Fluorine­based Zn salan complexes

We synthesised and characterised the racemic and chiral versions of two Zn salan fluorine-based complexes from commercially available materials. The complexes are susceptible to absorbing H 2 O from the atmosphere. In solution (DMSO-H 2 O) and at the millimolar level, experimental and theoretical studies identify that these complexes exist in a dimeric-monomeric equilibrium. We also investigated their ability to sense amines via 19 F NMR. In CDCl 3 or d 6 -DMSO, strongly coordinating molecules (H 2 O or DMSO) are the limiting factor in using these easy-to-make complexes as chemosensory platforms since their exchange with analytes requires a high excess of the latter.

or the presence of a base or aggregates (Scheme 1 B), non-coordinating solvents. The former process is well and explicitly discussed and recently reviewed by Di Bella. 52,56 In these reviews, the coordination number that the metal centre adopts depends on the solvent system and the presence (absence) of substrates, varying from four to six. Notably, the reduced version of salen, which means salan ligands (Scheme 1, upper right) and their corresponding Zn complexes, have been less investigated. [57][58][59][60][61][62][63][64] These compounds are susceptible to oxidative dehydrogenation, depending on the solvent medium. 65,66 Only 31 crystal structures have been deposited in the CSD. 67 The sp 3 carbon atoms of the salan framework impose flexibility and, consequently, possible alterations in the geometry of the metal centre, deviating from the dominant square planar and/or square pyramidal conformation. Previous studies identify the ability of Zn(II) complexes to sense biologically important small molecules 68 such as amino acids, 69,70 amines, 71 saccharides 72,73 and ribonucleosides. 74 Recently, Zhao suggested that Zn(II) salen complexes, ideal models for catalysis, are inappropriate for sensing purposes because their structural changes (Scheme 2, species A', B' and C') interfere chronically with the sensing process, thereby averaging the 19 F NMR signals of the interconverting species. 75 With all these in mind, we embarked on a project examining if the easy-to-make, low-cost, chiral Zn(II) salan complexes can be used as chemosensory platforms with 19 F NMR. We detail our ligand-complex design criteria for this purpose (Scheme 2). The use of salan frameworks will increase the flexibility of the organic framework and consequently affect the coordination geometry of the metal centre. However, incorporating the rigid cyclohexane backbone instead of the en moiety will impede the flexible character. Next, we wanted to investigate the impact of the different transducers (F vs OCF 3 ) but also varying its position, adjacent or away from the sensing point (metal centre). Thirdly, we investigated if altering the chirality of the host will impact the sensing process. Lastly, we envisaged these complexes existing as dimers in the solid state and possibly in the solution state; therefore, we hypothesised that saturated analyte (complex: analyte ratios 1:20, 1:40, 1:50 and above) solutions will favour the sole formation of species C' (Scheme 2), thus advancing a new sensing process (appearance of one peak corresponding to the complex+analyte species C'); the scope of this hypothesis is discussed and presented.    Complex synthesis. With the ligands in hand and bulk, we performed several reactions for synthesising the corresponding Zn complexes (Scheme 3). We screened several parameters such as metal salt (Zn(NO 3 ) 2 6(H 2 O), ZnCl 2 , Zn(OTf) 2 , Zn(BF 4 ) 2, Zn(ClO 4 ) 2 , solvent (MeOH, CH 2 Cl 2 , EtOH, CH 3 CN), metal : ligand ratio (3:1 to 1:3), temperature (25 o C, 50 o C, 75 o C), base (Et 3 N, Na 2 CO 3 , K 2 CO 3 ) and identified the optimum conditions as : Zn(NO 3 ) 2 6(H 2 O) : Ligand : Na 2 CO 3 in a molar ratio 1:1:1, on a mixture of solvents MeOH/H 2 O (10/2mL). The metal salt and ligand were dissolved in methanol, and the solution turned milky upon adding an aqueous Na 2 CO 3 solution. After 1 hr of reflux, the solution was filtered, and the filtrate was kept for slow evaporation. Shiny block-shaped colourless crystals were collected in good to moderate yields between four and fourteen days. Characterisation in solid state. Single crystal X-Ray diffraction studies (Table