On the molecular basis of H 2 O/DMSO eutectic mixtures by using phenol compounds as molecular sensors: a combined NMR and DFT study †

NMR and DFT studies of phenol compounds as molecular sensors were carried out to investigate H 2 O/ DMSO eutectic mixtures at a molecular level. The experimental 1 H NMR chemical shifts of the OH groups, d exp (OH), of phenol, paracoumaric acid, and vanillic acid show maximum deshielding and, thus, hydrogen bond interactions in the range of mole fractions 0.20 o w (DMSO) o 0.33. In the mole fractions w (DMSO) o 0.2, a progressive decrease in d exp (OH) was observed which demonstrates a decrease in hydrogen bond interactions at infinite dilution in H 2 O, despite the increase in the number of available hydrogen bond acceptor and donor sites. DFT calculated d calc (OH) of minimum energy solvation clusters were shown to be in reasonable agreement with the pattern in experimental d exp (OH) data. The chemical shift deshielding and, thus, increased hydrogen bond interactions in the natural product + DMSO + n H 2 O ( n = 2, 3) solvation clusters, relative to complexes in DMSO or H 2 O solutions, cannot be attributed to a single structural parameter of the cooperative interactions between H 2 O and DMSO molecules with the phenol OH groups of the natural products. The minimum energy conformers of phenol compounds + 2H 2 O + DMSO complexes are in excellent agreement with a recent low temperature neutron diﬀraction experiment of 3D 2 O + DMSO and demonstrate a general structural motif of solvation complexes. The combined use of 1 H NMR and DFT studies with emphasis on d (OH) of phenol compounds, as molecular sensors, can provide an eﬀective method for the study of solute–solvent interactions at the atomic level.


Introduction
Several mixtures of water with organic solvents show properties deviating from linearity, such as freezing point, density, viscosity, and adiabatic and isothermal compressibility. 1 Typical examples are dimethyl sulfoxide (DMSO), glycerol, and several other hydrophilic non-electrolytes, which protect several types of cells from damage during freezing and thawing, allowing them to be stored for very long periods at low temperatures. 2This was attributed to the prevention of damaging crystal formation, and reduction of the concentration of electrolytes during freezing. 3,4][7][8][9][10][11] Hydrogen bonding was identified as the main driving force of this phenomenon.In general, the strength of the hydrogen bonds can be correlated with the phase-transition temperature stability and solvent properties of the respective mixtures. 9or a system consisting of water and DMSO, extreme deviation from additivity was observed.At a mole fraction w(DMSO) C 0.33, a very low freezing point was measured (À140 1C).A stable DMSO/H 2 O (1 : 2) cluster was suggested to be responsible for this unusually low melting point. 12Each such subunit is rather loosely hydrogen bounded to the next subunit, thus, resulting in an overall structure that prevents crystallization.Further research revealed the formation of a rather complex phase diagram, and the dominance of stable DMSO/H 2 O (1:3) clusters. 13umerous experimental and theoretical studies have been reported on DMSO/water aqueous mixtures including neutron scattering, 14,15 investigation of the structure and dynamics of hydrogen bonding, 16 solvation dynamics and vibrational relaxation. 17MD simulations at ambient temperature have been carried out to evaluate three models of pure liquid DMSO and in DMSO/H 2 O mixtures. 18Statistical analysis of the hydrogen bond network revealed the presence of DMSO + 2H 2 O and not DMSO + 3H 2 O complexes.MD simulation estimated two regions of co-solvent DMSO-water clusters: w(DMSO) = 0.12-0.17and w(DMSO) = 0.27-0.35.In the second region, there is an anomalous solvation trend observed due to high viscosity while in the first region the solvation increases due to the formation of strong hydrogen bonded DMSO/H 2 O (1:2) complexes. 19DFT and polarizable force field have been utilized to investigate energetic and structural properties of DMSO-water clusters. 20Quantum chemical studies were also utilized to investigate inter-cluster interaction modes of the freezing behavior, particularly of DMSO/H 2 O (1:3) clusters. 