Phenol Recovery from Aromatic Solvents by Formation of Eutectic Liquids with Trialkyl-2,3-dihydroxypropylammonium Chloride Salts

: Trialkyl-2,3-dihydroxypropylammonium chloride salts have been investigated as liquid eutectic-forming salts for the extraction of phenol from aromatic-rich model oil (toluene), demonstrating how the increased partitioning of phenol from oil-phases can be combined with reduced co-miscibility of the salt with aromatic hydrocarbons through the introduction of the dihydroxypropyl-function. decomposition for each sample across the compositional range. Glass transition temperatures were measured by differential scanning calorimetry (DSC) performed using a TA instruments Q2000 DSC with an RCS 90 cooling system attached. Tzero aluminium pans were ﬁlled and sealed with Tzero hermetic lids within the glove box under an argon dry atmosphere before being transferred outside the glovebox to the DSC. The DSC measurements were repeated using cooling and heating cycles at 5 ◦ C min − 1 between − 90 and 50 ◦ C (salt/phenol mixtures) or 120 ◦ C (pure trialkyl-2,3-dihydroxypropylammonium chloride salts). The visual melting points were also conﬁrmed by polarising optical microscopy on samples that were placed between glass microscope cover-slips that were placed on the heating-stage of a Olympus BX50 polarising optical microscope that was equipped with a Linkam TH600 hot stage and TP92 temperature controller. The reported observed melting points, Tg determined from DSC, and thermal decomposition points from TGA are estimated ± 2.5 ◦ C based on the uncertainty in clearly deﬁning second order (Tg) transitions or onsets to decomposition in dynamic TGA. These results indicate how the inclusion of lipophobic/hydrophilic (propanediol) groups can overcome the challenge of co-miscibility of longer-chain functionalised ammonium salts with aromatic-rich feeds, enabling the successful and efﬁcient extraction by simultaneously inhibiting the miscibility of eutectic forming salt extractants with aromatics while retaining the ability to generate eutectic liquids with phenol across a wide composition range.


Introduction
Phenols are used on a megatonne scale worldwide as major chemical feedstocks for the production of phenolic and synthetic resins, caprolactam and bisphenol A, alkylphenols, and adipic acid [1,2]. The bulk of which is synthetically produced by peroxidation of cumene that is obtained from benzene [3]. The high phenolic contents of low grade coals, coal tars, and oils [4,5], as well as bio-oils produced through biorefining and biomass liquefaction [6], make these important alternative sources from which phenolic compounds can be directly accessed.
The concept that the formation of a polar eutectic liquid between phenol and an organic salt could be used to drive the non-aqueous extraction of phenol from oil was first described by Wu, Marsh, and co-workers [7,8]. They showed that the addition of organic tetraalkylammonium salts to phenol-containing hexane and toluene model oil feeds generated eutectic liquids that contained around 1:1-1:2 salt:phenol molar ratios that separate from the non-polar carrier hydrocarbon. Importantly, and key to this approach, is the ability to use an organic ether as an antisolvent to precipitate the organic salt component of the formed eutectic liquid, facilitating the separation and recovery of the phenols without generating the contaminated aqueous waste streams that are associated with conventional alkali/acid extraction/neutralisation pathways.
It is also worth noting that deep eutectic solvents have also been applied as bulk phases for the extraction of phenolic compounds from olive oils, agri-food byproducts, and polyphenol antioxidants from plant sources [9][10][11][12].
A range of tetraalkylammonium halides and choline chloride (ChCl) salts were initially studied and shown to be effective [7,8] with the phenol extraction efficiency increasing from ca. 2 M solutions in toluene as the organic salt added was changed from tetramethylammonium chloride ([NMe 4 ]Cl) to tetraethylammonium chloride ([NEt 4 ]Cl), giving up to 99.9% extraction efficiency. However, as the alkyl-chain size on the ammonium cations increases, so does solubility in aromatic solvents [13]. Tetrabutylammonium chloride ([NBu 4 ]Cl), miscible with toluene, does not form a separate salt/phenol phase [8], which makes this, and other lipophilic salts, unsuitable for treating aromatic-rich oils, such as pyrolysis bio-oil.
To address these restrictions, a range of ionic liquids [14][15][16][17][18], zwitterions [19][20][21], pre-formed deep eutectic solvents [22], and other eutectic-forming hydrogen bond acceptors [23][24][25] have been explored. In particular, gemini dicationic salts [17,18] and choline chloride derivatives that incorporate a single longer alkyl-substituent (ethyl to octyl) [26] have been examined in order to enhance phenol extraction efficiency and selectivity while retaining the immiscibility of the formed eutectic with aromatic oils. Most recently, Yi et al. [27] has reported that the addition of a ChCl/glycerol deep eutectic liquid was more effective than using ChCl alone to extract phenol from coal-based oils, with better selectivity for phenolics over hydrocarbons. It is worth noting that glycerol is polyhydoxylated and the structure of choline chloride/glycerol eutectic liquids is dominated by glycerol-glycerol hydrogen bonding [28].
The key motifs within DES-forming phenol extractants are shown in Table 1.
Studies of eutectic mixtures of phenol with aromatic ionic liquid salts using neutron scattering [29,30] identified the solvation of phenol molecules by both the cation and anion components of the salt, and that phenol extraction is enhanced when more weakly coordinating anions are used. This has been interpreted as a consequence of a reduction in the strength of cation-anion association, allowing for increasingly favourable cation-phenol interactions to facilitate improved extraction. It seems likely that the impact on changing the size of cations on phenol extraction efficiency with tetraalkylammonium salts [7,8] is also a consequence of an equivalent reduction in the strength of the cation-anion interactions. However, unfortunately, as reported, an extension of this is an increased miscibility with aromatic hydrocarbons. Thus, it is necessary to promote phenol-salt association, while, at the same time, restricting co-miscibility of the resultant DES with aromatic-rich feeds in order to obtain good selective phenol extraction.
We speculated whether these two competing challenges could be reconciled by introducing both lipophilic (N-alkyl) and lipophobic (poly-hydroxyl) groups into the organic salt eutectic former, and here we report on the successful demonstration of the validity of this design strategy using trialkyl-2,3-dihydroxypropylammonium chloride salts ( Figure 1) in order to extract phenol from hexane and toluene as models for aliphatic and aromatic oils.

