2.1 Reagents

Analytically pure indole, naphthalene, quinoline, acenaphthylene, choline chloride (ChCl), tetraethylammonium chloride (TEAC) and anhydrous formic acid (FA) were provided by the Shanghai Aladdin Biochemical Technology Co. Analytically pure toluene, anhydrous oxalic acid (OA), and p-toluenesulfonic acid (TsOH) monohydrate were provided by the China Pharmaceutical Group Chemical Reagent Co. The 20 wt% ammonium bisulfate aqueous solution was provided by the Tianjin Zhiyuan Chemical Reagent Co. With the exception of the ammonium bisulfate solution, the materials were dried before use. The water used was deionized. The real wash oil at 230 ∼ 250 °C was supplied by one steel factory, Wuhan, China, and its composition was detected by GC–MS (TRACE1300 ISQ Series, Thermo Fisher Scientific (China) Co). The composition of real wash oil is shown in Table S1.

2.2 Preparation of DESs and model wash oil

Preparation of DESs. The DESs systems used herein were prepared from TEAC or ChCl as hydrogen bond acceptor, and TsOH, FA, or OA as hydrogen bond donor. The preparation of TEAC/2TsOH is described herein as an example. TEAC and TsOH were added to a covered flask in a molar ratio of 1:2, and magnetically stirred at 90 °C until a uniform liquid was obtained. ¹H NMR results indicated the formation of hydrogen bonding between TEAC and TsOH (Fig. S1).

The TEAC/2FA, TEAC/OA, and ChCl/2TsOH DESs were prepared in the same way. All the prepared DESs were kept in a desiccator to prevent the absorption of moisture from the air.

Preparation of model wash oil. The composition of wash oil is complex, with bicyclic and tricyclic aromatic hydrocarbons, and nitrogen-containing compounds as main components. Herein, indole, naphthalene, acenaphthylene, and quinoline were selected as representative components to prepare the model wash oil. With the exception of quinolone, all the three compounds are solids at room temperature. Toluene is one of the important components of coal tar and has a very small solubility in DES (Larriba et al., 2018). Therefore, toluene was chosen as the solvent to dissolve the above four compounds for the preparation of the model wash oil. To this end, 10 g of indole, 30 g of naphthalene, 35 g of acenaphthylene, and 5 g of quinoline were added to a beaker and dissolved in a small amount of toluene until the mixture was clear. The mixture was transferred to a 250-mL volumetric flask and diluted with toluene to obtain a 40 g/L indole solution. Model oils with indole concentrations of 15, 25, and 60 g/L were prepared by the same method.

2.3 Separation process

Extraction of indole. A separation experiment was completed as follows: Equal mass of DESs was loaded in a graduated test tube with the model wash oil and a magnetic stirring rotor. The tube was placed in a water bath at a set temperature and stirred for 30 min. The tube was then allowed to settle at the water bath, and soon two layers were clearly visible. The upper layer was wash oil with the indole removed, and the lower layer was the DESs containing the indole. The volumes of upper and lower layer were recorded as V_U and V_L, respectively. These two layers were separated in a separatory funnel, and a small aliquot of the upper layer was removed for further analysis. The extraction process used was identical when TEAC alone was used to extract indole. In addition to replacing the tube with a beaker, the extraction process of indole from the real wash oil was the same as that of the model wash oil. A small amount of the upper layer was taken in a beaker and the pH was measured followed by the addition of AgNO₃. No change in pH or white precipitation was found, indicating that tetraethylammonium chloride and p-toluenesulfonic acid were not present in the residue of model or real wash oil.

Regeneration of DESs. Deionized H₂O was applied to regenerate the DESs. Twice the mass of deionized water was added to the DESs phase and the eutectic mixture was stirred at 45 °C for 20 min. The yellowish-brown crude indole settled at the test tube bottom. The DESs aqueous solution was assembled and subjected to rotary evaporation at 60 °C. The regenerated DESs and recovered deionized H₂O were stored for reuse.

