New family of Type V eutectic solvents based on 1,10-phenanthroline and their application in metal extraction

Although a theoretically extensive number of hydrophobic eutectic mixtures suitable for solvent extraction are obtainable, to date reported mixtures rely on a limited number of chelating groups, namely phosphine oxide, carboxylic acid, and ketones. Herein, a new family of Type V Deep Eutectic Solvents (DES) based on the polypyridyl ligand 1,10-phenanthroline (phen) is presented with the aim of facilitating the extrapolation of the disclosed findings to other potential systems based on phen or bipyridine derivatives. The hydrophobic eutectic solvent composed of the natural phenolic extract thymol with phen exhibits a solid-liquid phase diagram with a useable liquidus compositional range on the thymol-rich side. The physical properties of the mixture are dependent on the presence of solutes liable to change the nature of the existing hydrogen-bonded network, most notably through the protonation of phen. The metal chelating ability of phen are retained when included as a DES component. Quantitative extraction of various transition metals cations (Cd(II), Mn(II), Co(II), Ni(II), Cu(II) and Zn(II)) as well as the platinum group metal Pd(II) was obtained across a range of solvent compositions. The extraction of Li(I), Mn(II), Co(II) and Ni(II) was further studied as a function


Introduction
Polypyridyl ligands, of which 1,10-phenanthroline and 2,2′-bipyridine are the best-known base units, exhibit a high complexation affinity with many metal ions and represent versatile starting blocks for molecular recognition and supramolecular chemistry (Bencini and Lippolis, 2010;Hancock, 2013;Sammes and Yahioglu, 1994).The ligand 1,10phenanthroline (thereafter referred to as phen) is a hydrophobic and electron-poor heteroaromatic bidentate ligand with the two nitrogen atoms forced in permanent juxtaposition due to its rigid planar structure.A further peculiarity of phen and its substituted derivates is the poor σ-donor ability of the heteroaromatic nitrogen atoms and their π-electron deficiency, making them excellent π acceptors (Anderegg, 1963a(Anderegg, , 1963b)).These structural features, namely its planarity, rigidity, hydrophobicity, and electron deficiency, determine the metal coordinating ability of phen and the resulting application of its chelates.For example, the planarity of phen allows its intercalation or groove binding with DNA or RNA whilst the [Cu(phen) 2 ] 2+ complex is commonly used as DNA marker and cleaving reagent (Meijler et al., 1997).
Phen presents a high complexation constant with d, and to a lesser extent f, block cations due to their stabilisation by favourable enthalpic contributions arising from the formation of strong coordination bonds with the nitrogen atoms (Anderegg, 1963a(Anderegg, , 1963b;;Hancock, 2013).Furthermore, the rigidity of phen and the permanent juxtaposition of the nitrogen donor atoms provides it an entropic advantage relative to its 2,2′-bipyridine analogue, resulting in enhanced kinetics of complex formation as well as increased surface activity at the oil-water interface due to its hydrogen bonding with water molecules (Burchett and Meloan, 1972;Lewis et al., 2011).The presence of low-energy π* orbitals gives rise to strong metal-to-ligand charge-transfer (MLCT) absorption bands with the aforementioned metals resulting in a range of luminescent phen compounds with emission bands spanning from the UV to the near infrared region (Accorsi et al., 2009).These properties can be further adjusted by the judicious grafting of substituents on the phen backbone and modification of the J o u r n a l P r e -p r o o f Journal Pre-proof aqueous phase composition, namely pH (pK b (phen) = 4.86) (Accorsi et al., 2009;Bencini and Lippolis, 2010).
Increased organisation of phen derivatives and the covalent attachment of moieties to yield pincer-type ligands presenting size selective apertures were shown to increase the complexation constant and selectivity for a target cation (Carolan et al., 2012;Lewis et al., 2011;Tucker et al., 2020;Xu et al., 2016).An unexplored potential avenue to tune the extraction behaviour is by altering the energetic barrier to complexation through the inclusion of phen in pre-structured hydrogen-bonded assemblies.Type V deep eutectic solvents (DES) are binary mixtures of non-ionic compounds presenting an extreme eutaxia manifested as an important melting point depression relative to that predicted assuming thermodynamic ideality (Abranches et al., 2019).This non-ideal behaviour is assigned to the stronger intermolecular interactions between the DES components relative to those present in the pure precursors and is reflected in their solid-liquid equilibrium (SLE).Contrary to the dominant ionic interactions in Type I to IV DES (Smith et al., 2014), weak interactions, namely hydrogen bonding or other non-covalent interactions such as halogen bonding, dictate selfassembly in Type V DES (Abranches and Coutinho, 2022;Peloquin et al., 2021;Percevault et al., 2020;Schaeffer et al., 2021).Type V DES incorporating at least one component with a chelating polar group and natural extract (terpenes, fatty acids, etc) show great promise for solvent extraction (SX) application as benign alternatives to conventional petroleum-derived diluents or ionic liquids.A range of desirable properties are obtainable through the appropriate selection of the hydrogen bond donor (HBD) and acceptor (HBA) including a wide liquidus composition, synergistic decrease in water solubility, moderate viscosities, and increased metal loading due to extractant concentrations of ≥ 50 mol% (Byrne et al., 2020;Gilmore et al., 2018;Schaeffer et al., 2020;Vargas et al., 2021).Furthermore, the varied energetic landscape provided by the extensive number of hydrogen bonded configurations in the DES phase accommodates the significant extraction of polar solutes and suppresses the third-phase formation that hinders traditional SX at higher metal loadings (Vargas et al., 2021).Type V DES were successfully applied to the extraction of transition metals (Schaeffer et al., 2020(Schaeffer et al., , 2018)), Pd(II) and Pt(IV) (Vargas et al., 2021), U(IV) (Gilmore et al., 2018), As(III) (Rajput et al., 2021), In(III) (Edgecomb et al., 2020), Li(I) (Hanada and Goto, 2021), and Sm(III) (Ni et al., 2021).
Importantly for their application, Vargas et al. (2021) demonstrated that metal extraction in Type V DES presents a non-linear behaviour dependent on the extractant molar fraction and J o u r n a l P r e -p r o o f Journal Pre-proof the extent of the thermodynamic deviation ideality.As such, extraction in a given DES must be rationalised with regards to its phase diagram.
Although a theoretically extensive number of Type V DES are obtainable through variations in HBD and HBA selection, reported mixtures for SX rely on a limited number of chelating groups, namely phosphine oxide, carboxylic acid, and ketones.Whilst the solid-liquid phase diagram of phenol and pyridine was reported over a century ago (Bramley, 1916), to the best of our knowledge no DES with polypyridyl ligands were reported despite the numerous relevant properties of their complexes.The incorporation of these chelating agent paves the way for a new family of Type V DES and the development of functional fluids with industrially relevant properties, from the SX of metal ion for analytical pre-concentration to luminescent materials.In this work, the DES composed of phen with the prototypical HBD thymol is introduced and its relevant physical-chemical properties characterised (Abranches et al., 2019).Its use in SX for a range of metal ions is presented and the extraction mechanism elucidated.Additionally, the luminescent properties of the loaded DES phases after extraction are briefly addressed.The results herein provide a framework for the development of new and versatile class of Type V DES.