21Water dynamics in H 2 O/DMSO binary mixtures were studied with the use of polarization selective pump-probe experiments, 2D IR vibrational echo spectroscopy and IR absorption spectroscopy. 22IR spectroscopy of the SQO stretching mode combined with MD and quantum chemistry models were used to quantify hydrogen-bond populations in DMSO/H 2 O mixtures. 23MD studies of a DMSO/H 2 O (1 : 3) cluster using the AMBER force field suggested a value of B2.89 AE 0.11 water molecules around each DMSO molecule. 24Nevertheless, the driving forces behind non-covalent interactions in hydration of DMSO which is a dipolar aprotic solvent with both hydrophilic and hydrophobic properties, are not sufficiently understood. 25he phenols are major constituents of many biological and naturally occurring compounds and are ubiquitous in the plant and animal kingdom.Interestingly, around 7% of the current drugs contain phenol groups.7][28][29][30][31][32] Linear correlation between X-ray diffraction r(OÁ Á ÁO) distances and d exp (OH) 1 H NMR chemical shifts was suggested for a variety of b-diketone enol compounds with intramolecular O-HÁ Á ÁO hydrogen bonds. 33A linear relationship has also been presented between d exp (OH) and HÁ Á ÁO distances, determined from neutron diffraction. 34][37][38][39][40][41][42] The objective of the present study was to address three important issues.First to investigate the importance of hydrogen bonds in DMSO/H 2 O eutectic mixtures by using, as molecular sensors, phenol (Ph, 1), paracoumaric acid (PCA, 2), and vanilic acid (VA, 3) (Fig. 1).As discussed above, d(OH) of phenol compounds are very sensitive to hydrogen bonding interactions and, furthermore, they are significantly deshielded relative to the chemical shifts of H 2 O. Overlapping, therefore, with the strong H 2 O peak can be avoided which results in accurate d exp (OH) values, contrary to the case of e.g., alcohol type OH groups.
Second, to utilize the agreement between experimental and DFT computed 1 H NMR chemical shifts of phenol OH groups as a tool for investigating solute-solvent interactions and accurate structural models at the atomic level.Third to investigate whether there is a general structural pattern of phenol compound + (2, 3) H 2 O + DMSO solvation complexes.The molecules selected are secondary metabolites derived from plants and contain a single phenol OH group.Compounds with multiple phenol OH groups were not used in order to avoid strong overlapping of the resonances and, thus, uncertainties in the determination of the chemical shifts.Phenol (1), occurs in the essential oil from tobacco leaves (Nicotiannatabacum, Solanaceace) and pine needles (Pinus Sylvestris, Pinaceae) 43 and has the ability to encounter the gut microbes. 44aracoumaric acid (2) was found in many plants, such as plumbaginaceae's root, leaf and flower, 45 and has been used to treat diabetes, hypertension and cardiovascular diseases. 46Vanillic acid has been isolated from origanum vulgare and acts as antimelanogenesis agent. 47

NMR spectroscopy
The feasibility of obtaining high resolution 1 H NMR resonances of phenol OH groups was demonstrated by the use of dry nonprotic solvents, addition of acids, such as trifluoracetic acid (TFA), and or by decreasing the temperature. 29,48,49Solvents with significantly different dielectric constants, solvation, and hydrogen bonding ability were shown to induce large 1 H NMR chemical shift differences, and that the solvation state of the phenol OH group is a key factor in determining the value of d exp (OH).Solvent effects, therefore, on d exp (OH) have been used as a direct measurement of solvation changes and inter-and intramolecular hydrogen bond interactions. 31,36,49he use of     Similar experiments were performed at 280 K and 270 K (Fig. 2(Ba) and (Ca)) in order to reduce the proton exchange rate and, thus, linewidths which would allow a better definition of the OH chemical shifts for mole fractions w(DMSO) o 0.3.For mole fractions w 4 0.88 and 0.83 the solutions were frozen at 280 K and 270 K, respectively, therefore the OH chemical shifts were obtained through extrapolation of the linear dependence of d exp (OH) vs. T 31 in the temperature range of 292-318 K.By decreasing the temperature, the proton exchange is reduced, however, even at 270 K the resonances are still very broad which prohibit accurate estimation of the chemical shifts for w(DMSO) r 0.1.
