A green hydrophobic deep eutectic solvent for extraction of phenol from aqueous phase

Deep eutectic solvents (DESs), have been recognized as effective materials for the extraction of different compounds. In this study, the performance of a novel hydrophobic DES was evaluated for the extraction of phenol from aqueous solutions. Octanoic and dodecanoic fatty acid precursors with a definite molar ratio of 3:1, respectively, were used for the DES having a low melting point of 8.3 °C. The purity and stability of the product were confirmed via characterizing by FTIR, 1H and 13C NMR methods. The liquid–liquid equilibrium of the water + phenol + DES ternary system at different temperatures of 293.2, 298.2 and 308.2 K was accordingly studied through cloud point titration method and refractive index measurement. Interestingly, the important parameters of the solute distribution coefficient and the separation factor were, respectively, within the high levels of (6.8321–9.7787) and (895.76–2770.17), indicating the amazing capability of the DES. Reasonably, both of these parameters decreased with temperature. The NRTL and UNIQUAC thermodynamic models were employed to reproduce the obtained tie-lines and to determine the interaction parameters at each temperature. The low level root mean square deviations for the mentioned models were, respectively, within (0.0014–0.0027) and (0.0045–0.0063); confirming satisfactorily agreement with the experimental data.

properties of non-toxicity and biodegradability 12,13 are preferred as precursors for preparing DESs 14 .It is while, those made of short hydrocarbon chain fatty acids are usually unstable in water 15 .
The desired properties of DESs such as low viscosity and density, flammability and vapor pressure, water immiscibility and stability are remarkable.On the other hand, in order to optimize and simulate the liquid-liquid extraction process, ternary diagrams of the liquid-liquid equilibrium (LLE) must be presented.Here, a novel DES, prepared from octanoic acid (C 8 ) and dodecanoic acid (C 12 ) precursors was prepared and characterized via FTIR spectroscopy as well as 1 H and 13 C NMR analyses.Afterward, the phenol extractability by the proposed DES was scrutinized at different temperatures.Othmer-Tobias, Hand, and Bachman well-known equations were employed to confirm the tie-lines consistency.Using the Aspen Plus simulator, the tie-line data were regressed using the well-known NRTL and UNIQUAC thermodynamic models.For this aim, regarding the composition complexity of the DES; a pioneering novel approach, based on group contribution, was employed for correlating the tie-lines and obtaining the binary interaction parameters.

Experimental Materials
The fatty acid precursors of dodecanoic acid and octanoic acid were purchased from Sigma − Aldrich.Phenol was supplied from Merck and used without additional purification.A water deionizer apparatus (Hastaran, Iran) produced fresh deionized water with conductivity of less than 0.08 μS/cm, utilized for preparing solutions.The chemical names along with other related information are listed in Table 2.

Preparation of DES and solid − liquid phase diagram
Different molar ratios of C 8 and C 12 carboxylic acids were gradually heated in a jacketed glass vessel up to 70 °C while magnetic stirring until reaching a homogeneous and clear solution of single phase 21,24 .The samples were prepared by mass using an Ohaus balance (Adventurer, Pro AV264, Switzerland, uncertainty 0.0001 g).Consequently, temperature was gradually reduced until the first evidences of solid phase was observed at a certain temperature.The corresponding solid − liquid phase diagram of the DES, illustrated in Fig. 1, shows obtaining a specific DES from octanoic and dodecanoic fatty acids with certain molar ratio of 3:1 (0.25 mol fraction of dodecanoic acid).It can be seen that the melting point of the DES (8.3 °C) is much lower than octanoic acid (16.5 °C) and dodecanoic acid (43.2 °C) bringing about a stable liquid state product for operations under conventional temperatures.
The eutectic product was characterized by FTIR (Perkin Elmer FTIR spectrometer, USA) , 1 H and 13 C NMR analysis 16,21 (Varian − Inova 500 MHz NMR, USA) using dimethyl sulfoxide (DMSO-d6) as solvent.The density and refractive index of the prepared DES and water were measured, respectively, with an oscillating densimeter (Anton Paar DMA4500, Austria) with a relative uncertainty of 0.001, and a refractometer (Abbe AR4 Kruss, Germany) with uncertainty of 0.0007.The obtained values and those reported in literatures are compared in Table 3, indicating a close agreement.Ensuring the stability of the DES in aqueous media, 50% volume ratio in water were stirred for 8 h and was left settling overnight.Therefore, both phases were thoroughly in contact and were separated.

