Bis (2-pyridylmethyl)alkyl(thioalkyl)diamines as promising scaffolds for the construction of ﬂuorescent and redox chemosensors for transition and post-transition metal ions Inorganica Chimica Acta

( L 2 , L 4 ) and ferrocenylmethyl ( L 6 , L 7 ) pendant arms with the intention to study their potentiality as receptor units in molecular sensors. The optical response of L 1 – L 4 to the presence of the metal ions Cu 2+ , Zn 2+ , Cd 2+ , Hg 2+ and Pb 2+ has been investigated in MeCN/H 2 O (4:1 v/v) solution. The elec- trochemical response of L 6 and L 7 to the presence of the same metal ions has been investigated in anhydrous MeCN/CH 2 Cl 2 10:1 (v/v) solution. Results are compared and discussed with the aim to clarify the mutual role played by the bis (2-methylpyridyl)alkyl(thioalkyl)diamines and the signalling units attached to them in reaching the selectivity of the responses observed.


Introduction
The development of selective and sensitive analytical tools for rapid monitoring of transition and post-transition metal ions is of particular interest, since many of these metals are either pollutants in environmental systems or essential trace elements in biological systems. In particular, luminescence quenching or enhancement following the interaction between chemically engineered fluorescent molecules and the targeted metal ions is revealing an effective and simple solution to complex analytical problems .
From a structural point of view, according to the simple ''receptor-spacer-fluorophore'' supramolecular modular scheme, the most common class of fluorescent chemosensors (conjugated chemosensors) consists of a fluorogenic unit (signalling site) covalently linked, through an appropriate spacer, to a guest-binding site (receptor unit). The selective host-guest interaction of the target species with the receptor unit (recognition event) is converted into an enhancement or quenching of the fluorophore emission brought about by the perturbation of such photoinduced processes as energy transfer, charge transfer, electron transfer, or the formation or disappearance of excimers and exciplexes. However, the choice of the ''read-out'' or signalling unit can be critical to both the performance and the selectivity of the sensor, especially if a direct interaction of the fluorophore with the target species is possible (for example, the signalling unit possesses appropriate coordinating donor atoms for the metal ion species).
Important aspects of the design of fluorescent chemosensors for metal ions therefore include analyte affinity, choice of chromophore or fluorophore, binding selectivity and transduction signalling mechanism. All these aspects can in principle be targeted by changing the receptor, the signalling and the spacer units, taking advantage of the extensive knowledge available from classical coordination chemistry, as well as by choosing the medium (the solvent or the pH of the solution, for instance) in which the host-guest interaction takes place .
Similar considerations apply to any other type of chemosensor for metal ions based on the synthetic modular scheme outlined above, but making use of signalling units not necessarily involving an optic transduction mechanism of the host-guest interaction.
In this context, we have considered N,N 0 -bis(2-pyridylmethyl)propylendiamine (1) and N,N 0 -bis(2-pyridylmethyl)-1,5diamino-3-thiapentane (2) as metal ion binding platforms for the construction of the fluorescent chemosensors L 1 -L 4 (Scheme 1) via functionalization of the secondary amine groups with coordinating dansylamidoethyl (L 1 , L 3 ) or 2-quinolinylmethyl (L 2 , L 4 ) pendant arms. The compound L 5 featuring the 1,7-diaza-4-thiacyclononane ( [9]aneN 2 S, 3) was also synthesized for comparison purposes. Both tetradentate 1 and pentadendate 2 belong to the well known class of aminopyridine ligands whose coordination chemistry has been widely studied [22][23][24][25][26][27], but the use as receptor units in fluorescent chemosensors is limited to only few members of the family such as bis(2-picolyl)amine [28]. They present relatively soft donor centres at the pyridine rings (borderline base) with predisposition, therefore, to combine with late transition and posttransition heavy metals. In this respect, the optical response of L 1 -L 5 to the presence of Cu 2+ , Zn 2+ , Cd 2+ , Hg 2+ and Pb 2+ has been investigated in MeCN/H 2 O (4:1 v/v) solution. Furthermore, the potential use of 1 and 2 as binding units in redox chemosensors has also been considered by the synthesis of L 6 and L 7 featuring ferrocenylmethyl pendant arms at the secondary amino groups. The electrochemical response of L 6 and L 7 to the presence of the above-mentioned metal ions has been investigated in anhydrous MeCN/CH 2 Cl 2 10:1 (v/v) solution, and compared to the optical response of L 1 -L 4 in order to possibly analyze the role played by the receptor units represented by 1 and 2 in the responses observed for the two types of chemosensors, to the presence of the considered metal ions.

