Physicochemical properties of novel cholinium ionic liquids for the recovery of silver from nitrate media

Abstract

Cholinium based ionic liquids (IL), i.e. N-(2-hydroxyethyl)-N,N-dimethyl-N-octylammonium bis(trifluoromethanesulfonyl)imide ([C₈linChol]⁺[NTf₂]⁻) and N-(2-hydroxyethyl)-N-(2-ethylhexyl)-N,N-dimethylammonium bis(trifluoromethanesulfonyl)imide ([C₈ramChol]⁺[NTf₂]⁻) were synthesized and characterized by ¹H NMR, ¹³C NMR, ATR-FTIR and ESI-MS. Activation energy of the viscous flow, molar volume, molar entropy, lattice potential energy and isobaric expansion coefficients were deduced from density and viscosity measurements. These calculations led to the conclusion that neat [C₈ramChol]⁺[NTf₂]⁻ exhibits a slightly more compact ion arrangement than [C₈linChol]⁺[NTf₂]⁻. After characterizing physicochemical properties of these two neat ILs, extractive properties of [C₈linChol]⁺[NTf₂]⁻ and [C₈ramChol]⁺[NTf₂]⁻ saturated with water were also evaluated for silver recovery from aqueous nitrate solutions. Investigation of Ag(I) extraction as a function of pH showed a maximum of extraction efficiency at pH 5 (98.6%) for [C₈linChol]⁺[NTf₂]⁻ and good extraction selectivity of Ag(I) and Cu(II) towards Fe(III). High stripping efficiency was achieved by using 0.44 mol L⁻¹ nitric acid.

---

Introduction

Ionic liquids (ILs) are at the center of attention since they exhibit unique properties and promising applicability as solvent media in the fields of catalysis,¹ biomass processing,² petroleum refining,³ and electrochemistry,^4–6 and they can be used as magnetic fluids⁵ and energetic materials.^7,8 Their incomparable physical characteristics such as low volatility, low melting point, large electrochemical window and thermal stability have opened unlimited opportunities to scientists from all branches for further exploration. However, high cost, high viscosity, high water solubility and toxicity of ILs narrow down the fields of applicability in the industrial processes and, therefore, stimulate the research of new ILs. Over the past few decades, many new ILs have been synthesized and studied including, among others imidazolium, pyridinium, pyrrolidinium, piperidinium, phosphonium,^9–20 ammonium²¹ and more recently triazolium.²²

Cholinium salts, a quaternary ammonium cation containing an alkoxyl chain, have recently received attention as a new class of ILs because choline is believed to be environmentally benign and biodegradable.²³ However, previous studies showed that toxicity of cholinium-based ILs must be drawn carefully since toxicity depends on both the choice of the anion and the size/functionality of the alkyl side chains of the cation.^24,25

In this work, two novel cholinium ILs, i.e. N,N-dimethyl-N-octylammonium bis(trifluoromethanesulfonyl)imide ([C₈linChol]⁺[NTf2]⁻) and N-(2-hydroxyethyl)-N-(2-ethylhexyl)-N,N-dimethylammonium bis(trifluoromethanesulfonyl)imide ([C₈ramChol]⁺ [NTf₂]⁻) (Fig. 1) have been synthesized. Viscosity, density and thermal properties of these ILs have been measured in order to characterize their physicochemical properties and to calculate thermodynamic functions such as activation energy of the viscous flow, molar volume, molar entropy, lattice potential energy and isobaric expansion coefficients.

Fig. 1 Chemical structures of ILs synthesized in this work; (a) N-(2-hydroxyethyl)-N,N-dimethyl-N-octylammonium cation [C₈linChol]⁺, (b) N-(2-hydroxyethyl)-N-(2-ethylhexyl)-N,N-dimethylammonium cation [C₈ramChol]⁺ and (c) bis(trifluoromethanesulfonyl)imide anion [NTf₂]⁻.

Afterwards, extraction properties of these ILs towards silver, copper and iron have been investigated for the first time by measuring distribution ratios of Ag(I), Cu(II) and Fe(III) between nitrate aqueous media and the two ILs. To the best of our knowledge, there is no paper concerning the use of cholinium ILs in liquid–liquid extraction of metals. Only few studies concern the use of this class of IL in liquid–liquid extraction of organic compounds.^26–30 Finally, extraction equilibria is discussed based on metal distribution ratios and spectroscopic data including ¹H NMR, ¹³C NMR and ATR-FTIR.

Experimental section

Synthesis of ILs

All reagents were purchased from Fluka, Aldrich, Solvionic, VWR and TCI and were used as received.

Syntheses of cholinium ILs were performed by using a procedure adapted from a methodology previously described by Domańska et al.:³¹ (i) synthesis of halide precursor and (ii) synthesis of ([C₈linChol]⁺[NTf₂]⁻) and [C₈ramChol]⁺[NTf₂]⁻ by anionic metathesis. The overall synthesis reaction is given in Scheme 1.

Scheme 1 Synthetic pathway of cholinium derivated IL.

Water content, thermal properties, viscosity and density at 298.15 K for both ILs synthesized in the present paper are reported in Table 1. It is interesting to highlight that these ILs exhibit viscosity ten times lower than those reported for phosphonium ILs such as Cyphos IL 101.¹⁸

Table 1 Molecular weight, dynamic viscosity and density at 298.15 K, glass transition temperature (T_g) and decomposition temperature (T_d), and water content of synthesized ILs

IL	Molecular weight (g mol^−1)	Viscosity (mPa s)	Density (kg m^−3)	Tg (K)	Td (K)	Water content (mmol L^−1)
[C8linChol]^+[NTf2]^−	306	236.3	1303.97	194	447	4.3
[C8ramChol]^+[NTf2]^−	306	322.3	1320.68	195	498	16.6

---

N-(2-hydroxyethyl)-N,N-dimethyl-N-octylammonium chloride. A mixture of 2-dimethylaminoethanol (Fluka, purity ≥98%) (17.83 g, 0.2 mol) and 1-chlorooctane (Aldrich, purity = 99%) (32.7 g, 0.22 mol) in a stoichiometric ratio plus 5% for the halide was prepared and diluted in toluene (VWR, purity = 100%) (60 mL). The reaction mixture was then stirred and heated at 403.15 K for 14 h. After cooling at room temperature, the solution was filtered. The residual solvent was removed under reduced pressure. The crude product was recrystallized from a cyclohexane (VWR, purity ≤100%) + 5% ethanol (VWR, purity = 70%) mixture. The product was filtered and dried under vacuum. The resulting N-(2-hydroxyethyl)-N,N-dimethyl-N-octylammonium chloride ([C₈linChol]⁺[Cl]⁻) was obtained as a white hygroscopic solid with an average yield of 47%.

