Optically pure heterobimetallic helicates from self-assembly and click strategies †

Single diastereomer, diamagnetic, octahedral Fe(II) tris chelate complexes are synthesised that contain three pendant pyridine proligands pre-organised for coordination to a second metal. They bind Cu(I) and Ag(I) with coordination geometry depending on the identity of the metal and the detail of the ligand structure, but for example homohelical (ΔFe,ΔCu) configured systems with unusual trigonal planar Cu cations are formed exclusively in solution as shown by VT-NMR and supported by DFT calculations. Similar heterobimetallic tris(triazole) complexes are synthesised via clean CuAAC reactions at a tris(alkynyl) complex, although here the configurations of the two metals differ (ΔFe,ΛCu), leading to the first optically pure heterohelicates. A second series of Fe complexes perform less well in either strategy as a result of lack of preorganisation.


Introduction
Metallo-helicate chemistry is almost exclusively confined to D n -symmetric systems (n = 2-4) self-assembled from symmetric ditopic ligands as shown in Fig. 1(a).Heterobimetallic helicates, although relatively rare, provide an opportunity to synthesise C n or lower symmetry systems in which two different bidentate coordinating groups are connected via a linker, and using two metals which have a preference for one end of the ligand over the other [1][2][3][4][5][6] [Fig.1(b)].Alternatively, in a multi-step synthesis of heterobimetallics, one (inert) metal is coordinated before a second coordination site and metal are introduced. 7ith fully inert systems, enantiomers can sometimes be resolved using e.g.chromatography to isolate (highly) optically enriched helicates 2,7 but to our knowledge, the self-assembly of optically pure (or at least highly enantiomerically enriched) heterobimetallic complexes has not been reported.This could provide strategies for the synthesis of large complexes in which precise placement of functionality and charge can be achieved i.e. towards systems which mimic natural self-assembling systems such as protein α-helices.
We have recently shown that a range of single diastereomer diimine complexes of Fe(II) can be made very readily using 2-pyridinecarboxaldehyde and simple phenylethanamines as the source of chirality (Fig. 2). 8,9The diastereomeric ratios were determined by 1 H NMR spectroscopy to be >200 : 1 in all cases and this unique system has since been exploited by ourselves 10 and others 11,12 to generate novel architectures.Since these monometallic Fe(II) units are relatively inert, the opportunity arises to carry out further reactions as one might with say a ruthenium tris(chelate). 7n this report we describe the coordination of a second metal to either a preformed second coordination site or one created in situ by copper(I)-catalysed Huisgen 1,3-dipolar cycloaddition (CuAAC).

Results and discussion
In principle, suitable trifunctional metallo-ligands based on the complexes of Fig. 2 might be formed by functionalising at the phenylglycinol-derived site R or at the pyridine.We address these two approaches with ligands L 1-6 and L 7-10 respectively (Fig. 3).
An X-ray molecular structure of fac,Δ Fe ,R C -[FeL 1 3 ](ClO 4 ) 2 was determined (Fig. 4).The Fe tris(diamine) unit is in the expected fac arrangement with three sets of π-stacks between phenyl and pyridine units. 8,9The pendant pyridine substituents are arranged in a 3-fold symmetric array but with opposite helicity to the Fe unit.This is as we expected since the stereogenic centres at C(7) act as "corners" effectively switching the sense of twist.A similar phenomenon was displayed in a bimetallic structure based on this tris(diimine) unit. 10It proved difficult with the X-ray data available to discriminate between the structure shown and an alternative with pyridine N atoms at the inward facing positions.Nevertheless we were encouraged that the system appeared to be preorganised to bind a second metal.
While a range of reactions of fac,Δ Fe ,R C -[FeL  3 ](ClO 4 ) 2 with Fe(II), Co(II/III), and Zn(II) sources failed to give stable bimetallic species, the addition of CuI or AgClO 4 to acetonitrile solutions resulted in the formation of Fig. 2 The self-assembling diastereomerically pure tris-chelate system used in this work. 8,9g. 3 Ligands L 1 -L 10 assembled in situ via condensation or CuAAC reactions.3 Cu](ClO 4 ) 2 I.
