The Complex Story Behind a Deep Eutectic Solvent Formation as Revealed by L ‑ Menthol Mixtures with Butylated Hydroxytoluene Derivatives

,


■ INTRODUCTION
Deep eutectic solvents (DESs) are currently the ruling research topic about alternative media to traditional solvents owing to key properties like simple preparation, no need for purification, versatility, and an environmentally benign pedigree. 1−3 DESs also belong to the "designer" solvents category, meaning that their chemico-physical properties can be tuned to meet specific requirements through a judicious choice of the precursors and of their relative composition. 1 These qualities can potentially provide materials ideally suited for application fields among which the energy production and storage, 4,5 electroplating, 6,7 separation procedures, 8,9 CO 2 sequestration, 10 catalytic processes, 11,12 and drug delivery 13,14 represent a long but inevitably spurious list. The definition of these systems has been for a long time a source of debate in the scientific community, but at present an agreement seems to have been found in defining DESs as mixtures of two or more compounds showing a melting point (MP) depression that is significantly deeper than the ideally predicted one for at least one composition. 15,16 Moreover, the MP of the eutectic is lower than those of the individual parent compounds, thus allowing the achievement of a liquid phase even from solid starting materials. This behavior was described for the first time by Abbott and co-workers, who reported that by mixing choline chloride and urea, both solids at room temperature, a liquid phase with an MP of 12°C was obtained for the 1:2 molar ratio, which was called "reline". 17 Nevertheless, the extent of deviation from thermal ideality required to provide a "true" DES remains an ambiguous task.
The majority of the currently available DESs, classified in the type I−IV categories, is formed by at least one ionic component, so that the origin of the MP depression is typically attributed to complexation of a hydrogen-bond (Hbond) acceptor from a H-bond donor. 2,3,18−24 More recently, type V DESs have been presented and are formed by neutral molecular components only, very often based on terpenoid compounds like menthol and thymol. 25,26 Being nonionic, they typically display lower viscosities, are chloride-free, and often hydrophobic, so that they are suited for the removal of target analytes from aqueous solutions through liquid−liquid biphasic extractions. 27−29 However, also, this category deals with species possessing H-bonding capabilities. For these reasons, DESs are often indicated as extensively H-bonded media, which, although not being a strict definition for these systems, has become a common way to identify them. Examples of alternative interactions concurring as a driving force for the MP depression are scarce in the literature and only consist of a few cases pointing out the establishment of halogen or chalcogen bonding between the components, 30−32 while apolar−apolar attractions led by dispersion interactions have also been invoked 1−3,33 but never clearly demonstrated.
In this way, many key scientific questions remain unanswered. On the one hand, the individuation of H-bonding as the only parameter to take into account in DESs' designing seems to start crackling, even if the overwhelming literature about salt-based DESs condemns this argument as still being in its infancy. On the other hand, the achievement of a clearer predictive understanding of the structure−property relationship necessary to design new task-specific DESs is a legitimate goal. Here, we start answering these questions by proposing an extensive study about the mixtures formed by L-menthol (MEN) with the butylated hydroxytoluene (BHT), 2-tertbutyl-p-cresol (TBC), and p-cresol (PC) compounds ( Figure  1). It was recently demonstrated that mixing of BHT and MEN at a 1:3 molar ratio (or, equivalently, at BHT molar fraction x BHT = 0.25) provides a liquid phase at room temperature suitable for liquid−liquid microextractions of pesticides and fat-soluble micronutrients from food matrices. 34,35 This system is very intriguing as the molecular structure of BHT shows an evidently encumbered hydroxyl group because of two adjacent tert-butyl substituents ( Figure  1). 