Self-assembled supramolecular structures of O,N,N′ tridentate imidazole–phenol Schiff base compounds

Three imidazole-derived Schiff base compounds comprising an N-methyl imidazole group coupled to a phenol ring through an imine bond were synthesised. The structures differ by the substituent on the phenol ring at the 4-position: methyl (1), tert-butyl (2) and hydrogen (3). The compounds were synthesised using both a traditional reflux in solvent as well as an environmentally friendly solid-state reaction. Compounds (1)–(3) as well as the hemihydrate of (3) were all studied by single crystal X-ray diffraction. The asymmetric unit of compound (1) consists of two nominally planar molecules linked by hydrogen bonds to form a dimeric supramolecular structure. This dimeric structure was ubiquitous for the anhydrous forms of (1)–(3). The complementary hydrogen bonding motif between the imidazole N atoms and the phenol OH results in a stable 16-membered hydrogen-bonded ring. The asymmetric unit of (3) comprises two symmetry-independent molecules one of which has co-planar imidazole and phenol rings while the other shows a significantly oblique orientation. The hemihydrate of (3) similarly forms extensive hydrogen bonds, though in the form of a water-bridged dimeric structure. The hydrogen bond lengths (D⋯A) for compounds (1)–(3) are relatively short, ranging from 2.662(1) to 2.688(1) Å. DFT was used to understand the relative stability of the monomeric and dimeric species. These showed the hydrogen-bonded supramolecular structures were ca. 101 kJ mol−1 lower in energy than the non-interacting monomers. Scan simulations were used to calculate the total energy of the molecule as a function of phenyl ring rotation and showed why the expected planar configuration for a conjugated π-system was not observed experimentally. The barrier to rotation was found to be relatively low, 7.97(6) kJ mol−1, with the lowest energy conformations subtending dihedral angles of 22.319, 24.265 and 25.319° for molecules (1), (2) and (3), respectively. The electrostatic potential maps are able to succinctly explain the stability of the hydrogen bonds through the partial charges of the interacting atoms. TD-DFT simulations and analysis of the simulated and experimental UV/visible spectra suggest that the dimeric supramolecular structure is a stable species in solution. This was confirmed through 1H NMR titrations and an equilibrium constant of 0.16(5) M−1 was estimated.


Introduction
Imidazole-derived Schiff bases are readily synthesised via a condensation reaction between the corresponding imidazolecarboxaldehyde and amine derivative. In general, aromatic amine and aromatic aldehyde precursors result in higher yields than Schiff bases synthesised from aliphatic precursors. This is attributed to the electron dense aryl groups stabilising the imine bond through electron delocalisation. 1 This ease of synthesis coupled with the ubiquitous nature of imidazole moieties in enzymes, proteins and pharmaceuticals render it a favourable choice in biological studies. 1,2 Additionally, Schiff base ligands are able to coordinate and stabilise metals in a wide range of oxidation states. [2][3][4][5][6][7][8] Metal complexes of imidazole-based Schiff base ligands have therefore found application in a wide variety of elds including medicinal and biological chemistry as well as catalysis. 1 In recent years, derivatives of the 1-methyl-imidazol-2-yl methylidine Schiff Base ligand have been co-ordinated to various metals for a wide variety of applications. In contrast to imidazole ligands, the N-methyl group prevents ionisation upon metal ion coordination, ensuring the ligand remains neutral. 9 A search of the Cambridge Structural Database (CSD) shows that no free ligand crystal structures containing the 1-methyl-
NMR spectra for ligands (1) and (2) were recorded with a Bruker Avance III 400 MHz spectrometer equipped with a Bruker magnet (9.4 T) at frequencies of 400 MHz for 1 H and 100 MHz for 13 C. NMR spectra for ligand (3) were recorded with a 500 MHz Varian Unity Inova spectrometer equipped with an Oxford magnet (11.7 T) at frequencies of 500 MHz for 1 H and 125 MHz for 13 C. The spectra were recorded at 30 C. All NMR spectra were processed through Topspin 3.2, patch level 7. 26 The proton and carbon shis were calibrated to the residual DMSO solvent signal at 2.50 ppm for 1 H and 39.51 ppm for 13 C. The CDCl 3 used for the NMR spectroscopic titration was dried over CaH 2 for 24 hours prior to use. FTIR spectra were recorded using a Bruker Alpha FTIR spectrometer equipped with an ATR platinum Diamond 1 reectance accessory. Data were collected using 36 scans with a resolution of 1 cm À1 . Electronic spectra were recorded from 700-200 nm using a PerkinElmer Lambda 25 double beam spectrometer (1.0 cm path length sample cell) equipped with a PerkinElmer PTP-1 Peltier temperature controller set to 25.0 C. Elemental analysis data were collected using a Thermo Scientic Flash 2000 Organic Elemental CHNS-O Analyser. High Resolution mass spectra were acquired with a Waters Micromass LCT Premier time-of-ight mass spectrometer using electrospray ionisation in positive mode.
