Design of a double-decker coordination cage revisited to make new cages and exemplify ligand isomerism

The complexation study of cis-protected and bare palladium(II) components with a new tridentate ligand, i.e., pyridine-3,5-diylbis(methylene) dinicotinate (L1) is the focus of this work. Complexation of cis-Pd(tmeda)(NO3)2 with L1 at a 1:1 or 3:2 ratio produced [Pd(tmeda)(L1)](NO3)2 (1a). The reaction mixture obtained at 3:2 ratio upon prolonged heating, produced a small amount of [Pd3(tmeda)3(L1)2](NO3)6 (2a). Complexation of Pd(NO3)2 with L1 at a 1:2 or 3:4 ratios afforded [Pd(L1)2](NO3)2 (3a) and [(NO3)2@Pd3(L1)4](NO3)4 (4a), respectively. The encapsulated NO3– ions of 4a undergo anion exchange with halides (F–, Cl– and Br– but not with I–) to form [(X)2@Pd3(L1)4](NO3)4 5a–7a. The coordination behaviour of ligand L1 and some dynamic properties of these complexes are compared with a set of known complexes prepared using the regioisomeric ligand bis(pyridin-3-ylmethyl)pyridine-3,5-dicarboxylate (L2). Importantly, a ligand isomerism phenomenon is claimed by considering complexes prepared from L1 and L2.


Introduction
Coordination-driven self-assembly is a convenient strategy for the construction of supramolecules of desired dimensions via simple synthetic procedures. Well-defined metal-ligand coordination bonds enable the construction of designer targeted molecules with ease. The use of a palladium(II) component for complexation with a non-chelating bi-or polydentate ligand (usually N-donor ligands) is particularly advantageous for the construction of a variety of metallocages [1][2][3][4][5]. Complexation of cis-pro-tected palladium(II), i.e., (PdL') or bare palladium(II) with nonchelating bidentate ligands is known to afford a series of (PdL') m L m or Pd m L 2m -type self-assembled coordination complexes [5]. Pd 2 L 4 -type cages are the simplest representatives among the Pd m L 2m -type complexes, yet most utilised [5,6]. The Pd 2 L 4 -type cages are well explored for the encapsulation of guests that are anionic [7][8][9][10][11], neutral [12][13][14][15][16], radical initiators [17], and drug molecules [18,19]. It is necessary to emphasize here that Pd 2 L 4 -type cages contain a cavity. McMorran and Steel reported the first Pd 2 L 4 -type cage and the anion binding ability of the cavity [20]. They also used a non-chelating tridentate ligand in an attempt to prepare a Pd 3 L 4 -type double-decker coordination cage that would possess two cavities, if formed. However, the plan could not be executed as one of the coordinating atoms of the ligand remained unutilized [21]. Instead of the desired Pd 3 L 4 architecture, they observed a PdL 2 -type spirometallomacrocycle where bare palladium(II) is the juncture between two metallomacrocyclic rings. We report here a Pd 3 L 4 -type cage prepared from Pd(NO 3 ) 2 and pyridine-3,5diylbis(methylene) dinicotinate (L1) . We reported earlier the first Pd 3 L 4 -type double-decker coordination cage using a simple tridentate "E" shaped ester-based ligand bis(pyridin-3ylmethyl)pyridine-3,5-dicarboxylate (L2) [22,23]. An additional feature in our design using L2 is the stoichiometrically controlled formation of PdL 2 -type spiro and Pd 3 L 4 -type doubledecker complexes that is reversible under appropriate conditions. Subsequently, other research groups (Chand, Clever, Crowley and Yoshizawa groups) published Pd 3 L 4 -type cages [24]. This design has been further explored by Crowley et al. for the synthesis of a Pd 4 L 4 -type triple-decker cage [25]. The Clever research group reported a system in which two units of a double-decker cage are interlocked [26]. In this context, we revisited our earlier design of Pd 3 L 4 -type cages to prepare a corresponding analogues cages using the new ligand L1 (that is a positional isomer or regioisomer of L2) in order to exemplify ligand isomerism.