36−39 Given that the hydroxyl group should be the main thing responsible for the H-bond interaction, we decided to study also MEN mixtures with TBC and PC counterparts, which are analogues of BHT but with respectively one and no tert-butyl groups in ortho-position to the hydroxyl. Note that the PC compound possesses acute toxicity, so that the PC/ MEN mixtures are far from meeting the requirements of sustainable solvents, but it was selected to solve the scientific case as it is the structural analogue species to the BHT and TBC ones but with no steric hindrance around the hydroxyl group. In this way, BHT/MEN, TBC/MEN, and PC/MEN mixtures covering the full composition range of the components have been studied with an array of complementary experimental and theoretical methods able to deliver information from the molecular length-scale aggregation to the macroscopic behavior. The employed techniques include thermal analysis aimed at building solid−liquid equilibrium (SLE) diagrams by means of differential scanning calorimetry (DSC) and polarized optical microscopy (POM), proton ( 1 H) nuclear magnetic resonance (NMR) and Raman spectroscopy, and molecular dynamics (MD) simulations at both the classical and density functional tight-binding (DFTB-MD) levels of theory. Altogether, our combined methods and the peculiarity of the chosen systems provided an archetypal study able to shed light onto various steric and electronic contributions influencing the structure−property correlation in DESs. The impact of this work is that of providing new tools for a more conscious DES design and to possibly expand the currently accepted view of these systems. ■ METHODS Chemicals and Sample Preparation. MEN (natural source, food grade, ≥99%) and BHT (food grade, ≥99%) were purchased from Merck Life Science S.r.l. (Milan, Italy), while TBC (≥98%) and PC (≥99%) came from TCI EUROPE N.V. (Zwijndrecht, Belgium) and were used as received. BHT/MEN, TBC/MEN, and PC/MEN mixtures at different molar fractions x BHT , x TBC , and x PC of the components covering the full composition range at ∼0.1 intervals were prepared by mixing the desired amount of the pristine compounds in glass test tubes and heating at 343 K until homogeneous liquids were obtained. The samples were then stored in glass sealed vials at room temperature before the measurements. Under these conditions, crystallization was observed for the systems whose MP was above room temperature, which were melted again before the withdrawal for the instrumental measurements. The full list of the prepared samples is reported in Table S1 of the Supporting  Information (SI). DSC and POM Experiments. The MPs of the various BHT/ MEN, TBC/MEN, and PC/MEN mixtures were determined by means of DSC and POM measurements to build the SLE diagrams for these systems. DSC thermograms were acquired with a Mettler Toledo DSC 822e differential calorimeter (Mettler Toledo, Greifensee, Switzerland), equipped with a ceramic FRS5 sensor and a liquid nitrogen cooler. A sample quantity of 5−10 mg was sealed in Figure 1. Molecular structures of the butylated hydroxytoluene (BHT), 2-tert-butyl-p-cresol (TBC), p-cresol (PC), and L-menthol (MEN) components within the employed atom nomenclature (white, hydrogen; gray, carbon; red, oxygen). a 40 μL aluminum pan and subjected to a temperature program consisting of a first cooling ramp at −2 K min −1 from 353 to 193 K, followed by a heating ramp from 193 to 353 at 1 K min −1 . During the measurements, the furnace was purged with dry nitrogen with a flow rate of 30 mL min −1 . The MP of each mixture was taken as the maximum value of the melting peak occurring at the highest temperature in the heating ramp.
POM analysis was performed on the samples showing no melt or cold crystallization during the cooling or heating scans in the DSC experiments (vide inf ra). An Optiphot2-Pol light microscope (Nikon) equipped with a Linkam HFS 91 hot stage was employed, driven by a Linkam TP 92 temperature controller and connected to a digital camera (Nikon D5000). All of the experiments were carried out under a N 2 flow. A sample quantity of about 10 mg was positioned on a glass slide, and the temperature program consisted of a first cooling ramp from 353 to 173 K at −90 K min −1 , followed by a heating ramp from 193 to 353 K at 5 K min −1 . The MP of each mixture was taken as the average temperature between the start and the end point of the crystal melting process. For the sake of comparison with the DSC results, POM measurements were carried out in the whole composition range.
Details about the building of the SLE phase diagrams are given in the SI.