Refining of indole. The yellowish-brown crude indole may contain a small amount of quinoline, which was removed by adding an equal volume of 20 wt% ammonium bisulfate aqueous solution. The mixture, stirred magnetically at room temperature for 30 min, then separated into two layers after standing for 10 min. The upper layer contained the refined indole, which was dried and analyzed. The same refining process was applied when using TEAC alone to extract the indole.

2.4 Analysis methods

The indole content in the model wash oil was measured by gas chromatography analysis (GC, PerkinElmer Clarus 580), using an HP-5 capillary column (60 m × 0.25 mm × 0.25 mm) and a flame ionization detector. High purity nitrogen served as carrier gas. The operating temperature was 280 °C for the injector and 300 °C for the detector, and the split ratio was 5:1. The heating program was to maintain an initial temperature of 80 °C for 1 min, then raise it to 220 °C at a rate of 10 °C/min and hold for 3 min. o-Nitrotoluene was an internal standard substance, and the calibration curve was shown in Fig. S2.

The separation performace for indole was represented by the extraction efficiency and the distribution coefficient, calculated from equations (1) and (2), respectively.

FTIR spectroscopy (Bruker Tensor 27 spectrometer) was conducted in the wavenumber range of 400–4000 cm⁻¹. ¹H NMR spectroscopy (Agilent 600 spectrometer) was conducted at 25 °C with a ¹H sensitivity of ≥ 500:1.

2.5 DFT calculations

Density functional theory (DFT) calculations were performed using the Gaussian 09 program package (Frisch et al., 2009). The optimization of the structure and frequency analysis for the compounds and the conformations were performed at the M06-2X/6–31+G(d,p) level. The optimized structures without imaginary frequencies corresponded to the minimum energy structures. The single point energy was calculated at the level of M06-2X/6–311++G (d,p) with consideration of the counterpoise procedure (Yi et al., 2021) (basis set superposition error, BSSE). The interaction energy ( $\mathrm{\Delta }$ E) was calculated with equation (3), where E(A⋅⋅⋅B), E(A) and E(B) represent the energies of complex AB, compound A and compound B, respectively.

The localized orbital locator integrated π over plane (LOLIPOP) index (Lu and Chen, 2020) was used to quantify the π-stacking ability of aromatic compounds, including indole, toluene, naphthalene, quinoline, and acenaphthylene. The independent gradient model based on Hirshfeld partition (IGMH) (Lu and Chen, 2022) was used to visualize the interactions between indole and the other aromatic compounds or TEAC/2TsOH. Energy decomposition analysis based on forcefield (EDA-FF) (Lu et al., 2021) was used to analyze the non-bonded interactions between indole and the other aromatic compounds or TEAC/2TsOH. Electrostatic potential (ESP) (Qu et al., 2022) analysis was performed to determine the electrophilic and nucleophilic sites of indole and TEAC/2TsOH. The wave function was analyzed using Multiwfn 3.8 (Lu and Chen, 2012).

3.1 Association structures of indole with aromatic compounds

The components in wash oil are mostly aromatic compounds, with the electrons in the aromatic rings delocalized to form a conjugated system. Bicyclic and tricyclic aromatic hydrocarbons (naphthalene, acenaphthylene) are isoelectronic conjugated systems; quinoline is also an isoelectronic conjugated system, but the lone pair electrons of nitrogen participate in the conjugated system, which increases its electron density. In contrast, indole is a multi-electron conjugated system. The π−π interactions are one of the most important intermolecular interaction in aromatic compounds with a conjugated system. The localized orbital locator integrated π over plane (LOLIPOP) (Lu and Chen, 2020) method is useful to measure the π-stacking ability of aromatic systems. A ring with a smaller LOLIPOP value has stronger π-depletion (namely lower π-delocalization), and hence shows stronger π-stacking ability. The calculated LOLIPOP values and delocalized π electron orbitals for indole, toluene, naphthalene, quinoline, and acenaphthylene are compared in Fig. 1. There are three disconnections (shown with the red arrows in Fig. 1a) in the π-electron delocalized orbital of the indole six-membered ring, while that in the five-membered ring is continuous, indicating that the six-membered ring has lower π-delocalization than the five-membered ring. The calculated LOLIPOP values for indole also illustrate this result: The value for the six-membered ring (4.1) is lower than for the five-membered ring (4.6), showing that the six-membered ring of indole has better π-stacking ability than the five-membered ring. According to their LOLIPOP values, the π-stacking ability decreases in the order quinoline (2.7 ∼ 4.0) > naphthalene (3.7) > indole (4.1 ∼ 4.6) > acenaphthylene (6.9 ∼ 8.0) > toluene (8.5).