Fig. 6 VT 1 H NMR spectra of [FeL 2 1 H NMR spectra that the peaks corresponding to the phenyl rings (Fig. 6, peaks marked #) broaden as the temperature decreases with the extent of broadening ortho > meta > para.This is consistent with restricted CH-Ph bond rotations which cause averaging of the magnetic environments at room temperature. 13he 1 H NMR spectrum of Δ Fe ,R C -[FeL 2 3 Ag(CH 3 CN)](ClO 4 ) 3 is consistent with coordination of Ag(I); no broadening of signals associated with the second binding site was observed at accessible temperatures and only one diastereomer was observed.
Density functional theory (DFT) calculations using ADF2009 (version 2009.01) 14were used to investigate the structures of the Fe-Cu bimetallic complexes with L 1 and L 2 .In addition to the helicity at Fe which is essentially fixed, the pitch sense of the propeller arrangement of the pendant pyridine ligands about the Cu(I) centre needs also to be taken into account, and it was found that both homohelical Δ Fe Δ Cu and heterohelical Δ Fe Λ Cu isomers for both the 2-and 3-pyridinyl complexes could be optimised (Table 1).In both cases the input geometries around Cu(I) were based on the related X-ray structure of [Cu(3-methylpyridine) 3 I] which has a distorted tetrahedral geometry. 15or the 2-pyridinyl (L 1 ) complex, as the optimisations progressed the Cu-I bond was broken in both enantiomers, resulting in approximately trigonal planar geometries.As a result, the iodide anion was deleted from these calculations to allow convergence.The calculations reveal that the homohelical Δ Fe Δ Cu isomer is lower in energy by ca. 4 kcal mol −1 compared to the heterohelical Δ Fe Λ Cu isomer.Fig. 7 shows that the Δ Fe Δ Cu isomer in (a) is more longitudinally extended than Δ Fe Λ Cu (b) as further indicated by the Fe-Cu distances (ca.6.15 Å vs. 6.44 Å) and this is accompanied by; (i) reduction of unfavourable steric interactions between the pyridinyl ethers and the phenyl groups on the neighbouring ligands; and (ii) an increase in the magnitude of the propeller pitch ‡ of the (Py) 3 Cu units from ca. 36°to 49°.Whatever the origins of the energy differencethere are undoubtedly a number of competing factorsthe calculations suggest the Δ Fe Δ Cu isomer will be formed exclusively, at least at the level that can be determined by NMR spectroscopy, and indeed this is consistent with observations.
We were interested to investigate further the unusual 16 preference for trigonal planar Cu(I) geometry, specifically if this is a result of inherent chelate constraints in the ligand system [FeL 1 3 ] 2+ or from steric effects arising from 2-substitution of the pyridine.For these purposes we studied the model monometallic systems [Cu(2-methylpyridine) 3 I] and [Cu(3-methylpyridine) 3 I].For both isomers, structures were optimised with a range of fixed Cu⋯I distances.The results in Fig. 8 show that for the 2-methylpyridine complex (filled squares) the energy decreases as the iodide ligand dissociates and the Cu(I) approaches the trigonal planar arrangement (i.e. as the out of plane angle θ → 0°).In contrast, the 3-methylpyridine system  (open squares) increases in energy as the iodide dissociates, albeit more gradually, and indeed a tetrahedral structure is observed for [Cu(3-methylpyridine) 3 I] by X-ray crystallography. 15It would appear then that the preference for trigonal planar geometry about Cu(I) in the L 1 system is expected from a steric effect of the 2-substitution alone, notwithstanding any influence of conformational strain.Consistent with the observed and calculated structure of [Cu(3-methylpyridine) 3 I], DFT calculations on [FeL 2 3 CuI] 2+ isomers subsequently predicted tetrahedral geometries around Cu(I) in both Δ Fe Δ Cu and Δ Fe Λ Cu isomers, with the homohelical isomer being the more stable (ca. 3 kcal mol −1 ).