J o u r n a l P r e -p r o o f
Journal Pre-proof All systems were gravimetrically prepared through careful weighing of each component using an analytic balance Mettler Toledo XP205 (USA; uncertainty 0.0001 g).Proton nuclear magnetic resonance ( 1 H-NMR) analysis was performed using a Bruker Avance 300 (USA), operating at 75 MHz.Deuterated water with tetramethylsilane as the reference was added to the sample in a co-axial insert.Raman spectra were acquired on a Bruker MultiRAM (USA) equipped with a 1064 nm Nd-YAG laser source and a nitrogen cooled germanium detector.
All spectra were recorded in the 3600 to 100 cm -1 range with a resolution of 4 cm -1 (laser power of 100 mW and 400 scans) and deconvoluted into Gaussian components using the Origin software package.Fluorescence analysis of the loaded DES phase following extraction of Co(II), Ni(II) or Mn(II) was performed using a Jasco FP -8300 fluorimeter (USA) under emission mode (excitation band of 425 nm) in the spectral range 435 -750 nm, scan 200 nm min -1 .Samples were diluted in acetonitrile except for manganese samples, which were diluted in methanol due to the appearance of a precipitate.Powder X-ray diffraction (P-XRD) was determined using a Malvern Panalytical Empyrean (UK) with Cu Kα radiation operating with a Cu anode (K α1 = 1.5406Å; K α2 = 1.5444Å) at 45 kV and a spinning discs sample holder.Diffractograms were analysed using the X'pert Highscore Plus software package and the patterns matched against those in the International Centre for Diffraction Data (ICDD) database.
Metal quantification was performed using a Picofox S2 (Bruker Nano, USA) total reflection X-ray fluorescence spectrometer (TXRF) with a molybdenum X-ray source.The voltage of the X-ray tube was 50 kV and the current 600 μA.All carriers were first treated with 10 μL of silicon in isopropanol solution and dried at 353 K for 30 min.Aqueous phase samples were diluted in 0.5 wt.% polyvinyl alcohol acidified at pH = 2.0 using HCl and spiked with a known concentration of YCl 3 (1000 ppm Certipur ® standard, Sigma-Aldrich) to obtain a final standard element concentration of 1 to 10 ppm (mg L -1 ) depending on the sample metal content.Ten microliters of each solution containing the metals and yttrium standard were added onto a treated carrier and dried on a hot plate at 353 K for 30 min.The acquisition time was varied from 120 to 1000 seconds in function of the sample concentration.Lithium was quantified after dilution using an ion-selective electrode (Mettler Toledo Dx-series, USA) using a LiCl calibration curve from 0.1 to 10.0 mmol L -1 .Magnesium sulfate was used to maintain a constant ionic strength.