¹H NMR (CDCl₃, 400 MHz): δ (ppm) = 0.86 (t, J = 6.4 Hz, 3H, CH₃), 1.30 (m, 10H, CH₂), 1.74 (m, 2H, NCH₂CH₂), 3.36 (s, 6H, CH₃N), 3.53 (m, 2H, NCH₂), 3.70 (m, 2H, NCH₂CH₂OH), 4.11 (m, 2H, CH₂OH), 4.63 (s, 1H, OH).

N-(2-hydroxyethyl)-N,N-dimethyl-N-octylammonium bis(trifluoromethanesulfonyl)imide. To a solution of the IL [C₈linChol]⁺[Cl]⁻ (35.67 g, 0.15 mol) in water (75 mL), lithium bis(trifluoromethanesulfonyl)imide [LiNTf₂] (Solvionic, purity = 99.9%) (45.86 g, 0.16 mol) was added. The mixture was vigorously stirred at room temperature overnight. After the stirring process was stopped, an IL lower phase was clearly separated from the water phase. The IL phase was separated and dissolved in 30 mL of CH₂Cl₂ (VWR, purity = 99%). The organic phase was washed with water till the AgNO₃ test was negative. The organic phase was then dried with MgSO₄, and filtered. Afterwards, CH₂Cl₂ was eliminated under reduced pressure, the product was lyophilized (CHRIST® Alpha 1-2 LD Plus) and dried under vacuum with P₂O₅ to reduce water content and determinate physicochemical properties. [C₈linChol]⁺[NTf₂]⁻ IL was a yellowish liquid obtained with an average yield of 75%.

¹H NMR (D₂O, 400 MHz): δ (ppm) = 0.86 (t, J = 6.9 Hz, 3H, CH₃), 1.20–1.35 (m, 10H, CH₂), 1.71 (m, 2H, NCH₂CH₂) 3.06 (s, 6H, CH₃), 3.29 (m, 2H, NCH₂), 3.40 (m, HOCH₂CH₂N); 3.45 (s, 1H, OH) 3.97 (m, 2H, CH₂OH).

¹³C NMR (CDCl₃, 100 MHz): δ (ppm) = 9.23, 13.10, 22.35, 25.69, 27.64, 32.43, 33.44, 50.60, 55.96, 65.75, 70.95, 114.89, 118.08, 121.26, 124.45.

ESI-MS: cation m/z (%): 202 (M⁺, 100), 187 ((M − CH₃)⁺, 4), 117 ((M–CH₃–(CH₂)₅)⁺, 8). Anion m/z (%): 280 (M⁻, 100), 211 ((M − CF₃)⁻, 22), 146 ((M-SO₂CF₃)⁻, 97).

IR-ATR (cm⁻¹): 3534, 2961, 2933, 2874, 2862, 1481, 1468, 1347, 1328, 1224, 1182, 1134, 1053, 972, 928, 790, 764, 741, 654.

N-(2-hydroxyethyl)-N-(2-ethylhexyl)-N,N-dimethylammonium bromide. The procedure for the synthesis of the ramified product was similar to the one previously described for the linear product adding 5% excess of the 2-ethylhexylbromide (TCI, purity ≥95%) (0.22 mol, 38 mL) to the 2-dimethylaminoethanol (Fluka, purity ≥98%) (0.2 mol, 19 mL) in toluene (VWR, purity = 100%) (60 mL). After heating to 403.15 K for 20 hours, the crude product N-(2-hydroxyethyl)-N-(2-ethylhexyl)-N,N-dimethylammonium bromide [C₈ramChol]⁺[Br]⁻ was then filtered, dried, recrystallized and dried again under vacuum following the already described procedure. The pure bromide IL was an orangish solid obtained according to the same purification steps with an average yield of 67%.

¹H NMR (CDCl₃, 400 MHz): δ (ppm) = 0.92 (t, J = 6.8 Hz, 3H, CH₃), 0.97 (t, J = 7.4 Hz, 3H, CHCH₂CH₃), 1.34 (m, 4H, CH₂), 1.44–1.60 (m, 4H, CH₂), 1.83 (m, 1H, CH), 3.38 (s, 6H, NCH₃), 3.48–3.54 (m, 2H, NCH₂) 3.81 (m, 2H, NCH₂CH₂OH), 4.17 (m, 2H, CH₂OH), 4.39 (s, 1H, OH).

N-(2-hydroxyethyl)-N-(2-ethylhexyl)-N,N-dimethylammonium bis(trifluoromethanesulfonyl)imide. The [C₈ramChol]⁺[Br]⁻ (0.1 mol, 28.23 g) dissolved in water (40 mL) was added to a solution of lithium bis(trifluoromethanesulfonyl)imide (Solvionic, purity = 99.9%) (0.105 mol, 30.14 g) in water (20 mL). The mixture was vigorously stirred overnight after which, as previously, two phases were observed. The IL phase was separated and purified as previously described to afford the [C₈ramChol]⁺[NTf₂]⁻ IL as an orange liquid with an average yield for the metathesis of 74%.

¹H NMR (D₂O, 400 MHz): δ (ppm) = 0.84 (m, 6H, CH₃), 1.24 (m, 4H, CH₂), 1.36–1.42 (m, 8H, (CH₂)₃CH₃ and CH₃CH₂CH), 1.78 (m, 1H, CH), 3.03 (s, 6H, NCH₃), 3.19 (m, 2H, NCH₂), 3.39 (m, 3H, HOCH₂CH₂N and OH), 3.95 (m, 2H, CH₂OH).

¹³C NMR (CDCl₃, 100 MHz): δ (ppm) = 15.39, 22.15, 22.24, 25.72, 28.56, 28.70, 31.37, 50.99, 55.89, 65.76, 114.94, 118.13, 121.31, 125.34.