Materials
Phenol, hexane, toluene, and tetraalkylammonium halide salts used in extraction screening were obtained from Sigma-Aldrich.
Phenol was sublimed before use, while all other materials were used as received.
2.1.1. Trimethyl-2,3-dihydroxypropylammonium Chloride, [Me 3 NCH 2 CH(OH)CH 2 (OH)]Cl (1) 3-Chloro-1,2-propanediol (0.1265 mol, 10.58 mL) was added dropwise to a stirred ethanolic solution of trimethylamine (0.19 mol, 11.2309 g) that was maintained at 0 • C in an ice bath under an inert nitrogen atmosphere. After addition, the temperature of the mixture was raised to reflux (65-68 • C) and heated overnight. The solvent was then removed under reduced pressure to afford a yellow oil. This crude product was washed four times each with diethyl ether and acetone (50 mL), respectively, until a white solid precipitated from the acetone wash. The product was dried under vacuum, yielding a white hygroscopic powder (13.7 g, 51% yield). NMR: (

Characterisation
Thermal stability of the salts was examined by thermogravimetric analysis (TGA) using a TA instruments Q5000 TGA, with Tzero aluminium pans and Tzero aluminium hermetic lids. Dynamic heating was applied at 10 • C min −1 from room temperature to 400 • C to determine the onset of thermal decomposition for each sample across the compositional range. Glass transition temperatures were measured by differential scanning calorimetry (DSC) performed using a TA instruments Q2000 DSC with an RCS 90 cooling system attached. Tzero aluminium pans were filled and sealed with Tzero hermetic lids within the glove box under an argon dry atmosphere before being transferred outside the glovebox to the DSC. The DSC measurements were repeated using cooling and heating cycles at 5 • C min −1 between −90 and 50 • C (salt/phenol mixtures) or 120 • C (pure trialkyl-2,3-dihydroxypropylammonium chloride salts). The visual melting points were also confirmed by polarising optical microscopy on samples that were placed between glass microscope cover-slips that were placed on the heating-stage of a Olympus BX50 polarising optical microscope that was equipped with a Linkam TH600 hot stage and TP92 temperature controller. The reported observed melting points, Tg determined from DSC, and thermal decomposition points from TGA are estimated ±2.5 • C based on the uncertainty in clearly defining second order (Tg) transitions or onsets to decomposition in dynamic TGA.