S1) for Zn-1 and Zn-2 families identified the formation of dimers for all cases; however, minor differentiations could be identified in the crystallised lattice molecules for Zn-1-RR and Zn-1-SS. No lattice molecules could be determined for compound Zn-1-rac, however the actual formula for Zn-1-RR is [ZnL 1(CH 3 OH) 0.375(H 2 O)] and for Zn-1-SS is [ZnL 2(CH 3 OH)]. For the Zn-2 family, all compounds crystallise as [ZnL 2(H 2 O)]. All lattice molecules form hydrogen bonding interactions with the metalloligand (ZnL) moieties. Zn-1-rac and Zn-2-rac crystallise in achiral space groups (P2 1 /n and P-1), while the remaining four complexes crystallise in chiral space groups. The Zn centre inclines to adopt a trigonal bipyramidal geometry ( Figure S1 & Table 1). Notably, for the chiral species, two different indexes could be determined; however, their average is close to the index calculated for the non-chiral species (Table 1). In all species, the C-N bond is within the range of single bond values, discarding the occurrence of oxidative dehydrogenation, 65,76 while the phenoxido C-O bond is within the range of a single bond value. The Flack parameter value (Table S1) for all four compounds is close to zero, thus determining enantiomeric purity. The compounds were further characterised by Thermogravimetric (TG) and elemental (CHN) analysis, which slightly deviated from the expected calculated values. They are consistent with additional lattice solvent molecules. (Fig S2).  (7) 0.74 Trigonality index indicates the geometry of the coordination center. 77 When τ = 0 the geometry corresponds to square pyramidal, when τ= 1 corresponds to trigonal bipyramidal.

Zn-1-rac
Zn-2-rac Characterisation in solution state. With the complexes in hand, we attempted to elucidate if they retain their structure in solution; therefore, we recorded 1 H and 19 F NMR, CD and ESI-MS. The ESI-MS data for Zn-1-RR ( Fig S3) validate the formation of monomeric and dimeric species; a characteristic peak with corresponding isotropic distribution can be identified for both species. CD studies of selected samples in DMSO validate the retention of the enantiomeric form ( Figure 3A). Then, we recorded 1 H and 19 F NMR data using a coordinating solvent (d 6 -DMSO) and compared them with the free ligands ( Figure 3 B&C and ESI). From these data, complexation is evident; characteristic peaks in the aromatic and cyclohexane backbone regions shift at different values in 1 H NMR. In 19 F NMR, the sole peak shifts by 0.1 ppm (Fig 3C), whereas for the Zn-2 family, the peak shifts almost by 1 ppm (0.87ppm, Figure S5). Minor peaks can be observed in the 1 H NMR of Zn-1-SS; however, this may be attributed to different species formed in solution (see Scheme 2, A', B' and C'), rather than impurities since the 19 F NMR data suggests the presence of only one species. To validate the existence of a monomeric or a dimeric species, we performed 19 F NMR diffusion studies ( Figure 3D). However, the data was inconclusive as the apparent molecular weight was intermediate between the putative monomeric and dimeric species. We performed potentiometric studies for 2-rac and its corresponding Zn-2-rac complex at a millimolar level in two concentrations (0.2 and 1.2 mM) in a mixed DMSO-H 2 O solvent system ( Figure S6). Four deprotonation constants can be calculated for the two phenolic OH and two NH groups. The lowest pK value belongs to the deprotonation of one NH group, while the other deprotonation processes overlap. The lowest pK values in all measurements agree; however, the basic pH range values slightly differ in the samples with different ligand concentrations. The solution at 1.2 mM concentration becomes turbid above pH 9, prohibiting further evaluation. The higher pK values were set from the 0.2mM measurements. The interaction of the ligand with Zn(II) was studied at a 1:2 metal-to-ligand ratio in 0.2 and 1.2 mM (for the ligand) solutions; the data were evaluated considering the monomeric and dimeric species (Figure 4). From the distribution curves (Table 2), we note that (a) the higher ligand concentration shifts the complex formation to lower pH, and (b) the formation of the diprotonated dimeric complex in the 1.2 mM solution is favoured. Notably, upon ligand concentration increase, the data fitting improves with the dimeric model, and at a concentration of 1.2 mM, the formation of the dimeric complex is more favourable.  