From the above it is evident that, in the three temperatures investigated, d exp (OH) values strongly deviate from linearity for w(DMSO) o 0.35 and the extrapolated value of d exp (OH) at infinite dilution in H 2 O is smaller than that in the range of mole fractions 0.2 r w(DMSO) r 0.33.This demonstrates weaker hydrogen bond interactions at infinite dilution in H 2 O, despite the increase in the available hydrogen bond acceptor and donor sites, provided that d exp (OH) is not affected by proton exchange.Assuming that the spin  S3 in ESI †) can be attributed to increased hydrogen bonding interactions of Ph(1) + nH 2 O + DMSO (n = 2-4) subunits, which are loosely bound to each other, thus, resulting in reduced proton exchange (see also discussion on DFT calculations).Fig. 3 illustrates the OH 1 H NMR spectral region and correlation of d exp (OH) vs. w(DMSO) of PCA(2) at 300 K, 280 K and 270 K.The resulting linewidths were found to be broader, than those obtained with Ph(1).This can be attributed to the presence of the carboxyl group which increases proton exchange rate constant and, thus, linewidths.S4, ESI †), it can be concluded that the broad maximum of d(OH) corresponds to a maximum in hydrogen bonding interactions, as in the case of Ph (1).Again, the extrapolated value of d exp (OH) at infinite dilution in H 2 O is smaller than that in the region of the mole fractions 0.20 r w(DMSO) r 0.33.Similar behavior has been observed for VA(3) (Fig. S1 and Table S5, ESI †).The resulting linewidths were found to be broader than in the case of Ph(1) due to the presence of the carboxyl group which enhances proton exchange rate constants.Nevertheless, again a broad maximum at w(DMSO) = 0.35-0.2was observed.The OH resonance at mole ratios w(DMSO) o 0.12 cannot be distinguished from the baseline, therefore, the extrapolated value of d exp (OH) at infinite dilution in H 2 O could not be obtained accurately.

DFT calculations
It has been demonstrated that the calculated phenolic OH 1 H NMR chemical shifts, d calc (OH), using the conductor-like polarizable continuum model (CPCM), deviate significantly from the experimental values, especially in the case of solvents having high dielectric constants and hydrogen bond strengths. 35This is due to the fact that this model is the quantum chemical formulation of the Onsager reaction field model for bulk solvent effects which does not include specific solute-solvent interactions. 52,53In contrast, the use of discrete phenol + solvent complexes results in an excellent agreement between the calculated OH chemical shifts and the experimental values. 35,37,42he present computational study was performed by using the Gaussian 09 with the DFT method 54 with the structures optimized in the gas phase by using the standard functional B3LYP as well as the APFD with corrections for dispersion interactions, with the 6-31+G(d) basis set.The 1 H NMR chemical shifts were calculated with the GIAO (Gauge-Independent Atomic Orbital) and CSGT (Continuous Set of Gauge Transformations) 55 methods by using the B3LYP/6-311+G(2d,p) level in the gas phase.
The optimized structures of the phenol molecules with various discrete water and DMSO molecules, were obtained successively starting with the phenol type molecules by adding various numbers of water molecules followed by DMSO molecules in various possible positions, in such a way that they act either as donors or acceptors.The resulting molecular geometries were verified as energy minima on the potential energy surface (PES) by performing frequency calculations at the same level (zero imaginary frequencies).Four solvation clusters were examined for each of the three natural products of Fig. 1: (i) with a single solvation molecule of DMSO, (ii) with DMSO + 2H 2 O, (iii) with DMSO + 3H 2 O, and (iv) with 3H 2 O.In the case of VA(3), due to the proximity of hydroxy and methoxy groups at the ortho position, the importance of intramolecular hydrogen bonding was investigated also in VA + nH 2 O (n:1, 2) solvation clusters.
Due to the existence of different configurations at each mole ratio, it is necessary to identify the key hydrogen bond interactions, particularly in the minimum energy configurations.Selected structural and conformational properties of Ph(1) + solvent complexes, energy minimized using the APFD/6-31+G(d) and Selected structural and conformational properties of Ph(1) + solvent complexes with energy minimization using the B3LYP functional with the 6-31+G(d) basis set are shown in Fig. S2 (the ESI †).The hydrogen bond properties of the Ph + DMSO complex are very similar to those obtained by using the APFD functional.There are minor differences in bond angles and torsion angles; the hydrogen bond lengths are invariably shorter using the APFD functional due to the empirical dispersion correction term for noncovalent intermolecular interactions (Fig. 4 and Fig. S2, ESI †).