Solubility and LLE measurements
The basis of experiments in LLE experiments was the cloud-point titration and refractometry methods.Although various analytical methods have been developed to measure the concentration of components e.g., gas chromatography (GC) and high performance liquid chromatography (HPLC) 29,30 , the most applicable approach to determine the solubility of liquids is cloud point titration, also known as "turbid point titration".The method has advantages of easily operation, high accuracy and cost effective.An extensive discussion about the conventional methods in LLE analysis has been recently reported by Arce et al. 31 .This approach has been used extensively in recent publications 32,33 .Noteworthy, the visual detection of turbidity and end-point of titration, and the requirement of a calibration curve are the limitation of this analytical method.
For analysis, a set up consisted of a miniature jacketed vial with approximated volume of 5 cm 3 , refractometer, thermostat (Julabo, Germany, uncertainty of 0.01 K) and magnetic stirrer were used.A water stream was circulated by a thermostat driving pump through the jacket and refractometer for maintaining temperature at a  www.nature.com/scientificreports/certain value.Temperature of the thermostat was calibrated via a reference thermometer (Amadigit, Germany, uncertainty 0.01 K).Firstly, to determine the calibration curves (relationships between the concentration of each component and the refractive index), the binary mixtures of (water + phenol) or (DES + phenol) with specific amounts were prepared by means of the Ohaus balance.The next step was titrating of the binary mixtures, being stirred at a constant temperature, by precise addition of the remained component (DES or water) via a micro-siring until threshold of a stable cloudy solution.Then, the content was weighed to determine the amount of the third component and refractive index for the cloudy solution was measured immediately.In this way, the calibration curves were attained under atmospheric pressure of 81.5 kPa and different temperatures.In the Supporting materials (Table S1), the calibration curves and related data are presented.It can be seen in Figs.S1 and S2 that the refractive index of cloudy solutions increases with the phenol mass transfer which is attributed to its higher refractive index value.
The equilibrium cells containing different samples with specific and accurate amounts of involved components (water, phenol and DES) were tightly closed.These were agitated in a shaking water bath (N − BIOTEK − 304, South Korea, uncertainty of 0.1 K) at constant temperature and 175 rpm under ambient pressure for 4 h, then the phases were completely separated by resting more than 12 h time.
Collecting the organic phase (top) and aqueous phase (bottom) samples with a syringe, the related refractive indices were determined.It is worth noting that ensuring the accurate experimental data, the analysis of each sample was repeated three times.Using aqueous and organic phase calibration curves and the corresponding refractive indices, the mass fraction of each component with the uncertainty of 0.0002 except u(w 31 ) and u(w 11 ) as 0.0003 was determined.Table 4 presents the tie-line data at different temperatures.
The DES capability for selective separation of phenol was evaluated from separation factor (S) criterion: where D 1 is the water distribution coefficient and D 2 the phenol distribution coefficients obtained from: (1)  The greater than unity values of the separation factor and distribution coefficient confirms the capability of the DES for extraction of phenol.A higher D 2 is desired because of a higher extracted solute.In other words, less solvent is required to achieve a specific extraction.Also, S indicates the separation possibility and the solvent selectivity for the solute extraction 5,34 .

DES characterization
The FTIR spectra of the DES just after preparation and after mixing with water, are presented in Fig. S3.As can be seen, there is no change in DES structure after mixing with water. 1 H and 13 C NMR spectra are also given in Supporting materials (Figs.S4 and S5), respectively.Considering no additional peak after mixing with water, it can be concluded that there was no chemical reaction or dissociation of fatty acids, i.e. stable in contact with water.Also, the lack of the precursor peaks in the FTIR and 1 H NMR of water phase (Fig. S6) confirms again the stability.The summary of the DES analysis data is given in Table 5.

Tie-line results
The tie-line data for the ternary system of water + phenol + DES at different temperatures of 293.2, 298.2 and 308.2K are listed in Table 4. Corresponding phase diagrams are presented in Fig. 2. In general, mass fractions of the ith component of water = 1, phenol = 2 and DES = 3 are presented as w i1 and w i in the aqueous and organic (DES) phases, respectively.Regarding immiscibility of water and DES, partial miscibility of water and phenol as well as DES and phenol, the investigated system is assigned as a type − 2 LLE system 35 .
As illustrated in Fig. 2, the higher solubility of the solute in the DES phase is consistent with the slope of the obtained tie-lines and can be attributed to the potential hydrogen bonding between phenol and the DES 36,37 .With respect to temperature effect, a gradual diminishing of the binary area is observed with increasing temperature.This can be attributed to the little miscibility tend of water and DES and to some extent, decreasing the phenol hydrogen bound with the DES 38 .It is necessary to mention that the effect of temperature is more evident at higher concentrations of the solute.The same results were previously reported for several chemical systems, for instance, water + phenol + imidazolium ionic liquid 36 , water + phenol + cumene 27 and water + acetone + HMIMPF 6 IL 39 .