Determination of phase diagrams and SLE modeling
J o u r n a l P r e -p r o o f Journal Pre-proof Binary mixtures of phen with a HBD (thymol, chlorothymol or BHT) were prepared in different proportions covering the full composition range (at 0.1 mole fraction intervals) using an analytic balance Mettler Toledo XP205 under ambient conditions.Samples were heated under stirring at 353 K until a homogenous liquid was obtained and then stirred for an additional 30 min.Samples were left to crystallise at room temperature.Two experimental methods were applied to determine the melting temperatures depending on the physical state of the resulting mixture.For fully re-crystallised mixtures, samples were ground and packed into a glass capillary and the melting temperature measured using an automatic glass capillary device model M-565 from Buchi (Switzerland) with a temperature resolution of 0.1 K min -1 .For mixtures presenting a pasty texture (partial re-crystallisation), these were gradually heated until complete melting and the temperature was controlled with a PT100 probe with a precision of ±0.1 K.The probe was previously calibrated against a certified platinum resistance thermometer, SPRT100 (Fluke-Hart Scientific 1529 Chub-E4), traceable to the National Institute of Standards and Technology (NIST).Each procedure was repeated at least three times with the average result reported.
The SLE of the eutectic mixtures with complete immiscibility in the solid phase can be described by equation (1) (Prausnitz et al., 1999): where x i is the mole fraction of component i and γ i l its activity coefficient in the liquid phase, T is the absolute temperature, R is the ideal gas constant, T m,i , ∆ m H i and ∆ m Cp i are the melting temperature, enthalpy and heat capacity change of the pure solute, respectively.
When the differences between the eutectic and the melting temperatures of the pure compounds are small (< 100 K), as is the case in the studied systems (see Figure 1), the term related with heat capacities may be neglected due to its small contribution to the phase equilibrium calculations (Coutinho et al., 1995), yielding Equation (2): When ideality is assumed, γ i l equals to 1 and the solubility curves are derived from Equation (2).The experimental activity coefficients are obtained through Equation ( 2) using the experimental temperature data.

J o u r n a l P r e -p r o o f
Journal Pre-proof The stability and properties of the thymol+phen eutectic system for a phen molar fraction of x Phen = 0.2 was assessed as a function of the aqueous phase composition after contacting the eutectic phase with an aqueous phase of varying acidity from 0.0 to 1.0 mol L -1 of HCl in the absence of metal ions.The samples were left to equilibrate for 2 h at 298 K and an organic to aqueous phase ratio (O:A) of 0.5 prior to centrifugation at 10,000 rpm for 10 min and isolation of the DES phase.The water content of the DES phase was determined by Karl

Metal extraction in the thymol+phen eutectic
Metal ion extraction in the thymol+phen eutectic was assessed as a function of the aqueous phase acid concentration (0.0 to 3.0 mol L -1 ), counter anion selection (chloride, nitrate, sulphate), eutectic mol fraction (x Phen = 0.2-0.3)and metal to phen ratio.Unless otherwise specified, standard conditions were: metal concentrations of 0.02 mol L -1 , [HCl] = 0.05 mol L -1 , a phen molar fraction 0.2 and an organic to aqueous phase ratio of 0.50 (total volume of 1.5 mL).The systems were mixed under constant agitation of 1000 rpm for 1 min using an Eppendorf Thermomixer Comfort equipment (Germany) and thereafter left to rest for 45 min at room temperature.The systems were centrifuged for 10 min and rotated at 10,000 rpm and the phase separated with the aqueous phase analyzed by TXRF and ion-selective electrode potentiometer following appropriate dilution.All analyses were performed in triplicates with the average result reported.Metal ion (M n+ ) distribution coefficients (D M ), extraction efficiencies (%EE M ), and the separation factor between two given metals (α M1/M2 ) are calculated using Equations (3) to (5) respectively: J o u r n a l P r e -p r o o f Journal Pre-proof where the subscripts ini and f indicate the sampling timeframe before (ini) or after (f) extraction whilst DES and aq indicate the DES and aqueous phase, respectively.The DES phases after extraction of Co(II), Ni(II) or Mn(II) were further analyzed by fluorescence analysis following phase separation.
The detection limit of thymol+phen DES (x Phen = 0.2) for FeCl 3 was performed by UV-vis analysis of the DES phase.Briefly, 0.25 mL of DES were agitated for 1 min with various aqueous FeCl 3 solutions at pH = 5.5 and O/A = 0.25 (total volume 1.25 mL) and centrifuged thereafter.The DES phase was analysed by UV-vis using a Synergy HT microplate reader from Biotek Instruments (USA).The DES phase volume per sample well was 0.15 mL.