Table 1 shows calculated, d calc (OH) 1 H NMR chemical shifts of Ph + DMSO, Ph + 2H 2 O + DMSO, Ph + 3H 2 O + DMSO, and Ph + 3H 2 O solvation complexes, with energy minimization at the B3LYP/6-31+G(d) and APFD/6-31+G(d) levels.In practice, the various configurations of each solvation complex that exist within a 0.0 to 3.0 kcal mol À1 Gibbs energy window, can contribute to the overall computational chemical shift.For a given solvation species of Table 1 a Boltzmann analysis was performed in order to determine the relative contribution of each configuration to the overall computational chemical shift, d calc , Wg , using the equation 56,57 where i and j are configurations of a given solvation complex, G is the Gibbs energy of a particular configuration, R is the universal gas constant, and T is 298 K.The d calc (OH) 1 H NMR chemical shifts by the GIAO method, weighted by the respective Boltzmann factors (Table 1), d calc,Wg , were aligned in accord to the tendencies of the experimental chemical shifts.Thus, a maximum in the chemical shift was obtained with the Ph + 3H 2 O + DMSO solvation cluster, and a small reduction in the chemical shift of the Ph + 3H 2 O solvation cluster.By using the CSGT method, the Ph + 3H 2 O + DMSO solvation cluster has a slightly smaller chemical shift (9.84 ppm) than the Ph + 3H 2 O cluster (9.87 ppm), contrary to the experimental data.With minimization of the structures at the APFD/6-31+G(d) level, the calculated OH 1 H NMR chemical shifts, for both the GIAO and CSGT methods, weighted by the respective Boltzmann factors (Table 1) Conformational analysis of PCA(2) has been performed with a variety of DFT methodologies by several groups. 58,59Four minimum energy conformers A, B, C and D were obtained with PES calculations of the torsion angles C5-C4-C3-C2 and C3-C2-C1-O1a (Fig. 5).The global minimum energy conformer d calc,Wg (ppm) d calc,Wg (ppm)  is A; conformer B, with energy very similar to that of A, has been observed in the X-ray crystal structure of the PCA. 60The stronger intramolecular interaction between the electron lone pair of the carboxylic oxygen,QO1a, and the olefinic hydrogen, QC3-H, is responsible for the stabilization of conformers A and B vs.  Selected structural and conformational properties of PCA + solvent complexes, using the B3LYP/6-31+G(d) energy minimization method, in the gas phase, are shown in Fig. S3 (the ESI †).The hydrogen bond lengths are longer with the B3LYP functional, as in the case of phenol (1) + solvent complexes.
The calculated 1 H-NMR chemical shifts, d calc,g , of the four conformational states A, B, C and D (Fig. 5) and various solvation configurations of PCA, in the gas phase, are shown in Table 2.The 1 H NMR chemical shifts with the GIAO and CSGT methods, of the PCA + DMSO + 3H 2 O and PCA + 3H 2 O complexes, weighted by the respective Boltzmann factors, were aligned in accord to the tendencies of d exp , with energy minimization at the APFD/6-31+G(d) level.Similar tendencies were obtained at the B3LYP36-31+G(d)/CSGT level (Table 2).Comparison of the PCA + 3H 2 O and PCA + 3H 2 O + DMSO solvation complexes using both B3LYP and APFD functionals shows that: (i) the incorporation of the DMSO molecule increases the size of the cycling structure and (ii) the increase in the chemical shift in the PCA + 3H 2 O + DMSO complex cannot be attributed to a single structural parameter of the cooperative interactions between H 2 O and DMSO molecules with the PCA OH group.
The proximity of hydroxyl and methoxy groups at the ortho position in VA(3) may result in the formation of intramolecular hydrogen bonds. 61,62Thus, the X-ray structure determination of vanillic acid and theobromide co-crystal hydrate (VAÁTHBRÁ2 H 2 O) 63,64 shows that the hydroxyl and methoxy groups of VA adopt an in-out conformation which results in the formation of an intramolecular hydrogen bond.A water molecule acts as a bifurcated acceptor to the OH group of VA (O-HÁ Á ÁO(W 1 ) = 2.719( 2   The X-ray structure determination of VA(3) and isonicotinamide hydrate (2VAÁ2INMÁ2H 2 O) 63,64 shows that the hydroxyl and methoxy groups of VA adopt the out-out conformation In the VA + 2H 2 O + DMSO solvation complex (Fig. S4b, ESI †), the hydroxyl and methoxy groups adopt an anti-conformation.