Close

Fig. 1

LOLIPOP analysis of indole and other aromatic compounds (: H,: C).

Export to PPT

The independent gradient model based on Hirshfeld partition (IGMH) (Lu and Chen, 2022), an important graphical method to display intermolecular interactions, was used to visualize the interactions of indole with toluene, quinoline, naphthalene, and acenaphthylene, and the colored isosurface results of IGMH analysis are shown in Fig. 2. The isosurfaces represent regions where interactions occur. Since toluene contains a methyl group and a benzene ring, it forms C−H···π and π−π interactions with indole (Fig. 2a). The three other aromatic compounds, namely naphthalene, quinoline, and acenaphthylene, form face-to-face π−π interactions with indole (Fig. 2b to 2d). Obviously, indole mainly forms association structures with aromatic compounds in wash oil through π−π interactions. According to the isosurface area representing the interaction strength, the interactions between indole and toluene are weakest and those between indole and acenaphthylene are strongest.

Close

Fig. 2

IGMH analysis of interactions between indole and other aromatic compounds (Isovalue = 0.009).

Export to PPT

The nature of intermolecular interactions is divided into electrostatic, dispersion and repulsive interactions, among which electrostatic and dispersion interactions are attractive (Emamian et al., 2019). Energy decomposition analysis based on force field (EDA-FF) (Lu et al., 2021) was used to reveal the nature of interactions between indole and the aromatic compounds present in model wash oil, and the results are presented in Fig. 3. The electrostatic interaction energy between indole and the aromatic compounds (toluene, naphthalene, quinoline, and acenaphthylene) is less than −5.0 kJ/mol, and the dispersion interaction energy is distributed from −47.7 to −73.5 kJ/mol, indicating that association between indole and the aromatic compounds in wash oil is mainly induced by dispersion interactions. Therefore, the extraction of indole should overcome these dispersion interactions.

Close

Fig. 3

Energy decomposition analysis of interactions between indole and aromatic compounds.