X-Ray data were recorded on a Bruker Apex Duo diffractometer equipped with an Oxford Instruments Cryojet operating at 100(2) K and an Incoatec microsource operating at 30 W power. Crystal and structure renement data are given in Table  1. Selected bond lengths and angles are presented in Table 2. The data were collected with Mo Ka (l ¼ 0.71073 A) radiation using omega and phi scans with exposures taken at 30 W X-ray power and 0.50 frame widths using APEX2. 27 The data were reduced with the programme SAINT 27 using outlier rejection, scan speed scaling as well as standard Lorentz and polarisation correction factors. A SADABS semi-empirical multi-scan absorption correction was applied to the data. Direct methods, SHELX-2016 (ref. 28) and WinGX, 29 were used to solve all structures. All non-hydrogen atoms were located in the difference density map and rened anisotropically. All hydrogen atoms were treated with the standard riding model in SHELX-2016 with C-H aromatic distances of 0.93 A and U iso ¼ 1.2 Ueq and C-H methyl distances of 0.98 A and U iso ¼ 1.5 Ueq. The O-H atoms were located in the density map and allowed to rene isotropically.

General experimental method for (1)-(3)
Compounds (1) and (2) were synthesised based on a generic condensation reaction in ethanol. 30 A solution of 1-methyl-2imidazolecarboxaldehyde (3.00 mmol) in ethanol (10 mL) was added to a solution of the corresponding aminophenol (3.00 mmol): 2-amino-4-methylphenol and 2-amino-4-tert-butylphenol for (1) and (2), respectively. The resulting yellow solution was heated to reux for 90 minutes. The solvent was removed via rotary evaporation under reduced pressure and the resulting orange oil cooled to À20 C for 12 hours, yielding yellow crystals. Compound (3) was synthesised via a solid-state reaction, which has been used to prepare related imine compounds, 31 followed by a re-crystallisation process using toluene. 1-Methyl-2-imidazolecarboxaldehyde (4.42 mmol) and 2-aminophenol (2.99 mmol) were ground to a paste in an agate pestle and mortar for approximately 10 minutes. Upon standing, a yellow powder formed which was then dissolved in toluene (15 mL) with activated 3 A molecular sieves. The solution was heated to reux for 30 minutes. The molecular sieves were ltered from the hot solution. The yellow ltrate was le to stand at room temperature overnight. Dark yellow hexagonal-shaped crystals (0.352 g) formed upon slow evaporation of the solvent. The crystals were collected by ltration, crushed and placed in a vacuum oven at 50 C for 2 hours. A second re-crystallisation gave a nal yield of 71%. (1). Cooling the oil of (1) yielded yellow crystals suitable for X-ray diffraction. Yield: 0.605 g, 94%. UV-vis: (CH 3

Synthesis of ligands (1)-(3)
Three O,N,N 0 tridentate imidazole-imine Schiff base ligands have been designed and synthesised for chelation to the VO 2+ metal centre. The ligands comprise a 1-methyl-1H-imidazole ring linked by an imine bond at the 2-position to either 2-amino-4methylphenol (1), 2-amino-4-tert-butylphenol (2), or 2-aminophenol (3). Compounds (1) and (2) are novel. Compound (3) is known, 20 however no characterisation data or crystal structure of the free ligand has been reported. All ligands were synthesised by a condensation reaction of 1-methyl-2-imidazolecarboxaldehyde with the corresponding aminophenol. A common synthetic route for this condensation reaction is to reux the carboxaldehyde and amine in ethanol or methanol and allow the imine ligand to precipitate upon cooling. 30,33 This method did not produce a precipitate from the mother liquor and so all solvent was removed; this yielded the target compound as an oil. The formation of an oil is not unusual for this class of compound. 10,14,16,18,19 Upon cooling to À20 C, crystals of the respective ligand formed which remained stable once warmed to room temperature. The product yields were high (94% and 86% for (1) and (2), respectively) and of excellent purity (see ESI † 1 H and 13 C NMR spectra). Attempts to synthesise (3) utilising this same procedure did not yield the target compound. Synthesis of (3) has been reported. 