Simulation
Molecular dynamic (MD) simulations of the thymol+phen system for x phen = 0.2 (400 molecules of thymol and 100 molecules of phen) in the presence of water and HCl were carried out using Gromacs 5.1 package (Abraham et al., 2015) within the NpT ensemble by adopting the leapfrog algorithm to integrate the equations of motion at a fixed temperature (298 K) and pressure (1 bar) (Hockney et al., 1974).The non-polarizable OPLS all atom force field was used for phen and thymol, the optimised OPLS-AA forcefield was used for H 3 O + and Cl - (Bonthuis et al., 2016), and water molecules were represented by the SPC/E model (Berendsen et al., 1987).Phen and thymol partial charges were refined by DFT calculations using Gaussian09 (Frisch et al., 2009) employing Becke's three-parameter exchange (Becke, 1988) in combination with the Lee, Yang, and Parr correlation functional (Lee et al., 1988) at the B3LYP/6−311+G(d,p) level of theory.Atomistic charges were determined using the CHELPG methodology and applied with no charge scaling factor (Breneman and Wiberg, 1990).Hydrogen bonds were constrained by the LINCS algorithm (Hess et al., 1997) whilst LJ and Coulombic interactions were computed up to a cut-off radius of 1.2 nm.Long-range Coulombic interactions were evaluated by particle mesh Ewald (PME) (Darden et al., 1993).The temperature and pressure were controlled through the Nose-Hoover thermostat (Evans and Holian, 1985) and the Parrinello-Rahman barostat (Parrinello and Rahman, 1981), respectively.All simulations were started from a randomly distributed configuration and production runs were carried for 100 ns.The MD simulation outputs were visualized using the VMD software package (Humphrey et al., 1996).Domain analysis was

J o u r n a l P r e -p r o o f
Journal Pre-proof obtained using the TRAVIS post-processing package (Brehm et al., 2015a;Brehm and Kirchner, 2011).
The Conductor like Screening Model for Real Solvents (COSMO-RS) is a statistical thermodynamics model that predicts the chemical potential (and activity coefficients) of individual components in liquid mixtures (Eckert and Klamt, 2002;Klamt, 1995;Klamt et al., 1998).To do so, it relies on pair-wise interactions between screened charge surface patches of each molecule (the so-called σ-surface), obtained using DFT and the COSMO continuum solvation model.Given its ability to predict activity coefficients, COSMO-RS can be coupled with Equation 2 to estimate SLE phase diagrams.This was explored in this work to predict the SLE phase diagrams of thymol+2,2'-bipyridine.The necessary σ-surfaces were obtained using the COSMO-BP-TZVP template of the software package TURBOMOLE (TURBOMOLE V7.1 2016), which utilizes the def-TZVP basis set, the BP-86 DFT functional, and the COSMO solvation model with ideal screening (infinite permittivity).All COSMO-RS calculations were performed in the software package COSMOtherm (BIOVIA COSMOtherm, 2021) using the BP_TZVP_21 parametrization.

Phase diagram
Considering the breadth of phen-based DES applications that can be modulated through rational HBD selection, the following discussion summarizes the HBD influence on the HBD+phen phase diagram thereby providing a guideline for the design of phen-based Type V DES.The presence of hydrogen bonding in-itself is not a guarantee of "deep" eutectic solvent formation especially if the individual components already establish hydrogen bonds in their pure state.Whilst the ∆pK a difference between the hydrogen bond donor and acceptor couple provides an initial estimation of the hydrogen bond strength (Gilli et al., 2009), components presenting an important degree of self-association (e.g.carboxylic acid dimers) tend to yield ideal mixtures as was reported for alcohol+carboxylic acid systems (∆pK a > 10) (Martins et al., 2018).The addition of electron-withdrawing groups by resonance effect in the case of phenolic compounds (Abranches et al., 2019)  Journal Pre-proof the concept of "anti-crystal engineering" design rule when selecting HBA/HBD combinations to promote low melting temperatures and avoid the formation of co-crystals (Dean et al., 2009).This is relevant in the case of phen-based systems as its lack of configurational entropy imposed by its symmetry and rigidity makes the formation of co-crystal likely due to the directionality of the resulting HBA-HBD pair.
To illustrate these principles, the solid-liquid phase diagrams of phen with three phenolic compounds as HBD, namely thymol, chlorothymol or 2,6-di-tert-butyl-4-methylphenol (BHT) as HBD (structures shown in Figure 1A), were determined and presented in Figure 1B.The SLE data and derived activity coefficients are available in Table S1 and Figure S2 of the ESI, respectively, whilst the melting properties used to calculate the ideal solid-liquid phase diagrams (dashed lines) are available in Table S2.Importantly, anhydrous phen was used in this study; phen is often available in its monohydrated form and the presence of water is known to induce strong changes on the SLE (Abranches et al., 2021).Although the systems were prepared under ambient conditions, care was taken to minimise water exposure and the maximum water content of the prepared systems never exceeded 0.32 wt.% H 2 O.
Thymol, in addition to being a natural extract, was previously identified as an excellent HBD in systems with alcohols (Abranches et al., 2019) or phosphine oxide (Schaeffer et al., 2020) and serves as the starting point of this study.For comparative purposes, the reported SLE diagram of phenol+pyridine is shown in Figure 1C (Bramley, 1916).The thymol+phen system is characterised by two eutectic points at approximately x phen ~ 0.3 and 0.5, similarly to that in the phenol+pyridine system.A quasi-ideal behavior is observed on the phen-rich side of the phase diagram whilst important negative deviations from ideal behavior on the thymol-rich side (γ < 0.5, Figure S2) yields a useable liquidus range of x phen = 0.2 to 0.3 at 298 K.The XRD of the thymol+phen eutectic for x phen = 0.5 in Figure S3 shows the emergence of additional peaks relative to that of the starting components or x phen = 0.9, suggesting the presence of a different solid phase at that composition.