Raman Spectroscopy Measurements. Raman spectra were collected at the HPS laboratory of Sapienza University of Rome using a Horiba HR-Evolution microspectrometer in backscattering geometry equipped with a solid state laser (λ = 532 nm, 90 mW power at the sample surface) and a set of neutral attenuating optical filters to avoid laser heating and sample degradation. The elastically scattered light was removed by a state-of-the-art optical filtering device based on three BragGrate notch filters, 40 which also allowed us to reach very low frequencies (down to ∼50 cm −1 from the laser line). The detector was a Peltier-cooled charge-coupled device (CCD), and the spectral resolution was better than 3 cm −1 thanks to a 600 groove/mm grating with an 800 mm focal length. The spectrometer was coupled with a confocal microscope supplied with a set of interchangeable objectives. To obtain the best signal-to-noise ratio, an objective with 20× magnification and an N.A. of 0.35 was employed to collect the full spectra in the 50−4500 cm −1 range. Raman spectra in the O−H stretching (ν OH ) region (3100−3800 cm −1 ) were collected with an acquisition time of 50 s. The measurements were carried out by including the liquid samples in a 1 mm quartz cuvette at 323 K thanks to a Peltier thermostated cell to ensure that the samples remained at the liquid state. The data on BHT were collected on a supersaturated solution in n-heptane, since the MP of the compound was above the temperature allowed by the instrumental setup. This solvent was chosen as it is expected to give no interference in the ν OH spectral region. For the same reason, data on the BHT/MEN mixtures were acquired only up to the x BHT = 0.70 sample. A polynomial baseline subtraction has been applied to all the spectra. NMR Spectroscopy. All samples were transferred into 5 mm NMR tubes, equipped with a capillary containing deuterated dimethyl sulfoxide (DMSO-d 6 ), and then flame-sealed. NMR measurements were performed at 343 K to ensure that all samples were at the liquid state, without sample spinning with a Bruker NEO 500 console (11.74 T) equipped with a direct observe BBFO (broadband including fluorine) iProbe and a variable-temperature unit. The instrument was carefully tuned, shimmed, and the 90°pulses calibrated. For all samples, 1D 1 H spectra were recorded with 32 scans using 32 768 points.
Computational Details. Classical MD simulations have been carried out on BHT/MEN, TBC/MEN, and PC/MEN systems covering the full composition range. Cubic boxes were built with ∼50 Å side lengths and a number of species chosen to reproduce the simulated density at 343 K. This temperature was chosen to ensure that all samples were in the liquid state. Simulations for density estimation were carried out on systems built starting from the experimental densities at 298 K for the samples with x BHT/TBC/PC = 0.25 and resulting in 0.902, 0.955, and 0.966 g cm −3 , respectively. NPT simulations were carried out at 1 atm and 343 K for 10 ns employing the Nose−Hoover thermostat with a relaxation constant of 0.5 ps and the Parrinello−Rahman barostat with a coupling constant of 1.0 ps. New simulation boxes were built with random atomic positions with the Packmol program 41 in accordance with the obtained densities, which are listed in Table S3 together with the systems details. After an energy minimization, each box was equilibrated under NVT conditions at 500 K for 20 ns and then simulated for data collection at 343 K for an additional 50 ns. The equations of motion were integrated with a leapfrog algorithm with a 1.0 fs time step, and coordinates were saved every 100 steps. Stretching vibrations involving hydrogen atoms were constrained with the LINCS algorithm. 42 Structures and interactions of the BHT, TBC, and PC molecules were represented with the all-atom optimized potentials for liquid simulations (OPLS-AA) force field, 43 while for the MEN, the OPLS-compatible parameters developed by Jasik et al. were employed. 44 Mixed terms were constructed with the Lorentz− Berthelot combining rules, and a cutoff radius of 12 Å was applied for all nonbonded interactions, while long-range electrostatic forces were taken into account with the particle mesh Ewald method. 45,46 Simulations were performed with the Gromacs 2020.2 program, 47 while the VMD 1.9.3 software 48 was used for trajectories visualization and the TRAVIS code for the analysis. 49 DFTB-MD simulations of the BHT/MEN, TBC/MEN, and PC/ MEN systems at the x BHT/TBC/PC = 0.25 composition were carried out using the DFTB+ 21.2 program 50 at the third order variant of the density functional tight-binding (DFTB3) level of theory, which was previously demonstrated to improve the description of H-bond interactions, 51−53 in combination with the third-order-parametrization for organic and biological systems (3OB) 54 Slater−Koster files. Cubic boxes were initially built with random positions with the Packmol program, 41 with the size according to the experimental density at 298 K. The systems details are listed in Table S4. For each box, a geometry optimization was carried out and followed by a 2 ps NVE run and 10 ps NVT equilibration at 500 K. 100 ps NVT runs for data collection were carried out at 298 K. Such a relatively long simulation time for an ab initio MD simulation was possible thanks to the ∼2−3 orders of magnitude faster DFTB Hamiltonian than for the full DFT approach. 