20,25 However, an efficient solid-state synthesis adapted from Akerman and Chiazzari 31 was employed to synthesise (3) due to the reduced number of synthetic steps and minimal solvent use, i.e. a green synthetic method. In the adapted method, the 2-aminophenol and a slight excess of 1methyl-2-imidazolecarboxaldehyde were ground to a paste in an agate pestle and mortar with no additional solvents. The formation of the paste is indicative of the elimination of water, concomitant with imine bond formation. The target compound was most effectively recrystallized from toluene. The imidazole starting material, which is soluble in toluene, was added in excess to ensure complete reaction of the 2-aminophenol for which the solubility in toluene is similar to that of the target compound. Using the solubility differences was a simple and effective purication technique. When the re-crystallisation process was performed at room temperature, the crystals formed were the hemihydrate ((3)$0.5H 2 O), even with the use of molecular sieves in the re-crystallisation process. The anhydrous compound (3) could be obtained by dissolving the yellow powder formed from the solid-state reaction in toluene with activated 3 A molecular sieves and heating the resulting solution for 30 minutes. Crystals of (3) formed upon slow cooling the hot solution.

NMR spectroscopy
The imine group has similar chemical shis in all three compounds in both the 1 H (8.54 ppm for (1) and (3) and 8.56 ppm for (2)) and 13 C NMR spectra (150.70, 150.94 and 150.93 ppm for (1), (2) and (3), respectively). The same is true for the imidazole 1 H and 13 C chemical shis. The 1 H NMR chemical shis for the imine and hydroxyl group showed that, as in the solid state, the OH hydrogen atom remained on the oxygen in solution and did not migrate to the imine nitrogen to switch from the enol to the keto tautomer as has been reported for similar compounds. 30 The chemical shi for the OH group is similar for compounds (1) and (2) (8.76 ppm and 8.78 ppm, respectively), but is further downeld for compound (3) at 8.98 ppm. The electron-donating effect of the methyl and tertbutyl moieties para to the OH group have a shielding effect while the un-substituted phenol ring which has the electron withdrawing effect of the phenyl ring is deshielded in comparison. A similar effect is noted in the 13 C chemical shis of the C-OH atom (148.21, 147.99 and 150.53 ppm for compounds (1), (2) and (3), respectively). Fully assigned 1 H and 13 C spectra are available in the ESI. † Compound (1) crystallises in the monoclinic space group P2 1 /n with two hydrogen-bonded molecules comprising the asymmetric unit. The complementary hydrogen bonding between the imidazole N atom and the OH group of the adjacent molecule leads to a 16-membered hydrogen-bonded ring. The hydrogen bond parameters are summarised in Table 3. Despite the extended aromaticity of the ligands, the molecules exhibit a notable deviation from planarity. This deviation from planarity is indicated with the angle subtended by the veand six-atom mean planes of the imidazole and phenyl rings, respectively. This measures ca. 15 for both molecules (1a) and (1b) of the asymmetric unit. The most signicant difference in geometry between the two molecules in the asymmetric unit of (1) is the direction of the relative rotation between the phenyl and imidazole rings. With the phenyl ring as a reference, the methylimidazole moiety is rotated below the mean plane of the phenyl ring in one case and above in the second. The bond lengths and angles describing the imine bond for each of the molecules are summarised in Table 2. Compound (2) crystallised in the C2/c space group with a single molecule in the Table 3 Hydrogen bond lengths ( A) and bond angles ( ) describing the stabilising intermolecular interactions for (1)- (3) and (3) This journal is © The Royal Society of Chemistry 2020 RSC Adv., 2020, 10, 7867-7878 | 7871 asymmetric unit. Despite the signicant difference in steric bulk between the methyl and tert-butyl substituents of compounds (1) and (2), the geometry of the molecules is remarkably similar. The same distortion from planarity is noted in compound (2) with the angle subtended by the imidazole and phenyl rings measuring ca. 22 . The energetics of these out-ofplane rotations is further explored using molecular simulations (vide infra).