J o u r n a l P r e -p r o o f
Journal Pre-proof Substitution of the para hydrogen of thymol by the electron-withdrawing chloride in chlorothymol results in a markedly different SLE diagram (Figure 1B) with an increased liquidus compositional range from x phen = 0.3 to 0.5 at 298 K and the likely existence of a peritectic around x phen = 0.7.Contrary to the thymol+phen system, the chlorothymol+phen system presents a maximal negative deviation from ideality on the phen-rich side for x phen = 0.7 (Figure S2) resulting in a melting point depression of over 40 K at that composition.
Finally, the SLE of the BHT+phen system is defined by a eutectic point at x phen = 0.8 and the clear formation of a co-crystal with a 1:1 stoichiometry confirmed by XRD analysis (Figure S4).This is to be expected due to the steric constraints imposed by the two tert-butyl groups of BHT positioned either side of the phenolic proton and forcing a 1:1 perpendicular interlocking arrangement of BHT with phen.The analysis of the systems in Figure 1B is

J o u r n a l P r e -p r o o f
Journal Pre-proof simplified by the lone HBA nature of phen and the permanent juxtaposition of its nitrogen atoms.Although not experimentally studied within this work, COSMO-RS simulation suggests that replacement of phen with the more configurationally flexible 2,2'-bipyridine presents a more complex SLE that depends on the equilibrium of the trans and cis isomer in the eutectic at a given composition as shown in Figure S5.As such, care must be taken when transposing phen-based solid-liquid diagrams to the equivalent 2,2'-bipyridine systems.

Characterisation of the thymol+phen systemviscosity and density
In the following sections, the discussion focuses on the thymol+phen system at the molar fraction x phen = 0.2 with the system characterization revolving around the relevant properties for SX application, namely viscosity, density, chemical stability and co-extraction of water and acid.The composition of x phen = 0.2 was chosen as compromise between expected performance ([phen] = 1.28 mol kg -1 ) and toxicity of the DES, with phen presenting a LD50 (oral, rat) of 132 mg kg -1 compared to 980 mg kg -1 for thymol (SigmaAldrich, 2022).Three different systems were studied to understand the evolution of the viscosity and density profiles in the DES phase as a function of the aqueous phase composition in SX: (i) the "dry" DES after preparation and after equilibration with (ii) water or (iii) 1.0 mol L -1 HCl.These are presented in Figure 2 as a function of temperature from 293.15 to 323.15 K and summarized in Table S3-S4.The viscosity (η) was fitted with the Arrhenius equation (Equation 6) whilst a linear fit was employed to fit the density data.The fit parameters are summarised in Table S5.
where A is the pre-exponential factor (Pa s), E A the activation energy (J mol -1 ), R is the universal gas constant (J mol -1 K -1 ) and T is the temperature (K).Whilst the fit of the viscosity data was satisfactory at the studied conditions (R 2 > 0.99), it is likely that the Vogel-Tamman-Fulcher equation would provide a better estimation for temperatures nearing the solid-liquid transition as the system non-ideality becomes more pronounced.