ESI-MS: cation m/z (%): 202 (M⁺, 100), 187 ((M − CH₃)⁺, 6), 158 ((M–CH₃–(CH₂)₂)⁺, 14). Anion m/z (%): 280 (M⁻, 100), 211 ((M − CF₃)⁻, 18), 146 ((M-SO₂CF₃)⁻, 99).

IR-ATR (cm⁻¹): 3535, 2958, 2931, 2875, 2860, 1482, 1470, 1420, 1345, 1329, 1225, 1182, 1134, 1053, 972, 925, 791, 764, 741, 654.

ILs characterization

¹H and ¹³C NMR were recorded on a Bruker Spectrospin 400 spectrometer using CDCl₃ or an external insert of D₂O. Infrared spectra were registered using a Thermo Scientific FT-IR Nicolet 6700 spectrometer equipped with an ATR on an horizontal ZnSe crystal coated with diamond (A = 2.54 mm²) with a single reflection and an angle of incidence of 45° (Smart Miracle®) at a 28% of power. ESI-MS spectra were recorded with an Agilent 1100 Series – API 3000 Triple Quadrupole spectrometer.

Differential scanning calorimetry was performed with a Mettler Toledo DSC822e model. Heat flow measurements were carried out by cooling the sample from 298.15 K to 153.15 K and backward by heating from 153.15 K to 298.15 K and then up to 573.15 K at 3.5 K min⁻¹. Water contents in ILs were determined by using a Mettler Toledo V20 volumetric Karl Fisher titrator filled with Hydranal 5 Composite Reagent.

Measurement techniques

Density, viscosity and water content. Density measurements were performed with an Anton Paar (DSA 5000M) in the temperature range of 293.15 to 323.15 K at atmospheric pressure. A U-shaped tube placed in a metallic block was filled through the inlet port with each solution in a bubble free condition. The temperature in the cell was controlled precisely with a Peltier device (accuracy < 10⁻³ K). Prior to measurements, the internal calibration was verified by measuring with air and ultrapure water.

Viscosity was determined with an Anton Paar Automated Micro Viscometer AMVn instrument in the range of 293.15 to 323.15 K by using the falling ball model. The system used was a glass capillary of 3 mm inner diameter and a 2.5 mm diameter steel ball. Temperature was controlled with a Peltier device (accuracy <0.05 K). The system used was first calibrated with reference Cannon Instrument Company standard oil N44 provided by Anton Paar.

Liquid–liquid extraction. Silver nitrate (AgNO₃ VWR International, purity ≥99.8%), iron nitrate (Fe(NO₃)₃·9H₂O, Sigma-Aldrich, purity ≥98%) and copper nitrate (Cu(NO₃)₂·3H₂O, Prolabo purity ≥99%) were used to prepare aqueous phases.

Before liquid–liquid extraction of metal ions, ILs were pre-equilibrated three times with water in order to saturate them with water. Afterwards, ILs were contacted at 298.15 K with nitrate aqueous solutions containing 1000 mg L⁻¹ metals (silver, copper or iron). The phase volume ratio between aqueous phase and IL was equal to V_aq/V_IL = 1. The biphasic system was shaken with a Thermoshake shaker (Gerhardt) at 90 rpm for 90 minutes (60 minutes were required to reach constant value of the extraction efficiency vs. time).

Extraction efficiency is defined as:

where [M]_ini,aq, [M]_eq,aq, and [M]_eq,IL denote metal concentrations in aqueous and IL phases initially and at the equilibrium, respectively. V_IL and V_aq represent volume of aqueous and IL phases, respectively. No significant volume variation was observed before and after silver extraction as ILs were contacted with water and nitric acid solutions before silver extraction.

In order to study the influence of pH on metal extraction efficiency, pH value of each aqueous phase was adjusted with nitric acid or sodium hydroxide. Stripping experiments were performed by contacting the metal-loaded IL phase with 0.44 mol L⁻¹ HNO₃. IL phase and stripping aqueous phase solutions were shaken under the same conditions as during extraction experiments. Aqueous and IL phases were then settled and residual metal concentrations in aqueous phase were determined by ICP-OES (ICAP 6000 Series, Thermo Scientific) at 328.068 nm for silver, 259.940 nm for iron and 324.754 nm for copper. Metal ion concentration transferred from the aqueous phase into the IL was then deduced by mass balance. Finally, stripping efficiency (%S_eff) was calculated by using the following equation:

Water uptake in each stage of the extraction process was determined in IL by Karl-Fisher titration. Weight measurements were realized by using an AG 285 Mettler-Toledo balance.

Results and discussion

Physicochemical properties of synthesized ILs

Viscosity. Viscosity in an important intensive property for any IL since it can define its potential applications. Dynamic viscosity for the synthesized ILs was determined in the range of 293.15 to 323.15 K (Table 2). These results show that an increase of viscosity occurs when linear cation is replaced by ramified cation probably due to an increase of IL bulkiness and van der Waals interactions.

Table 2 Dynamic viscosity (η) of [C₈linChol]⁺[NTf₂]⁻ and [C₈ramChol]⁺[NTf₂]⁻ vs. temperature

Temperature (K)	[C8linChol]^+[NTf2]^−	[C8ramChol]^+[NTf2]^−
	η (mPa s)	η (mPa s)
293.15	330.3	453.1
298.15	236.3	322.3
303.15	178.0	231.1
308.15	136.3	171.7
313.15	105.7	130.6
318.15	80.6	99.7
323.15	63.8	76.9

---

Viscosity measurements as a function of temperature can be used to calculate activation energy for the viscous flow with the Arrhenius equation that is generally used to describe common liquid properties:³²

where E_a,η is the activation energy for the viscous flow (in J mol⁻¹), η is the dynamic viscosity (in mPa s), T is the temperature (in K), R is the universal gas constant (in J K⁻¹ mol⁻¹), and η^∞ is the viscosity at infinite temperature.

E_a,η values calculated from the slope of ln(η) vs. 1/T for [C₈linChol]⁺[NTf₂]⁻ and [C₈ramChol]⁺[NTf₂]⁻ are equal to 42.8 kJ mol⁻¹ and 46.4 kJ mol⁻¹, respectively (Fig. 2). These values are much greater than those reported in other works for compounds such as imidazolium or lactam based ILs^32–34 and close to activation energy of phosphonium ILs (for instance, E_a,η = 34.89 kJ mol⁻¹ for hexyl(tetradecyl)phosphonium chloride).¹⁷ As the activation energy for the viscous flow is greater, ions movement is more difficult in cholinium-based ILs due to stronger interactions in solution. A small increase of anion–cation interactions is observed when alkyl chains of the cation is ramified.