In the minimum energy configuration of the VA + 3H 2 O + DMSO solvation cluster (Fig. S4c, ESI †), the hydroxyl and methoxy groups adopt an anti-conformation.Two molecules of water, W 1 and W In the minimum energy configuration of the VA + 3H 2 O solvation cluster, the hydroxyl and methoxy groups adopt an anti-conformation (Fig. S4d, ESI †).The three water molecules and the OH group of VA are hydrogen bonded in a cyclic order with hydrogen bond lengths in the range of 1.589-1.763Å and hydrogen bond angles 152.91-164.21.The water molecule W 3 forms a very short hydrogen bond with O 4 of VA (1.589 Å), however, the hydrogen bond angle O w 3 -H w 3 Á Á ÁO 4 strongly deviates from linearity (152.91).The torsion angle j(C 5 C 6 C 4 O 4 ) = À14.51 is larger than that of the VA + 3H 2 O + DMSO solvation cluster.
Selected structural and conformational properties of VA + solvent complexes with energy minimization using the B3LYP functional with the 6-31+G(d) basis set in the gas phase, are shown in Fig. S5 (the ESI †).The structures and hydrogen bond properties are very similar to those obtained by the use of APFD functional.The hydrogen bond lengths are longer and the j torsion angles smaller by using the B3LYP functional.
Fig. S6 (ESI †) illustrates minimum energy structures of VA + 2H 2 O and VA + H 2 O complexes.In the VA + 2H 2 O solvation cluster the hydroxyl and methoxy groups adopt anticonformation as in the case of VA + 3H 2 O.The two molecules of water and the OH group of VA are hydrogen bonded in a cyclic order with hydrogen bond lengths of 1.745-1.957Å, which are significantly longer than those in the VA + 3H 2 O complex.This results also in hydrogen bond angles which strongly deviate from linearity (138.21-152.51).In the VA + H 2 O complex the hydroxyl and methoxy groups adopt an inout conformation which results in a rather weak intramolecular hydrogen bond (O 3 Á Á ÁH 4 = 2.376 Å).The water molecule W 1 plays a bridging role, forming a hydrogen bond interaction with the OH group (O w 1 Á Á ÁH  From the above it can be concluded that in the case of VA + DMSO complexes the most stable configurations are those with an intramolecular hydrogen bond interaction (Table S6 The computational 1 H NMR chemical shifts of the solvation complex VA + 3H 2 O + DMSO, with energy minimization using the B3LYP/6-31+G(d) and the APFD/6-31+G(d) methods, weighted by the respective Boltzmann factors, d calc,wg , values were found to be smaller than those of the VA + 3H 2 O complex for both the GIAO and CSGT methods in the gas phase.This is not in agreement with the tendencies of the experimental chemical shifts (Table S6, ESI †).Since the computational 1 H NMR chemical shifts of the VA + 2H 2 O and VA + H 2 O complexes are significantly smaller than those of the VA + 3H 2 O + DMSO complex, very probably, in solution state, there is an equilibrium of various VA + nH 2 O (n = 3-1) solvation species.
From the comparison of experimental and computational NMR chemical shifts of the phenol OH groups of the natural products of Fig. 1, it appears to be conclusive that hydrogen bond interactions in the solvation clusters of natural product + 3H 2 O + DMSO are stronger than in DMSO or H 2 O.The enhanced hydrogen bond interactions can be attributed to the significant cooperative nature of the interaction between the H 2 O and DMSO molecules and the OH groups of the natural products.