55 Self-consistent charge (SCC) optimization was carried out with 10 −5 tolerance, and the k-points meshes corresponded to a Γ-point sampling, while periodic boundary conditions were applied in all directions. Thermalization was performed with the Nose−Hoover thermostat with a coupling strength of 3200 cm −1 . The equations of motion were integrated with a velocity Verlet algorithm with a 1.0 fs time step and trajectories saved every 10 steps. Grimme's DFT-D4 empirical dispersion corrections were applied to improve the description of van der Waals interactions. 56 The BHT, TBC, PC, and MEN pristine compounds were also simulated with the same level of theory. Production runs for data collection were performed in NVT conditions at 348 K for 50 ps. This temperature was chosen as above the determined MP of BHT (343 K). Details about these systems are listed in Table S5. ■ RESULTS AND DISCUSSION Thermal Characterization. Access to the SLE phase diagram between the components is imperative for the thermodynamic characterization of a eutectic mixture, as it allows one to compare the experimental MP depression to the ideally predicted one and classify a system as either a DES or a eutectic solvent. In Figure 2, we report the phase diagrams for the BHT/MEN, TBC/MEN, and PC/MEN mixtures, where the best-fit curves obtained after interpolation of the experimental MP temperatures are compared with the ideal SLE diagrams. Table 1  value obtained after interpolation of the experimental melting temperatures is 295.6 K and thus almost matches the ideal one (Table 1). A behavior close to thermal ideality is observed for the whole composition range, as the obtained MPs are found to lie close to the ideal phase diagram (Figure 2a). This result, alongside the determined χ values, which are close to zero (  Figure 3, we report the magnified ν OH contribution of the spectra recorded in the 3100−3800 cm −1 range, while the complete Raman profiles are shown in Figure S1. The ν OH signal of pristine MEN is dominated by a broad band centered at about 3400 cm −1 (Figure 3a−c). This low-frequency contribution has been previously associated with hydroxyls behaving as H-bond donors by means of the hydrogen atom, while contemporaneously employing the oxygen atom as a Hbond receptor through its lone pairs, and labeled as δ-OHs. 33,57−63 This spectral evidence therefore indicates the formation of H-bonded networks or aggregates and a high extent of self-association of the MEN molecules in the precursor compound. BHT addition to MEN (Figure 3a) is translated into the enhancement of a high-frequency contribution appearing in the 3600−3650 cm −1 region, up to the formation of a distinct narrow peak with a maximum at 3647 cm −1 for the BHT n-heptane solution. This feature has been previously connected with "free" hydroxyl groups not involved in H-bonding and labeled as α-OH. [57][58][59]64,65 Note that β-OH also resonate in the same spectral region and are referred to hydroxyls behaving as proton acceptors with the oxygen atom, while contemporaneously not behaving as Hbond donors with the hydrogen atom. A pictorial representation of these hydroxyl types is given in Figure 3d. In both cases, the hydrogen atom is free from H-bonding, resulting in a similar frequency of the O−H oscillator. 59 It is worth noting that such a spectral evolution occurs at the expense of the δ-OH band characterizing the pristine MEN, which is sequentially blue-shifted and flattened for increasing BHT concentration (Figure 3a). Altogether these findings suggest that the addition of the sterically hindered BHT to MEN is   able to perturb the H-bonded association present in the pristine compound and to promote an increasing isolation of the BHT and MEN molecules in terms of H-bonding. As a consequence, the hindering effect provided by the two tertbutyl substituents is highly detrimental for the establishment of H-bond interactions by the BHT component. A similar flattening of the ∼3400 cm −1 band characterizing pristine MEN is also evident upon addition of the sterically semifree TBC compound (Figure 3b). However, at variance with the BHT/MEN case, this spectral evolution occurs alongside the enhancement of a marked peak appearing at ∼3540 cm −1 . This contribution lies in the middle of the spectral region connected with γ-OHs, representing terminal proton-donating hydroxyls with the lone pairs of the oxygen atom not involved in H-bonding (Figure 3d), being thus responsible for the formation of H-bonded dimers. 57,58,65 Note that the phenolic nature of TBC makes its hydroxyl group more positive due to resonance effects and the hydrogen atom a better donor, while the oxygen atom becomes a worse acceptor ( Figure S2b). On the other hand, MEN possesses no aromaticity, so that the TBC-MEN donor−receptor interaction should be the most favored one at least from an electrostatic point of view. This also applies for BHT ( Figure  S2a), but the steric exclusion in both the ortho-positions of the hydroxyl group is evidently able to overwhelm this favorable electronic effect. Conversely, the TBC molecule carries only one tert-butyl substituent, thus leaving partial access to the hydroxyl group and allowing the establishment of the favorable TBC−MEN interaction. The obtained Raman profiles are therefore compatible with the replacement of the H-bonded aggregates among MEN molecules through the formation of TBC-MEN dimers due to the competition of the more acidic TBC proton.