Phenol Extraction Screening
Gas chromatography (GC) analysis was carried out using an Agilent 7820A GC with flame ionisation detection (FID) that was fitted with a HP-5 column with the dimensions of 30 m × 320 µm and particle size 0.25 µm. The mobile phase was hydrogen at a flow rate of 30 mL min −1 and air at a flow rate of 400 mL min −1 with a makeup flow of 25 mL min of helium. The oven temperature was set at 350 • C and the sample size was 1 µL. The extraction efficiency was determined by the difference between phenol content in the model oil before, and after, contact with the organic salts, which calibrated the GC response to an internal dodecane standard (0.01 M).
The extraction experiments followed the methods described in the literature [8]. Stock solutions containing phenol (0.1 M) and dodecane (0.01 M) in hexane or toluene were prepared. Typically, a 20 mL aliquot of model oil (containing phenol) was added to a glass tube that was equipped with a magnetic follower and then immersed in an oil bath held at 30 • C. An accurately measured mass of extractant (2 mmol, which corresponded to 1:1 molar ratio to the phenol content), was added. The tube was sealed and the mixture stirred vigorously for 30 min and then allowed to settle for 30 min The upper oil phase was then sampled and analysed by GC. The concentration of phenol post treatment was determined by comparing the integrals for the signals corresponding to phenol (3.196 min) and the dodecane internal standard (5.316 min) with the response factor being determined across the concentration range 0.0012-0.1 M phenol with 0.01 M dodecane and the extraction efficiency being calculated by comparing the initial and final phenol concentrations in the oil. Uncertainties in the overall extraction results are estimated to be ±2% derived from duplicate measurements of phenol concentrations with reproducibility better than 1% and the LOD in the analysis of [phenol] to better than 0.0005 M and the corresponding error bars are shown in the figures below.
The initial observation that mixtures of phenol with each of the three dihydroypropylammonium chloride salts (1-3) generate liquid compositions strengthened the proposition that these could be used for phenol-extraction through DES formation. The phase behaviour of mixture of the triethyl-substituted 2 with phenol was examined both visually and by DSC. The results are shown in Table 2 and Figure 2. Pure phenol and 2 both show well defined freezing points at 97 and 41 • C, respectively, from DSC measurement. In contrast, combining mixtures of solid 2 with phenol across the mole fraction composition range χ phenol = 0.33-0.75 formed room temperature liquids that did not crystallise upon chilling in a refrigerator, and only glass transition points were observed on thermal cycling by DSC between −38 to −54 • C, with a significant expansion of the liquidus range justifying definition as eutectic liquids.
This depression and suppression of melting as compared to both pure components gives rise to a broader, more extensive liquid range for 2/phenol mixtures (Figure 2) than that for ChCl/phenol [35], although only glasses, rather than crystalline solids, were obtained upon cooling. When comparing 2/phenol with ChCl/phenol, the minima in the solidification point appears to be shifted from χ phenol = 0.75 (3:1) for ChCl/phenol [35] to around 0.50 (1:1), which could reflect the greater number of hydroxyl groups present in each cation, allowing for a greater number of hydrogen-bonding sites between cations and phenol molecules. The expansion of the liquid range for 2/phenol mixtures is also evident from the freezing point depression of ca. 95 • C as compared to pure phenol and 151 • C from 2.