Effects of higher Ph + nH 2 O (n 4 4) solvation clusters
Preliminary calculations of Ph(1), as a molecular sensor, in the presence of nineteen (19) discrete water molecules were performed using the semi empirical AM1 method.The H 2 O molecules were distributed uniformly around the phenyl ring and the hydroxyl group (Fig. S8 in ESI †) which can act both as a hydrogen donor and acceptor.In order to increase the accuracy of the calculation the force constants and frequencies at every optimization step (by using the Opt = CalcAll keyword) were computed, since it is known that the calculation of frequencies defines the curvature of the Potential Energy Surface (PES).From the final optimized structure, the nine nearest to the hydroxyl group water molecules were chosen and reoptimized again using the AM1 methodology (Fig. S8(c

Comparison with neutron diffraction data
Recently, neutron powder diffraction data have been reported 65 from a number a flash-frozen aqueous solutions of DMSO with mole fractions w(DMSO) = 0.25-0.67.Three stoichiometric hydrates were obtained which were crystalized on warming between 175 K and 195 K.A DMSO + 3D 2 O complex and two different polymorphs of DMSO + 2D 2 O complexes have been identified.Of particular interest is the structure of DMSO + 3D 2 O complex in an asymmetric unit, which has no structural analogue in the CSD database (Fig. 8).Two of the symmetrynon-equivalent water molecules O W 1 and O W 2 are involved in hydrogen bonding only with other water molecules.The third water molecule forms two hydrogen bonds with the DMSO oxygen and accepts one hydrogen bond via O W 2 .Of particular interest are the HÁ Á ÁO contacts between methyl hydrogens and both water and DMSO oxygen atoms which represent a significant fraction of the Hirshfeld surface area. 65The water molecule W3 of the neutron powder diffraction structure is replaced by the phenol compounds in the (1,2,3) + 2H 2 O + DMSO solvation complexes (Fig. 8).The hydrogen bond distances and hydrogen bond angles are very similar in all cases, which demonstrate a common structural motif of the solvation species.A main difference was found in the two HÁ Á ÁO W 3 contacts of the methyl hydrogens of DMSO with W3 in the neutron diffraction study (2.561 Å and 3.245 Å, Fig. 8a) while those in the computational study are very similar and close to 2.5 Å (Fig. 8b-d).This can be attributed to the ability of W3 to from further hydrogen bond interactions with additional D 2 O molecules which are not possible in the case of the aromatic OH group.

Experimental procedure
NMR experiments were performed on Bruker AVANCE III HD 400 and 500 spectrometers.Samples of the natural products were dissolved in 0.6 ml of mixtures of H 2 O with DMSO-d 6 with solute concentration of B25 mM. 5 mm NMR tubes were used.In variable temperature NMR studies with solutions of various molar concentrations, the selection of the appropriate referencing process is of critical importance.For mobile isotropic media there are two main methods of referencing: (a) internal referencing, where the reference compound is added directly to the system under study and is almost invariably used for 1 H and 13 C NMR.(b) External referencing, involving sample and reference contained separately in coaxial cylindrical tube.These methods have various advantages and disadvantages.For (a) the shielding of the reference nucleus depends on the solvent and on the concentration of both solute and reference owing to the effects of intermolecular interactions.The chemical shifts of 3-(trimethylsilyl) propane-1-sulfonate (DSS) in dilute aqueous and DMSO-d 6 solutions were found to be 0.0173 ppm and À0.0246 ppm, respectively, on the scale with TMS as zero. 66In our study, these differences are very small, falling well below the anticipated range of solvents effects (Dd B 0.52 to 0.73 ppm of Fig. 2, 3 and Fig. S1 in ESI †).Furthermore, due to low molar concentrations of solutes (c = 25 mM) and DSS (c = 0.8 mM), intermolecular interactions would be expected to have negligible effects on chemical shifts.External reference procedures (b) generally require corrections arising from differences in bulk magnetic susceptibility between sample and reference.For the usual coaxial cylindrical arrangement, the correction for superconducting magnets is given by: where k refers to the relevant volume susceptibility of the sample and reference.In our case the magnetic susceptibility corrections at room temperature of H 2 O is À9.035 ppm and of DMSO-d 6 is À7.73 ppm, therefore, a correction of d true À d obs = 0.43 ppm is required which is comparable to the range of solvent effects (Dd B 0.52 to 0.73 ppm of Fig. 2, 3 and Fig. S1 in ESI †).Furthermore, it has been shown that in the case several binary mixtures of organic liquids the value of magnetic susceptibility is not an additive property because of the influence of various factors, such as intermolecular forces.This is especially the case for H 2 O/DMSO mixtures which deviate from linearity and estimates for magnetic susceptibility at variable temperatures would require corrections for variations for solvent density which are not negligible. 67,68It can, therefore, be concluded that the external reference procedure (b) has several disadvantages relative to internal procedure (a).The procedure (a), therefore, was utilized in the present study with standard DSS, d( 1 H) = 0.00 ppm.All spectra were acquired with an acquisition time of 2.1 s, relaxation delay of 5 s, 32 K data points and 901 pulse length.

