Efficient and recyclable removal of imidazolium ionic liquids from water using resorcinol–formaldehyde polymer resin

Abstract

Resorcinol–formaldehyde (RF) resin is used for the first time to remove imidazolium-based ionic liquids (ILs) from water via adsorption. 1-Butyl-3-methylimidazolium chloride (BMIMCl) was selected as a representative imidazolium IL, and the adsorption kinetics and isotherm of BMIMCl to RF resin were determined. The pseudo first order rate law appeared to be a more appropriate model to interpret the adsorption kinetics, whereas the adsorption isotherm data was well described by the Langmuir isotherm model. The activation energy (E_a) was determined to be 10.8 kJ mol⁻¹, and the maximum adsorption capacity was estimated as 96 mg g⁻¹ at 30 °C. The elevated temperature facilitated adsorption of BMIMCl to RF resin and the adsorption was also much enhanced under alkaline conditions. Charged surfactants could compete with BMIMCl during adsorption or lessen the electrostatic attraction between BMIMCl and RF resin. When a non-ionic surfactant was present, BMIMCl might be confined within the micelles of the non-ionic surfactant, thereby limiting the adsorption of BMIMCl onto RF resin. Cations of ILs with longer alkyl chains showed a higher affinity towards RF resin, whereas the adsorption of BMIMBF₄ to RF resin was much improved owing to the higher hydrophobicity of BF₄⁻ than Cl⁻. The regenerated RF resin also exhibited a stable and efficient adsorption capacity for BMIMCl and the desorption of ILs from RF resin can be also achieved within relatively short periods. These results reveal that RF resin could be a promising and recyclable adsorbent for removing imidazolium ILs from water.

---

1. Introduction

Room temperature ionic liquids (ILs) are melting salts with relatively low melting points, making them free-flowing liquids even at ambient temperature. ILs possess many advantageous properties, such as negligible vapor pressure, high thermal stability, non-flammability as well as high solvating capacity.^1,2 Therefore, ILs have been used in reaction and separation applications to substitute for conventional solvents³ and employed in various other fields, including catalysis,⁴ energy production,⁵ pharmaceuticals,⁶ analytical chemistry,⁷ electrochemical applications.⁸ ILs are becoming more popular in industry and many processes involving water streams.⁹ Therefore, the release of ILs into water environments is inevitable. However, many studies reveal that a number of common ILs can be hazardous to aquatic ecology,^10,11 indicating that it is necessary to develop effective techniques to remove ILs from water environments.

To date, several techniques have been proposed to remove ILs, such as bio-degradation,^12,13 advanced oxidation,¹⁴ filtration,¹⁵ and adsorption.^16–19 In view of operation cost and convenience,²⁰ adsorption remains as one of the most widely employed techniques to remove ILs from water. Thus, several adsorbents have been evaluated for removing ILs from water, for example, carbonaceous materials (e.g., activated carbon and graphene oxide),^16–18 polymers,^19,21 minerals,²² and bio-derived materials.^19,23 Although carbonaceous materials can be preferred owing to its abundance, adsorption of ILs to carbonaceous adsorbents was relatively low without additional modifications.^16–18 In contrast, polymeric materials can exhibit much higher adsorption capacities. However the evaluated polymeric materials were limited to divinylbenzene-based commercial resins and certain functional groups (i.e., strong acids) were required in order to give appreciable adsorption capacities. Without strong acidic groups, polymeric resin appeared to be ineffective for adsorption of ILs.²¹ In addition, recyclability of these polymeric resins, which could pose considerable challenges, has not been evaluated. Therefore, alternative polymeric resins which are effective for removing ILs even in the absence of strong acidic functional groups should be developed and the corresponding recyclability should be demonstrate for practical usage of polymeric resins for removing ILs.

Here we propose to use another common polymeric adsorbent, resorcinol–formaldehyde (RF) resin to remove ILs from water. RF resin, although it does not consist of strong acids, has also been proven to exhibit high affinity towards cations, such as cesium ions.²⁴ Nevertheless, to our knowledge, adsorption of ILs to RF resin has not been investigated; therefore in this study RF resin was used to adsorb ILs in water. Because alkylimidazolium-based ILs are the most extensively investigated ILs in the literature,^25,26 1-butyl-3-methylimidazolium chloride (BMIMCl) was selected as a representative alkylimidazolium IL.

Adsorption behaviors of BMIMCl to RF resin were investigated by determining adsorption kinetics and isotherm. Factors influencing adsorption of BMIMCl were examined including temperature, pH and co-existing surfactants. In addition to BMIMCl, 1-hexyl-3-methylimidazolium chloride (HMIMCl) and 1-ethyl-3-methylimidazolium chloride (EMIMCl) were also tested to study effect of alkyl chain length of alkylimidazolium cation. Adsorption of 1-butyl-3-methylimidazolium tetrafluoroborate (BMIMBF₄) to RF resin was evaluated to examine effect of hydrophobicity of anions of ILs. Recyclability of RF resin was also evaluated by regenerating used RF resin with NaCl solution and the desorption kinetics of IL from RF resin in NaCl solution was also determined and analyzed by a desorption model.