Fig. 2 ln(η) vs. 1/T for [C₈linChol]⁺[NTf₂]⁻ (□) and [C₈ramChol]⁺[NTf₂]⁻ (○) with linear regression according to eqn (3), insert: experimental viscosity vs. temperature.

Density. Experimental densities were evaluated in the temperature range of 293.15 to 323.15 K and at atmospheric pressure for the two ILs (Fig. 3). As expected, density decreases linearly with an increase in temperature. It is also clearly observed that density increases when the linear cation is substituted with the branched chain. Given the structural similarities between the two ILs, an increase in density can only occur when a better arrangement of the ions in the liquid takes place thus allowing a greater number of ions to be located in a volume unit.

Fig. 3 Temperature dependence of density for the studied ILs. □: [C₈linChol]⁺[NTf₂]⁻ ○: [C₈ramChol]⁺[NTf₂]⁻. Straight lines: calculated density obtained from eqn (6).

Hence, we considered that the ramification in alkyl chains increases the ions interactions and thus, promotes an increase in density. The experimental density can be represented by the following linear equation:

ρ = A + BT (4)

where ρ is the density (in kg m⁻³) and A and B are adjusting parameters calculated from experimental data with a 95% confidence level: (A = 1549.9 kg m⁻³; B = −0.8248 K⁻¹) for [C₈linChol]⁺[NTf₂]⁻ and (A = 1567.6 kg m⁻³; B = −0.8242 K⁻¹) for [C₈ramChol]⁺[NTf₂]⁻ (R² = 0.9999993 for the linear IL and 0.999996 for the ramified IL).

As it was stated, density data was used to calculate the molar volume of each IL by means of the following equation:

where V_m is the molar volume (in m³ mol⁻¹), M is the molar weight of the IL (in kg mol⁻¹). The calculated V_m values of the ILs are listed in Table 3. It can be observed that molar volumes of the linear IL were approximately 4.7 cm³ mol⁻¹ greater than those of the ramified IL whatever the temperature.

Table 3 Density (ρ), molar volume (V_m) and isobaric expansion coefficient (α_P) for [C₈linChol]⁺[NTf₂]⁻ and [C₈ramChol]⁺ [NTf₂]⁻ at different temperatures

Temperature (K)	[C8linChol]^+[NTf2]^−			[C8ramChol]^+[NTf2]^−
	ρ (kg m^−3)	Vm (cm^3 mol^−1)	αP × 10^4 (K^−1)	ρ (kg m^−3)	Vm (cm^3 mol^−1)	αP × 10^4 (K^−1)
293.15	1308.11	368.04	6.45	1324.84	363.39	6.31
298.15	1303.97	369.27	6.52	1320.68	364.60	6.33
303.15	1299.84	370.50	6.58	1316.54	365.82	6.35
308.15	1295.70	371.73	6.64	1312.40	367.03	6.37
313.15	1291.58	372.96	6.70	1308.26	368.25	6.39
318.15	1287.46	374.19	6.77	1304.12	369.46	6.41
323.15	1283.38	375.42	6.83	1299.99	370.67	6.43

---

It is interesting to be able to predict density and molar volume for conditions differing from those used experimentally. In a previous work, Ye and Shreeve³⁵ developed an additive model to predict density though it was restricted to ambient temperatures and pressures.

Based on that model, Gardas and Coutinho³⁶ proposed an expansion to relate the density of an IL with its molar volume in a wide range of temperatures (273.15–393.15 K) and pressures (0.10–100 MPa). This last model was used to calculate density in the experimentally tested range of temperatures in order to establish the reliability of the model from which it was possible to extrapolate density values at different conditions. The model equation is represented as:

where ρ_calc is the calculated density (in kg m⁻³), V_m is the estimated molar volume (in m³ mol⁻¹) at the reference temperature (T_ref in K) and pressure (P_ref in MPa), T is the temperature in K and P is the pressure (in MPa). The coefficient values for a, b and c were found from optimized literature data, i.e. a = 0.8005 ± 0.0002, b = (6.6520 ± 0.0069) × 10⁻⁴ K⁻¹ and c = (−5.919 ± 0.024) × 10⁻⁴ MPa⁻¹.^36–39

Fig. 3 shows a good agreement between experimental values of densities and those calculated from eqn (6). Molar volume can be deduced by rewriting eqn (6) as follows:

V_m,calc(T, P) = (a + bT +cP)V_m(T_ref, P_ref) (7)

A very good agreement is found between calculated densities or molar volumes and experimental values in the range of 293.15 K to 323.15 K at atmospheric pressure for both ILs.

Fig. 4 compares the molar volumes of [C₈linChol]⁺[NTf₂]⁻ and [C₈ramChol]⁺[NTf₂]⁻ as a function of temperature with those calculated with eqn (7) by taking V_m,ref (298.15 K, 0.1013 MPa) = 369.27 cm³ mol⁻¹ and 364.60 cm³ mol⁻¹ for [C₈linChol]⁺[NTf₂]⁻ and [C₈ramChol]⁺[NTf₂]⁻, respectively.

Fig. 4 Molar volumes calculated by eqn (7) and compared with experimental values for [C₈linChol]⁺[NTf₂]⁻ (□) and [C₈ramChol]⁺[NTf₂]⁻ (○).

The agreement between experimental and calculated values confirms that eqn (7) proposed by Gardas and Coutinho can be used to calculate molar volumes of cholinium ILs with a good accuracy and possibly be used to extrapolate values for specific conditions outside from the present range of temperature and pressure.

Density values were also employed to calculate the isobaric expansion coefficient, α_P, according to eqn (8).