Computational methods
The computational study was performed by using the Gaussian 09 with the DFT method. 54The structures were optimized by using two functionals, B3LYP and APFD, and the 6-31+G(d) basis set.The optimized structures of the phenol compounds with various discrete water and DMSO molecules were obtained successively starting with phenol (the sensor) and adding first various numbers of water molecules followed by DMSO in various possible positions in such a way that they act either as donors or acceptors.From the resulting structures, those having the lowest energies, were selected.The 1 H NMR chemical shifts were calculated with the GIAO and CSGT methods 55 by using the B3LYP/6-311+G(2d,p) level in the gas phase.Comparative studies using the CPCM 52,53 were not performed since in the case of mixed solvation complexes there is no appropriate continuum model.The computed geometries were optimized by performing frequency calculations at the same level (zero imaginary frequencies).DSS was optimized at the same levels of the theory and was the reference for computed 1 H NMR chemical shifts.The computational chemical shifts of DSS and TMS were found to be identical, in excellent agreement with a minor experimental chemical shift difference of 0.02 ppm. 68

Conclusions
This paper presents an extensive NMR and DFT analysis of hydrogen bonding interactions in H 2 O/DMSO eutectic mixtures by using phenol OH groups of natural products as molecular sensors.The 1 H NMR chemical shifts of the OH groups are strongly dependent on specific solute-solvent interactions, demonstrating maximum hydrogen bonding interactions in the range of mole fractions w(DMSO) = 0.33 to 0.20 for Ph(1) PCA(2), and VA(3).DFT calculated 1 H NMR chemical shifts of various solvation species were found to be in reasonable agreement with experimental chemical shifts even with computationally less demanding level of theory.The increased hydrogen bond interactions in the natural product + nH 2 O (n = 2, 3) + DMSO solvation clusters, relative to complexes with DMSO or H 2 O, cannot be attributed to a single structural parameter of the cooperative interactions between the phenol OH groups of the natural products with H 2 O and DMSO molecules.The excellent agreement of the recent neutron powder diffraction data of a DMSO + 3D 2 O complex 65 with our minimum energy conformers of [Ph(1), PCA(2), VA(3)] + 2H 2 O + DMSO complexes demonstrates a common structural motif of the above solvation species.We would, therefore, expect that the combined use of 1 H NMR studies and DFT calculations of phenol OH groups will provide a very effective method for the structural and electronic description of solute-solvent interactions at an atomic level, 69 and more specific in the field of eutectic mixtures and ionic liquids. 70,71

Fig. 2 (
Aa) illustrates the OH1 H NMR spectral region of (1) for various mole fractions, w(DMSO), of DMSO/H 2 O mixtures at 300 K.For mole fractions w(DMSO) = 1-0.35relatively sharp resonances were observed which allowed the accurate determination of chemical shifts.A linear correlation of d exp (OH) vs. w(DMSO) was observed for w(DMSO) Z 0.35 which shows an increase in deshielding and, thus, in hydrogen bond interactions upon increasing the concentration of H 2 O.This is to be expected since an increasing number of water molecules will increase the number of available sites of hydrogen bond
acceptors, and donors.In the range of mole fractions 0.2 r w(DMSO) r 0.33, a broad maximum of d exp (OH) and, thus, in hydrogen bond strength was observed.This range of mole fractions corresponds to mole ratios of H 2 O/DMSO of 2/1-4/1 (Table S1, ESI †).For mole fraction w(DMSO) o 0.2, a progressive decrease in d(OH) was observed which demonstrates a decrease in hydrogen bond interactions at infinite dilution in H 2 O.The OH resonance at mole fractions w(DMSO) o 0.1 cannot be distinguished from the baseline.