2. Experimental

RF resin was prepared according to a reported protocol;²⁷ the detailed preparation procedure can be found in ESI (Fig. S1†). Adsorptive removal of BMIMCl from water using RF resin was evaluated by batch-type experiments. In a typical experiment, 0.1 g of RF resin was added to 200 mL of BMIMCl solution with an initial concentration (C₀) of 50 mg L⁻¹. The mixture was then placed on a temperature-control heating plate with mechanical stirring. After a given time t, sample aliquots were withdrawn from the mixture and the remaining concentration of BMIMCl at a given time t (C_t) was analyzed using a UV spectrophotometer at 210 nm (Chrome-tech S-2200, Taiwan).

The adsorption capacity of RF resin for BMIMCl at a given time t (q_t, mg g⁻¹) was determined using the following equation (eqn (1)):

where v (L) denotes the volume of reaction solution and W (g) denotes the mass of RF resin used for adsorption. When the IL concentration reached equilibrium (i.e., C_e, mg L⁻¹), the adsorption capacity of RF resin at equilibrium was represented as q_e (mg g⁻¹). Adsorption isotherm of IL to RF resin was determined by adding a fixed amount of RF resins (0.01 g) to 20 mL of IL solution with increasing C₀ ranging from 0 to 100 mg L⁻¹. Experiments of adsorption isotherm were performed for 3 h.

Adsorption kinetic and isotherm experiments were conducted at different temperatures (i.e., 30, 40 and 50 °C) to examine the effect of temperature on adsorption behaviors and to determine thermodynamic parameters. The pH values of IL solutions were also varied to be 3–11 to examine the effect of pH. To further investigate the effect of pH, zeta potentials of RF resin was also measured by a zetasizer (Nano-ZS, Malvern Instruments Ltd, Malvern, UK).

Since IL-containing wastewater may contain other compounds, effects of co-existing compounds was examined by adding a cationic surfactant, CTAB, an anionic surfactant SDS and a non-ionic surfactant P123 to IL solutions. Recyclability of RF resin was also evaluated; used RF resin was regenerated by washing IL-rich RF resin with NaCl aqueous solution (5 wt%) or ethanol. The regenerated RF resin was then used in a subsequent adsorption experiment.

3. Results and discussion

3.1 Characterization of RF resin

Fig. 1(a) shows the morphology of the as-prepared RF resin, which consists of micrometer-sized irregular particles. A closer view of its surface (Fig. 1(b)) reveals that the surface of the as-synthesized RF resin was smooth. In addition, Fig. 1(c) shows the XRD pattern of the as-prepared RF resin which agrees with reported amorphous XRD patterns of RF polymer.²⁸ The surface chemistry of RF resin was determined using FT-IR, and its spectrum is shown in Fig. 1(d), which is consistent with reported spectra of RF resin in the literature.²⁹ Specifically, the peaks at 1092 and 1215 cm⁻¹ can be attributed to the C–O stretch and deformation of benzyl ether groups, whereas the peaks at 1479, 2846 and 2929 cm⁻¹ are associated with the CH₂ stretching and bending vibrations. The peak at 1610 cm⁻¹ is derived from aromatic ring stretches. These characterizations indicate that RF resin was well prepared. The average molecular weight of the as-prepared RF resin was measured as 500 (±100) g mol⁻¹.

Fig. 1 Characteristics of RF resin: (a) and (b) SEM images of RF under different magnifications; (c) XRD pattern and (d) FT-IR spectrum.

3.2 Adsorption kinetics of BMIMCl to RF resin

To investigate adsorption behaviors of BMIMCl to RF resin, adsorption kinetics was first measured and analyzed. Fig. 2(a) shows q_t of RF resin as a function of reaction time at different temperatures. At 20 °C, q_t increased gradually as the reaction time proceeded and approached an equilibrium q_t of 50 mg g⁻¹ after 120 min. When temperature increased to 40 and 60 °C, q_t approached 70 and 80 mg L⁻¹, respectively, revealing that higher temperatures improved adsorption capacity of RF resin for BMIMCl. Nevertheless, the effect of temperature on the adsorption kinetics could not be easily visualized in Fig. 2(a). Therefore, the pseudo first order and second order rate laws were employed to determine rate constants as follows (eqn (2) and (3)):

Fig. 2 Adsorption behaviors of BMIMCl to RF resin at different temperatures: (a) kinetics of BMIMCl adsorption to RF resin and fitting results using the pseudo first order rate law (dotted lines) and the pseudo second order rate law (solid lines). (C₀ = 50 mg L⁻¹, RF = 500 mg L⁻¹) (b) adsorption isotherms of BMIMCl adsorption to RF resin and fitting results using the Langmuir isotherm model (solid lines) and the Freundlich (dashed lines). (RF = 500 mg L⁻¹).

The pseudo first order rate law

The pseudo second order rate law

where k₁ (min⁻¹) and k₂ (g mg⁻¹ min⁻¹) are the rate constants for the pseudo first and second rate laws, respectively.