Export to PPT

3.2 Interactions between indole and TEAC/2TsOH

Molecular electrostatic potential (ESP) analysis is a useful tool to predict the nucleophilic and electrophilic sites in molecules (Qu et al., 2022). The strength and orientation of intermolecular interactions, such as hydrogen bonds and π-π interactions, can be well-predicted and explained by analyzing the magnitude and positions of minimum and maximum ESP. The ESP analysis of indole and TEAC/2TsOH was performed, as shown in Fig. 4. For indole, the minimum ESP (−100.1 kJ/mol) and maximum ESP (225.0 kJ/mol) sites appear on the benzene ring and the H atom connected to the N atom, respectively, indicating that the electron-rich benzene ring and the electron-deficient H in indole molecule can interact easily with other molecules. For TEAC/2TsOH, there are three ESP minimum sites (−172.4, −180.1, and −182.1 kJ/mol) and three ESP maximum sites (147.3, 149.5, and 153.3 kJ/mol) that can interact with other molecules. These results indicate that the contact of indole with TEAC/2TsOH may be most favorable when the benzene ring of indole is close to the choline cation of TEAC/2TsOH, and when the −NH of indole is close to the Cl or O atoms of TEAC/2TsOH. This was confirmed by structural optimization of the conformation (Fig. 5), that is, the benzene ring approaching the choline cation and −NH approaching Cl atom make the complex most stable. The indole-TEAC/2TsOH complex has five hydrogen bonds, including four C−H···C (ranging from 2.53 to 2.87 Å) and one N−H···Cl (2.44 Å), which are illustrated by black dashed lines in Fig. 5. Incidentally, the sum of the Bondi radii of C and H is 2.90 Å, while that of H and Cl is 2.95 Å (Yi et al., 2019). To visualize graphically the interactions between indole and TEAC/2TsOH, IGMH analysis was performed and the result is shown in Fig. 6. The interactions between −NH and the Cl ion are stronger than those between the benzene ring and the choline cation, because the value of sign(λ₂)ρ between −NH and the Cl ion ranges from −0.02 to −0.01 a.u, while that between the benzene ring and the choline cation is near −0.01 a.u. The farther the value of sign(λ₂)ρ is from zero, the stronger the interactions (Lu and Chen, 2022).

Close

Fig. 4

ESP analysis of indole and TEAC/2TsOH.

Export to PPT

Close

Fig. 5

Optimized complex structure of indole with TEAC/2TsOH.

Export to PPT

Close

Fig. 6

IGMH analysis of interaction between indole and TEAC/2TsOH (Isovalue = 0.005).

Export to PPT

There are more hydrogen bonds and C−H···π interactions between indole and TEAC/2TsOH than between indole and the aromatic compounds in wash oil. Accordingly, the results of energy decomposition analysis (Fig. 6) show that the electrostatic interaction energy between indole and TEAC/2TsOH is −52.8 kJ/mol, which is 10 ∼ 30 times that between indole and the other compounds in wash oil. The dispersion interaction energy is −72.5 kJ/mol, which is equivalent to the maximum dispersion interaction energy (−73.1 kJ/mol) between indole and the compounds in wash oil. The multiple interaction sites and larger interaction energy imply that it is feasible to extract indole from wash oil with TEAC/2TsOH DES.

3.3 Extraction performance of indole by TEAC/2TsOH

Temperature. Temperature is an important parameter in liquid–liquid extraction. Five operation temperatures were selected herein to assess the influence of temperature on the extraction of indole, and the results are provided in Fig. 7a. As shown, the extraction efficiency of indole was not sensitive to temperature and remained almost constant at 96.8 ± 2% in the temperature range studied. This temperature-independent extraction efficiency indicates that the separation process is nonreactive, in agreement with results for the separation of nitrogen/sulfur/oxygen-containing compounds in coal-based liquids or fuel by DESs (Yi et al., 2019; Xu et al., 2022; Wang et al., 2022). Meanwhile, the concentration of indole in wash oil remained low (1.2 ± 1 g/L) at the different temperatures. From an economic viewpoint, a temperature of 30 °C would be optimal and was used in subsequent investigations.

Close

Fig. 7

Effects of experimental conditions on the extraction performance of indole: (a) extraction temperature; (b) extraction time; (c) mass ratio of TEAC/2TsOH to indole; (d) initial concentration of indole in the wash oil.

Export to PPT

Stirring time. The stirring time is another important parameter in liquid–liquid extraction. The shorter the time to reach equilibrium, the more conducive is the process to large-scale use. To assess the influence of time, the stirring time was varied from 5 to 30 min, and the results obtained are summarized in Fig. 7b. As can be seen in the figure, there was only a slight increase in indole extraction efficiency for increasing stirring time, indicating that indole can be transferred quickly from the wash oil to the TEAC/2TsOH DES. This result is due to the stronger interactions between indole and TEAC/2TsOH than between indole and the aromatic compounds in wash oil (Fig. 3). Correspondingly, the content of indole in wash oil only decreased slightly from 3.0 to 1.2 ± 1 g/L as the stirring time increasing from 5 to 30 min. To separate indole quickly and maintain a balanced separation procedure, the optimal stirring time was set to 20 min.