The Raman spectrum of the sterically free PC compound also consists of a marked γ-OH contribution, plus a δ-OH one similar to that characterizing pristine MEN (Figure 3c). This fingerprint indicates that the molecular association in the precursor compound consists of both H-bonded agglomerates among PC molecules and PC−PC donor−acceptor dimers. Once PC is mixed with MEN, a retention of both the γand δ-OH contributions is observed for all of the explored compositions. As in the case of the sterically semifree TBC, also the PC species has better H-bond donor and worse receptor capability due to its phenolic nature ( Figure S2c (Figure 2a). Nevertheless, a liquid phase is obtained at room temperature for compositions close to the eutectic one, and at the x BHT = 0.25 molar fraction this mixture has been previously employed under room temperature operative conditions, 34−36 suggesting that some sort of interplay must occur upon mixing of the parent compounds. To deepen this aspect and check if additional interactions are present, we inspected the spectral region associated with the C−H stretching modes in the Raman profiles of the BHT/MEN mixtures (see Figure 3e). The wavenumber region between 2800 and 3000 cm −1 comprises a complex set of overlapped peaks due to the symmetric and antisymmetric CH 2 and CH 3 stretching, 66−68 which can be associated with the tert-butyl, isopropyl, and methyl substituents of the BHT and MEN components. Nevertheless, the entire group of C−H vibrations is found to shift at lower wavenumbers upon BHT addition to MEN. This effect can be better appreciated from the positions of the two main peak maxima reported as a function of the BHT concentration in Figure 3f. A decrease of the C−H stretching frequency has been previously invoked as an indicator for the increase of the alkyl chain interactions and conformational order. 66,67 This result is compatible with the establishment of apolar−apolar attractions among the alkyl substituents of the BHT and MEN components, suggesting that this may be also a driving force for the BHT/MEN eutectic formation. As such dispersion forces are known to be weak, the experimentally determined MP results are close to the ideality, and this system is classified as an ideal eutectic. Note that a systematic shift of the C−H stretching bands frequency is observed also upon MEN mixing with TBC and PC ( Figure S3a and b, respectively), even if of a smaller extent with respect to the BHT/MEN case. This evidence suggests that apolar−apolar attractions could also play a role in the TBC/MEN and PC/MEN mixtures, even if the H-bonding contribution in these systems is overwhelming with respect to these interactions. NMR Characterization. The 1 H NMR spectra collected on the BHT/MEN, TBC/MEN, and PC/MEN mixtures are reported in Figure S4, while in Figure 4 we present the change in the chemical shift of selected protons (Δδ H = δ mixture − δ neat ) of the constituents across the full composition range with respect to their pure state. 33 A rise in δ H (downfield shift) is indicative of protons in a less shielded magnetic environment, thus as far as the hydroxyl protons are concerned this descriptor delivers information on the extent of H-bonding formation, since an increase in this interaction is expected to lower the electron density and make the nucleus less shielded with respect to the magnetic field. 69 Therefore, a downfield shift of the OH signal (Δδ OH > 0) has to be interpreted here as due to the formation of H-bonds in the mixture that are stronger than the existing ones in the neat compounds. Similarly, a low-frequency (upfield) shift (Δδ OH < 0) indicates the weakening of H-bonds in the pure component and their replacement with relatively less strong H-bonds in the mixture. 23,70−72 Concerning the BHT/MEN system, the gradual addition of BHT to MEN is translated into a marked upfield shift of the MEN hydroxyl proton chemical shift ( Figure S4), with Δδ MEN−OH reaching a value of −2.9 ppm for the highest BHT concentration (Figure 4a). This behavior can be attributed to a weakening of the H-bond interactions among the MEN molecules with respect to the pristine compound. On the other hand, this evolution is not mirrored by an equal downfield shift of the δ H of the BHT hydroxyl proton ( Figure  S4), showing very low Δδ BHT−OH values through the full composition range (Figure 4a). The whole result therefore indicates that the progressive breaking of the MEN−MEN intermolecular association provoked by the addition of the sterically hindered BHT is not compensated by a significant establishment of further H-bonds, confirming the increasing isolation of the two compounds in terms of H-bonding in agreement with the Raman spectroscopy results (Figure 3a).