Synthesis of 4,7-bis(dansylamidoethyl)-1-thia-4,7diazacyclononane (L 5 )
A solution of dansylamidoethyl chloride (0.55 g, 1.76 mmol) in dry MeCN (150 mL) was added dropwise to a mixture of 1,7-diaza-4-thiacyclononane (3) (0.13 g, 0.88 mmol) and K 2 CO 3 (0.60 g, 4.4 mmol) in dry MeCN (40 mL). The mixture was maintained at room temperature for 48 h under nitrogen. The solid was filtered off, the solvent was removed under reduced pressure and the residue was dissolved in CH 2 Cl 2 and washed with water. The organic extracts were dried over Na 2 SO 4 , filtered and the solvent removed in vacuo to give a yellow oil which was purified by flash-chromatography on silica gel using a CH 2 Cl 2 /MeOH (9:1 v/v) mixture as eluant. A yellow solid was obtained (0.363 g, 59% yield

Synthesis of N,N
A solution of (ferrocenylmethyl)trimethylammonium iodide (4.72 g, 12.28 mmol) in anhydrous MeCN (100 mL) was added dropwise to a refluxing solution of 1 (1.17 g, 4.56 mmol) and K 2 CO 3 (4.44 g, 32.12 mmol) in anhydrous MeCN (40 mL). The resulting mixture was stirred overnight at 80°C. K 2 CO 3 was then filtered off and washed with hot MeCN (20 mL). The solvent was removed under reduced pressure and the residue was dissolved in CH 2 Cl 2 and washed with water. The organic extracts were dried over Na 2 SO 4 , filtered and the solvent removed in vacuo.

Synthesis of N,N
A solution of (ferrocenylmethyl)trimethylammonium iodide (1.51 g, 3.90 mmol) in anhydrous MeCN (100 mL) was added dropwise to a refluxing solution of 2 (0.50 g, 1.66 mmol) and K 2 CO 3 (1.83 g, 13.28 mmol) in anhydrous MeCN (40 mL). The resulting mixture was stirred overnight at 80°C. K 2 CO 3 was then filtered off and washed with hot MeCN (20 mL). The solvent was removed under reduced pressure and the residue was dissolved in CH 2 Cl 2 and washed with water. The organic extracts were dried over Na 2 SO 4 , filtered and the solvent removed in vacuo to yield a brown oil (0.37 g, 32.0% yield). Elem. Anal. Calc. for C 38

Cyclic voltammetry
Cyclic voltammetry experiments were recorded at a scan rate of 100 mV s À1 , using a conventional three-electrode cell, consisting of a combined working and counter platinum electrode and a standard Ag/AgCl (in KCl 3.5 mol dm À3 ; 0.2223 V at 25°C) reference electrode. The experiments were performed at 25°C in anhydrous MeCN/CH 2 Cl 2 10:1 (v/v) mixture. The solutions were about 1.0 Â 10 À3 mol dm À3 in the electroactive species (L 6 or L 7 ) with n-Bu 4 NBF 4 (0.1 mol dm À3 ) as supporting electrolyte. A stream of argon was passed through the solution prior to the scan. For each redox-responsive ionophore (L 6 or L 7 ), different solutions were prepared containing increasing amounts of the metal guest cation as hydrated perchlorate or tetrafluoroborate salt (molar ratio ranging from 0 to 1:1) and the cyclic voltammogram was recorded for each solution. Data were recorded on a computer-controlled Autolab PG STAT 100 potentiostat-galvanostat using model GPES electrochemical analysis software. Intensities were corrected for Lorentz-polarization effects and for absorption using multi-scan corrections (T min 0.470, T max 0.661) [35]. The structure was solved by direct methods using SIR92 [36], followed by difference Fourier synthesis and refined by fullmatrix least-squares on F 2 using SHELXL [35]. All non-H atoms were refined anisotropically and H atoms were introduced at calculated positions and thereafter incorporated into a riding model with U iso (H) = 1.2 U eq (C). At final convergence R 1 [I > 2r(I)] = 0.0362 and wR 2 [all data] = 0.0890 for 379 refined parameters; S = 0.95.