Distribution coefficients and selectivity criteria
The distribution coefficients are presented in Figs. 3 and 4, and the separation factor in Fig. 5.The phenol distribution coefficient lies within the range (6.8321-9.7787)and the separation factor within (895.76-2770.17).These confirm the amazing capability of the DES for phenol extraction.The comparison of separation factor ranges of DES with several other solvents, examined for separation phenol from water, at a typical temperature of 298.2 K, is provided in Table 6.There is significantly higher separation factor for the DES than ILs and some  www.nature.com/scientificreports/other conventional solvents.Indeed, the higher separation factor of methyl tert-butyl ketone should be regarded with the environmental issues which limits the applications.The separation factor decreases with increasing phenol mass fraction in aqueous phase (Fig. 5).The reason is relevant to the fact that the water content in the organic DES phase increases with solute concentration.On the other hand, increasing temperature leads to a lower separation factors because of weakening the hydrogen www.nature.com/scientificreports/bond between phenol and the DES 38 .Figure 6 presents the phenol equilibrium distribution between the organic (DES) and aqueous (water) phases at different temperatures, showing rather linear variations.

Consistency tests
The reliability of the experimental data was evaluated based on the Othmer − Tobias, Hand and Bachman wellknown correlations as  where A 1 and B 1 , A 2 and B 2 , and A 3 and B 3 are, respectively, intercepts and slopes of the Othmer − Tobias, the Hand and the Bachman correlations 31 .The value of intercepts, slopes and related determination coefficients (R 2 ) are tabulated in Table 7.The close to unity determination coefficient verifies the tie-line data consistency.

Correlation models
The consistency of the attained ternary data against the NRTL and UNIQUAC thermodynamic models was checked by utilizing Aspen Plus (V.8.4) simulator through introducing DES as a pseudo component.Hence, the physico-chemical properties of the DES including normal boiling point, molecular weight and density were introduced to the software.For this aim the method proposed by Mirza et al. 45 was employed.Relevantly, simultaneous utilization of the modified Lydersen − Joback − Reid (LJR) method along with the mixing rules has been proposed for critical properties estimation as well as normal boiling points.The accuracy of the method has been tested for 39 different DESs 46 .In this regard, group contribution method and the type and frequency of the groups of atoms were considered.The contributions are summated to obtain the final estimate of the boiling point as where T b (K), n i and T bMi (K) are, respectively, the normal boiling point of a DES, the frequency of appearance of the ith group of atoms in the molecule, and group contribution in boiling point for the modified LJR method 45 .Based on Eq. ( 7) the DES normal boiling point was obtained as 551.0 K.
In the NRTL thermodynamic model, the non-randomness parameter (α ij ) was considered to be 0.2 and 0.3 47 .In the UNIQUAC model, the r i parameter of the number of segments in each molecule and the q i parameter of the relative surface area per molecule are listed in Table 8 48 .
The corresponding interaction parameters of the both models at different temperatures are listed in Table 9.In a previous study, the proper description of the binary interaction parameters are provided 49 .The LLE phase diagram of ternary system based on the obtained data and those of the predicted values by the models are depicted in Fig. 2. Obviously, the models reasonably fit the experimental data.The comprehensive root mean square deviation (RMSD) values were determined from the following equation for each of the investigated models and listed Table 9.
where w exp imn and w cal imn represent the experimental and the model predicted mass fractions, respectively.Here, the subscripts of i = 1, 2, 3 are for the components and m = I, II are for the aqueous and organic phases.Further, the n = 1, 2, …, N subscript stand for the number of tie-lines.As shown in Table 9, the RMSD values are very low for the NRTL model within (0.0014-0.0027) and for the UNIQUAC within (0.0045-0.0063).Therefore, the more appropriate model for predicting the tie-line data is the NRTL model

Conclusions
Utilizing a novel hydrophobic DES, consisted of octanoic and dodecanoic acid precursors, the extraction of phenol from aqueous phase was feasible at different temperatures.The liquid state DES was stable for satisfactory extraction operations.Liquid-liquid equilibrium experiments were performed based on the cloud point titration method and refractive index measurements.The results revealed that low amounts of the DES was required for a specific task and that the tie-lines were with positive slopes since phenol tends to be more in the organic phase.Relevantly, the distribution coefficient and separation factor confirmed the high capacity and selectivity of the DES compared to the conventional organic solvents.Higher separation factors were corresponding to the lower temperature which can be attributed to forming hydrogen bond between phenol and the DES.Meanwhile, the consistency of tie line data was appropriately assessed by the Othmer − Tobias and Bachman equations.Finally, the data were correlated by employing a pioneering approach in group contribution calculations relevant to the thermodynamic models of NRTL and UNIQUAC.The appropriate low RMSD values, particularly for the NRTL model, confirmed good agreement with the experimental data.
To extend investigations, capability of the DES in mass transfer of phenol extraction could be scrutinized.For this aim, the viscosity of the solvent and the molecular diffusivity of phenol in the solvent have to be considered.

Table 1 .
The list of studies on the using of hydrophobic DESs in liquid-liquid extraction.

Table 2 .
Specification of the utilized chemicals.

Table 3 .
Refractive index and density for water and the DES at temperature of 298.2 K and under atmospheric pressure of 81.5 kPa a .

Table 6 .
Separation factor for several solvents in phenol extraction from aqueous phase at 298.2 K.