When the pseudo first order rate law was used to model the adsorption kinetic data, fitting results can be represented as the dotted lines in Fig. 2(a) and fitting parameters are summarized in Table 1. As correlation coefficients (R₁²) were all higher than 0.995 at the testing temperatures, the pseudo first order raw law appears to appropriately describe the adsorption kinetics of BMIMCl to RF resin. In addition, the pseudo first order rate constant increased from 0.023 to 0.030 min⁻¹, revealing that higher temperatures led to faster adsorption kinetics.

Table 1 Kinetic parameters for BMIMCl adsorption to RF resin at various temperatures

Temp. (°C)	Conditions
	Pseudo-first-order			Pseudo-second-order
	k1 (min^−1)	qe, estimated (mg g^−1)	R1^2	k2 × 10^3(g mg^−1 min^−1)	qe, estimated (mg g^−1)	R2^2
30	0.023	53.34	0.980	0.236	76.27	0.995
40	0.026	70.92	0.990	0.253	91.97	0.997
50	0.030	80.37	0.983	0.282	102.62	0.996

---

When the adsorption kinetic data were modeled by the pseudo second order rate law, fitting results can be represented as the solid lines seen in Fig. 2(a) and fitting parameters are also summarized in Table 1. The correlation coefficients (R₂²) were equivalent or higher than 0.995, indicating that the pseudo second order rate law could be also an appropriate model to interpret the adsorption kinetics. The pseudo second order rate constant also increased from 0.236 × 10⁻³ to 0.282 × 10⁻³ g mg⁻¹ min⁻¹ as the temperature changed from 30 to 50 °C, validating the positive effect of elevated temperature on the adsorption kinetics.

While the two rate laws seemed to be suitable kinetic models to interpret the adsorption kinetics, the estimated adsorption capacities at equilibrium (q_e, estimated) obtained using the pseudo first order rate law were much closer to the experimental values, suggesting that the pseudo first order rate law should be a preferred kinetic model to analyze the adsorption kinetic data of BMIMCl to RF resin.

3.3 Adsorption isotherm of BMIMCl to RF resin

To determine maximum adsorption capacity of BMIMCl to RF resin, adsorption isotherms of BMIMCl to RF resin at different temperatures were measured and analyzed. Fig. 2(b) shows that as C_e increased, q_e gradually increased and then approached saturation. When the temperature increased, q_e approached to a higher saturation value, confirming the positive effect of elevated temperature on adsorption of BMIMCl. This also suggests that the adsorption of BMIMCl to RF was an endothermic reaction.

The adsorption isotherm data were then first modeled by the Langmuir isotherm. In the Langmuir isotherm, adsorption is assumed to occur as a monolayer on a homogenous surface. Because a limited number of adsorption sites on the homogeneous surface, a maximum capacity would be reached. The Langmuir isotherm can be described as follows (eqn (4)):

where q_max is the estimated maximum adsorption capacity and K_L denotes the Langmuir isotherm constant which is related to the adsorption bonding energy.

Fitting results of the isotherm data using the Langmuir isotherm are represented as the solid lines seen in Fig. 2(b) and fitting parameters are listed in Table 2. The correlation coefficients (R_L² > 0.990) reveal that the Langmuir isotherm seemed a satisfactory model to interpret the isotherm data. The q_max value increased significantly from 96 to 135 mg g⁻¹ as the temperature varied from 30 to 50 °C, validating that adsorption capacity was greatly improved at high temperatures. In comparison with q_max values of reported adsorbents for BMIMCl (Table 3), RF resin actually exhibited relatively high adsorption capacities compared to several carbonaceous adsorbents and a few commercial resins, showing that RF resin can be a quite promising adsorbent for BMIMCl.

Table 2 Modelling parameters of adsorption isotherms derived from Langmuir model, Freundlich model, and Langmuir–Freundlich (Sips) models at various temperatures

Ionic liquid	Temp. (°C)	Langmuir			Freundlich			Langmuir–Freundlich (Sips)
		qmax (mg g^−1)	KL (L mg^−1)	RL^2	KF (mg g^−1) (L mg^−1)^1/n	n	RF^2	qm (mg g^−1)	KLF (L mg^−1)^1/n	n	RLF^2
BMIMCl	30	96	0.076	0.991	21.6	3.1	0.973	83.1	0.090	0.651	0.996
BMIMCl	40	124	0.086	0.999	26.0	2.8	0.983	110.9	0.104	0.766	0.998
BMIMCl	50	135	0.105	0.997	31.0	2.9	0.981	121.5	0.125	0.784	0.998
EMIMCl	30	75	0.039	0.992	8.2	2.2	0.984	49.9	0.397	0.064	0.999
HMIMCl	30	124	0.363	0.997	53.4	4.4	0.976	123.2	0.974	0.366	0.997
BMIMBF4	30	113	0.532	0.996	56.6	5.3	0.983	120.7	1.303	0.483	0.998

---

Table 3 Comparison of BMIMCl adsorption capacity of RF resin with the reported adsorbents

