 60 °C for 1 h, and then 1-iodobutane (1.50 g, 8.15 mmol) was added. The reaction mixture was stirred at 65 °C for 48 h, and then 5 mL of ethanol were added. The solvent was evaporated to dryness, and the residue was treated with water (50 mL) and extracted with CH₂Cl₂ (3 × 50 mL). The organic solution was dried over anhydrous Na₂SO₄, filtered and the solvent was eliminated under vacuum. A mixture of the previous yellow oil (1.34 g, 6.62 mmol) and 1-iodobutane (1.23 g, 6.70 mmol) was stirred under argon atmosphere at 85 °C for 20 h. The reaction mixture was washed with dry diethyl ether (3 × 25 mL) in an ultrasonic bath, providing the pure 2·I as a white solid (2.47 g, 93% yield). M.p. 150–1 °C. δ_H (300 MHz; CDCl₃; Me₄Si) 0.99 (6H, t, J = 7.4 Hz, N-C₃H₆-CH₃), 1.45 (4H, m, N-C₂H₄-CH₂-CH₃), 2.02 (4H, m, N-CH₂-CH₂-C₂H₅), 2.47 (6H, s, C(5,6)-Me), 4.56 (4H, t, J = 7.4 Hz, N-CH₂-C₃H₇), 7.42 (2H, s, C(4,7)-H) and 11.01 (1H, s, C(2)-H). δ_C (75.4 MHz, CDCl₃) 13.5, 19.8, 20.7, 31.3, 47.3, 112.8, 129.8, 137.5, 140.4. HRMS-ESI(+) Calcd for C₁₇H₂₇N₂ [M]⁺ 259.2169, found 259.2167.

Melting points of compounds 2·A: 2·MeSO₃, 62–3 °C; 2·Bu₂PO₄, 56–7 °C; 2·PF₆, 85–6 °C; 2·BF₄, 109–110 °C; 2·CF₃SO₃, 78–9 °C; 2·SCN, 64–5 °C; 2·AcO, 2·BzO and 2·Lact are oily compounds at room temperature