Investigating amine response. With the complexes in hand, we attempted to identify the best solvent system, with CDCl 3 being our first choice. Previous works used this solvent and the complexes at 1mM level. [38][39][40] The data presented herein are the outcome of mixing the analyte and the complex within the NMR tube, avoiding stirring or sonication for prolonged periods; the reason for this choice is our aim to develop an efficient, convenient and rapid sensing method. Our first trial for Zn-1RR ( Figure S7), identifies that this noncoordinating solvent system levies monomeric-dimeric competition, 52 therefore, we discarded its use. Then we used a binary solvent system (CDCl 3 -DMSO) in different ratios. However, the same behaviour was observed again (results are not presented). Thus, we concluded that DMSO would be the ideal solvent to proceed. The 19 F NMR data of the complexes (Figure 4) identify a single broad peak, indicating that an equilibrium (monomer vs dimer) is favoured or that different types of interactions occur. 36 Initial studies of the complex:analyte in 1:1 to 1:4 ratios give an unaltered complex spectrum. (results not presented) To our disappointment, titrations of Zn-1RR and Zn-1SS at millimolar scale with limited excess of phenylglycinol in ratios 1:5 to 1:50, as this was our testing hypothesis, show minimal differences, thus prohibiting us from further continuing with this study. Given that we incorporated a strongly coordinating solvent (DMSO) which may occupy the coordinating site(s) on the Zn centre, we tried experiments with 1:100 and 1:150 ratios to favour the DMSO-analyte exchange. The latter experiment (1:150) suggests that a ligand (1-RR vs analyte) exchange process starts, and the Zn(analyte) x complex forms (Scheme 1, lower).
Then we tested Zn-1SS with other amines ( Figure 5 A-C). Our first choice was to incorporate a diamine at different ratios (1:20, 1:50 and 1:100, Figure 5A). The data from the first trial shows the main peak slightly shifted and two minor peaks, which could be indicative of the formation of the complex+analyte species; however, given that none of the two peaks is of the same intensity or integral with the main peak, we discarded this set of experiments from future studies. We also tried a different amine but noted a slight shift in the principal peak and the appearance of a minor peak ( Figure 5B). The same observation was noted when an amino acid was incorporated ( Figure 5C). Last, we used Zn-2RR and phenyl glycinol at high complex analyte ratios (1:100 and 1:150), but again, one single peak could be observed ( Figure 5D). DFT studies. We performed a DFT study to investigate the species of Zn-1-RR formed in solution, to support the interpretation of the observed experimental signals (Scheme 4). Calculations were carried out to model the system in non-coordinating (chloroform) and coordinating (DMSO) solvent. To find the most relevant species, we screened the conformational space using CREST, 79 and re-optimised in Gaussian 80 the lowest and most representative minima at the B3LYP 81-83 -D3 84 /6-31g(d,p) 85,86 & SDD 87 level. The solvent (chloroform or DMSO) was modelled as a continuum media using PCM; 88,89 in some calculations, up to two DMSO explicit molecules were included (see Computational Details in ESI). Results in chloroform show that in the absence of an analyte or coordinating solvent, the optimised Zn-1-RR structure resembles that (Dimer) defined by single X-Ray diffraction. The Dimer is thermodynamically stable compared to two molecules of the corresponding monomer by 20.5 kcal.mol -1 , suggesting the absence of the Monomer species in these conditions. However, two distinctive signals can be observed in the 19 F NMR experimentally ( Figure S7). For this reason, we decided to reinvestigate the system, considering the possible presence of some water molecules, as this notion was noted during the preparation of the complexes (see Figure S2). Different conformers of the Monomer and Dimer were optimised with one, two or three coordinating water molecules. This selection is because the Zn(II) coordination sphere can immediately change from 4 to 5 to 6 due to the presence of water molecules and/or Lewis bases. This flexibility has been explicitly described in biological and catalysis-related examples. 