Material	T (°C)	Maximal adsorption capacity	Reference
Macroporous resorcinol–formaldehyde	30	96	In this study
	40	124
	50	135
Functional carbon sphere (derived from cellulose)	35	22	Qi et al.^17
Oxygenated carbonaceous material	25	56	Qi et al.^17
Graphene oxide	—	123	Longlong et al.^16
Amberlyst 15	30	395	Li et al.^21
Cellulose microsphere	25	64–188	Xu et al.^23
Activated charcoal	35	30	Palomar et al.^18
Sawdust-based activated carbon	25	21	Vijayaraghavan et al.^19
Coal-based activated carbon	25	20	Vijayaraghavan et al.^19
Amberlite IR-120	30	361	Li et al.^21
Lewatit CNP 80	30	124	Li et al.^21
Amberlite IRC-84	30	34	Li et al.^21
Amberlite IRN-150	25	179	Vijayaraghavan et al.^19
NKA-9	30	12	Li et al.^21

---

Furthermore, the isotherm data was analyzed using another isotherm model, the Freundlich isotherm. In the Freundlich isotherm, physisorption (mono-layer adsorption) and chemisorption (multi-layer adsorption) both can simultaneously occur during adsorption. The Freundlich isotherm is typically described in the following equation (eqn (5)):

where K_F represents the Freundlich isotherm constant, and n denotes the heterogeneity factor of RF resin, which is associated with the surface heterogeneity. The corresponding fitting results are represented as the dashed lines seen in Fig. 2(b). Fitting parameters are also summarized in Table 2; the n values at the testing temperatures are higher than the unity, suggesting that the adsorption isotherm of BMIMCl can be considered as a L-type isotherm and that there was a strong affinity between BMIMCl and RF resin.³⁰ Nevertheless, the fitting lines seemed less satisfactory to fit the data points compared to the solid fitting lines of the Langmuir isotherm in view of correlation coefficients.

In addition, an integrative model combining the Langmuir and Freundlich isotherms, called Langmuir–Freundlich (L–F) isotherm (also called Sips isotherm), was employed to analyze the isotherm data. The L–F isotherm is established on the basis of the Freundlich isotherm with an asymptotic characteristic that adsorption approaches saturation eventually. Therefore, the L–F isotherm is typically described in the following equation (eqn (6)):

where K_LF (L mg⁻¹)^1/n denotes the L–F isotherm constant and q_m represents the estimated maximum adsorption capacity. If the heterogeneity factor (n) is the unity, the L–F isotherm in fact becomes the Langmuir isotherm. Fitting parameters using the L–F isotherm were summarized in Table 2. The corresponding correlation coefficients are very comparable to those obtained using the Langmuir isotherm, but much higher than those obtained using the Freundlich isotherm. These analyses using the three models suggest that the adsorption of BMIMCl to RF resin was more likely attributed to physisorption and adsorption would eventually reach saturation.

3.4 Activation energy and thermodynamic parameters of BMIMCl adsorption

As the rate constants (either k₁ or k₂) noticeably increased at elevated temperatures, the relationship between the rate constants and temperature can be further correlated via the Arrhenius equation to determine activation energy (E_a, kJ mol⁻¹) using the following equation (eqn (7)):where A denotes the pre-exponential factor (g mg⁻¹ min⁻¹); R represents the universal gas constant; and T represents temperature (K). k₁ was selected in particular because the pseudo first order rate law appears to be a more satisfactory model to interpret the kinetics as discussed in the Section 3.2. A plot of ln k₁ versus 1/T can be seen in Fig. S2 (see ESI†), in which the data points are perfectly-fit by the linear regression line with R₂ = 0.996. This demonstrates that the adsorption kinetics of BMIMCl to RF resin at different temperatures can be estimated by the Arrhenius equation. The E_a of adsorption of BMIMCl to RF is determined as 10.8 kJ mol⁻¹.

Additionally, one can also notice that the adsorption isotherm constant (e.g., K_L) also increased significantly at elevated temperatures. Correlations between adsorption isotherm constants and temperature can determine essential thermodynamic parameters. Specifically, the Langmuir isotherm constants (K_L) at different temperatures can be correlated to obtain the adsorption enthalpy (ΔH°) and entropy (ΔS°) using the following equations (eqn (8) and (9)):

ΔG° = −RT ln K_LF (8)

ΔG° = ΔH° − TΔS° (9)

where ΔG° is the free energy of the adsorption at a given temperature and K_L is the Langmuir isotherm constant. Table S1† lists ΔG° values at different temperatures; ΔG° becomes more negative at a higher temperature, suggesting that the adsorption of BMIMCl to RF resin can occur spontaneously at the testing temperatures and has a greater chance to occur if temperature increases. Based on eqn (9), a plot of ΔG° versus T is shown in Fig. S3,† in which the data points are also well-fit with R₂ = 0.997. The slope and intercept of the linear regression line are the adsorption entropy (ΔS°) and the adsorption enthalpy (ΔH°), respectively. As (ΔH°) is determined as 13.2 kJ mol⁻¹, the adsorption of BMIMCl to RF resin can be validated to be endothermic. The relatively low adsorption enthalpy also suggests that the adsorption of BMIMCl to RF resin was a physical adsorption phenomenon. Additionally, as ΔS° is calculated as 0.12 kJ mol⁻¹ K⁻¹, a noticeable change in system entropy (TΔS°) of 36.4 kJ mol⁻¹ could occur at 30 °C. This change in entropy can be attributed to mobility of adsorbates (i.e., BMIMCl) and the degree of destruction of solvation shells upon removal of hydrophobic parts of BMIMCl from solution owing to the high cohesion energy density of water.^22,31 The positive value of ΔS° suggests that the disorder at the RF-solution interface was increased, because water molecules were desorbed from pores or surface of RF resin as BMIMCl was adsorbed to RF resin.