Extraction of Phenol from Hexane and Toluene
Phenol extraction from hexane and toluene solutions as model oils was examined following the procedure that was described by Guo et al. [8].
Examining data from the screening of extraction with a broad range of tetraalkylammonium halide salts ([NR 4 ]X, R = Me, Et, Pr, Bu, and X = Cl, Br, I) [29] and comparison with the literature, some clear trends in the impact of the different eutectic forming organic salts on phenol extraction are apparent. There is a need for a hard, basic anion (supporting phenol-OH hydrogen bond donation to the anion as a primary association mode [30,36]), and the extraction efficiency increases, in general, with increasing size of the N-alkyl substituents from Me-Bu, as seen for the [NR 4 ]I salts, where phenol extraction from hexane increased from 7% to 96%. Increasing the size of the N-alkyl substituents makes the ammonium cation larger, which reduces competing coulombic cation-anion associations, allowing for eutectic compositions to form more readily. However, as noted by Guo et al. [8] and confirmed here, increased lipophilicity of the organic salt also enhances co-miscibility with aromatic-rich oil-phases, such as toluene. These data support a strategy for improving the performance of eutectic forming phenol-extractants by adding lipophobic diol functions into the cation to reduce the co-miscibility of the organic salt, and salt/phenol eutectic liquid, with toluene, leading to greater extraction efficiency and the partitioning of the eutectic liquid formed from the oil phase.
Most reported phenol extraction studies using the eutection formation approach are screened using concentrated aromatic-rich oils (typically 90-200 g L -1 , equivalent to 1-2.5 M), whereas measurements from hexane are usually made with lower phenol concentrations. In previous work, examining the use of ammonium salts, it was shown that the extraction efficiency for phenol is reduced at low phenol contents [31]. Here, we examined the extraction from 2 M solutions of phenol in toluene and 0.1 M solutions in hexane in order to compare the results with literature [7,8], and from more dilute 0.1 M solution in toluene to assess the changes in partition efficiency with phenol content.
Having established, by examination of the phase behaviour, that mixtures of 2 with phenol do form a eutectic liquid at compositions around χ phenol = 0.5 (1:1), the three trialkyl-2,3-dihydroxypropylammonium chloride salts with methyl-, ethyl-, and butyl-substituents were tested as eutectic liquid-forming extractants for phenol from hexane and toluene. When appropriate quantities (1:1 to phenol) of the three trialkyl-2,3-dihydroxypropylammonium chloride salts were added to the hexane/phenol and toluene/phenol feed solutions, small volumes of a dense eutectic liquid was formed, and, on standing, settled at the bottom of the flasks. The phenol concentration in the bulk hydrocarbon phase before, and after, contacting was determined by GC analysis that was calibrated to an internal dodecane standard, and the extraction efficiency calculated as the percentage phenol removed: where [phenol] init and [phenol] f inal are, respectively, the initial and post extraction concentrations of phenol determined in the model oil phase. The results and a comparison with the data that were obtained for representative tetraalkylammonium chlorides (from Turner et al. [29]) and from choline chloride are shown in Table 3. All of the salts tested showed good extraction of phenol from hexane, with greater than 87% removal while using 1 and 2. These results are comparable with those that were obtained using ChCl, [NMe 4 ]Cl, and [NEt 4 ]Cl (89% in each case). We did obtain quantitative phenol removal with the slightly more lipophilic [NPr 4 ]Cl and [NBu 4 ]Cl salts. However, the extraction efficiency is dominated by the effect of the strong HBA chloride anion, and it is difficult to obtain clear insight into the relative influence of the cation on extraction of phenol from hexane. Similarly, the results obtained from 2 M solution in toluene (190 g L -1 ) using 1 (92%), 2 (99%), and ChCl (96%) are both high and qualitatively comparable to the 95% extraction with ChCl reported in the literature. Guo et al. [8] reported an extraction efficiency of 96% with [NMe 4 ]Cl and better than 99.9% with [NEt 4 ]Cl, showing a small increase in performance as the N-alkyl chain length was increased. This is consistent with the enhancement of the extraction efficiency between 1 (methyl) and 2 (ethyl) from 2 M phenol in toluene.
Meng et al. [31] reported that the phenol extraction efficiency using the eutectic liquid-forming extraction approach decreased as the concentration of phenol in the organic phase was reduced and the limiting distribution coefficient for phenol between the two phases is reached. At the lower initial phenol concentration of 0.1 M in toluene, all three organic salts with N-methyl substitutents (1, ChCl and [NMe 4 ]Cl) performed poorly as extraction agents. Almost no extraction (2%) was observed while using 1, whereas greater, but still only moderate, extraction of phenol was observed with ChCl (33%) and [NMe 4 ]Cl (32%), giving residual phenol concentrations around 0.065 M (ca. 6 g L -1 ).
As the length of the N-alkyl chain substituents of the trialkyl-2,3-dihydroxypropylammonium chloride salts was increased from methyl (1) to ethyl (2) to butyl (3), the efficiency of the extraction of phenol significantly improved (Figure 3). The ethyl and butyl-substituted salts (2 and 3) both showed greater extraction than ChCl, which was the preferred extractant [26,35,37,38], where only 33% of the phenol was removed from the low phenol concentration system. These results illustrate two characteristics that arise from the addition of the dihydroxypropyl-chain. First, efficient phenol extraction can be achieved from the 2 M phenol solutions in toluene, and that the extraction efficiency increases with increasing the length of the N-alkyl substituents from 1-3, with an overall enhancement of extraction and recovery of phenol from low concentration (0.1 M) toluene. With [Bu 3 NCH 2 CH(OH)CH 2 OH]Cl (3), over 80% phenol extraction was achieved from this lower concentration aromatic feed, which approaches the results that were obtained with [NEt 4 ]Cl and [NPr 4 ]Cl as eutectic forming components, and it is more effective than ChCl as a deep extractant, delivering good partitioning and extraction of phenol even from low concentration aromatic feeds.
Thus, the key components in these salts as eutectic liquid-forming extractants are identifiable: the incorporation of butyl-chains on the cation to phenol π-cation association and addition of the lipophobic propane-diol group to induce phase separation of the formed eutectic liquid from aromatic-rich hydrocarbons.