Ligand isomerism includes metal complexes (at least two) having the same molecular formula but are composed of different positional isomers (regioisomers) of the ligand. Positional isomers (regioisomers) of a non-chelating ligand system capable of forming palladium(II) complexes of same molecular formula is a rare phenomenon [27][28][29][30][31][32][33][34]. Such palladium(II) complexes should represent the phenomenon of "ligand isomerism". We reported a family of Pd 2 L 4 -type complexes that fits under the definition of ligand isomerism [34]. In the pursuit of ligand isomerism in Pd 3 L 4 -type double-decker cages we intended to include our reported cage (prepared from palladium(II) and L2) [22,23] and construct a new isomeric Pd 3 L 4 -type complex. The complexation study of cis-protected and bare palladium(II) components with the new tridentate ligand, i.e., pyridine-3,5diylbis(methylene) dinicotinate (L1) is the focus of this work. In addition, the dynamic behavior as well as anion binding abilities of selected complexes are also probed. Ligand L1 is a constitutional isomer of ligand L2 [22,23] and is expected to exhibit similarities but also some differences in complexation behavior with palladium(II) components. There are also some similarities and some differences in the related properties of these complexes.

Results and Discussion
Design and synthesis of ligand L1 The new ligand L1 was designed as a positional isomer (regioisomer) of the known ligand L2 (Figure 1). Each of these ligands has three pyridine moieties separated by two spacer moieties (-CH 2 OC(=O)-). Both ligands are semi-rigid/semiflexible due to the spacers' conformational mobility. The "E-shaped" conformation of the ligand that is suitable for the formation of the targeted Pd 3 L 4 -type complex is shown here for clarity of discussion. In a given ligand, two of the pyridine rings are substituted in the 3-position and are terminal and symmetrically disposed with respect to the central/internal 3,5-disubstituted pyridine ring. The spacers are identical in both ligands, however, their orientations are reversed in ligand L1 as compared to the known ligand L2. The primary intention of the design of L1 was to have a positional isomer (regioisomer) of the ligand L2. The tridentate ligand pyridine-3,5-diylbis(methylene) dinicotinate (L1, Figure 1) was prepared by condensation of pyridine-3,5-diyldimethanol [35] with nicotinoyl chloride hydrochloride in dry dichloromethane in the presence of triethylamine. The reaction mixture was stirred at room temperature for 24 h followed by aqueous work-up and column chromatography purification to afford pyridine-3,5-diylbis(methylene) dinicotinate (L1) as a white solid. The ligand was fully characterized by NMR spectroscopy and ESIMS techniques (Supporting Information File 1, Figures S1-S11). In addition, NOESY analysis was helpful in distinguishing the protons H a and H f .
It is assumed that the electron density at the central pyridine ring in L1 (that is a lutidine derivative) should be higher than that at the central pyridine ring of L2 (that is a dinicotinate derivative) having electron-withdrawing carbonyl substituents. Also, the electron density at the terminal pyridine ring in L1 (that is a nicotinate derivative) should be lower than that at the Scheme 1: (i)/(ii) Complexation of Pd(tmeda)(Y) 2 with the ligand L1 at 1:1 and 2:3 metal-to-ligand ratios, respectively; (iii)/(iv) complexation of Pd(Y) 2 with the ligand L1 at 1:2 and 3:4 metal-to-ligand ratios, respectively. For terminal pyridine ring of L2 (that is a picolyl derivative). The electrostatic potential maps at the terminal and internal pyridine nitrogen calculated using DFT methods (Supporting Information File 1, Table S2), however, are found to be comparable. Nevertheless, it seemed interesting to check whether or not the subtle difference in the electron density at the pyridine N centers has any influence on the coordination behavior of the ligands.

Complexation of palladium(II) components with ligand L1
Complexation of cis-protected palladium(II) was carried out with the ligand L1 at two different metal-to-ligand ratios (1:1 and 3:2). We also carried out the complexation of bare palla-dium(II) with the ligand L1 at two different metal-to-ligand ratios (1:2 and 3:4). The complexation reactions performed in DMSO-d 6 allowed the monitoring of complex formation and those performed in DMSO were used for isolation of the complex by precipitation methods. The resulting complexes at specified ratios of the reactants are depicted in Scheme 1 and the details of the complexation behavior are described below.
Complexation of cis-protected palladium(II) with ligand L1 at 1:1 metal-to-ligand ratio  rt for 15 min. The reaction was repeated in DMSO and the (PdL')L-type complex [Pd(tmeda)(L1)](NO 3 ) 2 (1a, Scheme 1(i)) was isolated from the reaction mixture by a precipitation method that is described in the experimental section. The 1 H NMR spectrum of the solution showed formation of a single discrete complex (Figure 2(ii)). Counter-anion (BF 4 − , ClO 4 − and OTf − ) variation was also carried out to suc-  Figure S17). One of the coordination sites of the ligand L1 thus remained unutilized in these mononuclear complexes.