A different situation is obtained for the TBC/MEN system ( Figure 4b). Here, the addition of the sterically semifree TBC to MEN provokes a decrease of the MEN hydroxyl Δδ H of a similar extent to what is found for the BHT/MEN mixtures (Δδ MEN−OH = −2.4 ppm for the highest TBC concentration). However, such an evolution goes hand in hand with a marked increase of the Δδ H of the TBC hydroxyl proton, with Δδ TBC−OH reaching a value of 3.5 ppm for the highest MEN concentration. This finding indicates that the breaking of the H-bonded agglomerates initially present in pristine MEN is accompanied by a strengthening of the H-bonds played by the TBC molecules with respect to the precursor state. This massive change can be explained with the strong H-bond donor and poor receptor capabilities of the TBC species due to resonance effects ( Figure S2b). Indeed, when mixed with MEN, the TBC molecules encounter a "regular" H-bond acceptor with no mesomeric effects, so that the preferential TBC-MEN interaction is established, which according to the Raman spectroscopy results is mostly played through TBC-MEN H-bonded dimers (Figure 3b). Such a concept has been previously indicated as "polarity asymmetry" and used to explain the MP depression in the archetypal type V DES formed by thymol and MEN. 16,33 However, in the case of TBC, this molecule adds to the polarity asymmetry also the steric effect, as the tert-butyl substituent in the ortho-position to the hydroxyl group is a further factor preventing contemporaneous behavior as a H-bond donor and acceptor. This exasperated asymmetry leads to poor H-bond interactions in the precursor state, where only TBC molecules are present, and explains the marked increase in the Δδ H of the TBC hydroxyl proton upon MEN addition (Figure 4b) and ultimately the deviation from thermal ideality of the TBC/ MEN system (Figure 2b).
Addition of the sterically free PC to MEN also provokes a decrease of the Δδ H of the MEN hydroxyl proton (Figure 4c), even if to a lower extent with respect to the BHT/MEN and TBC/MEN cases (Δδ MEN−OH = −1.1 ppm for the highest PC concentration). This provides evidence that the H-bonded network of pristine MEN is less perturbed by the PC introduction as compared to the BHT and TBC ones. Such evolution is accompanied by an increase of the Δδ H of the PC hydroxyl, which is greater than that obtained for BHT ( Figure  4a) but lower than the TBC one ( Figure 4b). As for TBC, PC is also phenolic in nature and thus a stronger H-bond donor and weaker receptor ( Figure S2c). However, in this case, there is no steric hindrance in the ortho-positions to the hydroxyl group, so that the H-bond donor/receptor asymmetry only relies on electronic effects. For this reason, the intermolecular aggregation in the precursor is less prevented than in the TBC case, and H-bonded agglomerates among PC molecules are also present, as evidenced by the δ-OH contribution in the Raman profile of the pristine compound (Figure 3c). When mixed with MEN, the PC-MEN donor−receptor H-bond is in any case established as the most favored interaction, leading to a significant increase of the Δδ H of the PC hydroxyl proton (Figure 4c) and the achievement of a type V DES (Figure 2c Table 2 we list the maximum positions of the first peak and the coordination numbers N computed integrating each curve up to a cutoff distance chosen at the first minimum. As can be observed, in the BHT/MEN system, the H MEN −O MEN g(r) associated with the H bonds among MEN molecules is overwhelming with respect to the others (Figure 5a). The maximum distance of 1.98 Å is in line with those typically observed for strong H bonds, [18][19][20][21][22]36,73 while the obtained integration number is 0.34 (Table 2). Conversely, the g(r)'s where the sterically hindered BHT molecule behaves as a Hbond acceptor, namely, the H MEN −O BHT and H BHT −O BHT ones, do not show distinguishable peaks, confirming the hampered access to the hydroxyl group of this component. A special comment must be dedicated to the H BHT −O MEN distribution. Here, a distinct peak is obtained, suggesting that BHT behaves as a H-bond donor toward the MEN molecule. However, careful inspection of this g(r) reveals that the maximum position is 2.18 Å (Table 2), thus relatively longer than a strong H bond. Provided that the bond distance is not the only criterion to define the H-bond strength, 73 we carried out the analysis of the bond angle formed between the H-bond donor and acceptor sites, which we correlated with the H BHT − O MEN g(r) in a combined distribution function (CDF). The reference system for this analysis is depicted in Figure 5d, while the obtained CDF is shown in Figure 5e. Here, a high probability spot is located around an angle value of about 135°, thus far away from the 180°angle usually associated with strong H-bonds. 