DFT calculations
Quantum-chemical calculations based on Density Functional Theory (DFT) [37] were performed on the 1:1 Zn 2+ complex of L 2 in the gas phase with the ligand adopting three possible conformations in the coordination to the metal centre. The GAUSSIAN09 (Rev. A.02) suite of programs was used [38], adopting the mPW1PW [39] hybrid functional. Schäfer, Horn and Ahlrichs double-f plus polarisation all-electron basis sets [40] were used for C, H, and N, whereas the LanL08 BS's with relativistic effective core potentials (RECP) were adopted for the heavier Zn, providing in addition dtype polarization functions [41]. The geometry optimizations were performed without introducing any structural simplification or symmetry restraint, and by adopting a tight SCF convergence criterion (SCF = tight keyword). All calculations were carried out on a 64 bit E4 workstation equipped with four quadcore AMD Opteron processors and 16 Gb of RAM, and running the OpenSuSE 10.3 Linux operating system.
A molecular modelling investigation of the 1:1 Zn 2+ complex of L 2 with an implicit simulation of the MeCN environment was also carried out. In this case a conformational search by means of a molecular dynamics procedure (T = 800 K, 200 saved conformations) was first carried out on each complex by means of an empirical force field method (OPLS2005) [42]. The lowest-energy conformers so obtained were then minimized at the DFT/B3LYP [43] level of theory using the LAV3P ⁄⁄ + basis set, which uses an effective core potential for metal atoms, as implemented in the software JAGUAR [44], and an implicit simulation of the MeCN environment by the Poisson-Boltzmann equation [45]. In particular, the band at around 338 nm shifts to lower wavelengths (330-332 nm) and increases slightly in intensity, while for the band at 254 nm only a slight increase is observed (see Fig. 1a for the spectrophotometric titration of L 1 with Cu 2+ ). No significant changes were observed in the UV-Vis spectra of the three ligands upon addition of Zn 2+ or Cd 2+ . For all three dansylamidopropyl derivatives, a quenching of the luminescence intensity was observed only upon addition of Cu 2+ [I rel = 12%, 4%, and 17% for L 1 , L 3 and L 5 , respectively] or Hg 2+ [I rel = 40%, 34%, and 27% for L 1 , L 3 and L 5 , respectively, I rel (%) represents the percentage of the residual fluorescence intensity in correspondence of a M 2+ /L molar ratio of 2], with the formation of 1:1 metal-to-ligand complexes as suggested by the inflection point in the fluorescence intensity/molar ratio plots (see Figs. 1b and 2 for the case of L 1 , and Supplementary material, Fig. S4, for the cases of L 3 and L 5 ). In no case was a shift of the emission band observed upon addition of the metal ions.

Results and discussion
The strong quenching effect observed with Cu 2+ can be ascribed to either a dansyl-to-metal energy-transfer (ET) or to a metal-todansyl electron-transfer (eT) mechanism [47]. In the case of Hg 2+ , a contribution to fluorescence quenching from the heavy-atom effect cannot be excluded, while the ET mechanism cannot be effective for this d 10 ion [47].