In comparison with other reported adsorption entropies for adsorption BMIMCl (i.e., 0.031–0.055 kJ mol⁻¹),^22,32 the ΔS° for BMIMCl adsorption to RF resin is relatively high. This indicates that BMIM cation might exchange with more mobile ions originally present on the surface of RF resin, leading to the increase in entropy during the adsorption process.²² Another possibility for the relatively high ΔS° is that both the head group and alkyl side-chain of BMIM cation interacted with the surface of RF resin and higher degree of destruction of solvation shells was achieved, liberating more water molecules.²²

3.5 Effect of pH on adsorption on BMIMCl to RF resin

Furthermore, the effect of initial pH of BMIMCl solutions was examined in Fig. 3. A distinct trend can be observed that adsorption capacity was relatively low at low pH and increased gradually as pH became higher. Specifically, q_e remained quite similar at pH = 5–7 but started decreasing noticeably at pH = 4, followed by a huge drop in q_e at pH = 3. This indicates that the adsorption of BMIMCl to RF resin could be stable under neutral and weakly acidic conditions but was unfavorable under highly acidic conditions. When pH was raised up to 8 or above, q_e was substantially increased, suggesting that the alkaline environments facilitated adsorption of BMIMCl to RF resin. As the surface charge of RF resin might be varied under different conditions, the zeta potentials of RF resin at pH ranging 3 to 11 were measured (Fig. 3). RF resin inherently exhibited negatively charged surface in this pH range, even under the neutral condition. The zeta potential became even more negative when pH increased from 3 to 11. One can note that the variation of zeta potentials from pH = 3 to 11 actually could be associated with the variation of q_e in the same pH range. The q_e of RF resin significantly decreased when its zeta potential became less negative under acidic conditions. In contrast, q_e of RF resin was much enhanced as the zeta potential became more negative at pH = 8 or higher. This suggests that the adsorption of BMIMCl to RF resin was strongly associated with the electrostatic interaction between RF resin and BMIMCl, in which the positive BMIM ions could be attracted to the negatively charged RF resin. As the molecular weight of RF resin can be varied, its surface properties (e.g., surface charges) may be changed, leading to different adsorption capacity of RF resin for BMIMCl. This can be an interesting follow-up study to reveal such a relationship.

Fig. 3 Effect of pH on BMIMCl adsorption to RF resin and zeta potentials of RF resin under various pH values at 30 °C. (C₀ = 50 mg L⁻¹, RF = 500 mg L⁻¹).

3.6 Effects of surfactants on adsorption on BMIMCl to RF resin

Considering that wastewater may contain other compounds, for example, artificial surfactants, we also examine the effects of charged and uncharged surfactants on the adsorption of BMIMCl to RF resin. First, when a typical cationic surfactant, CTAB, with an equivalent concentration to BMIMCl was present in a BMIMCl solution, q_t of RF resin for BMIMCl was almost zero even after 150 min adsorption. This demonstrates that the presence of the cationic surfactant, CTAB, significantly hindered the adsorption of BMIMCl to RF resin possibly because the adsorption competition between cetyltrimethylammonium (CTA) cations and BMIM cations. On the other hand, the anionic surfactant, SDS, with the equivalent concentration was also added to a BMIM solution. The q_t also decreased substantially; however the decrease in q_t in the presence of SDS was much less than that in the presence of CTAB. When SDS was present, sodium ions from SDS might also neutralize the negative charge of RF resin. As a result, the electrostatic attraction between RF resin and BMIMCl was lessened, thereby leading to a reduced q_t.

In addition to CTAB and SDS, we also examined how non-ionic surfactants affected the adsorption capacity by selecting Pluronic P123, an ether-based triblock copolymer. Although P123 is a non-ionic surfactant, the adsorption was also significantly affected in the presence of P123. A number of studies reveal that when ILs are mixed with P123, IL molecules can be confined within micelles of P123.^33,34 Therefore, the attraction of BMIMCl to RF resin was significantly limited, leading to the lower q_t seen in Fig. 4.

Fig. 4 Effect of co-existing surfactants on BMIMCl adsorption to RF polymer at 30 °C (C₀ = 50 mg L⁻¹, RF = 500 mg L⁻¹).

3.7 Adsorption isotherms of other imidazolium ILs to RF resin

As RF resin exhibited promising adsorption capacities for BMIMCl, we also evaluated adsorption capacities of RF resin for other imidazolium ILs. Fig. 5 and Table 2 show adsorption capacities of RF resin for EMIMCl and HMIMCl, which consist of similar chemical structures to BMIMCl with a short and longer alkyl chain, respectively. The adsorption isotherm of EMIMCl reveals that the saturation q_e of RF resin for EMIMCl was much less than that for BMIMCl. A longer alkyl chain has been proven to enhance the attraction between ILs and resin via stronger van der Waals and polar interactions.^35,36 It has been also reported that adsorption of ILs with longer alkyl chains to resin can be improved via hydrophobic ionic liquid–water interactions.²¹

Fig. 5 Adsorption isotherms of various ionic liquids to RF at 30 °C and fitting results using the Langmuir isotherm model (solid lines) (C₀ = 50 mg L⁻¹, RF = 500 mg L⁻¹).