Compound (3) has been studied in the solid state as both the hydrated and anhydrous forms: (3)$0.5H 2 O and (3), respectively. Some signicant variations in the geometry are evident. In compound (3), which crystallised in the P2 1 /c space group, the asymmetric unit comprises two independent molecules with signicantly different molecular congurations. Molecule (3a) shows a more perpendicular orientation of the phenyl and imidazole rings while (3b) is approximately planar. Using the same six-and ve-membered phenyl and imidazole mean planes to describe the deviation from planarity, the angles subtended by molecules (3a) and (3b) measure approximately 66 and 7 , respectively. The hydrated form crystallised in the monoclinic C2/c space group and has adopted a conguration with a slight distortion from planarity (as indicated by an angle of 6.1 between the phenyl and imidazole mean planes), consistent with compounds (1), (2) and (3b).
Selected bond lengths for the compounds are summarised in Table 2. The C5-N3 imine bond lengths, which range from 1.277-1.284 A and the C3-C5-N3 bond angles, which range from 122.3-123.4 are indicative of the double bond character of the azomethine group and sp 2 hybridisation of the imine carbon atom. The isomerisation about the imine bond is exclusively trans for all compounds studied. The trans conguration is seemingly favoured as there would be non-bonded repulsion between the hydroxyl and imidazole N-CH 3 group in a cis conguration. The trans conguration also allows for weakly stabilising intramolecular C-H/O interactions. This trans conguration has been reported for similar compounds. [10][11][12][13][14][15][16][17][18][19][20]32 The bond distances and angles show little variation within the present library of compounds and compare favourably with those previously reported for related compounds. [10][11][12][13][14][15][16][17][18][19][20] The data in Table 2 show that the N3-C6-C7 and N3-C6-C11 bond angles are (unexpectedly) signicantly different, measuring ca. 127 and 114 , respectively. This deviation from the ideal angle of 120 for an sp 2 hybridised carbon atom is likely a consequence of steric repulsion between the phenol OH group and the imine C-H. This distortion is less pronounced for molecule A of compound (3) as the steric strain is released by the out-of-plane rotation of the phenyl ring.
Compounds (1)-(3) form complementary hydrogen bonds between the un-substituted imidazole nitrogen and the OH group of the neighbouring molecule. This hydrogen bonding motif yields a dimeric supramolecular structure supported by a sixteen-membered hydrogen bonding ring. In the case of (1) the asymmetric unit consists two molecules which are linked by hydrogen bonds while ligand (2) has a single molecule in the asymmetric unit. The asymmetric unit of the anhydrous compound (3) comprises two symmetry-independent molecules which are not hydrogen-bonded to each other, but rather to neighbouring molecules. The most notable difference between the two independent molecules in the asymmetric unit in compound (3) is the C5-N3-C6-C7 torsion angle. For the relatively planar molecule A this torsion angle is À7.7(2) and for molecule B, with the out-of-plane twisting of the phenyl ring, it measures À59.1 (2) . Despite the differences in molecular geometry, the same complementary hydrogen bonding motif is evident in both structures. The dimeric supramolecular structures are shown in Fig. 2. The hydrogen bonding parameters are summarised in Table 3.