It is worth noting that most of the ligands reported in the literature containing the dansyl fluorophore manifest an ON-OFF type fluorescence response to Hg 2+ , Cu 2+ or both, as in the case of L 1 , L 3 and L 5 , with the complexation process being concomitant with the deprotonation of the sulfonamide group [47][48][49][50][51]. In some cases an enhancement of the fluorescence intensity in the presence of Zn 2+ is also observed [50,51]. These data, therefore, indicate that the dansyl group is an appropriate signalling unit to choose when approaching the design of selective fluorescent chemosensor for Hg 2+ or Cu 2+ . The selectivity in the optical response is, therefore, heavily determined by the cooperation of the receptor unit in the binding process. In our case, the lack of influence exerted by the sulfur donors in L 3 and L 5 on the selectivity of the proposed chemosensors is particularly surprising. In principle, the presence of soft donor atoms in the binding unit should favour interaction with Hg 2+ and consequently a preferential optical response toward this metal cation. Presumably the sulfur donor in both L 3 and L 5 is not much involved in metal coordination. The metal ions essentially experience an N 4 coordination mode from the aminopyridine frameworks of L 3 similar to that imposed by L 1 . Furthermore, on comparing the optical responses of L 1 , L 3 and L 5 towards Cu 2+ and Hg 2+ , it appears that the metal coordination of the pyridine units in the case of L 1 and L 3 significantly increases the quenching effect of Cu 2+ with respect to Hg 2+ . These results, together with data from the literature [52][53][54][55][56][57][58][59][60], clearly indicate the quinoline group is an appropriate signalling unit for the construction of selective fluorescent chemosensors for Zn 2+ over Cd 2+ ; these two metal ions have similar chemical behaviour due to their electronic closed-shell d 10 configuration and, therefore, they usually cause similar spectral changes after interactions with fluorescent chemosensors. Very recently, the idea has emerged that for Zn 2+ quinoline-based chemosensors, the CHEF effect for the small Zn 2+ as compared to the larger Cd 2+ ion might be determined by the ''steric crowding'' in the corresponding complexes: an elongation of the Zn-N bond due to ''steric crowding'' would partially restore the PET quenching mechanism and hence cause a lower CHEF effect for Zn 2+ relative to Cd 2+ [55]. Although this hypothesis appears to be confirmed on comparing the optical response of a limited range of quinoline-based chemosensors, extending the comparison to a larger family of molecules reveals that this ''steric crowding'', if real, is not the only factor to be considered and investigated to fully understand both the [log K ass = 7.20(2) and 5.40 (2) for [ZnL 2 ] 2+ and [CdL 2 ] 2+ , respectively, evaluated from the spectrofluorimetric data using HypSpec [46]. ability of quinoline-based fluorescent chemosensors to discriminate Zn 2+ and Cd 2+ and the relative intensity of emission in the presence of these two metals.
On going from a ''sterically crowded'' ligand such as TQEN (Table 1) to less sterically crowded ones such as TQA or 1 iso-TQEN for which no significant Zn-N quinoline bond length distortions are observed in the corresponding Zn 2+ complexes in the solid state (as compared to the mean length for the Zn-N quinoline bonds to quino-line and quinoline-like ligands found in the CSD [52], Table 1) the I rel (Zn 2+ )/I rel (Cd 2+ ) ratio increases by a factor of 2.8 or 4.5, respectively. This level of increase, while not negligible, is not outstanding either.
Also quite intriguing is the higher CHEF effect observed for Cd 2+ versus Zn 2+ with T(MQ)EN, where the 1:1 complex with Zn 2+ shows Zn-N quinoline bond length distortions comparable to those observed in the analogous complex with TQEN (Table 1, Fig. 5a).
We have performed DFT calculations and optimized in the gas phase the geometries of the three most likely pseudo-octahedral conformers of the 1:1 Zn 2+ complex with L 2 [C1 (quinolines occupy axial positions, with pyridines and tertiary nitrogen atoms occupying the equatorial plane), C2 (quinoline groups are equatorial), C3 (the two quinoline groups occupy one axial and one equatorial position)] (see Supplementary material, Tables S1-S3, for the orthogonal Cartesian coordinates of C1, C2 and C3). The stability of the three conformers follows the order C1 > C2 > C3, but with energy differences between them of less than 1.  (Table 1). However, according to the hypothesis of the ''steric crowding'' described above, only significantly shorter bond distances could explain the high CHEF effect observed for L 2 in the presence of Zn 2+ . Interestingly, on trying to simulate the presence of a MeCN environment around the complex cation, the optimized metric parameters for very similar conformers of the 1:1 Zn 2+ complex with L 2 resulting from a conformational search by means of a molecular dynamics procedure, appear to indicate a slight lengthening of the interactions between the quinoline moieties in L 2 and the Zn 2+ metal ion (bond distances are however comparable with that observed for [Zn(BQDMEN)Cl 2 ] in the solid state (Table 1)), being the conformation with the quinoline groups in cis position still more stable than the other two, but only of less than 0.6 kcal/mol (Fig. 6).
