The heterometallic one-dimensional solvated coordination polymer [NiPt2Cl6(TRIP-Py)4] n

The heterofunctional ligand TRIP-Py cooordinates to PtII cations with its phosphatriptycene P-atom donor and to NiII cations with its pyridyl N-atom donor, giving a one-dimensional heterometallic coordination polymer.


Introduction
The research area of coordination polymers (CPs) has become an established field in modern inorganic and coordination chemistry over recent decades (Batten et al., 2008). CPs offer the possibility to adjust the material properties not just through the design of the ligand and the choice of the metal cation, but also through the dimensionality and topology of the CP. This allows a tailoring for a vast range of applications from catalysis, magnetism and optics to chemical separation, medicine and electrochemistry (Wang et al., 2020;Zhong et al., 2022;Yu et al., 2022;Zhou et al., 2022;Zhang et al., 2021;Indra et al., 2018). Controlling and understanding the properties of a CP requires information on its structure, making diffraction techniques indispensable for the field. As the growth of large single crystals of CPs can be quite challenging due to their inherent insolubility, the field profits heavily from high-flux X-ray sources like synchrotron facilities and modern techniques like electron diffraction (Balestri et al., 2019;Huang et al., 2021).
While the vast majority of CPs contains a single type of metal cation, interest in heterometallic CPs containing two or more different metal cations in an orderly fashion is steadily growing (Kremer & Englert, 2018;Kuwamura & Konno, 2021). This inherently increases the synthetic challenge but opens an even larger playground to tune and combine properties. Gaining control over the position of the two different cations is frequently achieved by using heterofunctional ligands with distinctly different coordination sites. These can, for example, differ in their Pearson hardness (Pearson, 1963) and preferably coordinate metal cations of matching Pearson character.
In this article, we address the selectivity of a soft phosphorus and a harder nitrogen donor. This combination has been demonstrated to give selective heterometallic coordination compounds for a long list of discrete metal complexes (Hara et al., 2021;Schroers et al., 2021). In CP chemistry, however, the same pair of donor sites has only very recently been used for a heterometallic CP (Gildenast et al., 2022a). The ligand used in this previous report on heterometallic Zn II / Hg II polymers and also in the construction of the title compound is a rigid linear linker combining a pyridyl moiety with a phosphatriptycene, abbreviated as TRIP-Py (Fig. 1).
The phosphatriptycene belongs to the family of caged phosphines and has unique properties due to its special geometry (Shet et al., 2021;Tsuji et al., 2006). The introduction of the secondary bridgehead forces the phenylene propellers to be parallel to the phosphorus lone pair. Thus, the H atoms are pointing in the same direction increasing steric demand. Accordingly, until our recent publication (Gildenast et al., 2022b), no metal complex with more than two phosphatriptycene ligands bound to a single metal cation had been reported. At the same time, the geometry forces acute C-P-C angles which increases the s-character of the lone pair, lowering its basicity and �-donor strength while increasing the � acidity (Agou et al., 2004;Freijee & Stam, 1980;Jongsma et al., 1974;Drover et al., 2018;Hu et al., 2019;Mahaut et al., 2022). This strengthens the bond, especially towards electron-rich metal cations (Cao et al., 2019;Hu et al., 2021).
In this article, we present the crystallization and particularly challenging structural investigation of a desolvation-labile heterometallic CP in which TRIP-Py connects the softer Pt II and the harder Ni II cations. In contrast to our previously reported structures involving TRIP-Py, the halides coordinated at either metal cation are not engaged in polymer expansion and remain strictly terminal.