Table 3 summarizes experimental density values, calculated molar volumes and expansion coefficients for both ILs. In the temperature range of 293.15 to 323.15 K, isobaric expansion coefficient value falls in small ranges from 6.45 10⁻⁴ to 6.83 10⁻⁴ K⁻¹ for [C₈linChol]⁺[NTf₂]⁻ and from 6.31 10⁻⁴ to 6.43 10⁻⁴ K⁻¹ for [C₈ramChol]⁺[NTf₂]⁻ which are lower than the values reported for other ILs such as imidazolium but bigger than the values reported for water.^25,33,34,40 The temperature dependence of the isobaric expansion coefficient is very small, with a negative slope within the studied temperature range. This behavior is not anomalous since other ILs also present α_P values with very modest temperature-dependences.⁴⁰ As it can be seen in Table 3, branching of the alkyl chain of cholinium cation decreases α_P values as compared to the linear IL. It was possible to conclude that α_P was less temperature-sensitive for [C₈ramChol]⁺[NTf₂]⁻ than [C₈linChol]⁺[NTf₂]⁻ possibly because branching increased short-range interactions.

A linear equation relating standard entropy and molecular volume can be applicable to ILs since the equation of Glasser considers the Coulomb interactions as the main contributors to the lattice energy. Hence, it can be considered that ILs consisting of a large organic cation have the average properties of an ionic solid and an organic liquid. The equation of Glasser is expressed as:⁴¹

S^° ≈ 1246.5V_molec + 29.5 (9)

where S° (in J K⁻¹ mol⁻¹) is the standard entropy for ILs in liquid state at ambient temperature (298 K) and V_molec (in nm³ per molecule) is the molecular volume.

Standard entropies are equal to 793.8 J K⁻¹ mol⁻¹ for [C₈linChol]⁺[NTf₂]⁻ and 784.2 J K⁻¹ mol⁻¹ for [C₈ramChol]⁺[NTf₂]⁻. Entropy values over the range of temperatures experimentally tested are shown in Table 4.

Table 4 Standard entropy (S°) and lattice energy (U_POT) of [C₈linChol]⁺[NTf₂]⁻ and [C₈ramChol]⁺[NTf₂]⁻ at different temperatures

Temperature (K)	[C8linChol]^+[NTf2]^−		[C8ramChol]^+ [NTf2]^−
	S° J K^−1 mol^−1	UPOT (kJ mol^−1)	S° J K^−1 mol^−1	UPOT (kJ mol^−1)
293.15	791.3	321.6	781.7	323.0
298.15	793.9	321.0	784.2	322.7
303.15	796.4	320.9	786.7	322.3
308.15	798.9	320.2	789.2	322.0
313.15	801.5	320.3	791.7	321.7
318.15	804.0	320.0	794.2	321.3
323.15	806.6	319.6	796.7	321.0

---

The slight decrease of standard entropy due to the presence of branching in the cation of cholinium IL suggests that branching increases slightly IL organization. Conversely, a significant reduction of IL organization was observed by Panda et al.³⁴ when the length of alkyl chains in cations of IL such as pyrrolidinium increases.³⁴

Finally, lattice potential energy (U_POT) was also calculated from density values by the following Glasser equation⁴¹ (Table 4):

where γ (in kJ mmol^−2/3) and δ (in kJ mol⁻¹) are fitted coefficients of the salts, ρ is the density (in kg m⁻³), and M is the molecular weight (in kg mol⁻¹). Glasser suggested that, as the density of the condensed melt and solid phases are similar, the calculated lattice potential can be applied to both phases (the constants given as γ = 1981.2 kJ mmol^−2/3 and δ = 103.8 kJ mol⁻¹). The lattice energies of ILs calculated from the above eqn (10) are listed in Table 4. The calculated potential energies were found to be lower than the ones of other fused salts as CsI (613 kJ mol⁻¹)⁴² which explains the liquid state of both cholinium based ILs at lower temperature. As it has been previously reported^34,42,43 the lattice energy decreases with the increase in the carbon chain length, which is attributed to a lower packing efficiency. In our case, the lattice energy for the ramified IL was slightly greater than for the linear IL, which is consistent with previous conclusions proposed in the calculation of density and entropy.

In structural terms, the ramified IL presents a more compact arrangement between the alkyl chains, which explains the increase in density and viscosity and the lower molar volume and entropy when compared to the linear IL.

Liquid–liquid extraction of silver from nitrate media

Before Ag(I) extraction, ILs were preequilibrated three times with water and one time with 0.44 mol L⁻¹ HNO₃ at phase volume ratio V_aq/V_IL = 1. Table 5 shows that both ILs extract about 1.8 mol L⁻¹ of water. Such a concentration of water in ILs significantly changes physicochemical properties of cholinium ILs as well as their organization (vide infra). Therefore, metal extraction by ILs cannot be directly connected with interactions in solution deduced from physicochemical measurements in neat ILs as those reported previously in the present work.

Table 5 Water concentration in [C₈linChol]⁺[NTf₂]⁻ and [C₈ramChol]⁺[NTf₂]⁻ after being contacted with water and nitric acid (0.44 mol L⁻¹) at phase volume ratio V_aq/V_IL = 1

Ionic liquid (IL)	[H2O] after 1st contact with water (mol L^−1)	[H2O] after 3rd contact with water (mol L^−1)	[H2O] after contact with HNO3 (mol L^−1)
[C8linChol]^+[NTf2]^−	1.90	1.89	1.80
[C8ramChol]^+[NTf2]^−	1.79	1.65	1.82

---

Extraction efficiency of Ag(I) from nitrate aqueous solutions has been investigated as a function of pH for both ILs (Fig. 5). It is clear that pH-dependence of silver extraction efficiency showed that ion exchange between proton and Ag(I) might be involved in the extraction process.

Fig. 5 Extraction efficiency of silver (% E_eff) from nitrate media by [C₈linChol]⁺[NTf₂]⁻ (□) and [C₈ramChol]⁺[NTf₂]⁻ (○) as a function of pH at equilibrium. Phase volume ratio V_aq/V_IL = 1; T = 298.15 K: initial silver concentration = 9.3 10⁻³ mol L⁻¹.