The hydrogen-bonded dimers in (1) are symmetry independent, but in (2) and (3) inversion dimers are formed. The waterbridged dimer of (3)$0.5H 2 O is of C 2 symmetry. These hydrogen bonds are considerably shorter than the sum of the van der Waals radii of the interacting atoms. This short bond length coupled with the fact that the bonds are approaching the ideal bond angle would suggest that they are moderate to strong interactions. The compounds also all show intramolecular interactions between the imine C-H (donor) and the phenol OH (acceptor) group. The dimers of (3a) and (3b) are linked by C-H/O interactions between the phenol OH group and the imidazole C-H group to form one-dimensional columns which transverse the ab plane. The one-dimensional column is shown  in Fig. S7. † In the case of (3)$0.5H 2 O, the water-bridged dimers are linked through C-H/N interactions to form a one dimensional column co-linear with the c-axis (Fig. 3). The same imidazole N atom therefore acts as an acceptor for two intermolecular interactions. For compounds (1), (2) and (3) the imine N atom is precluded from participating in intermolecular interactions because of steric crowding by the imidazole methyl group and ortho hydrogen atom of the phenol ring.

Solution studies
The spontaneous dimerization of compounds (1)-(3) in the solid state was established through X-ray crystallography. This then begs the question whether the same process is apparent in the solution phase. To this end, the dimerization process at 30 C was followed by 1 H NMR titration. Compound (2) was selected for the solution phase study as it shows the highest solubility in chloroform owing to the tert-butyl functional group. 1 H NMR spectra for compound (2) were recorded at 30 C over a concentration range of ca. 3 Â 10 À3 -4 Â 10 À1 M based on the mass of the monomeric unit. The line shapes and chemical shis of the OH group were followed over this concentration range (Fig. 4). The chemical shi of the OH group varies signicantly over this concentration range, moving downeld from 6.62 ppm to 7.11 ppm. This shi is likely a consequence of the deshielding effect of hydrogen bonding. Importantly, the chemical shis and line widths of the remaining peaks are unchanged over this concentration range. This shows that their magnetic environment does not change with concentration. These signals are therefore a reference point for the 1 H NMR spectrum of (2) and highlight the marked change of the phenolic OH group as a consequence of intermolecular hydrogen bonding. In addition to the deshielding of the OH group, line broadening is also noted. This is a consequence of the greater rate of exchange for the monomer 4 dimer equilibrium, in relation to the timescale of the 500 MHz 1 H NMR. The observed signal is the weighted time average of the two species in solution. The observed resonance may then be explained by the simple sum of the products of the line widths and fractions of the monomer and dimer as shown in eqn (1).
In eqn (1), d obs is the observed chemical shi, d M and f M are the chemical shi and fraction of the monomeric species, respectively. Similarly, d D and f D are the chemical shi and fraction of the dimeric species, respectively, in solution at a given concentration.
The concentration-dependent 1 H NMR spectra strongly suggest that the dimerization process does take place in the solution state. To quantify this, the equilibrium constant for the dimerization process was established by monitoring the change in chemical shi of the OH group as a function of concentration. By tting eqn (2), the equilibrium constant for the dimerization process was determined. 33 Eqn (2) is a 1 : 1 binding equation, which is appropriate for a dimerization process, 33 and was tted by a least squares process using Origin 9.1.
In eqn (2), d is the weighted average chemical shi of the phenolic OH hydrogen (i.e. the observed chemical shi), d D is the chemical shi of the OH group in the dimeric species, d M is the chemical shi of the OH hydrogen atom in the monomeric species, K is the equilibrium constant and C is the formal concentration of the monomer in solution. The tted data are presented in Fig. 5.