90,91 When one water molecule coordinates with the Dimer, it produces an intermediate species Dimer(H 2 O), which is 7.8 kcal.mol -1 more stable; hence water coordination is favoured. For the monomer, the Monomer(H 2 O) is favoured by 13.8 kcal.mol -1 . However, the significant instability of the monomer compared to the Dimer, makes Monomer(H 2 O) less stable than Dimer(H 2 O) by 14.5 kcal.mol -1; hence this species could neither be experimentally observed. Calculations accounting for two coordinating water molecules result in Dimer(H 2 O) 2 and Monomer(H 2 O) 2 with relative energies of -16.0 and -8.6 kcal.mol -1, respectively respect to Dimer. The coordination of this second water molecule reduces the difference between monomeric and dimeric species. However, the dimer is still preferred by 7.4 kcal.mol -1 . The most stable isomer of Monomer(H 2 O) 2 presents a geometry with a water molecule coordinated to Zn and the other, forming a bridge between the oxygen of the ligand and the coordinated water. The geometry of this species suggests that a strong interaction with a third water molecule could be favoured. Calculations including a third water molecule yield Dimer(H 2 O) 3 and Monomer(H 2 O) 3 with relative energies of -24.2 and -21.9 kcal.mol -1 , respectively respect to Dimer; see Scheme 4.  Please note that as small errors in computational free energies will impact the equilibrium constants, we can only provide a qualitative explanation. The monomer-dimer equilibrium observed for initial water concentrations between 0.1 and 0.5 M could be at concentrations lower or higher. 93 The geometries of Dimer(H 2 O) 3 and Monomer(H 2 O) 3 are presented in Figure 7. The most stable dimeric species presents a water molecule strongly coordinated to one of the Zn and another weakly coordinated to . Adding more water molecules has not been considered as more strong interactions seem unlikely, and conformational complexity increases We also investigated the structures related to amines, considering the coordination of one and two analyte molecules, (R) 2-Phenylglycinol, to the monomer and the dimer as hypothesised (Scheme 2). We found that Dimer(analyte), Monomer(analyte) are located at -13.3 and -6.7 kcal.mol -1 and Dimer(analyte) 2 , and Monomer(analyte) 2 at -28.7 and -29.5 kcal.mol -1 , respectively; see Scheme 5 and Figure 7. In this case, there is no space for the coordination of a third analyte molecule. Formation of Monomer(analyte) 2 species is primarily favoured. The practical difficulties for sensing analytes at low concentrations seem to be due to the need for water-analyte exchange. The main chemical equations of the equilibria, in this case, are Dimer(H 2 O) 3 + 4 analyte ⇌ 2 Monomer(analyte) 2 + 3 H 2 O and Monomer(H 2 O) 3 + 2 analyte ⇌ 2 Monomer(analyte) 2 + 3 H 2 O. The energy difference between the aqua and analyte species is significant, 5.3 kcal.mol -1 and 3.8 kcal.mol -1 , respectively. In both cases, the expressions depend upon the concentration of water and analyte. The two equilibriums are related, as well as to the previously mentioned species. Hence, a system of 12 equations needs to be solved to find the concentration of all species at equilibrium (see SI).  Following experimental results, we explored the equilibria in DMSO. The above-presented minima were computationally reoptimised using DMSO as the solvent, and results are presented in Scheme 5. The different conformers for the dimeric and monomeric species with one and two DMSO molecules coordinated were also searched. We found that the coordination of a DMSO molecule to form Dimer(DMSO) is favoured by -5.7 kcal.mol -1 and the inclusion of a second DMSO molecule slightly stabilises the resulting Dimer(DMSO) 2 to -8.0 kcal.mol -1 in respect to the Dimer. The most stable conformer for Dimer(DMSO) 2 is presented in Figure  8, DMSO coordinates through the oxygen. The coordination of DMSO molecules also stabilises each monomer; the first DMSO stabilises the monomer in 7.6 kcal.mol -1 and the second 5.4 kcal.mol -1 more (see, Scheme 5 and Figure 8). Hence, two Monomer(DMSO) 2 molecules have similar energy as one Dimer(DMSO) 2 (two monomers are 0.7 kcal.mol -1 more stable). However, both are significantly less stable