Additionally, as discussed in the previous sections, the adsorption of BMIMCl to RF resin could be also attributed to the electrostatic attraction between imidazolium cations and negatively-charged RF resin. Therefore, a given amount of RF resin should attract a certain number of imidazolium cations. Nevertheless, because the molecular weight of EMIMCl was lower than that of BMIMCl, the adsorption capacity of RF resin for EMIMCl was accordingly less than that for BMIMCl. Likewise, the adsorption capacity of RF resin for HMIMCl was higher than that for BMIMCl (Fig. 5) owing to a higher molecular weight of HMIMCl than BMIMCl.

On the other hand, we also measured adsorption capacity of RF resin for BMIMBF₄ to examine the effect of different counter ion of ILs in Fig. 5. While having the same cation, the adsorption capacity of RF resin for BMIMBF₄ was much higher than that for BMIMCl, showing that a strong affinity existed between BMIMBF₄ and RF resin. Since, BF₄⁻, is considered to be more hydrophobic and less basic than Cl⁻, this comparison suggests that higher hydrophobicity of ILs enhances adsorption of ILs to aromatic adsorbents.¹⁸ In a previous study, sulfonic and carboxylic divinylbenzene-based resins were evaluated for removing imidazolium ILs.²¹ However, the adsorption of ILs with anions of higher hydrophobicity to these resins is not significantly different from those with Cl⁻, possibly because the electrostatic interactions between these highly acidic resins and ILs were predominantly strong. As a result, even though the hydrophobicity was increased, enhancement in adsorption of ILs was not noticeably observed.

In addition, ΔG° of adsorption of these other imidazolium-ILs to RF resin was also calculated and summarized in Table S2.† As these ΔG° values were all negative, adsorption of these other imidazolium-ILs to RF resin can also occur spontaneously at the testing temperature. Adsorption capacities of RF resin for EMIMCl, HMIMCl and BMIMBF₄ are also compared with those of other reported adsorbents in Table S3 (see ESI†). While studies reporting adsorption capacities for EMIMCl, HMIMCl and BMIMBF₄ are very limited, one can note that RF resin still exhibited much higher adsorption capacities than other reported adsorbents.

3.8 Recyclability of RF resin and desorption of BMIMCl from RF resin

As RF resin has shown its promising adsorption capacity for imidazolium ILs, here we further examined whether RF resin can be reused for removal of ILs multiple times. To generate used RF resin, two solvents were used to wash IL-rich RF resin: the first solvent was ethanol and the second was aqueous NaCl solution (5 wt%).

Fig. 6(a) shows that the ethanol-washed RF resin was still able to remove BMIMCl from water at cycle #2; however q_e decreased gradually in the following cycles. This demonstrates that ethanol was, although capable, not highly efficient to regenerate used RF resin. On the other hand, when NaCl solution was employed to regenerate used RF resin, the adsorption capacity of RF resin, even though slightly decreased, could be recovered mostly at cycle #2. In the following cycles, the adsorption capacity of RF resin remained comparable, indicating that NaCl solution was much more effective for regenerating used RF resin. In addition, desorption kinetics of BMIMCl from used RF resin was also measured (Fig. 6(b)) which shows ratios of the amount of recovery (q_r) to the adsorption capacity (q_e) as a function of desorption time (t_r). One can note that more than 80% of adsorbed IL could be recovered from RF resin after merely 6 hours and more than 95% could be desorbed after 24 hours. This demonstrates that used RF resin can be easily regenerated within a relatively short period by NaCl solution and RF resin remains highly stable and efficient to remove BMIMCl from water.

Fig. 6 (a) Recyclability of RF resin for BMIMCl adsorption via the regeneration of used RF by NaCl-washing and ethanol-washing treatments (C₀ = 50 mg L⁻¹, RF = 500 mg L⁻¹); (b) desorption kinetics of BMIMCl from RF resin at 30 °C (C₀ = 50 mg L⁻¹, RF = 500 mg L⁻¹).

The desorption kinetics was also analyzed according to the following equation developed by Peppas (eqn (10)):³⁷

where k_r and n are empirical parameters. Fitting result for the desorption kinetics using eqn (10) can be represented as the dashed line in Fig. 6(b); the data points are well-fit by the fitting line with R² = 0.985. This suggests that the recovery of BMIMCl from IL-rich RF resin can be appropriately described by eqn (10). The fitting result determines k_r as 58.67 (±1.26) min^−0.15 and n as 0.15 (±0.01). With these estimated values, the detailed form of eqn (10) can be formulated as