The non-linear least squares t of eqn (2) to the concentration-dependent 1 H NMR chemical shi of the phenolic OH group in Fig. 5 shows that the equilibrium constant for the process is 0.16(5) M À1 at 30 C with d D and d M  values estimated to be 11.4(9) and 6.62(1) ppm, respectively, at the same temperature. The DG value for this reaction is ca. 4.6 kJ mol À1 . This shows that at this temperature the process is slightly endergonic and may explain why relatively high concentrations of compound (2) were required before a signicant change in the chemical shi of the OH group was observed. These data are in agreement with the energetics of similar systems. 33 The DG value is relatively close to zero in CDCl 3 at 30 C, without variable temperature studies it is not possible to know whether the dimerization process would remain endergonic at lower temperatures or become exergonic. In the solid state, the hydrogen bond parameters and molecular geometry are very similar irrespective of the functional group on the phenol ring. Speculatively, the dimerization processes for compounds (1) and (3) in the solution state should therefore have similar equilibrium constants to compound (2).
To ensure that the observed spectral changes were due to the dimerization process and not hydrogen bonding to water, the effects of water on the spectra were also examined through the addition of small aliquots (1 mL) of water to the NMR samples. This did lead to some line broadening, but importantly had minimal effect on the chemical shi which moved downeld (1%. This suggests, rstly, that the process of drying the CDCl 3 over CaH 2 was effective and that, secondly, the changes in chemical shi as a function of concentration are the result of Schiff base dimerization in solution.

Density functional theory
The solid-state structures showed some unexpected molecular congurations for the different compounds. This prompted a further study into the optimum geometry and relative energies of these molecular congurations. Molecular simulations using Density Functional Theory (DFT) were performed using Gaussian 09 W. 34 The X-ray coordinates were used for the input structures unless otherwise specied. Lowest energy conformations were determined through geometry optimisation of both the dimeric and monomeric structures. Frequency simulations were completed for both the monomeric and dimeric species of each compound. All simulations were run at the B3LYP/6-311G level of theory, single rst polarisation and diffuse functions (d,p) were also added to the basis set to improve accuracy. Input les were prepared, and output les analysed using GaussView 5.0. 35 The frequency simulations suggest the geometry optimisations are true minima on the global potential energy surface based on a lack of negative frequency eigenvalues. Transition energies and oscillator strengths were calculated for 24 excited states using the TD-DFT method [36][37][38][39][40][41][42][43][44] at the same level of theory applied to the geometry optimisations. A Polarizable Continuum Model (PCM) was included in the calculation of transition energies and oscillator strengths to account for any solvent effects. 43,44 The TD-DFT simulations were performed on the in vacuo geometryoptimised structures. The molecular orbitals were assigned by studying the spatial distribution of their isosurfaces. Only singlet excited states were considered. The experimental absorption spectra were recorded in acetonitrile and thus an acetonitrile solvent continuum was included in the simulations to account for any possible solvent effects.
The solid-state structure of compound (3) showed that two distinct molecular congurations exist in the solid state. This prompted a study of the relative energies of these congurations. In addition to the relative energies of these two congurations, the total energy of the molecules as a function of the relative angles of the phenyl and imidazole rings was performed (rotations were effected in 10 increments around the C5-N3-   (1) showing all atoms rendered at their van der Waals radii. The plot shows that the out-of-plane rotation reduces the non-bonded repulsion between the C-H and phenol oxygen atom.
C6-C7 torsion angle). These data showed an interesting relationship between energy and molecular geometry. This relationship is illustrated in Fig. 6. The same relationship between molecular geometry and total energy was noted for compounds (1)-(3), irrespective of the substitution on the phenyl ring. This result seems reasonable as the relative rotation of the phenyl ring with substitution at the 4-position will not be inuenced by steric interactions in vacuo.