For mole fractions w(DMSO) = 1-0.35,a linear correlation of d(OH) vs. w(DMSO) is evident.In the range of mole fractions 0.20 r w(DMSO) r 0.33, a broad maximum of d(OH) was observed.Since the slow exchange regime t OH À1 { |n OH À n H 2 O | is fulfilled (Table
, d calc,Wg , were aligned in accord to the tendencies of the experimental chemical shifts, d exp .Comparison of the Ph + 3H 2 O and Ph + 3H 2 O + DMSO solvation complexes using both B3LYP and APFD functionals shows that: (i) the incorporation of the DMSO molecule increases the size of the cycling structure and (ii) the increase in the chemical shift in the Ph + 3H 2 O + DMSO complex cannot be attributed to a single structural parameter of the cooperative interactions between H 2 O and DMSO molecules with the phenol OH group.
Computed 1 H NMR chemical shifts, ppm, of various configurations of different molecular cluster compositions of Ph(1), in the gas phase, d calc,g , relative Gibbs DG values, kcal mol À1 , and experimental chemical shifts, d exp , with energy minimization at the B3LYP/6-31+G(d), and APFD/6-31+G(d) levels with NMR calculations at the GIAO (A) and CSGT (B) level.d calc,Wg are the 1 H NMR chemical shifts in the gas phase of various solvation configurations, weighted by the respective Boltzmann factors Composition of solvation clusters B3LYP/6-31+G(d) APFD/6-31+G(d) DG d calc,g

Fig. 7
Fig.7The four minimum energy conformers of VA in the gas phase with energy minimization at the B3LYP/6-31+G(d) level.
) in ESI †).The next steps involved optimization of the Ph(1) molecule with the nearest 1, 3, 4 and 5 discrete water molecules, including also the complex with nine H 2 O molecules, by using the DFT method and employing the B3LYP functional and the 6-31+G(d) basis set.Representative optimized structures are shown in Fig. S9 in ESI.† Fig. S10 in ESI † shows GIAO calculations (at the B3LYP/ 6-311+G(2d,p) level) of the OH chemical shifts of various Ph(1) + nH 2 O (1-9) solvation clusters.The computational chemical shift of the Ph(1) + 3H 2 O complex shows the best agreement with the experimental value and, thus, this molecular cluster is a good approximation of the structure of Ph(1) in aqueous solution.

50 Tables S2 and S3 (ESI †) show that in all cases Dn/t OH À1 4 10 which demonstrates that the slow exchange regime t OH À1 { |n OH À n H 2 O | is fulfilled and no effect of chemical exchange on d exp (OH) is expected. The minor effect of increasing the water content on t OH À1
50,51 |, then, the line shapes reduce to two Lorentzian lines with maxima at n OH , the resonance frequency of the phenol OH site, and n H 2 O , the resonance frequency of the H 2 O, with intensities P OH and P H 2 O and widths t OH À1 and t H 2 O À1 .In this limit the only effect of the exchange is to produce a life-time broadening.50,51As the exchange gets more rapid and t OH À1 , t H 2 O À1 become comparable with n OH À n H 2 O , then, the peaks at n OH and n H 2 O move towards each other.Since the natural line width, especially at low temperatures, is dominated by field inhomogeneity broadening, it appears plausible to extract it from the internal reference signal of DSS and subtract it from the exchange-broadened lines, at a given temperature, using the equation t OH À1 = Dn 1/2 (OH) À Dn 1/2 (DSS) where Dn 1/2 (OH) and Dn 1/2 (DSS) are the linewidths of the OH group and DSS, respectively., for 0.2 o w(DMSO) o 0.4 (Table

Table 1
) a Denotes that H 2 O molecules are hydrogen bonded with the OH group of Ph(1).bDenotes that the DMSO molecule is hydrogen bonded with the OH group of Ph(1).
, ESI †).In the case of VA + 2H 2 O + DMSO solvation complexes most of the open configurations are significantly more stable than those with intramolecular hydrogen bond interactions.Similarly, in VA + 3H 2 O + DMSO solvation complexes the open configurations are the most stable with negligible population of the molecular clusters with intramolecular hydrogen bond interactions (Table S6, ESI †).The proximity of hydroxyl and methoxy groups in vanillic acid may result also in reduced solvent accessibility.Contrary to the case of VA + 3H 2 O and VA + 2H 2 O, low energy configurations of the VA + H 2 O complex with intramolecular hydrogen bond interactions were obtained.