than the complexes with water or analyte molecules coordinated. Indeed, the most stable minimum is that of the monomer (analyte) 2 , however the concentration of DMSO is several orders of magnitude larger than that of the analyte. The chemical equations, corresponding equilibrium constants, and mass balances were written to predict the concentration of all species at equilibria. Concentrations were obtained by solving a system of 19 equations and 19 unknown variables considering initial concentrations 14.06 M DMSO, 0.50 M water, 1 10 -3 M Dimer, no concentration of the rest of species except for the analyte for which we considered 0, 1 10 -3 , 1 10 -2 and 0.1 M. At none or low concentration of analyte the most abundant species is Mono(DMSO) 2 . Although this complex is less stable than the complexes with analyte and water, the large concentration of DMSO concerning the rest of the coordinating molecules ( < 0.5 M) makes the equilibria shift towards its formation. It is also interesting to note that although the stability of Mono(DMSO) 2   Discussion. The synthesis of the targeted families is straightforward, from commercially available resources and in two high-yielding steps. The compounds, as anticipated, are isolated as dimers; however, postsynthesis, they are susceptible to absorbing moisture (H 2 O). In the solution phase, in the presence of coordinating solvents (DMSO or H 2 O) and at a millimolar level (1mM), NMR diffusion, solution and ESI studies propose a dimer-monomer equilibrium; this finding is in contrast to known Zn-salen complexes which exist as monomers under similar conditions. 52,71 Variable temperature studies (NMR) would be insightful to elucidate at which temperature the monomeric or dimeric species will be favoured. Still, given that we aimed to develop an operational simple detection method, we did not proceed further.

Dalton Transactions Accepted Manuscript
We anticipated that coordinating solvent molecules, one or two, would occupy the vacant positions in the Zn coordination sphere and that slight or significant excess of analyte would favour the formation of monomeric species C' or C'', respectively (Scheme 6), avoiding the presence other species (A' or B', Scheme 2). By doing so, a new single distinct peak corresponding to C' or C'' would appear in the NMR data, thus providing a new detection method (not two different peaks, one for the complex and one for the complex+analyte). However, the solvent-analyte exchange requires an extreme excess of the latter for a new distinct peak to appear in the 19 F NMR spectrum. The above evidence (Figures 5&6) suggests that this approach may be suitable for diamines but cannot be generalised. For example, the excess of phenylglicinol results in complex decomposition, formation of the Zn(analyte)x complex and release of the fluorinated ligand ( Figure 5, Zn-1SS 1:150  Then, theoretical calculations were performed to shed light on the experimental findings. These calculations scrutinised the stability of all possible tetra-, penta-and hexa-coordinated Zn monomeric and dimeric species in the presence of coordinating and non-coordinating solvents and analytes. Different equilibriums are possible (Schemes 4 & 5), while speciation depends on concentration. The outcome favours unconventional hexa-coordinating species, which is a peculiar finding unsupported by the provided experimental evidence. However, these calculations explain that a) the solvent-analyte exchange is energetically unfavourable and b) the concentration of the H 2 O molecules significantly impacts this process and prohibits the development of a new sensing process at low concentrations.

Conclusions.
For the first time, we examined the ability of fluorine-based Zn-salan complexes to function as amine sensors with 19 F NMR. Being aware that these species will isolate as dimers, we aimed to obtain species C' or C'' varying the analyte excess, but this process is unfavourable due to the presence of coordinating solvents. Future synthetic efforts will focus on altering the organic framework and having the antenna close to the metal centre to ease the sensing process in non-coordinating solvents.

Authors Contribution
GEK devised the project with critical input from NBE. NBE synthesised and characterised the ligands and complexes and performed the 19