The energy scan shows that the highest energy conguration has the phenyl ring perpendicular to the molecule. In this orientation the p-orbitals may no longer overlap, the extended aromaticity of the ligands is thus broken, and the molecule is consequently destabilised. The lowest energy conformation has a C5-N3-C6-C7 torsion angle which measures 22.319 . Logically, a planar molecule which would allow for complete overlap of the p-orbitals of the conjugated p-system would be the lowest energy conformation. In practice, a slight out-of-plane rotation (as indicated by the above torsion angle) lowers the energy of the structure. A 'space-lling' plot which renders the atoms using their van der Waals radii (Fig. 7) provides insight into the potential reason for this. A planar conguration leads to increased steric repulsion between the imine C-H and phenol oxygen atom. An out of plane rotation increases the H/O distance from 2.123 to 2.205 A, reducing the steric strain and yielding a more stable molecule. The same torsion angle measures 24.265 and 25.319 for molecules (2) and (3), respectively. The barrier to rotation of the phenol group is low, averaging 7.97 kJ mol À1 for the three compounds with a standard deviation of 0.06 kJ mol À1 . This low energy barrier may indicate why compound (3) was able to adopt two quite different geometries in the solid state.
Structural overlays (least squares ts) of geometry-optimised and experimental structures (Fig. 8) were calculated using Mercury 4.1.3. 45 The root-mean-square deviations indicate the experimental and simulated structures are generally in good agreement for the monomers. The dimeric structures have larger deviations between the experimental and lowest energy conformations. This difference lies predominantly in the relative angle of rotation between the two molecules comprising the dimer. In compounds (1), (2) and (3b), the solid-state dimers could be considered approximately co-planar with the angles subtended by the two sixteen-atom mean plans of the non-H atoms measuring 2.0 for compound (1) and 0 for compounds (2) and (3), since they are inversion dimers. The relative rotation of the two molecules lowers the energy of the dimer by a modest 1.1 kJ mol À1 compared to a co-planar arrangement in the gas phase. The geometry-optimised structures show that in the absence of packing constraints imposed by a crystal lattice, the lowest energy conguration (albeit by a small margin) has the two molecules subtending angles of 39.39, 38.63 and 34.10 between the same 15-atom mean planes comprising all non-hydrogen atoms of the imidazole, phenol and imine groups. The differing geometries of the dimers are shown in Fig. 9. The hydrogen-bonded dimers formed by (2) and (3) are related by inversion symmetry while that of compound (1) is of C 1 symmetry in the solid state. The geometry optimised in vacuo structures of all three compounds are not related through inversion symmetry, but rather C 2 symmetry.
The stability gained by formation of the supramolecular structures was also calculated using the same level of theory. The hydrogen bonding motif comprises two complementary O-H/N hydrogen bonds. The hydrogen-bonded dimers are signicantly lower in energy than the non-interacting molecules, highlighting the stability of this motif. The hydrogen bonded compounds (1)-(3) are lower in energy by 101.8(2) kJ mol À1 . This corresponds to a bond energy of 50.9 kJ mol À1 per hydrogen bond of the motif. The small  standard deviation shows that in the gas phase the substituent on the phenol ring has effectively no inuence on the bond strength. The water-bridged supramolecular structure of (3)$ 0.5H 2 O is similarly stabilised by the hydrogen bonds which lower the energy by 135.1 kJ mol À1 compared to a model with two independent ligands and a water molecule.
In order to better understand the origins of the hydrogen bonds, the partial charge distribution (NBO charges, measured in electrons) were analysed. These indicate that the OH hydrogen atom has the highest positive partial charge in both the dimeric and monomeric species in all three compounds. The most negative partial charge is carried by the phenol oxygen atom in all species; the imidazole N atom has the second highest partial negative charge. There are examples in literature which exhibit hydrogen bonding solely between the electronrich oxygen atom (H-bond acceptor) and the OH hydrogen atom of a neighbouring molecule (CSD reference JUBKOG). 25 In the present structure, the added stability afforded by the complementary hydrogen bonding motif (i.e. two hydrogen bonds) is apparently more signicant than an interaction forming merely between the species with the highest partial negative and positive charges.
A plot of the electrostatic potential (ESP) in Fig. 10 highlights the favourable electrostatic interaction between the imidazole N-atom and the phenolic OH group. The ESP surface also indicates formation of the intramolecular C-H/O interaction between the imine C-H and phenolic OH group.