Experimental
Unless stated otherwise, all reagents and solvents were obtained from commercial sources and used without further purification. TRIP-Py and [PtCl 2 (COD)] were prepared according to published procedures (Gildenast et al., 2022a;Brauer, 1981). For the single-crystal X-ray diffraction measurement, the � goniometer at PETRA-III, P24, EH1, was used. The instrument was equipped with a Dectris CdTe area detector. For our experiment, synchrotron radiation (25 keV, � = 0.500 Å ) was used at a temperature of 100 (2) K (Oxford Cryostream 600 instrument, Oxfordshire, UK). Data were integrated with XDS (Kabsch, 2010) and corrected for absorption by multi-scan methods with SADABS (Bruker, 2014). The powder diffraction patterns were recorded at the Institute of Inorganic Chemistry, RWTH Aachen University, using a curved Stoe imaging-plate detector (IP-PSD). The diffractogram was recorded on a flat sample at ambient temperature in transmission using Cu K� 1 radiation. The ATR FT-IR spectrum was measured with a Shimadzu IRSpirit with a QATR-S ATR unit equipped with a diamond prism and is shown in Fig. 2. It immediately shows the presence of the ditopic ligand in the solid. In the range between 1600 and 500 cm À 1 , the spectrum reflects the pattern observed for uncoordinated TRIP-Py (Gildenast et al., 2022a). The elemental analysis (CHN) was measured using a HERAEUS CHNO-Rapid VarioEL. The thermogravimetric (TGA) measurements were carried out with a Netzsch STA 409 C/CD in a flux of air (60 ml min À 1 ) at a heating rate of 5 K min À 1 on a sample dried in air. The EDX measurement was performed in a Leo/ZeissFE-SEM Supra 35 VP instrument equipped with an OxfordINCA Energy 200 (SiLi crystal, 133 eV, 10 mm 2 ).

Synthesis and crystallization
TRIP-Py (17.6 mg, 0.04 mmol) and [PtCl 2 (COD)] (7.5 mg, 0.02 mmol) were each dissolved in dichloromethane (1 ml each) and the solutions were combined. NiCl 2 ·6H 2 O (2.4 mg, 0.01 mmol) was dissolved in ethanol (1 ml). The two solutions were layered with a layer of the mixed solvents (1 ml) in between. After several days, light-green crystals of 1 were obtained. For bulk analyses, they were isolated by filtration and washed with ethanol (yield: 14.6 mg, 60%).

Refinement
Crystal data, data collection and structure refinement details for 1 are summarized in Table 1 and the asymmetric unit is shown in Fig. 3.
H atoms attached to C atoms were introduced in calculated positions and treated as riding, with U iso (H) = 1.2U eq (C). For the pyridyl rings, split positions were refined for the C atoms in positions 2, 3, 5 and 6 with respect to the nitrogen. Only a site occupancy of 0.5 is compatible with reasonable interatomic distances between neighbouring pyridyl rings. The contribution of pore-contained solvent to the structure factors was treated with the bypass algorithm as implemented in SQUEEZE in PLATON (van der Sluis & Spek, 1990;Spek, 2015); a detailed discussion of alternative approaches is given in Section 3 (Results and discussion).

Results and discussion
The title compound, [NiPt 2 Cl 6 (TRIP-Py) 4 ] n , was prepared by reactive diffusion crystallization of an in-situ-generated dichloromethane solution of the complex [PtCl 2 (TRIP-Py) 2 ] with an ethanolic solution of NiCl 2 . The insoluble product is a heterometallic coordination polymer connected via covalent and coordinative bonds in one spatial direction (Fig. 4).
The Ni II cation is located on a crystallographic centre of inversion and resides in pseudo-octahedral coordination by two chloride ligands and four pyridyl donors of TRIP-Py ligands. Steric repulsion between ortho H atoms of adjacent ligands and between pyridyl H atoms and the halide ligands requires a tilt of the heteroaromatic rings. As a continuous windmill arrangement is incompatible with the inversion symmetry, disorder with alternative ring conformations of exactly half site occupancy is enforced. Each [NiCl 2 (TRIP-Py) 4 ] cross is connected to the next one via two PtCl 2 moieties with the P-atom donors in a cis configuration, resulting in a one-research papers 120 Gildenast  Computer programs: KAPPA (Paulmann, 2023), XDS2022 (Kabsch, 2010), SHELXT2018 (Sheldrick, 2015a), SHELXL2018 (Sheldrick, 2015b) and PLATON (Spek, 2020).

Figure 3
Displacement ellipsoid plot of the asymmetric unit of [NiPt 2 Cl 6 (TRIP-Py) 4 ] n in 1 (40% probability level), with labels for the atom sites. H atoms and alternative conformations for the disordered pyridyl rings have been omitted for clarity. The plot shows that the Pt-P distances for phosphatriptycenes are very comparable to those of regular uncaged phosphines. In contrast, the metal-ligand distances in Au I complexes of phosphatriptycenes (Gildenast et al., 2022a) are among the shortest of all phosphines in the Cambridge Structural Database (CSD; Version 5.43, with updates from November 2022; Groom et al., 2016). We speculate that � backbonding may play a less pronounced role in the case of the Pt II cation with its more positive formal charge. The P-Pt-P angle, however, is systematically at the larger end of the spectrum for phosphatriptycenes. The repulsion of the large triptycene moieties distorts the coordination sphere around the Pt II cation increasing the P-Pt-P angle and compressing the three remaining cis angles. Additionally, a reduction in planarity of the coordination sphere occurs compared to the cis-PtCl 2 complex of the uncaged phosphine PPh 3 (Table 2).