Effect of Mixing Time
The effect of mixing time on the extraction studies is shown in Figure 4 for extraction from 0.1 M phenol in toluene with 2. The phenol concentration in the feed solution is rapidly reduced over 5 min from 0.10 M to 0.051 M (49% extraction), before finally equilibrating to 0.040 M (60% extraction) after stirring for 20 min Pang et al. [7] reported equilibration after 2 min mixing, although Hou et al. [37] have subsequently shown that equilibration is mass transfer limited and is strongly dependent on the efficiency of mixing. Hence, these differences are likely due to either efficiency of mixing in different contactors or to variations in the relative interfacial tensions between the different eutectic liquids formed and toluene. In the screening studies, all of the mixtures were contacted and mixed for 30 min. It was observed that the salt/phenol liquid phases tended to form droplets, which implied high interfacial tension between the eutectic and hydrocarbons. This was especially evident for [Me 3 NCH 2 CH(OH)CH 2 OH]Cl, which coalesced to a small spherical ball that was not effectively dispersed and mixed with the toluene/phenol phase. This poor contacting, thought to be a result of the high interfacial tension, could be the origin of the poorer extraction performance of [Me 3 NCH 2 CH(OH)CH 2 OH]Cl (2% extraction) as compared to that of tetramethylammonium chloride and choline chloride, which incorporate a single -OH group in the cation (33 and 34%, respectively).

Effect of Extractant Mole Ratio
The extraction of phenol from the hydrocarbons requires an addition of at least the 'minimum' quantity of the organic salt to form the eutectic liquid. The 1:1 salt:phenol compositions sit within the large liquidus region of the trialkyl-2,3-dihydroxypropylammonium chloride/phenol phase diagram ( Figure 2). The impact of the amount of 2 added to the extraction efficiency from 0.1 M phenol in toluene was explored by varying the molar ratio of 2:phenol from 0-1.6 molar equivalents. The results are shown in Figure 5.
The residual phenol concentration in toluene was reduced from 0.1 M to ca. 0.04 M (corresponding to 3.8 g L −1 , 60% extraction efficiency) as the amount of 2 added was increased to ca. equimolar to the initial phenol content. Pang et al. [7] showed that, with ChCl, [EtNH 3 ]Cl, and [Et 3 NH]Cl, optimum extraction was obtained at the 1:1 salt:phenol equivalence point, with no enhancement of extraction on adding greater quantities of the organic salts. Here, when additional 2 above 1:1 2:phenol ratio was added to phenol/toluene solutions, a small additional reduction in the residual phenol concentration could be achieved to [phenol] f inal = 0.027 M (2.5 g L −1 ). This could correlate with the differences between the phase diagrams for 2/phenol and ChCl/phenol shown in Figure 2; however, it is most likely that these apparent differences are the statistical deviation between results. The variation between data at 1.2:1 to 1.6:1 2:phenol is 1.6%.

Conclusions
Overall, the effectiveness of trialkyl-2,3-dihydroxypropylammonium salts as eutectic liquid-forming extractants has been demonstrated. Synergistic improvement in performance, particularly from aromatic phases, was achieved by the incorporation of both ethyl-, or butyl-, with dihydroxypropyl-groups on the ammonium cation leading to simultaneously good association with phenol and inhibition of miscibility of eutectics with toluene.
Incorporating these two leads improved phase separation and consequently increased extraction efficiency with tributyl-2,3-dihydroxypropylammonium chloride as compared to tetrabutylammonium chloride.
It is notable that the phenol extraction efficiency of low concentrations of phenol (0.1 M) from toluene with tributyl-2,3-dihydroxypropylammonium chloride (3) is significantly greater than that found here with ChCl, or with the other two dihydroxypropylammonium chloride salts (1 and 2). The addition of one molar equivalent of tributyl-2,3-dihydroxypropylammonium chloride (3) to a 0.1 M phenol solution in toluene resulted in an 84% extraction of phenol as the eutectic and a residual phenol concentration in the toluene phase of 0.016 M ([phenol] f inal = 1.5 g L −1 ). This compares favourably with the best recent data from the literature, where dicationic salts ([phenol] f inal = 3.9 g L −1 ) [17,18], zwitterionic betaines ([phenol] f inal = 4.8 g L −1 ) [19], and tetraethylammonium amino acid ionic liquids ([phenol] f inal = 1.4 g L −1 ) [14] have been developed.
These results indicate how the inclusion of lipophobic/hydrophilic (propanediol) groups can overcome the challenge of co-miscibility of longer-chain functionalised ammonium salts with aromatic-rich feeds, enabling the successful and efficient extraction by simultaneously inhibiting the miscibility of eutectic forming salt extractants with aromatics while retaining the ability to generate eutectic liquids with phenol across a wide composition range.