The UV-visible spectra were calculated for both the monomeric and dimeric species using the TD-DFT method at the same level of theory as used for the geometry optimisations, but including an acetonitrile solvent continuum running the Polarizable Continuum Model (PCM). Table S1 † indicates the main transitions with oscillator strengths for both the monomeric and dimeric structures of (1).
The electronic transitions are summarised in Table S1, † these data show that the spectra of both the monomeric and dimeric species are dominated by high energy p / p* Fig. 10 Electrostatic potential (ESP) map from the total SCF density for the dimer of (1) highlighting the zones of positive and negative potential and illustrating the origin of the hydrogen bonding. Fig. 11 Superposition of the experimental UV-visible spectrum of (1) and the TD-DFT calculated spectra (CH 3 CN solvent continuum) of the monomeric and dimeric structures. A calculated spectrum of an equilibrium system with 60% contribution from the dimer and 40% from the monomer is also shown. transitions. To link the experimental and simulated data, a superposition plot of the spectra was generated (Fig. 11). The superposition plot shows good correlation between the experimental spectra and simulated data for the dimeric structure, particularly in terms of l max values. The extinction coefficients (the simulated data were not normalised) differ to some degree and suggest that the solution state is in reality an equilibrium between the monomeric and dimeric species. A calculated spectrum for a mixed system is indicated in Fig. 11, this shows excellent agreement with the experimental data for a system with 60% dimer and 40% monomer present. This result seems reasonable considering that the hydrogen-bonded supramolecular structure is more stable than the isolated molecules.
The highest occupied molecular orbitals (HOMO) and lowest unoccupied molecular orbitals (LUMO) for the dimers of (1)- (3) show they are all of p-symmetry (Fig. 12), but signicantly they span both molecules. This is interesting as it suggests the dimer is not simply two adjacent monomers, but rather a genuine supramolecular structure with p-electrons spread over both molecules, in accord with previous reports on H-bonded dimers of pyrrole-imine Schiff base derivatives. 31,46 Since the basis set has been augmented with diffuse functions, the results are likely to be reliable.

Conclusions
Three O,N,N 0 donor tridentate imidazole-imine Schiff base ligands have been synthesised and fully characterised. Ligands (1) and (2) are novel. The ligands (1) and (2) were successfully synthesised utilising a standard condensation reaction by heating to reux 1-methyl-2-imidazolecarboxaldehyde with the corresponding phenol. Compound (3) was synthesised from 1methyl-2-imidazolecarboxaldehyde and 2-aminophenol using a solid-state synthetic method followed by re-crystallisation from toluene. The synthetic methods proved to be rapid and high yielding and in the case of (3), environmentally friendly. The single crystal X-ray structures of (1), (2) and (3) were elucidated and showed hydrogen bonding between the unsubstituted imidazole nitrogen and the OH group of a neighbouring molecule. This complementary hydrogen bonding motif leads to dimeric supramolecular structures. By varying the synthetic method, it was possible to form either the anhydrous or hemihydrate form of (3). The hemihydrate molecule favours the formation of a water-bridged supramolecular structure. The stability of the dimers was conrmed by DFT simulations which showed the hydrogen-bonded structures to be lower in energy by ca. 101 kJ mol À1 (this equates to ca. 50 kJ mol À1 for each hydrogen bond). The solid-state structures are inversion dimers for (2) and (3) and C 1 symmetry for (1). The geometry optimised dimers are all of C 2 symmetry. The solidstate structure of (3) showed two signicantly different molecular congurations described by the relative rotation of the imidazole and phenol rings. This was probed using DFT methods which showed that the barrier to rotation for these two groups is relatively low (7.97(6) kJ mol À1 ) with the lowest energy conformations subtending angles of 22.319, 24.265 and 25.319 for molecules (1), (2) and (3), respectively. The ESP maps succinctly explain the stability of the hydrogen bonds through the partial charges of the interacting atoms. TD-DFT simulations suggest that the dimer is stable and the dominant species in solution.

Conflicts of interest
There are no conicts to declare.