36,73 This result suggests that the BHT molecule may tentatively behave as a H-bond donor toward the MEN molecules due to its phenolic nature, but the steric encumbrance around its hydroxyl group prevents this species from establishing a significant H-bond interaction. Since no relevant H-bonds are found among the components of this system, we also computed the g(r)'s between the centers of mass (COMs) of the alkyl substituents to get confirmation about interactions among the hydrophobic moieties suggested by the Raman experimental outcomes (Figure 3e,f). As shown in Figure 5f, all of the obtained curves present well-defined first peaks displaying a remarkable degree of structuration. This evidence subtends a high degree of structural order among the alkyl chains and pleads for apolar−apolar attractions as a driving force for the BHT/MEN eutectic formation.
Concerning the TBC/MEN system, the more relevant distribution is the H TBC −O MEN one (Figure 5b), which shows a remarkably short distance of 1.92 Å and an N value of 0.68 (Table 2). This g(r) is connected with the TBC-MEN donor− receptor H bond and confirms this interaction as the favored  Figure 5c). However, here it is worth noticing that the H PC − O PC distribution reflecting self-aggregation among PC molecules is slightly more marked than the correspondent H TBC −O TBC one in the TBC/MEN system (Figure 5b), providing a potential confirmation about the higher tendency of the sterically free PC to self-interact.
To better unveil these issues, we also performed DFTB-MD simulations of the pristine compounds BHT, TBC, PC, and MEN. The intermolecular H−O g(r)'s describing the interaction among the precursor molecules in their pure states are shown in Figure 6a, while the correspondent parameters are listed in Table 2. As expected, the H PC −O PC distribution shows a distinct first peak with a maximum at 2.02 Å, validating the self-association among the sterically free PC molecules in the pristine compound, and the same is true for the H MEN − O MEN one in pristine MEN. The H BHT −O BHT g(r) shows no peak at all, confirming the hampered aggregation among the sterically hindered BHT molecules also in the precursor state. A special look is deserved for the H TBC −O TBC distribution obtained for the sterically semifree TBC. Here, this function describing the H bond among TBC molecules does not show a distinct peak and seems to confirm the hampered self-aggregation of the TBC precursor. A more detailed analysis has been achieved by calculating the CDF between this g(r) and the angle distribution between the donor and acceptor sites, with the same reference system shown in Figure 5d. The obtained CDF displays a high intensity region around the ∼180°value, which is consistent with a H-bond interaction (Figure 6b), 36,73 but a broad distribution is found also at other angle values. For the sake of comparison, the analogous CDFs computed for the H-bond networking PC and MEN compounds are shown in Figure S5. Here, probability spots are highly localized around the ∼180°angle and are compatible with strong H-bond interactions. Further insights into the three-dimensional arrangement among TBC molecules in the pristine compound have been obtained by means of a spatial distribution function (SDF) calculations ( Figure  6c). Here, the three-dimensional density distribution of the TBC hydroxyl hydrogen and oxygen atoms, as well as a r max is the position of the first peak maximum, and N is the integration number calculated by integrating each curve up to the reported cutoff distance. The atom names are employed according to the nomenclature reported in Figure 1. No data are reported for the g(r)ρ's that showed negligible peaks.  