Mass ratio of TEAC/2TsOH to indole. The influence of the mass ratio of TEAC/2TsOH to indole on the separation performance was investigated in the range of 1:5 ∼ 1:0.5 and the results are presented in Fig. 7c. As expected, increasing the mass of TEAC/2TsOH used increased the extraction efficiency of indole; when a mass ratio of TEAC/2TsOH to indole below 1:2, the amount of TEAC/2TsOH was insufficient to extract all the indole. For mass ratios from 1:2 to 1:0.5, the extraction efficiency remained almost unchanged at 97.2 ± 2%, and the content of indole in wash oil only decreased from 13.1 to 2.2 ± 1 g/L. On the basis of the good extraction efficiency and the favorable distribution coefficient, a 1:2 mass ratio was deemed optimal.

Initial indole concentration. To assess the extraction performance of TEAC/2TsOH for wash oil with different indole concentrations, the influence of the initial indole concentration on the extraction efficiency was analyzed in a range from 15 to 60 g/L. The results presented in Fig. 7d show that when the amount of TEAC/2TsOH was insufficient, that is, the mass ratio of TEAC/2TsOH to indole was below 1:2, the residual indole content in the oil phase after extraction increased with the increasing of initial indole concentration in the wash oil. However, when the mass ratio of TEAC/2TsOH to indole was above 1:2, the wash oil with different initial indole contents eventually contained the same residual concentration of indole, around 1.0 ± 1 g/L. These results indicate that when the amount of TEAC/2TsOH used is sufficient, the extraction performance is not correlated with the initial indole concentration, demonstrating that TEAC/2TsOH can be applied to wash oils with different initial indole concentrations.

Adding water to TEAC/2TsOH. The influence of H₂O on the extraction of indole using TEAC/2TsOH is illustrated in Fig. 8. It can be seen that the extraction efficiency of anhydrous TEAC/2TsOH is highest, reaching 97.2 ± 2%. For increasing amounts of H₂O added from 2 ∼ 10 wt%, the indole extraction efficiency decreased rapidly from 91.4 to 37.6 ± 1%. For 10 ∼ 60 wt% of H₂O addition, the extraction efficiency decreased more slowly but was essentially lost at 60 wt%. Kaur et al. (Kaur et al., 2020) suggested that H₂O has a great influence on hydrogen-bonded DESs systems. The native hydrogen-bonded structure of the pure DESs is destructed upon dilution, and more or less destroyed to form an aqueous solution when the H₂O content reaches a certain level (> 40%). Because indole is insoluble in water, the TEAC/2TsOH aqueous solution cannot extract it from the wash oil, resulting in the indole extraction efficiency dropping to zero at 60 wt% of H₂O content.

Close

Fig. 8

Influence of water addition on the extraction efficiency of TEAC/2TsOH.

Export to PPT

3.4 DESs recycling

DESs regeneration is usually performed with organic solvents, such as ethyl acetate (Yi et al., 2019) and diethyl ether (Qu et al., 2022), both toxic and highly flammable. The experimental results reported above demonstrated that TEAC/2TsOH has no indole extraction ability when the H₂O content in TEAC/2TsOH reaches 60 wt%. The TEAC/2TsOH DES regenerated with H₂O instead of organic solvents was applied to separate indole under the same conditions for three additional cycles. The separation performance of TEAC/2TsOH in the successive cycles, provided in Fig. 10a, decreased slightly from 97.2 ± 2% to 95.3 ± 2, 94.7 ± 2, and 93.2 ± 1% for cycles 2 ∼ 4. The reason for this may lie in that when TEAC/2TsOH was regenerated with H₂O, some of it dissolved in H₂O, which reduced the amount of TEAC/2TsOH left for extracting indole, resulting in a gradual decrease in extraction efficiency. When comparing the fresh and recycled TEAC/2TsOH by FTIR analysis, there were no obvious changes observed in the spectra (Fig. 10b), indicating that TEAC/2TsOH can be regenerated without structural changes. The good stability of the extraction capacity and the structure are important from an economic perspective for the implementation of TEAC/2TsOH in commercial processes.