Fig. 5 shows that ramification in the alkyl chain of the cholinium cation was responsible for a decrease of extraction efficiency. It could then be inferred that cholinium cation is involved in silver extraction since extraction efficiency is different depending on the nature of the cholinium cation. This difference in extraction properties might come mainly from the increase of steric hindrance when [C₈linChol]⁺[NTf₂]⁻ is replaced by [C₈ramChol]⁺[NTf₂]⁻. Likely, previous works have shown that steric hindrance of extracting agents influences strongly their extraction properties.⁴⁴

However, it cannot be precluded that the difference of Ag(I) extraction by the two ILs is not connected with any change of IL organization due to other phenomena such as water coextraction (Table 5). In the last few years, interactions between water and ILs and the role of water on the nanostructural organization of ILs have been studied from both the experimental and the theoretical point of view.⁴⁵ The main conclusion, is that water molecules are preferentially hydrogen bonded to the ILs anions, even at high water concentration.⁴⁶ When increasing the water content, the polar network progressively collapses and water aggregates are formed, leading to a well-defined hydrogen-bonded network.⁴⁷ A recent paper concerning cholinium ILs aqueous solutions showed that hydrogen bonding is the driving force in these systems.⁴⁸ In particular, ¹H NMR studies showed that both cation and anion interact with water in the case of cholinium carboxylate ILs, but the anion counter-part presents strong interactions which may lead to the formation of domains or hydrophobic clusters.³⁰ Therefore, the difference in water content between [C₈linChol]⁺[NTf₂]⁻ and [C₈ramChol]⁺[NTf₂]⁻ may also play a role regarding the difference in extraction behavior towards Ag(I) reported in Fig. 5 when [C₈linChol]⁺[NTf₂]⁻ is used instead of [C₈ramChol]⁺[NTf₂]⁻.

Fig. 6 shows that water extraction is different for [C₈linChol]⁺[NTf₂]⁻ and [C₈ramChol]⁺[NTf₂] since the variation of water extraction in the range of pH 0.5 to 4.5 was less pronounced for the linear IL than for the ramified IL.

Fig. 6 Water concentration in ILs vs. pH at equilibrium after contacting [C₈linChol]⁺[NTf₂]⁻ (□) or [C₈ramChol]⁺[NTf₂]⁻ (○) with an aqueous phase containing initially 9.3 10⁻³ mol L⁻¹ silver. Phase volume ratio V_aq/V_IL = 1; T = 298.15 K.

The variation of water extraction vs. pH cannot be explained by the salting-out effect as ionic strength in the aqueous phase was kept constant (I = 0.44 mol L⁻¹) but rather by a change in ILs structuration due to proton co-extraction.

Proton extraction has been investigated by ¹H NMR during silver extraction. Fig. 7 displays ¹H NMR spectra of [C₈linChol]⁺[NTf₂]⁻ and [C₈ramChol]⁺[NTf₂]⁻ before any contact, after being contacted with water, with nitrate aqueous solution and after being contacted with nitrate aqueous solution containing silver at equilibrium pH 0.3 and 1.2. A peak located at 3.70 ppm appears when ILs is contacted with nitrate aqueous solution at equilibrium pH = 0.3. This peak does not appear at equilibrium pH = 1.2. Such a peak at 3.70 ppm might be attributed to the formation of HNTf₂ since pK_a of HNTf₂/NTf₂⁻ in aqueous phase is 1.2 (ref. 49) (surprisingly, such an observation suggests that pK_a value in these IL is close to that reported in aqueous solutions).

Fig. 7 ¹H NMR spectra of [C₈linChol]⁺[NTf₂]⁻ and [C₈ramChol]⁺[NTf₂]⁻. Black: ILs before any contact with aqueous phase; red: ILs after water pre-equilibrium (equilibrium pH = 8.6 for the linear and 7.7 for the ramified); green: ILs after being contacted with nitrate aqueous media (equilibrium pH = 0.3 and 1.2, respectively); blue: ILs after being contacted with nitrate aqueous media containing silver (equilibrium pH = 0.3 and 1.2, respectively). Signal for D₂O at 4.7 ppm.

Water and nitric acid were coextracted with silver.

{[C₈RChol]⁺[NTf₂]⁻}_IL + nH₂O → {[C₈RChol]⁺(nH₂O)[NTf₂]⁻}_IL (11)

Nitric acid extraction might occur according to the following reaction:

{[C₈RChol]⁺[NTf₂]⁻}_IL + HNO₃ → {[C₈RChol]⁺[NO₃]⁻}_IL + {HNTf₂}_IL (12)

where R = lin or ram for the cholinium cation containing a linear or ramified alkyl chain, respectively. Subscript “IL” indicates that the species are solubilized in IL.

Kurnia⁴⁹ determined the mutual solubility of water and ILs, methyl(trioctyl)ammonium [N₈₈₈₁]⁺and methyl(trioctyl)phosphonium [P₈₈₈₁]⁺ in the presence of NTf₂⁻ anion by means of conductivity measurements. They deduced that the solubility of these ILs in water is very low. The mole fraction solubility of these ILs in water was estimated to range between 10⁻³ to 10⁻⁴, i.e. between 6 10⁻² to 6 10⁻³ mol L⁻¹.

By comparing solubility values of ammonium and phosphonium ILs and initial silver concentration in aqueous phase (9.3 10⁻³ mol L⁻¹), it cannot be precluded that silver extraction may occur, at least partially, by the following cation exchange equilibrium as reported for strontium extraction:⁴⁶

{[C₈RChol]⁺[NTf₂]⁻}_IL + Ag⁺ → [C₈RChol]⁺ + {AgNTf₂}_IL (13)

The pH-dependence of extraction efficiency observed in Fig. 5 evidences that proton-silver exchange may also occur during silver extraction:

{[C₈RChol]⁺[NO₃]⁻}_IL + {HNTf₂}_IL + Ag⁺ → {[C₈RChol]⁺[NO₃]⁻ + AgNTf₂}_IL + H⁺ (14)

At pH greater than the pK_a value of HNTf₂, which is close to 1.2, silver might be extracted by the following reaction as well:

{[C₈RChol]⁺[NTf₂]⁻}_IL + AgNO₃ → {[C₈RChol]⁺[NO₃]⁻ + [AgNTf₂]}_IL (15)

The influence of phase volume ratio on extraction efficiency of silver by [C₈linChol]⁺[NTf₂]⁻ and [C₈ramChol]⁺[NTf₂]⁻ at 298.15 K was investigated: 96.2% of extraction is achieved for the linear cholinium IL and 68.8% of Ag(I) is extracted with the ramified IL at a V_aq/V_IL = 0.25.