Figure 1. (b) CDF between the H TBC −O TBC g(r) and angular distribution function computed from the DFTB-MD simulation of the TBC compound. The color box on the right side is relative to the probability function of finding the inspected particle at that distance and angle normalized to one. (c) SDFs around the TBC molecule computed from the DFTB-MD simulation of the TBC compound. The functions have been computed with respect to an internal reference system that is integral to the carbon ring. The observed species are shown according to the color code reported on the right. The hydroxyl hydrogen atom of the reference molecule is represented by an intramolecular SDF. Isosurfaces have been drawn with the same density/maximum density ratio according to the following isovalues in nm −3 : H TBC (6.4), O TBC (7.0), TBC COM (7.6). molecular COM, has been computed around each TBC molecule taken as a reference. The first feature capturing the attention is that most of the observed probability spots are found close to the carbon body of the TBC reference, while only a small volume corresponding to the observed O TBC atom is found close to the H TBC reference one. This result, together with the poorly structured H TBC −O TBC g(r) (Figure 6a) and with the correspondent CDF (Figure 6b), confirms that the TBC−TBC H-bond interaction in the pristine compound is likely to be tentatively formed, but the contemporaneous behavior of TBC as a H-bond donor/receptor is hampered because of a combined effect between the unfavorable electronic and steric factors. This picture is in agreement with the body of evidence provided by the experimental outcomes and serves as a rationalization of the thermal behavior of the TBC/MEN system in relation with the BHT/ MEN and PC/MEN ones.

■ CONCLUSIONS
A detailed study of the BHT/MEN, TBC/MEN, and PC/ MEN mixtures has been carried out with the aim of enlightening the various contributions ruling the intermolecular interactions among the components and the eutectic formation. The thermal characterization showed that the BHT/MEN system can be classified as an ideal eutectic, while the TBC/MEN and PC/MEN mixtures behave as type V DES for a wide range of compositions around the eutectic point. Addition of the sterically hindered BHT to MEN is detrimental toward the H-bonded agglomerates among MEN molecules present in the pristine compound, which are not compensated by the establishment of further H-bond interactions. Nevertheless, a liquid phase at room temperature is obtained for compositions close to the eutectic one. In the absence of relevant H-bonding, the high structural order observed among the alkyl functional groups of the BHT and MEN constituents subtends for apolar−apolar attractions as a driving force of the eutectic formation. Addition of the sterically semifree TBC to MEN also provokes the breaking of the H-bonded network of the pristine compound, but in this case it is replaced by the formation of the more favorable TBC-MEN donor−receptor H-bonds. This behavior derives from a concurrence of electronic and steric factors, as the better Hbond donor and worse receptor capabilities of the TBC molecule due to mesomeric effects are exasperated by the presence of one tert-butyl group in ortho-position to the hydroxyl. The whole effect is that of making the intermolecular interactions in the TBC/MEN mixtures much more favorable than in the precursor TBC compound, and a type V DES is obtained as a result. The sterically free PC molecule is also a better H-bond donor and worse receptor due to a phenolic nature, but, differently from the TBC case, a considerable amount of self-association is observed already in the precursor compound by means of both PC−PC donor−receptor dimers and H-bonded networks. This behavior is maintained also upon mixing with MEN, but the PC-MEN donor−receptor Hbond is established as a more favorable interaction, and this is the driving force for the type V DES formation. Evidence for apolar−apolar attractions among the components are also found for the TBC/MEN and PC/MEN mixtures, but they play a marginal role compared to the dominating H-bond contribution. The body of evidence that we obtained here was therefore able to clarify the different factors on the molecular level influencing the thermal behavior of the studied eutectic mixtures. Along with the peculiarity of the inspected systems, the impact of this work is that of providing new tools for a more conscious understanding of the complex correlations between local structure and macroscopic properties in hydrophobic DESs.
List of the prepared samples, details about the building of the SLE phase diagrams, details of the simulated MD systems, Raman and NMR spectroscopy additional results, discussion of classical MD results (PDF)

Notes
The authors declare no competing financial interest.

■ ACKNOWLEDGMENTS
Part of the calculations was performed on the Marconi100 system of the CINECA supercomputing center (grant IsC99_CODES). The authors acknowledge financial support from the Italian Ministry of University and Research (MIUR) through Grant "PRIN 2017, 2017KKP5ZR, MOSCATo" and from the University of Rome "La Sapienza" Grant RG11916B702B43B9.