4. Conclusion

In this study, RF resin was used for the first time to remove imidazolium ILs from water via adsorption. The adsorption kinetics of the model imidazolium IL, BMIMCl, to RF resin can be more appropriately modeled by the pseudo first order rate law with R² > 0.995. The rate constant also increased significantly from 0.023 to 0.030 min⁻¹ as temperature changed from 20 to 60 °C with the E_a of 10.8 kJ mol⁻¹. The maximum adsorption capacity was estimated using the Langmuir isotherm model as 96 mg g⁻¹ at 20 °C, revealing that RF resin exhibited a relatively high adsorption capacity compared to several carbonaceous and commercial adsorbents. The adsorption enthalpy (ΔH°) and entropy (ΔS°) for the adsorption BMIMCl to RF resin were also determined as 13.2 kJ mol⁻¹ and 0.12 mol⁻¹ K⁻¹, respectively. When pH was changed from 3 to 10, the surface charge of RF resin was found to decrease substantially, leading to higher adsorption capacities attributed to the stronger electrostatic attraction between RF resin and BMIMCl. The maximum adsorption capacities of RF resin for EMIMCl, HMIMCl, and BMIMBF₄ were also estimated as 75, 124 and 113 mg g⁻¹, respectively, suggesting that cations with longer alkyl chains and anions with higher hydrophobicity facilitated adsorption of imidazolium ILs to RF resin. RF resin was also regenerated easily by washing with NaCl solution; the regenerated RF resin exhibited stable and efficient adsorption capacity for BMIMCl. These results reveal that RF resin can be a promising and recyclable adsorbent for removing imidazolium ILs from water.

Acknowledgements

The authors thank Carey Russell at Children's National Medical Center (Washington, District of Columbia) for her assistance with the English editing.