The electronic spectra of these L 1 and L 2 systems have some unusual features.UV-Vis absorbance spectra shown in Fig. 9 show familiar strong bands in the region 220-320 nm, corresponding to π-π* transitions within the ligands, three further peaks corresponding to MLCT bands are observed at 380, 520 and 540 nm respectively.In both cases, Cu(I) coordination results in an increase in intensity of π-π* region of the spectra, a decrease in the intensity of the MLCT bands between 460-650 nm and an increase at 360 nm.These Fe(II) based CT transitions are likely to be strongly affected by structural changes at the metal, but no such perturbation is evident from crystallography or calculations.We suggest that the presence of the nearby cationic charge of the Cu(I) is responsible for this change in relative intensity.For the π-π* region the change may be a result of ordering by coordination of the heterocycles to Cu(I).In the corresponding CD spectra measured on the same cuvette (Fig. 10) the picture is of a general reduction in intensity for reasons unknown, but evidently from the UV spectra this is not an error from concentration measurement.

Dalton Transactions Paper
one-pot synthesis with the propargylic ether derivative of (R)-2phenylglycinol.This complex is of particular interest due to its potential to undergo copper(I)-catalysed Huisgen 1,3-dipolar cycloaddition (CuAAC) 'click' reactions between the three alkyne units and organic azides.Accordingly it was treated with benzyl azide in the presence of dry triethylamine and a catalytic amount of copper(I) iodide (0.3 eq. per complex) in dry acetonitrile to give a purple solid.The 1 H NMR spectrum at ambient temperature contained only one set of ligand resonances, however at low temperature (233 K), the 1 H NMR spectrum revealed that there were in fact two species present in the recrystallised sample in the ratio 1 : 1. Slow vapour diffusion of ethyl acetate into an acetonitrile solution of Δ Fe ,R C -[FeL 4 3 Cu](ClO 4 ) 2 I in air resulted in the formation of single crystals that were suitable for X-ray diffraction.The solid state structure is shown in Fig. 12 along with key bond lengths and angles.The asymmetric unit contains one complex and a mixture of perchlorate and triiodide (resulting from oxidation of iodide in air) as the counterions.The Fe(II) and Cu(I) cations lie on a threefold axis.The complex has an approximately octahedral arrangement around Fe(II) and a trigonal planar arrangement around Cu(I), with a distance between the metal centres of ca.6.14 Å.Most interestingly, the sense of the helicity at the Cu(I) centre is opposite to that at the Fe(II) centre (i.e.Δ Fe ,Λ Cu ).Nevertheless this is not a mesocate since the metals are different and thus the system is chiral.Hence, we would describe this as the first example of an optically pure heterohelicate. 33 addition to the usual 8,9 three sets of inter-ligand paralleloffset π-stacking interactions in the Fe unit there are three sets of CH-π interactions 34,35 present between the triazole benzyl  substituent and the CHCH 2 O group on a neighbouring ligand.The Cu(I) is coordinated to the triazole rings via the 3-position i.e. the N(3) nitrogen atom, as expected on the basis of its higher basicity and perhaps also here a more favourable chelate conformation.There is nevertheless also some precedent for the coordination of metals to the 2-position [i.e.N(4)] in 1-substituted-1,2,3-triazole rings. 36he geometries of the homohelical Δ Fe Δ Cu and heterohelical Δ Fe Λ Cu isomers of Δ Fe ,R C -[FeL 4 3 Cu] 3+ were optimised and their energies calculated (Table 2) from density functional calculations as above.While the benzyl substituents on the triazole rings point down towards Fe(II) in the X-ray structure, various rotamers were also considered and optimised.The observed heterohelical isomer was found to be the lower in energy than the homohelical diastereomer by at least 9.27 kcal mol −1 such that we expect this structure exist exclusively in solution at equilibrium.Low temperature NMR studies are in agreement and on the basis of these calculations and the X-ray structure we can assign the solution species as heterohelical Δ Fe ,Λ Cu ,R C -[FeL 4 3 Cu] 3+ .The origins of this stereoselection are not obvious, and indeed the sums of angles around Cu in the structures are all very close to 360°indicating a lack of strain here.While as for Δ Fe ,Δ Cu ,R C -[FeL 1 3 Cu] 3+ there are likely to be a number of contributing factors, examination of space-filling models reveals an absence here of steric compression between the triazole and the ether units in the neighbouring ligands, thus allowing the system to adopt the more favourable conformations in the linker chain of the heterohelical system.