The ESIMS data of the compounds 1b, 1c and 1d confirmed the formation of mononuclear complexes (Supporting Information File 1, Figures S18-S20). As an example, the ESIMS spectrum of compound 1b (Supporting Information File 1, Figure S18) showed isotopic peak patterns at m/z 658.14 and 285.57, respectively, which correspond to the cationic fragments [1b − BF 4 ] + and [1b − 2BF 4 ] 2+ that are formed due to the loss of one and two units of counter anions from 1b. The experimental and theoretical peak patterns were found to be in agreement. The data of 1c and 1d are given in Supporting Information File 1.
Complexation of cis-protected palladium(II) with ligand L1 at a 3:2 metal-to-ligand ratio The addition of three equivalents of cis-Pd(tmeda)(NO 3 ) 2 to a clear solution of two equivalents of ligand L1 in DMSO-d 6 produced a turbid mixture. However, a clear yellow solution was obtained upon stirring the mixture at 90 °C for 5 min. The progress of the complexation reaction was monitored by 1 H NMR spectroscopy. We targeted a Pd 3 L' 3 L 2 -type complex [36], i.e., [Pd 3 (tmeda) 3 (L1) 2 ](NO 3 ) 6 (2a, Scheme 1(ii)). However, the NMR spectrum showed the formation of [Pd(tmeda)(L1)](NO 3 ) 2 (1a) and uncomplexed cis-Pd(tmeda) 2+ . The reaction was allowed to continue where upon the integration ratios of the peaks corresponding to H f and H g were lower than expected and that of H d was higher than expected. In addition, a new peak was observed at around 9.15 ppm ( of the H g signal in 2a could not be induced by complexation and is best described by considering an anisotropy effect of the nearby carbonyl groups the ligand strand. and OTf − . It is important to note that Pd(Y) 2 solutions were prepared by reacting PdI 2 with AgY and the precipitated AgI was removed by filtration. Following this procedure, the presence of iodide as impurity could not be ruled out but its presence was found to not influence the formation of the targeted complex. In contrast, the presence of chloride remaining as impurity when PdCl 2 was reacted with AgY to prepare Pd(Y) 2 , contaminated Pd(Y) 2 and produced upon complexation with L1 complexes 3a-d along with some other products. The choice of PdI 2 is on the basis of our previous experience from related cages [23]. The complex 3a was characterized by various NMR techniques (Supporting Information File 1, Figures S27-S31). The 1 H NMR spectrum of compound 3a showed a single set of peaks ( Figure 3) featured with complexation-induced downfield shifts of protons belonging to the terminal pyridines (Δδ = 0.79, and 0.47 ppm for H a , and H b , respectively) as compared to the free ligand L1.
The peak position of H f did not change indicating that the central pyridine is not involved in the complexation. The 1 H NMR spectra of compounds 3b, 3c and 3d are very much comparable to that of 3a (Supporting Information File 1, Figure  S32). One of the coordination sites of the ligand L1 thus remained unutilized in these mononuclear complexes.
The ESIMS spectrum of compound 3a (Supporting Information File 1, Figure S33) showed isotopic peak patterns at m/z 866.10 and 402.05, respectively, which corresponds to the cationic fragments [3a − NO 3 ] + and [3a − 2NO 3 ] 2+ that are formed due to the loss of one and two units of counter anions from 3a. The experimental and theoretical peak patterns were found to be in agreement. The ESIMS spectrum of 3b is provided in Supporting Information File 1, Figure S34.  Figure S35b). The products in the mixture were identified as [(Cl)(NO 3 )@Pd 3 (L1) 4 ](NO 3 ) 4 (6a') and [(Cl) 2 @Pd 3 (L1) 4 ](NO 3 ) 4 (6a) and their proportion was found to depend on the amount of chloride as impurity. The influence of chloride on the product composition is discussed in a later section.