J o u r n a l P r e -p r o o f
Journal Pre-proof  S5.
Viscosity provides a macroscopic indication to the strength and extent of the intermolecular hydrogen bond network in DES.At 298.15 K, the thymol+phen DES has a viscosity of 75.31 mPa s, comparable to that in the strongly non-ideal thymol+trioctylphosphine oxide (x thymol = 0.5, η = 69.93mPa s) (Schaeffer et al., 2020) but more than for quasi-ideal mixtures such as thymol+decanoic acid (x thymol = 0.5, η = 12.16 mPa.s) (Martins et al., 2018).Water saturation of the DES decreases the viscosity to 63.94 mPa s whilst equilibration with a 1.0 mol L -1 HCl solution increases it to 99.63 mPa s.As will be discussed below, these changes are assigned to the (i) weakening of the HBD-HBA interaction upon inclusion of water and (ii) increased re-structuration of the DES phase to accommodate the presence of ionic solutes coming from the acid (H + and Cl -) and protonated phen ([phenH] + ).In all cases, the activation energies (E A ) exceed 50.0 kJ mol -1 with the pure DES exhibiting the highest E A = 57.1 kJ mol -1 (Table S5) and is consistent with that of viscous fluids.Problematically for its application in SX, densities of the tested DES systems are close to that of water making phase de-mixing difficult in the absence of centrifugation.The concentration of polar solute in the DES phase increases its density, resulting in an aqueous/organic phase inversion upon metal extraction in 1.0 mol L -1 HCl (vide infra).

J o u r n a l P r e -p r o o f
Journal Pre-proof

Characterisation of the thymol+phen system -pH dependency and structure relationship
In aqueous media, phen acts as a weak base with a pK b of 4.86 and presents pH-dependent properties which can translate into diminished metal complexation (Accorsi et al., 2009).
Furthermore, formation of the [phenH]Cl salt will have strong repercussions on the properties of the DES as the melting point of [phenH]Cl monohydrate is of approximately 497 K, far above that of neutral phen (Honeywell Research Chemicals, 2022).However, protonation constants are known to vary by several orders of magnitude in organic media (Sarmini and Kenndler, 1999) making the translation of pK b from the aqueous to the DES phase impossible.Although the pK b of phen-derivatives is often determined by titration using UVvis, the important phen concentration and molar attenuation coefficient poses a practical problem as excessive dilution breaks the defining structure of DES into its simple solvated components (Hammond et al., 2017).As such, the protonation of phen in the thymol+phen eutectic was followed by 1 H-NMR by equilibrating the DES phase with aqueous solutions of varying acidity from 0.0 to 1.0 mol L -1 (pH = 0.0) at an O:A = 0.50.By placing the deuterated solvent as a co-axial insert, the DES phase was not diluted for analysis.The resulting spectra, focusing on the aromatic and hydroxyl protons, are presented in Figure 3.
Complimentarily, the extracted water and acid content to the DES phase was determined for each system and are reported in Table S6.

J o u r n a l P r e -p r o o f
Journal Pre-proof  S6).For an initial aqueous phase HCl concentration of 1.0 mol L - 1 , a final HCl content in the DES phase of 0.18 mol kg -1 was obtained, representing a H + to phen ratio of ~0.14.Furthermore, the protonation of phen could be visually followed as the DES phase took on a strong yellow color when in contact with 1.0 mol L -1 of HCl consistent with the red-shifted spectra of [phenH] + relative to phen (Armaroli et al., 1992).Overall, it appears that partial protonation of phenanthroline in DES occurs for an aqueous phase pH of ≤ 1.0, almost four orders of magnitude lower than for the pure compound in water.A series of shift in Figure 3 identified by the triangle symbols do not appear in all spectra and likely reflect some broad ππ interactions and/or resonance effects upon protonation of phen.
The favorable free energy of complexation driving ion partition in SX is quenched by the energetic penalty from (i) solvent nanostructure reorganization around the extracted and coextracted species, (ii) the confinement of ionic solutes in small polar volumes and (iii) excess free energy of the water/ion mixture (Bourgeois et al., 2020).Considering the non-ideality of DES stemming from their pre-organised hydrogen bonded structure, any significant changes to the latter from the partition of acid and water is liable to affect the extraction mechanism.
In fact, Pd(II) extraction in the thymol+trioctylphosphine oxide was shown to occur only upon HCl co-extraction and the breaking of the DES hydrogen bonded-network (Vargas et al., 2021).As such, the concentration of extracted water and acid to the DES phase in Table S6 was after which a non-negligible amount of acid and water are co-extracted.These extracted polar solutes make up one third in mol.% of the final DES phase in equilibrium with an aqueous phase of [HCl] = 3.0 mol L -1 , yielding a final H + to phen ratio of 0.31.
To further probe the influence of these polar solutes on the DES structure, four MD simulations of the thymol+phen system at fixed ratio of 4:1 were carried out with increasing polar solute concentrations mirroring the experimental compositions, namely (1) pure DES, (2) x H2O = 0.12, (3) x H2O = 0.17 and (4) x H2O = 0.16 with x HCl = 0.02 (simulated as the dissociated Cl -and H 3 O + ions).The nano-structuring of the DES phase was estimated by calculating the number of polar domains and the volume of the largest domain based on the Voronoi tessellation method (Brehm et al., 2015b).The polar domains are defined as the subset composed of H 2 O, H 3 O + and Cl -molecules, the nitrogen atoms of phen and hydroxyl group of thymol, with the results presented in Figure 4B.A domain number greater than 1 is indicative of a dispersed subset.Additionally, representative snapshots of the equilibrated simulations with increasing polar solute fraction are provided in Figure 4C to illustrate the discussion.Addition of polar components in the DES phase results in the reduction in the number of discrete molecular-scale domains as these coalesce into larger assemblies as shown by the important polar domain volumes.Interestingly, a small addition of HCl did not change the domain number for a constant x H2O but yielded an increase in the maximum volume which is expectable upon the introduction of electrostatic interactions to a non-ionic system.Overall, the observed behaviour in Figure 4 parallels the hierarchical self-assembly observed in the organic phase of conventional SX systems, albeit without third-phase formation (Clark, 2019;Motokawa et al., 2019).