Further examples of heterohelical bimetallic complexes with L 5 and L 6 (Fig. 3) were synthesised from click reactions between fac,Δ Fe ,R C -[FeL 3 3 ](ClO 4 ) 2 and the appropriate organic azides (3,5-dimethylbenzyl azide and 4-nitrobenzyl azide) in the presence of 1 equivalent of CuI per complex.In all cases, low temperature 1 H NMR spectra show only a single set of peaks consistent with a single diastereomer in solution.
The ability to remove Cu(I) from these bimetallic complexes would be attractive so a number of methods were attempted to remove Cu(I) post-'click' reaction.The use of a copper scavenger, e.g.Smopex-111, Smopex 112, CupriSorb™, EDTA or sodium sulfide, either removed no copper or caused complete decomposition of the complex by removing both Cu(I) and Fe(II).As an alternative was also attempted the ruthenium(II) catalysed 'click' reactions [37][38][39] but no reaction occurred.Conversely, the use of copper wire and PMDETA 40,41 as the catalyst caused complete decomposition.These results led us to consider the scope for functionalisation from the pyridine side of the monometallic complex.
Pyridinyl ethers L [7][8] In order to compare the coordination behaviour above with that of pendant units attached to the pyridine ligands in the tris(chelate)Fe(II), the ligands L 7-8 (Fig. 3) were designed.Subsequently, the use of 5-(2-pyridinyloxy)picolinaldehyde in the standard one-pot synthesis with (R)-α-methylbenzylamine and Fe(ClO 4 ) 2 •6H 2 O led to the formation of diastereomerically pure fac,Λ Fe ,R C -[FeL 7 3 ](ClO 4 ) 2 (d.r.> 200 : 1).The X-ray molecular structure (Fig. 13) shows in contrast to that in Fig. 4 that the three pyridinyl ether groups are not preorganised for binding a second metal, but is otherwise conventional.
Addition of 1 equivalent of CuI to an acetonitrile solution containing fac,Λ Fe ,R C -[FeL 7 3 ](ClO 4 ) 2 •CH 3 CN led to no significant changes in chemical shifts, suggesting that coordination of Cu(I) to the second binding site is weak in solution.The same observation was made on addition of AgClO 4 .In further support of this, crystallisation of fac,Λ Fe ,R C -[FeL 7 3 ](ClO 4 ) 2 in the presence of AgClO 4 resulted in a partially refined X-ray molecular structure showing an extended lattice structure with each tris(chelate)Fe(II) unit coordinated to three different silver(I) units; the formation of this structure is presumably driven by lattice enthalpy.a Type A is as shown in Fig. 10 whereas in type B a rotation of ca.180°a bout the N-C bond in the benzyl groups was made before optimisation i.e. benzyl groups point "up".In contrast to the L 7 case above, reaction of three equivalents of both 5-(3-pyridinyloxy)picolinaldehyde and (R)-α-methylbenzylamine with Fe(ClO 4 ) 2 •6H 2 O gave a sample with broad NMR resonances.In solution the product slowly converted from purple to a paramagnetic red system on standing.This is likely to be due to the formation of a bis complex [FeL 8 2 ] 2+ as a result of the chelate effect; this 3-pyridinyl ligand is appropriately structured to coordinate in a tridentate fashion, as we have observed previously for a tert-leucinol derivative. 