Furthermore, extraction properties of [C₈linChol]⁺[NTf₂]⁻ towards Cu(II) and Fe(III) were investigated in order to assess the extraction selectivity of the linear IL. For this purpose, 9 × 10⁻³ mol L⁻¹ of Ag(I), 0.018 mol L⁻¹ of Fe(III) and 0.016 mol L⁻¹ of Cu(II) were mixed in 0.44 mol L⁻¹ HNO₃ solution. After contacting this aqueous phase with [C₈linChol]⁺[NTf₂]⁻ at a phase volume ratio equal to 1, equilibrium pH was equal to 0.3. Elemental analyses showed that 1.6% of Fe(III), 22.8% of Cu(II) and 16.8% of Ag(I) were extracted under these experimental conditions. Therefore, a good selectivity was achieved towards iron(III) and no selectivity was obtained between Cu(II) and Ag(I). Metal transfer from aqueous phase into ILs can be envisaged as taking place by steps in which the first one involves the freeing of an ion from its hydration before it is permitted to react with cations and anions of ILs in which it is transferred. This freeing requires the investment of work that is the negative of the standard Gibbs energy of hydration of the ion which are equal to 440, 2016 and 4271 kJ mol⁻¹ for Ag(I), Cu(II) and Fe(III), respectively.^50,51 These values are in agreement with the high selectivity observed for the recovery of Ag(I) and Cu(II) towards Fe(III).

Finally, stripping tests were performed by using 0.44 mol L⁻¹ nitric acid at various phase volume ratio. Table 6 shows that 100% of silver stripping efficiency was reached at a phase volume ratio V_aq/V_IL equal to 1 with the [C₈linChol]⁺[NTf₂]⁻. Stripping efficiency of Ag(I) by [C₈ramChol]⁺ [NTf₂]⁻ was achieved at a phase volume ratio of 2.0 (98.96%).

Table 6 Stripping efficiency (%) vs. phase volume ratios of Ag(I) for [C₈linChol]⁺[NTf₂]⁻ and [C₈ramChol]⁺[NTf₂]⁻. Stripping solution = 0.44 mol L⁻¹ HNO₃; T = 298.15 K

Phase volume ratio (Vaq/VIL)	Stripping efficiency (%) [C8linChol]^+[NTf2]^−	Stripping efficiency (%) [C8ramChol]^+[NTf2]^−
0.25	89.26	57.10
0.50	87.57	52.90
1.00	100.0	79.68
2.00	95.78	98.96
5.00	86.39	89.77

---

Conclusions

Cholinium-based ILs were synthesized and structurally characterized by ¹H NMR, ¹³C NMR, ATR-FTIR and ESI-MS. Thermodynamic calculations deduced from density and viscosity measurements, as the entropy or the lattice potential energy, revealed a higher level of arrangement for the ramified IL, whilst the isobaric expansion coefficient showed a lower effect of the temperature for the same IL which was interpreted as a better packing of the ions for [C₈ramChol]⁺[NTf₂]⁻ compared to [C₈linChol]⁺[NTf₂]⁻.

Concerning the extractive properties of the ILs without added extractant, extraction was believed to occur primarily by the effect of NTf₂⁻ anion recovering up to 98.6% of Ag(I) with the linear IL at an equilibrium pH = 5.8 whereas the ramified IL recovered only 40.5% at its maximum level at an equilibrium pH value of 4.4. Stripping was easily undertaken with 0.44 mol L⁻¹ nitric acid almost to a quantitative level (>99%) after two stages of stripping. A good selectivity for Ag(I) extraction towards Fe(III) was found.

Acknowledgements

Rafael Manuel Rios-Vera wishes to acknowledge the National Council of Science and Technology (CONACYT) for the PhD scholarship number 367900/339666.