References

1. J. S. Wilkes, P. Wasserscheid and T. Welton, Introduction, in: Ionic Liquids in Synthesis, Wiley-VCH Verlag GmbH & Co. KGaA, 2008, pp. 1–6.
2. T. Welton, Room-Temperature Ionic Liquids. Solvents for Synthesis and Catalysis, Chem. Rev., 1999, 99, 2071–2084.
3. X. Han and D. W. Armstrong, Ionic Liquids in Separations, Acc. Chem. Res., 2007, 40, 1079–1086.
4. H. Olivier-Bourbigou, L. Magna and D. Morvan, Ionic liquids and catalysis: Recent progress from knowledge to applications, Appl. Catal., A, 2010, 373, 1–56.
5. D. R. MacFarlane, N. Tachikawa, M. Forsyth, J. M. Pringle, P. C. Howlett, G. D. Elliott, J. H. Davis, M. Watanabe, P. Simon and C. A. Angell, Energy applications of ionic liquids, Energy Environ. Sci., 2014, 7, 232–250.
6. M. Smiglak, J. M. Pringle, X. Lu, L. Han, S. Zhang, H. Gao, D. R. MacFarlane and R. D. Rogers, Ionic liquids for energy, materials, and medicine, Chem. Commun., 2014, 50, 9228–9250.
7. J. Shao, Z. Wan, H. Liu, H. Zheng, T. Gao, M. Shen, Q. Qu and H. Zheng, Metal organic frameworks-derived Co₃O₄ hollow dodecahedrons with controllable interiors as outstanding anodes for Li storage, J. Mater. Chem. A, 2014, 2, 12194–12200.
8. M. Armand, F. Endres, D. R. MacFarlane, H. Ohno and B. Scrosati, Ionic-liquid materials for the electrochemical challenges of the future, Nat. Mater., 2009, 8, 621–629.
9. N. V. Plechkova and K. R. Seddon, Applications of ionic liquids in the chemical industry, Chem. Soc. Rev., 2008, 37, 123–150.
10. T. P. Thuy Pham, C.-W. Cho and Y.-S. Yun, Environmental fate and toxicity of ionic liquids: A review, Water Res., 2010, 44, 352–372.
11. D. Zhao, Y. Liao and Z. Zhang, Toxicity of Ionic Liquids, Clean: Soil, Air, Water, 2007, 35, 42–48.
12. E. Liwarska-Bizukojc, C. Maton and C. V. Stevens, Biodegradation of imidazolium ionic liquids by activated sludge microorganisms, Biodegradation, 2015, 26, 453–463.
13. D. Coleman and N. Gathergood, Biodegradation studies of ionic liquids, Chem. Soc. Rev., 2010, 39, 600–637.
14. E. M. Siedlecka, M. Gołębiowski, Z. Kaczyński, J. Czupryniak, T. Ossowski and P. Stepnowski, Degradation of ionic liquids by Fenton reaction; the effect of anions as counter and background ions, Appl. Catal., B, 2009, 91, 573–579.
15. J. F. Fernández, D. Waterkamp and J. Thöming, Recovery of ionic liquids from wastewater: Aggregation control for intensified membrane filtration, Desalination, 2008, 224, 52–56.
16. T. Longlong, Z. Xin, Q. Wei, L. Dan, J. Qiang, L. Jin, Y. Yuanlv, L. Zhan and W. Wangsuo, The adsorption of water-soluble ionic liquids on graphene oxide of different oxygen content, RSC Adv., 2014, 4, 58536–58545.
17. X. Qi, L. Li, Y. Wang, N. Liu and R. L. Smith Jr, Removal of hydrophilic ionic liquids from aqueous solutions by adsorption onto high surface area oxygenated carbonaceous material, Chem. Eng. J., 2014, 256, 407–414.
18. J. Palomar, J. Lemus, M. A. Gilarranz and J. J. Rodriguez, Adsorption of ionic liquids from aqueous effluents by activated carbon, Carbon, 2009, 47, 1846–1856.
19. K. Vijayaraghavan, T. P. T. Pham, C.-W. Cho, S. W. Won, S. B. Choi, M. Juan, S. Kim, Y.-R. Kim, B.-W. Chung and Y.-S. Yun, An Assessment on the Interaction of a Hydrophilic Ionic Liquid with Different Sorbents, Ind. Eng. Chem. Res., 2009, 48, 7283–7288.
20. M. N. Rashed, Adsorption Technique for the Removal of Organic Pollutants from Water and Wastewater, Organic Pollutants—Monitoring, Risk and Treatment, ed. M. N. Reashed, InTech, Rijeka, 2013.
21. L. Li, Y. Wang and X. Qi, Adsorption of imidazolium-based ionic liquids with different chemical structures onto various resins from aqueous solutions, RSC Adv., 2015, 5, 41352–41358.
22. W. Mrozik, C. Jungnickel, M. Skup, P. Urbaszek and P. Stepnowski, Determination of the adsorption mechanism of imidazolium-type ionic liquids onto kaolinite: implications for their fate and transport in the soil environment, Environ. Chem., 2008, 5, 299–306.
23. M. Xu, Y. Ao, S. Wang, J. Peng, J. Li and M. Zhai, Efficient adsorption of 1-alkyl-3-methylimidazolium chloride ionic liquids onto modified cellulose microspheres, Carbohydr. Polym., 2015, 128, 171–178.
24. C. A. Nash and S. T. Isom, Characterization of Spherical Resorcinol–Formaldehyde Resin Cesium Adsorption with Batch Contact Tests, Sep. Sci. Technol., 2010, 45, 1822–1827.
25. R. Hagiwara and Y. Ito, Room temperature ionic liquids of alkylimidazolium cations and fluoroanions, J. Fluorine Chem., 2000, 105, 221–227.
26. G. Law and P. R. Watson, Surface Tension Measurements of N-Alkylimidazolium Ionic Liquids, Langmuir, 2001, 17, 6138–6141.
27. G. Zhang, C. Ni, L. Liu, G. Zhao, F. Fina and J. T. S. Irvine, Macro-mesoporous resorcinol–formaldehyde polymer resins as amorphous metal-free visible light photocatalysts, J. Mater. Chem. A, 2015, 3, 15413–15419.
28. R. Gupta and B. Kandasubramanian, Hybrid caged nanostructure ablative composites of octaphenyl-POSS/RF as heat shields, RSC Adv., 2015, 5, 8757–8769.
29. M. C. Gutiérrez, F. Rubio and F. del Monte, Resorcinol–Formaldehyde Polycondensation in Deep Eutectic Solvents for the Preparation of Carbons and Carbon–Carbon Nanotube Composites, Chem. Mater., 2010, 22, 2711–2719.
30. B. B. K. Huat, S. S. Gue and F. H. Ali, Tropical Residual Soils Engineering, CRC Press, 2004, pp. 377–403.
31. E. J. Sullivan, J. W. Carey and R. S. Bowman, Thermodynamics of Cationic Surfactant Sorption onto Natural Clinoptilolite, J. Colloid Interface Sci., 1998, 206, 369–380.
32. L. Reinert, K. Batouche, J.-M. Lévêque, F. Muller, J.-M. Bény, B. Kebabi and L. Duclaux, Adsorption of imidazolium and pyridinium ionic liquids onto montmorillonite: characterisation and thermodynamic calculations, Chem. Eng. J., 2012, 209, 13–19.
33. R. L. Vekariya, V. K. Aswal, P. A. Hassan and S. S. Soni, Influence of N-Alkylpyridinium Halide Based Ionic Liquids on Micellization of P123 in Aqueous Solutions: A SANS, DLS, and NMR Study, Langmuir, 2014, 30, 14406–14415.
34. Y. Yu, W. Zhu, L. Hua, H. Yang, Y. Qiao, R. Zhang, L. Guo, X. Zhao and Z. Hou, Ionic liquid-Pluronic P123 mixed micelle stabilized water-soluble Ni nanoparticles for catalytic hydrogenation, J. Colloid Interface Sci., 2014, 415, 117–126.
35. S. Hassan, L. Duclaux, J.-M. Lévêque, L. Reinert, A. Farooq and T. Yasin, Effect of cation type, alkyl chain length, adsorbate size on adsorption kinetics and isotherms of bromide ionic liquids from aqueous solutions onto microporous fabric and granulated activated carbons, J. Environ. Manage., 2014, 144, 108–117.
36. J. Lemus, C. M. S. S. Neves, C. F. C. Marques, M. G. Freire, J. A. P. Coutinho and J. Palomar, Composition and structural effects on the adsorption of ionic liquids onto activated carbon, Environ. Sci.: Processes Impacts, 2013, 15, 1752–1759.
37. N. A. Peppas, Analysis of Fickian and non-Fickian drug release from polymers, Pharm. Acta Helv., 1985, 60, 110–111.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra09067b

---