8ick reactions of L 9 alkynyl complex; complexes of L 10 The diastereomerically pure compound fac,Λ Fe ,R C -[FeL 9 3 ]-(ClO 4 ) 2 (for L 9 see Fig. 3) was synthesised from the propargylic ether derivative of 5-hydroxy-2-picolinaldehyde.Copper(I)-catalysed Huisgen 1,3-dipolar cycloaddition (CuAAC) 'click' reactions between this complex and benzyl azide did not behave in the same way as for fac,Δ Fe ,R C -[FeL 3 3 ](ClO 4 ) 2 and the reaction with sub-stoichiometric (catalytic) amounts of CuI led to mixtures of starting material [FeL 9 3 ](ClO 4 ) 2 with [FeL 9 2 L 10 ](ClO 4 ) 2 , [FeL 9 L 10 2 ](ClO 4 ) 2 and [FeL 10 3 ](ClO 4 ) 2 (Fig. 3) according to 1 H NMR spectroscopy (see ESI †) and mass spectrometry.With one equivalent of CuI per Fe centre (i.e. the conditions that cleanly gave Δ Fe ,R C -[FeL 4 3 Cu](ClO 4 ) 2 I) the majority product was [FeL 10 3 ](ClO 4 ) 2 but the without Cu + coordination.Addition of three equivalents of CuI per Fe unit or more, essentially complete click conversion was apparent, and the chemical shift difference for the diastereotopic O-CH 2 -triazole group had widened to ca. 0.25 ppm indicating Cu + coordination i.e. [FeL 10 3 Cu] 3+ .Attempts to recrystallize this species for analysis led to slow decomposition and formation of red paramagnetic products.Presumably it suffers the same fate as the L 8 system above, and for the same reasons.

Conclusion
Optically pure (highly enantiomerically enriched) heterobimetallic Fe-Cu and Fe-Ag helicates can be synthesised via the pre-formed monometallic Fe(II) complexes of L 1 and L 2 .In both cases only a single isomer was observed by 1 H NMR spectroscopy and corresponding DFT calculations predict the ΔΔ homochiral isomer will form exclusively.Interestingly, formation of the trigonal planar geometry at Cu(I) is driven by the substitution pattern of the pyridine ligand, independently of any conformational effects in the bimetallic structure.
CuAAC 'click' reactions are most efficient on the monometallic complex of L 3 and give Fe-Cu helicates directly.Again, a single isomer is formed but in this case both X-ray crystallography and DFT calculations support the exclusive formation of the Δ Fe Λ Cu heterochiral isomers.These compounds are the first examples of what might be called heterohelicates since they have opposite configurations at the metal but are not mesocates.Similar reactions on the Fe(II) complex of L 9 are much less efficient presumably because the system, as demonstrated in the analogous pyridine series, is not preorganised for coordination of the Cu(I) ion; binding of the Cu(I) catalyst to the three triazoles appears to promote the reaction.
We are thus now in a position to confidently predict stereochemical properties in such bimetallic systems and will now move towards the synthesis of heterobimetallics designed for aqueous solubility. 33

Experimental
General considerations (R)-2-Phenylglycinol and 5-(hydroxy)picolinaldehyde were synthesised as previously described. 9,42,43Deuterated solvents were purchased from Sigma-Aldrich and Cambridge Isotope Laboratories.Sodium hydride dispersions in mineral oil were placed in a Schlenk vessel under an inert atmosphere and washed three times with diethyl ether to remove the oil.The sodium hydride powder was then dried and stored in the dry box.Suitable precautions should be taken in the synthesis of organic azides. 44here appropriate, reactions were carried out under argon using a dual manifold argon/vacuum line and standard Schlenk techniques or MBraun dry box.THF was pre-dried over sodium wire and then heated to reflux for 3 d under dinitrogen over potassium and degassed before use.Dried THF was stored in a glass ampoule under argon.All glassware and cannulae were stored in an oven at >375 K.