The complex 4a was characterized by various NMR techniques (Supporting Information File 1, Figures S37-S41). The ESIMS spectrum of compound 4a confirmed the formation of a trinuclear complex (Supporting Information File 1, Figure  S42).  Figure 4 (see Supporting Information File 1 for details). Geometry optimization and calculation of frequencies were performed using Gaussian 09 software package at the B3LYP/6-31G* level of theory [37]. Since the complex [Pd 3 (tmeda) 3 (L1) 2 ] 6+ could not be prepared exclusively, we looked into the energetics of the system. The overall Gibbs free energies (∆G) and the enthalpies (∆H) for the formation of the trinuclear complex [Pd 3 (tmeda) 3 (L1) 2 ] 6+ considering its formation from 1 equivalent of [Pd(tmeda)(NO 3 ) 2 ] and 2 equivalents of [Pd(tmeda)(L1)] 2+ were found to be not feasible (616.349 kcal mol −1 ) and endothermic (+537.727 kcal mol −1 ), respectively (see Figure S71 and Table S3 in Supporting Information File 1). However, a small amount of the trinuclear complex was formed experimentally. Probably, the counter anions stayed in the hemi-cage part of the trinuclear structure making it somewhat feasible. A detailed investigation of solvent and counter anion might help.
Complex-to-complex transformations: 3a versus 4a The in situ prepared mononuclear complex [Pd(L1) 2 ](NO 3 ) 2 (3a) was found to be stable at room temperature for days in Hydrogen atoms are omitted for clarity, red, blue, grey and cyan colors represent oxygen, nitrogen, carbon and palladium, respectively.

Scheme 2:
Reorganization of (i) a mixture of Pd(NO 3 ) 2 and 3a at a 2:1 ratio leading to 4a with a complete conversion and (ii) a mixture of L1 and 4a at a 2:1 ratio leading to 3a but with a partial conversion.
DMSO-d 6 (Supporting Information File 1, Figure S43) but not upon heating. The 1 H NMR spectrum recorded after heating the solution at 90 °C for 24 h revealed decomplexation and signals for the free ligand were observed (Supporting Information File 1, Figure S44). In addition, the solution turned dark and dark particles were observed. Upon cooling the solution, the free ligand should have undergone complexation to form 3a. However, no complexation was observed and it is assumed that palladium(II) got reduced to palladium(0). In another experiment, Pd(NO 3 ) 2 was added to a solution of 3a at a 2:1 ratio where upon complex-to-complex conversion was observed at room temperature or upon heating to afford 4a (Scheme 2(i)).
With the appropriate amount of Pd(NO 3 ) 2 a complete formation of 4a was observed within 20 min at rt or 5 min at 90 °C (Supporting Information File 1, Figures S45 and S46).
On the other hand, the in situ prepared trinuclear complex 4a was found to be stable at room temperature as well as at 90 °C for days (Supporting Information File 1, Figures S47 and S48). The free ligand L1 was added to a solution of 4a in DMSO-d 6 at room temperature and the sample was monitored by 1 H NMR spectroscopy. The calculated amount of ligand was added to the solution of 4a (at 2:1 ratio) to match the stoichiometric requirement for the formation of 3a. Although the formation of 3a was observed, it remained only as a minor product, and the added ligand L1 was partially consumed. Thus, the unbound ligand remained in its free state along with 4a. No further change was observed after 30 min (Supporting Information File 1, Figure  S49). Heating of the reaction mixture did not help in further pushing the conversion towards the formation of 3a (Supporting Information File 1, Figure S50). Prolonged heating could not help because the complex 3a is unstable under such conditions. This provided additional support on the higher stability of 4a as compared to 3a.
Each of the two cavities of cage 4a is loaded with one NO 3 -.
This phenomenon of NO 3 encapsulation by a related isomeric cage was established by us earlier [22,23]. Halide recognition by the complex 4a through anion exchange was studied by portionwise addition of freshly prepared solutions of tetra-nbutylammonium halide, i.e., TBA(X) (where X stands for F − , Cl − , Br − and I − ) in four separate experiments using DMSO-d 6 as the solvent. The anion exchange processes were monitored by 1 H NMR spectroscopy of the samples (Supporting Information File 1, Figures S51-S54). Is iodide not capable of replacing the preexisting nitrate in a competition or iodide is not suited at all for the cavity irrespective of any competition? The following argument might answer this question. The complexation reaction shown in steps (ii) of Scheme 3 suggest that the presence of BF 4 − , ClO 4 − or OTf − could not support the formation of the double-decker cage even though the required amount of palladium(II) was available. The addition of TBAI to any of these solutions containing Pd(Y) 2 and 3b, 3c or 3d, respectively, did not lead to double-decker cages indicates that I − is not suited for the cavity. However, addition of TBANO 3 , TBAF, TBACl and TBABr produced the corresponding anion encapsulated double-decker cages as shown in steps (iv), (v) and (vi) of Scheme 3. Representative 1 H NMR spectra for the conversion of 3b to corresponding products 4b, 5b, 6b and 7b are shown in Supporting Information File 1 ( Figure S55). the complex 5a was found to be unstable and it decomposed within a few hours. Thus the 1 H NMR and ESIMS spectrum of 5a were recorded from freshly prepared samples. The complexes 4a, 6a and 7a are quite stable and no decomposition was observed. Detailed characterization data of 4a-7a form a variety of NMR techniques are provided in Supporting Information File 1, Figures S56-S66.