J o u r n a l P r e -p r o o f
Journal Pre-proof

Co(II), Ni(II), and Mn(II) extraction in the thymol+phen system
Following its characterization, the potential of the thymol+phen eutectic for the SX of metal cations is now addressed.The extraction behaviour of Mn(II), Co(II), Ni(II) and Li(I) from multi-elemental solutions was followed in detail as a function of the HCl concentration in  with a small rise in partition observed as the x phen is increased.
Unexpectedly given that log(K 1 ) decreases from Ni(II) > Co(II) > Mn(II) (Hancock, 2013), ( Anderegg and Wanner, 1986;Hancock, 2013).The %EE M and D M values are available in Table S11.Unless otherwise specified, standard conditions were: metal concentrations of 0.02 mol L -1 , x phen = 0.2 and an O/A ratio of 0.50 (total volume of 1.5 mL).Additionally, the water content of the DES phase following metal cation partition was measured.However, due to the low metal concentration used, no significant difference between metal systems were observed and a similar average water concentration as for the equivalent system without metal (Table S6) of [H 2 O] DES = 1.49± 0.09 wt.% was obtained.(0.20 μmol L -1 ) (Hirayama et al., 2013), N-aryl-O-acylhydroxylamine (0.50 μmol L -1 ) (Xuan et al., 2016), arene-based fluorescent probes (8.54 μmol L -1 ) (Kumar et al., 2015), benzimidazolyl pyridine (0.28 μmol L -1 ) (Bhaumik et al., 2011), and phen functionalised polymers (0.05 μmol L -1 ) (Nawaz et al., 2018).A recent study applied the hydrophobic eutectic composed of tetrabutylammonium chloride and decanoic acid with phen as chelating agent for the microextraction and determination of Ni(II) and Zn(II) by flame atomic absorption spectroscopy from various foods (Elahi et al., 2022).This suggests that if properly optimised, the system studied herein can be extended to other metals after considering the metal analysis technique, the potential interference of other ions as well as the usable pH range.

Fischer
(KF) titration using a Titroline 7500 KF Trace (Metrohm, Switzerland) whilst the acid concentration was determined by acid-base titration before and after contacting using phenolphthalein as an indicator.The loss of phen from the DES to the aqueous phase was measured by UV-vis on a Shimadzu UV-1800 spectrophotometer (Japan), with the obtained calibration shown in FigureS1.Analyses were performed in triplicate with the average result reported.Viscosity was determined using a viscometer-densimeter SVM 3000 Anton Paar Rotary Stabinger (Austria), with a temperature uncertainty of ± 0.1 K and a relative uncertainty of dynamic viscosity of ± 0.35 %.A thermostatic bath (ME-18 V Visco-Thermostat, Julabo, (Germany) ± 0.1 K) was used for temperature control.
or by halogenation (hydrogenated vs fluorinated alcohols) (Isabel Cabaço et al., 2021) minimises self-association by increasing the acidity of the HBD group whilst reducing the basicity of the HBA.A final consideration is J o u r n a l P r e -p r o o f

Figure 3 .
Figure3. 1 H-NMR analysis of the initial aqueous phase acidity influence on the thymol+phen eutectic phase (x phen = 0.2; O:A = 0.50).Stars and triangles represent unassigned peaks that do not appear in all spectra.
converted to molar percentage to better follow the change in the DES composition with the results shown in Figure 4A.To simplify the calculations, it was assumed that no J o u r n a l P r e -p r o o f Journal Pre-proof significant transfer of the DES component to the aqueous phase takes place.As will be shown later, this assumption holds only for aqueous phase pH values above the protonation of phen, below this pH the partition of [phenH]Cl to the aqueous phase occurs.In agreement with the 1 H-NMR results in Figure 3, the DES composition is unchanged until [HCl] ≥ 0.1 mol L -1

Figure 4 .
Figure 4. A) Influence of the initial aqueous phase HCl concentration on the thymol+phen eutectic phase composition at equilibrium (x phen = 0.2; O:A = 0.50).The extracted water and acid content to the DES phase was determined for each system and are reported in Table S6.Square symbols represent the composition in the pure DES.B) MD-derived number of polar domains and volume of the largest continuous one in the DES phase as a function of the H 2 O and HCl molar fraction in the system.C) Representative examples of the change in the dominant aggregate obtained from MD simulations upon increased concentration of polar solutes.