NMR spectra were recorded on Bruker Spectrospin 300/400/ 500 MHz spectrometers.Routine NMR assignments were confirmed by 1 H-1 H (COSY) and 13 C-1 H (HMQC) correlation experiments where necessary.The spectra were internally referenced using the residual protio solvent (CDCl 3 , CD 3 CN etc.) resonance relative to tetramethylsilane (δ = 0 ppm).ESI mass spectra were recorded on Bruker Esquire 2000 and Bruker MicroTOF spectrometers.Infra-Red spectra were measured using a Perkin-Elmer FTIR spectrometer.Elemental analyses were performed by Warwick Analytical Services, Coventry, UK and MEDAC Ltd, Surrey, UK.
Optical rotation measurements were performed on a Perkin Elmer Polarimeter 341 by Warwick Analytical Services, Coventry, UK.In all cases the following parameters were used: solvent methanol, temperature 20 °C, path length 100 mm, wavelength 589 nm.
Density functional optimisations were carried out using the Amsterdam Density Functional program (version 2009.01). 14tarting points for geometry optimisations were taken from crystallographic data where available, and where unavailable, starting structures were created from existing crystallographic fragments.Solution structures were optimised relative to acetonitrile (vide infra) using a triple-ζ plus polarisation basis set (TZP) on all atoms with the OPBE functional and Grimme's empirical correction for dispersion. 47Small frozen cores 48 were used throughout.Calculations used integration level 5 (as defined by ADF) with convergence criteria of e = 0.0001 a.u., rad = 0.005 Å and grad = 0.001 a.u.Å −1 for the total binding energy, Cartesian displacement and energy gradient respectively.Acetonitrile solvent effects were included based on the conductor-like screening model (COSMO) implemented in ADF. 49Non-bonded radii used were H = 1.350Å, C = 1.700Å, N = 1.608Å, Fe = 1.858Å.A dielectric constant of 37.5 (acetonitrile) and an outer cavity radius of 2.76 Å were further used to parameterise the COSMO solvation cavity.

Dalton Transactions Paper
of sodium hydride (0.17 g, 7.3 mmol, 2.0 eq.) in dry THF (10 ml).The solution was stirred for 1 h at ambient temperature under partial vacuum.Propargyl bromide (80 wt% in toluene, 0.43 ml, 3.8 mmol, 1.05 eq.) was added dropwise and the solution was stirred for 1 h at ambient temperature under argon.At this point, the solution was heated to reflux (65 °C) under partial vacuum overnight before cooling to ambient temperature, followed by the addition of brine (30 ml).The product was extracted into diethyl ether (4 × 50 ml), dried over sodium sulfate and the solvent was removed under reduced pressure to leave a yellow oil (crude yield = 0.52 g).This crude product was purified by Kügelrohr distillation at rotary pump vacuum to give the product as a clear liquid (95 °C).Purified yield = 0.45 g, 2.6 mmol, 71%.Optical rotation −32.11°(6.082g per 100 ml).

Fig. 9
Fig. 9 UV/vis spectra of Fe and Fe/Cu complexes of L 1 3 and L 2 .Path length 1.0 cm and concentration 3.3 mM.

Fig. 10
Fig. 10 CD spectra of Fe and Fe/Cu complexes of L 1 3 and L 2 .Path length 1.0 cm and concentration 3.3 mM.

7 [
Fig.11(a)].The reaction was repeated using a stoichiometric amount of copper(I) iodide (1.0 equivalent per Fe).The resulting low temperature 1 H NMR spectrum showed only one set of peaks [Fig.11(b)] and microanalysis indicated that copper(I) iodide was present in a 1 : 1 ratio with the complex.Hence, the spectrum shown in Fig.11(a) arises from an equilibrium between two complexes fac,Δ Fe ,R C -[FeL4 3 ] 2+ and Δ Fe ,R C -[FeL 43 Cu] 3+ as shown in Scheme 1.