The molecular compositions of the halide encapsulated complexes were confirmed by recording ESIMS data for the systems. The double halide encapsulated complexes 6a and 7a were detected. Also, one of the mixed halide-nitrate encapsulated complexes, i.e., 7a' was also detected.
The ESIMS spectrum of compound 6a (Supporting Information File 1, Figure S67) showed isotopic peak pattern at m/z = 956.02 which corresponds to the cationic fragment [6a − 2NO 3 ] 2+ that was formed due to the loss of two counter anions from 6a. The ESIMS spectrum of compound 7a (Supporting Information File 1, Figure S68) Figure S69) showed isotopic peak pattern at m/z = 991.51 which corresponds to the cationic fragment [7a' − 2NO 3 ] 2+ that is formed due to the loss of two units of counter anions from 7a'. The experimental and theoretical peak patterns were found to be in agreement. (12a) was stable for a few hours. These differences are ascribed to the positional exchanged functionalities in the ligands L1 and L2. Probably, the coordination ability of the central pyridine ring is better than that of the terminal pyridine rings in case of L1. However, in the mononuclear complexes of L1 the central pyridine remained uncomplexed, which may be due to the formation of metallomacrocyclic rings. This behavior is not observed in the case of mononuclear complexes of L2. Thus, the mononuclear complexes of L1 are reluctant to form (e.g., from 4a and L1) and prone to decomposition. As far as trinuclear complex formation is concerned the central pyridine ring of L1 is in a relatively favorable situation, thus the complex 4a could form and 3a was a kinetic product.

Conclusion
A set of mononuclear and trinuclear complexes were prepared through complexation of cis-protected palladium(II) and bare palladium(II) components with the new tridentate ligand L1. A variety of counter anions were employed to broaden the scope of the choice of metal components. Mononuclear complexes with PdL'L composition could be prepared easily, however, Pd 3 L' 3 L 2 -type trinuclear complexes were obtained in only small amounts. Also, mononuclear complexes of PdL 2 and trinuclear complexes of Pd 3 L 4 -type compositions were prepared easily. The choice of the counter anion did not influence the formation of mononuclear complexes whereas the counter anion displayed a template role for the formation of trinuclear complexes, especially for Pd 3 L 4 -type complexes. The anions helped to screen the charge repulsion between the palladium(II) ions. The complexation behavior of palladium(II) components with the ligand L2 have been reported earlier [23]. The similarities and differences in the complexation behaviors of the ligands L1 and L2 were highlighted. A qualitative comparison indicated that ligands L1 and L2 are well suited for the formation of trinuclear only and mononuclear/trinuclear complexes, respectively. The Pd 3 L' 3 L 2 -type complexes could be prepared, though in small proportions, using ligand L1 but not L2. The ligands L1 and L2 are positional isomers (regioisomers) hence many of their complexes could be rightfully considered under ligand isomerism in coordination complexes.

Experimental
Synthesis of ligand L1: A mixture of pyridine-3,5diyldimethanol (282.6 mg, 2.03 mmol) and nicotinoyl chloride hydrochloride (500.0 mg, 4.06 mmol) in dry DCM (50 mL) was placed in a 100 mL round-bottomed flask. The flask was placed in an ice bath to cool the mixture followed by the dropwise addition of triethylamine (2 mL). Then, the reaction mixture was stirred at room temperature for 24 h followed by the addition of a saturated aqueous solution of sodium bicarbonate. The organic layer was separated and the solvent was evaporated using a rotavapor. The crude product was purified by column chromatography using EtOAc/