References

1. A. Brenno, D. Neto and J. Spencerb, J. Braz. Chem. Soc., 2012, 23, 987–1007.
2. S. R. E. Hassan, F. Mutelet and J. C. Moise, RSC Adv., 2013, 3, 20219–20226.
3. C. Shu, T. Sun, H. Zhang, J. Jia and Z. Lou, Fuel, 2014, 121, 72–78.
4. M. C. Buzzeo, R. G. Evans and R. G. Compton, ChemPhysChem, 2004, 5, 1106–1120.
5. D. Liu, H. Groult, L. Gaillon, C. Rizzi, N. Soulmi, C. M. Julien, E. Briot and D. Krulic, J. Solid State Electrochem., 2015 DOI:10.1007/s10008-015-2809-8.
6. V. Lair, J. Sirieix-Plenet, L. Gaillon, C. Rizzi and A. Ringuede, Electrochim. Acta, 2010, 56, 784–789.
7. R. E. Del Sesto, T. M. McCleskey, A. K. Burrell, G. A. Baker, J. D. Thompson, B. L. Scott, J. S. Wilkes and P. Williams, Chem. Commun., 2008, 4, 447–449.
8. S. Le Vot, D. Dambournet, H. Groult, A. T. Ngo, C. Petit, C. Rizzi, C. Salzemann, J. Sirieix-Plenet, O. J. Borkiewicz, E. Raymundo-Piñero and L. Gaillon, Dalton Trans., 2014, 43, 18025–18034.
9. R. P. Singh, R. D. Verma, D. T. Meshri and J. M. Shreeve, Angew. Chem., Int. Ed., 2006, 45, 3584–3601.
10. A. Andresovaa, J. Storchb, M. Traïkiac, Z. Wagnera, M. Bendovaa and P. Hussonc, Fluid Phase Equilib., 2014, 371, 41–49.
11. R. L. Gardas, H. F. Costa, M. G. Freire, P. J. Carvalho, I. M. Marrucho, I. M. A. Fonseca, A. G. M. Ferreira and J. A. P. Coutinho, J. Chem. Eng. Data, 2008, 53, 805–811.
12. M. Soucková, J. Klomfar and J. Pátek, Fluid Phase Equilib., 2012, 333, 38–46.
13. G. Adamova, R. L. Gardas, L. P. N. Rebelo, A. J. Robertsond and K. R. Seddon, Dalton Trans., 2011, 40, 12750–12764.
14. A. Chagnes, H. Allouchi, B. Carré and D. Lemordant, Solid State Ionics, 2005, 176, 1419–1427.
15. A. Chagnes, M. Diaw, B. Carré, P. Willmann and D. Lemordant, J. Power Sources, 2005, 145, 82–88.
16. A. Chagnes, A. Tougui, B. Carré, N. Ranganathan and D. Lemordant, J. Solution Chem., 2004, 33, 247–255.
17. D. Cholico Gonzalez, M. Avila-Rodriguez, J. Antonio Reyes-Aguilera, G. Cote and A. Chagnes, J. Mol. Liq., 2012, 169, 27–32.
18. D. Cholico-Gonzalez, M. Avila-Rodrigues, G. Cote and A. Chagnes, J. Mol. Liq., 2013, 187, 165–170.
19. V. T. Nguyen, J.-C. Lee, J. Jeong, B.-S. Kim, G. Cote and A. Chagnes, Ind. Eng. Chem. Res., 2015, 54, 1350–1358.
20. B. Chamiot, C. Rizzi, L. Gaillon, J. Sirieix-Plenet and J. Lelievre, Langmuir, 2009, 25(3), 1311–1315.
21. K. Machanova, A. Boisset, Z. Sedlakova, M. Anouti, M. Bendova and J. Jacquemin, J. Chem. Eng. Data, 2012, 57, 2227–2235.
22. M. A. Balazadeh and M. Foroutan, Fluid Phase Equilib., 2013, 356, 63–70.
23. M. Petkovic, J. L. Ferguson, H. Q. Nimal Gunaratne, R. Ferreira, M. C. Leita, K. R. Seddon, L. P. N. Rebeloa and C. Silva Pereira, Green Chem., 2010, 12, 643–649.
24. X. Wang, C. A. Ohlin, Q. Lu, Z. Fei, J. Hu and P. J. Dyson, Green Chem., 2007, 9, 1191–1197.
25. A. J. L. Costa, M. R. C. Soromenho, K. Shimizu, I. M. Marrucho, J. M. S. S. Esperança, J. N. Canongia Lopes and L. P. N. Rebelo, ChemPhysChem, 2012, 13, 1902–1909.
26. A. J. L. Costa, M. R. C. Soromenho, K. Shimizu, I. M. Marrucho, J. M. S. S. Esperança, J. N. C. Lopes and L. P. N. Rebelo, J. Phys. Chem. B, 2012, 116, 9186–9195.
27. X. D. Hou, N. Li and M.-H. Zong, ACS Sustainable Chem. Eng., 2013, 1, 519–526.
28. Z. Li, X. Liu, Y. Pei, J. Wang and M. He, Green Chem., 2012, 14, 2941–2950.
29. X. Liu, Z. Li, Y. Pei, H. Wang and J. Wang, J. Chem. Thermodyn., 2013, 60, 1–8.
30. D. J. S. Patinha, L. C. Tomé, H. Garcia, R. Ferreira, C. S. Pereira, L. P. N. Rebelo and I. M. Marrucho, J. Chem. Thermodyn., 2015, 84, 93–100.
31. U. Domańska and R. Bogel-Łukasik, J. Phys. Chem. B, 2005, 109, 12124–12132.
32. A. Chagnes, A. Tougui, B. Carré, N. Ranganathan and D. Lemordant, J. Solution Chem., 2004, 33(3), 247–255.
33. P. K. Chhotaray, S. Jella and R. L. Gardas, J. Chem. Thermodyn., 2014, 74, 255–262.
34. S. Panda and R. L. Gardas, Fluid Phase Equilib., 2015, 386, 65–74.
35. C. Ye and J. M. Shreeve, J. Phys. Chem. A, 2007, 111, 1456–1461.
36. R. L. Gardas and J. A. P. Coutinho, Fluid Phase Equilib., 2008, 263, 26–32.
37. R. L. Gardas, M. G. Freire, P. J. Carvalho, I. M. Marrucho, I. M. A. Fonseca, A. G. M. Ferreira and J. A. P. Coutinho, J. Chem. Eng. Data, 2007, 52, 1881–1888.
38. R. L. Gardas, M. G. Freire, P. J. Carvalho, I. M. Marrucho, I. M. A. Fonseca, A. G. M. Ferreira and J. A. P. Coutinho, J. Chem. Eng. Data, 2007, 52, 80–88.
39. M. Tariq, A. P. Serro, J. L. Mata, B. Saramago, J. M. S. S. Esperança, J. N. C. Lopes and L. P. N. Rebelo, Fluid Phase Equilib., 2010, 294, 157–179.
40. L. Glasser, Thermochim. Acta, 2004, 421, 87–93.
41. A. Xu, J. Wang, Y. Zhang and Q. Chen, Ind. Eng. Chem. Res., 2012, 51, 3458–3465.
42. D. W. Fang, W. Guan, J. Tong, Z. W. Wang and J. Z. Yang, J. Phys. Chem. B, 2008, 112, 7499–7505.
43. D. Beltrami, A. Chagnes, M. Haddad, H. Laureano, H. Mokhtari, B. Courtaud, S. Jugé and G. Cote, Sep. Sci. Technol., 2013, 48, 480–486.
44. R. M. Douglas, and A. F. Stewart, in Ionic Liquids as Green Solvents, American Chemical Society, 2003, vol. 856, ch. 22, pp. 264–276.
45. J. Lopes and A. A. H. Padua, J. Phys. Chem. B, 2006, 110, 3330–3335.
46. A. Stark, A. W. Zidell and M. M. Hoffmann, J. Mol. Liq., 2011, 160, 166–179.
47. C. E. S. Bernardes, M. E. M. da Piedade and J. N. C. Lopes, J. Phys. Chem. B, 2011, 115, 2067–2074.
48. I. Khan, K. A. Kurnia, T. E. Sintra, J. A. Saraiva, S. P. Pinho and J. A. P. Coutinho, Fluid Phase Equilib., 2014, 361, 16–22.
49. K. A. Kurnia, M. V. Quental, L. M. N. B. F. Santos, M. G. Freire and J. A. P. Coutinho, Phys. Chem. Chem. Phys., 2015, 17, 4569–4577.
50. M. L. Dietz and J. A. Dzielawa, Chem. Commun., 2001, 2124–2125.
51. J. Rydberg, M. Cox, C. Musikas and G. R. Choppin, in Solvent Extraction Principles and Practice, Marcel Dekker, 1992, ch. 2, p. 43.

---