In general, we believe that when the signalling unit can coordinate the metal centre a ''synergic'' cooperation between the signalling and the receptor units must operate to determine the sensitivity and selectivity in the optical response by a fluorescent chemosensor, even though the receptor unit or the chemosensor as a whole is not thermodynamically selective in the host-guest interaction.
What appears to be crucial in determining the emissive behaviour of a fluorescent chemosensor is the manifold of electronic levels associated to the given couple of receptor and signalling units (in this a ''synergic'' cooperation between the two units should be recognized). This manifold can be selectively perturbed in solution by a metal centre despite a small degree of thermodynamic selectivity in solution, and particular structural features in the solid state. The medium in which the host-guest interaction takes place is also important.
A clear example of this ''synergy'' between the signalling and receptor units in reaching a selective optical response towards a metal cation comes just from the family of quinoline-based chemosensors for Zn 2+ under consideration.
For QDTAPy (Scheme 2) a I rel (Zn 2+ )/I rel (Cd 2+ ) ratio of about 50 is reached with a remarkable sensitivity in the optical response (Fig. 5b) [53], despite the Zn-N quinoline bond distance in the complex [Zn(QDTAPy)H 2 O] 2+ (Table 1) being comparable to those in zinc(II) complexes with TQA, 1 iso-TQEN and the other less sterically crowded quinoline-based ligands in Scheme 2. This means that in QDTAPy the receptor unit and the signalling unit cooperate  ''synergically'' to give a highly selective and sensitive optical response to the presence of zinc(II) despite the fact that this ligand also binds to cadmium(II) (because of its sensing properties, QDT-APy has been recently used for the development of a novel flow injection analysis (FIA) system for the determination of traces amounts of Zn 2+ [61]). Interestingly, although the observed I rel (Zn 2+ )/I rel (Cd 2+ ) for L 2 is much lower than that observed for QDT-APy, the sensitivity of the optical response to Zn 2+ [I rel (Zn 2+ )] is much higher than that observed for all the other quinoline-based ligands considered, and second only to that recorded for QDTAPy, suggesting that L 2 also exhibits ''synergic cooperation'' between the signalling and the receptor units in sensing Zn 2+ efficiently. ). For both receptors, a one-wave electrochemical behaviour was in general also observed upon addition of 1 equiv. of the metal ions considered, with DE values very close to those observed in the presence of HClO 4 , suggesting that in the presence of metal ions the electrochemical shift is still due to the protonation of the ligands rather than to a metal coordination process. Only in the case of L 6 with Cu 2+ was two-wave electrochemical behaviour observed, indicating a stronger interaction of this ligand with this metal centre: upon addition of increasing amounts of Cu 2+ , the redox wave of the free L 6 was gradually replaced by a new reversible wave at more positive potential (E ½ = 594 mV), the anodically shifted electrochemical wave corresponding to the Fc/ Fc + redox couple of the [Cu II L 6 ] 2+ species. The current for the new redox couple [Cu II L 6 ] 2+ /[Cu II L 6 ] 3+ increased linearly until one equivalent of the guest metal cation had been added, at which point the reversible oxidation wave corresponding to the uncomplexed ligand disappeared.
The presence of two distinguishable waves in the cyclic voltammograms of L 6 upon addition of Cu 2+ can be accounted for in terms of an sufficiently higher stability constant for the complex [Cu II L 6 ] 2+ with respect to stability constants of the complexes Table 1 Comparison of the M-N quinoline bond lengths (M = Zn 2+ , Cd 2+ ) in the corresponding structurally characterized complexes of the ligands reported in Scheme 2 a .

Ligands
Zn-N quinoline Mean Zn-N quinoline Cd-N quinoline Mean Cd-N quinoline [M II L] 2+ (L = L 6 , M = Zn 2+ , Cd 2+ , Hg 2+ or Pb 2+ ; L = L 7 , M = Cu 2+ , Zn 2+ , Cd 2+ , Hg 2+ or Pb 2+ ) and a sufficiently large difference between the half-wave potentials for the two observed redox couples [62]. According to thermodynamic considerations, the following equation can be derived to describe two-wave behaviour of a redox chemosensor: E ½ complex À E ½ free ionophore = (RT/nF)ln(K neutral /K ox ) (K neutral is the metal binding constant for the neutral unoxidized redox-responsive ionophore and K ox is the metal binding constant for its oxidized form). K neutral /K ox and E ½ complex À E ½ free ionophore represent quantitative measures of the perturbation of the redox centre induced by the guest complexation to the receptor unit; both quantities express the ability of the redox chemosensor to selectively respond to the host binding of the guest species [62][63][64].