Close

Fig. 9

Extraction efficiency in successive indole extraction cycles (a) and FTIR spectra of fresh and regenerated TEAC/2TsOH (b).

Export to PPT

3.5 Comparison with other extraction systems

Other acidic DESs. Formic acid (FA) or oxalic acid (OA), and choline chloride (ChCl) are widely used as hydrogen bond donors and hydrogen bond acceptor (Lemaoui et al., 2021), respectively. They were also used herein to prepare three additional DESs for the separation of indole, and their separation performance is compared with TEAC/2TsOH in Fig. 11. The extraction efficiency and distribution coefficient were highest for TEAC/2TsOH among the investigated DESs, and lowest for TEAC/OA, which demonstrates that the combination of HBA and HBD selected for the preparation of DESs influences their extraction performance for indole. This may be due to the benzene ring in TsOH being more conducive to the formation of intermolecular interactions between the DES and indole than the alkyl carbon chains in FA and OA, while the –OH group in the ChCl cation increases the polarity of the DESs, which may adversely affect the phase transfer of indole.

Close

Fig. 10

Separation of indole with different acidic deep eutectic solvents.

Export to PPT

Other reported extractants. For comparison, a quaternary ammonium salt (TEAC) was applied to separate indole from wash oil, as first deliberated by Wu’s group (Ji et al., 2018). The experimental procedures used were the same as described for TEAC/2TsOH. The separation performance of TEAC and TEAC/2TsOH is compared in Fig. 12. There is no obvious difference in indole extraction efficiency when using TEAC (97.5 ± 3%) and TEAC/2TsOH (97.2 ± 2%). The similar extraction efficiencies may be due to the same driving forces for separation in both systems, namely hydrogen bond and van der Waals interactions, and the comparable interaction energies of indole and both extraction agents. The interaction energy between indole and TEAC/2TsOH is −83.0 kJ/mol, and that between indole and TEAC is −84.6 kJ/mol. The main difference between the two processes lies in the purity level of the isolated indole, which was lower for TEAC (90.2 ± 2%) than for TEAC/2TsOH (98.5 ± 1%). The lower indole purity for TEAC extraction may be ascribed to the fact that TEAC forms DESs with indole during its separation from the oil phase, and the DESs formed in situ acts as a solvent interacting and dissolving the other aromatic compounds in the wash oil, including naphthalene and acenaphthylene, resulting in more aromatic compounds being transferred to the extraction phase. These aromatic compounds lowered the purity level of the indole product.

Close

Fig. 11

Separation performance of indole with TEAC and TEAC/2TsOH.

Export to PPT

3.6 Extraction of indole from real wash oil

The extraction experiment on real wash oil via TEAC/2TsOH DESs was carried out. The amount of indole in the original wash oil was 5.2 wt%. TEAC/2TsOH with a 0.5 time mass amount of indole was added to wash oil, and result showed that the amount of indole in the extracted wash oil was 1.9 wt%. The extraction efficiency of indole in real wash oil was around 63.5% (The uncertainty of this extraction efficiency was ± 3%). The GC–MS results of refined indole product are shown in Table S2. It can be seen that the area percentage of aromatic compounds entrapped in the indole product was 42.3%, indicating that the purity of the indole product is not high. The main reason for the low purity is the very complex composition of the real wash oil (Table S1), with compounds such as pyridines, phenolic compounds, benzene or naphthalene derivatives (with unsaturated side chains), indanol, biphenyl, hexadecane, and chamazulene being entrained during the extraction process (Table S2). Designing DESs with higher selectivity is a future endeavor.