Figure 5A .
Figure 5A.These metals were selected due to their importance in achieving a green energy future and their forecasted demand in various associated applications including lithium ion batteries for energy storage(Herrington, 2021).Additionally, the influence of x phen molar

Figure 5 .
Figure 5. Extraction of Mn(II), Co(II) and Ni(II) in the thymol+phen eutectic as a function of the HCl concentration in the presence of [Li(I)] = 0.08 mol L -1 .Influence of B) x phen molar fraction and C) Co(II) to phen ratio on the extraction.Standard conditions of [M] = 0.02 mol L -1 , pH = 5.5, x phen = 0.2 and O/A = 0.5 were applied unless specified otherwise.D) Aqueous phase concentration of phen after equilibration of thymol+phen eutectic (x phen = 0.2 and O/A = 0.5) with an aqueous phase of varying acidity, shown in blue.For comparison, the aqueous solubility of pure phen and [phenH]Cl were taken from (Burgess and Haines, 2002) and (Honeywell Research Chemicals, 2022) respectively and are shown in orange.E) results indicate a preferential extraction of Co(II) > Ni(II) > Mn(II).To better understand the unexpected behaviour, Raman analysis of the loaded DES phase was performed for the individual transition metals ([M(II)] = 0.10 mol L -1 ) at pH = 5.5 and is presented in FigureS8.The analysis focuses on the ring deformation vibrations of phen in the 700 to 740 region cm −1 which were shown to be sensitive to the interaction with transition metals(Muniz-Miranda et al., 2010;Thornton and Watkins, 1991).The Raman bands observed at 711 cm −1 (green) and 740 cm −1 (orange) do not shift compared to the DES phase with no metal, but there is a change in their relative intensity and the appearance of a band at 725 -730 cm −1 (blue).The observed frequency change at ~730 cm -1 increases when going from Co(II) to Ni(II) and Mn(II) whilst the band at 711 cm −1 decreases.This is likely due to the strength of the interaction between the ligand and the metal and are consistent with the preferential extraction of Co(II) > Ni(II) > Mn(II) in the thymol+phen DES.The metal saturated DES-phase were also studied by fluorescence analysis as the emission spectra is sensitive to changes in the molecular structures of the complex(Carter et al., 2014), shown in Figure5E.The corresponding UV-vis spectra are available in FigureS9.The results show that after excitation at 425 nm corresponding to the MLCT band of the complexes, all metal complexes exhibit similar emissions in the spectral range between 480-496 nm with small but notable differences.The Mn(II) loaded DES phase exhibits emission bands at 467, 489 and at 530 nm while for Ni(II), two broad emission bands were observed centred at 480 nm and 536 nm.The Co(II) spectra is characterised by the splitting of the lower wavelength band resulting in four bands at 462, 477, 497 and 547 nm and a change in the relative intensity of the bands compared to the spectra of Ni(II) and Mn(II).Although not conclusive, spectroscopic results suggest the preferential extraction of Co(II) over Ni(II) and Mn(II) is driven by the greater strength of the Co(II)-phen interaction and small changes in the DES phase complex.However, it remains unclear as to what is the underlying reason for this divergence between the aqueous log(K 1 ) of complexes and the extraction profiles when phen is incorporated into a DES.Additional work probing the speciation of metals in the DES phase could provide further insight as to the differences between the extracted complexes.J o u r n a l P r e -p r o o f Journal Pre-proof Metal extraction in the thymol+phen systemextension to other systems In this final section, the metal extraction studies in the thymol+phen system are extended to other metal families and the potential of the eutectic for metal pre-concentration and analysis is demonstrated for Fe(III) quantification.The extraction of selected metals representing different groups, namely PGMs (Pd(II)), heavy metals (Cd(II)), transition metals, alkaline earth metals (Ba(II)) and lanthanides (La(III)), was determined at [HCl] = 0.05 mol L -1 .The extraction efficiencies are shown in Figure 6A and the derived distribution coefficients are plotted against the aqueous first formation constant for each metal with phen in Figure 6B

Figure 7 .
Figure 7. A) Naked eye and B) UV-vis quantification of FeCl 3 in the thymol+phen eutectic (pH = 5.5, O/A = 0.25, x phen = 0.2).The photograph in A) was taken using an iPhone 12 and presented without modification.Please refer to the online journal edition for the coloured version of this Figure.