In the case of L 6 with Cu 2+ the value obtained for K neutral /K ox indicates that in MeCN L 6 binds the Cu 2+ ion ca. 10 3 times more strongly than it does its oxidized form (L 6 ) 2+ . Crystals corresponding to the formulation [CuL 8 (ClO 4 ) 2 ] (see Scheme 1) were unexpectedly obtained from the reaction of L 6 with Cu(ClO 4 ) 2 Á2H 2 O in MeCN following partial removal of the solvent from the reaction mixture and slow diffusion of Et 2 O vapour into the remaining volume. This complex is formed via cleavage of one of the two ferrocenylmethyl pendant arms from the starting ligand L 6 to give L 8 as complexing agent. In the complex [CuL 8 (ClO 4 ) 2 ] (see Fig. 7) the metal centre occupies a distorted octahedral geometry with the equatorial coordination sites occupied by the four N-donor atoms from L 8 . The apical positions are occupied by oxygen atoms from two perchlorate counteranions. Due to steric repulsion the two pyridyl rings do not lie on the same plane; as indexes of this distortion, N(3) is displaced 0.98 Å out of the mean plane defined by N(1), N(2) and N(4), and the dihedral angle between the planes defined by the two pyridyl rings is 40.4°. We cannot provide an unambiguous mechanism of this unexpected cleavage of one ferrocenylmethyl pendant arm of L 6 upon complexation to Cu 2+ , but a similar decomposition has been observed for the ligand ferrocenylmethyl-bis(2-pyridylethyl)amine in the presence of ZnBr 2 [65]: here the authors propose a metal-promoted nucleophilic attack at the benzylic FcCH 2 moiety by water, to afford complexed bis(2-pyridylethyl)amine and ferrocenylmeth-anol. However, this mechanism is in doubt because neither in the case of ferrocenylmethyl-bis(2-pyridylethyl)amine in the presence of ZnBr 2 [65], nor in the case of L 6 in the presence of Cu 2+ , could ferrocenylmethanol be detected in the reaction mixture.
Cyclic voltammetry of the complex [CuL 8 (ClO 4 ) 2 ] in MeCN reveals only a Coulometric-tested one-electron reversible oxidation wave at E ½ 590 mV versus Ag/AgCl corresponding to the redox wave assigned to the couple [Cu II L 6 ] 2+ /[Cu II L 6 ] 3+ . These findings further support our conclusion that the two ferrocene groups in L 6 and L 7 are electrochemically independent of one another and are oxidized in a single step.

Conclusions
Following previous studies in the field of fluorescent chemosensor design, it is obvious that the developments of efficient fluorescent chemosensors having both binding and sensing selectivity for probing a targeted metal cation in a real matrix, is a challenging task; the kind of selectivity to privilege is the optical response selectivity that can be achieved much easier than the thermodynamic (binding) selectivity [66]. On the other hand, for applications in real matrices that do not contain species which interfere with the binding process of the targeted analyte, the chemosensor does not require strict binding selectivity; alternatively, should this selectivity be required, it could be acquired with the appropriate choice of the medium for the host-guest interaction. On the basis of these general considerations, we advocate that the quest for efficient fluorescent chemosensors for analytical applications follows the ''complementary receptor-spacer-fluorophore'' supramolecular synthetic approach: given a pre-defined receptor unit (not necessarily the best in the binding process) different signalling units are in turn linked to it in order to establish the best combination showing at least selectivity and sensitivity in the optical response towards a given analyte [66].
The results reported in this paper clearly show that multidentate aminopyridine ligands such as 1 and 2 can be successfully deployed as receptor units in both fluorescent and redox Although the responses to the presence of the metal ions considered [Cu 2+ , Zn 2+ , Cd 2+ , Hg 2+ , Pb 2+ ] by the chemosensors presented here appear to be determined more by the intrinsic properties of the signalling units than by the selective binding properties of the receptors, synergism between the receptor and signalling units in determining selectivity and sensitivity in the optical responses appears to be a real possibility.