Figure 5
Scatter plot for the geometry of [PtX 2 (PR 3 ) 2 ] complexes (X = halide or methyl and PR 3 = tertiary phosphine). The P-Pt-P angle is plotted against the Pt-P distance. The data for the shown fragment were extracted from the CSD (Groom et al., 2016). The search was limited to error-free data sets collected at T � 200 K with R 1 � 0.05. Polymers and disordered structures were excluded, as well as structures of chelating phosphines with a C 2 bridge. All Pt complexes of phosphatriptycenes are added to the plot with star-shaped data points, and their respective secondary heteroatom is noted as Y-TRIP, with Y being either B, N or Si. Additionally, the data from the structure presented in this article are noted with this article. The colours of the stars denote whether they are chloride or methyl complexes (Drover et al., 2018;Tsuji et al., 2006;Ube et al., 2017).
The pore contains strongly disordered solvent molecules. Based on preliminary distances between residual electrondensity peaks and in agreement with the solvents employed in the synthesis, both dichloromethane and ethanol molecules are present. A tentative refinement of solvent molecules was performed, and 5.6 dichloromethane and 10.8 ethanol molecules per unit cell could be assigned in this model A (Fig. 8).
On the one hand, the above-mentioned solvent model A is not fully satisfactory: it did not account for the complete pore space but left a discrete void and a thin solvent-accessible channel, with a combined volume of 871 Å per unit cell. Despite the combined use of rigid fragments and hard geometry restraints for the solvent part, this partial solvent model A did not converge without damping, most probably because of high correlation between refinement variables describing the solvent. On the other hand, the graphical Packing of two neighbouring polymer strands of [NiPt 2 Cl 6 (TRIP-Py) 4 ] n in 1 shown along [311]. H atoms have been omitted and the triptycene wings simplified for clarity. The yellow ellipse shows how adjacent polymers are interdigitated.

Figure 7
Packing of 1 � 3 � 2 unit cells of 1 with a void contact surface calculated with Mercury (Macrae et al., 2020) (1.2 Å probe radius, 0.3 Å grid spacing). The spheres represent the diameter of the largest possible sphere that can pass through the pore along the given unit-cell vectors (Willems et al., 2012).

Figure 8
The asymmetric unit of 1, with a partial molecular model of the solventfilled pore. H atoms have been omited for clarity. The dichloromethane molecule shown as dashed is only partially occupied and overlaps with the position of the adjacent ethanol molecule.

Figure 9
Plot for 1 of the agreement factor R 1 against the diffraction resolution for the original data and the data modified with the bypass algorithm as implemented in PLATON (van der Sluis & Spek, 1990;Spek, 2015). synopsis of the solvent-masking process in Fig. 9 indicates that 'squeezing out' the entire solvent-filled pore according to model B is an equally crude approximation. Fig. 9 shows that the contribution of the solvent molecules to the structure factors extends up to a resolution of 0.4 Å À 1 , i.e. almost into atomic resolution. The solvent part is at least in part associated with long-range order and cannot be well modelled by an electron gas. This explains why the solventsqueezed structure model B retains a significant number of disagreeable intensities in the intermediate resolution range. These unsatisfactory intensity data show better agreement with the calculated structure factors from the partial solvent model A. In conclusion, we decided to report the more straightforward model B because localization of individual solvent molecules is not a crucial feature for the title structure. The overall content of the pore can be estimated from the results of the bypass algorithm as summarized in Table 3.
For the estimation of the spatial demand of a disordered solvent molecule, we followed the suggestion of Mecozzi & Rebek (1998) and assumed a 1.3-fold of the volume of the molecules in their own crystal structure. A combination of 5 dichloromethane (DCM) and 20 ethanol molecules per unit cell represents a good fit to pore volume and electron count. This amount of dichloromethane molecules is slightly lower than in our tentative molecular solvent model A, but we recall that this model is rather unstable. The large number of volatile solvent molecules also makes the compound prone to rapid desolvation. This impairs a reasonable validation of the structure by powder diffraction. We measured powder patterns of both wet crystals taken directly from the mother liquor, as well as dried samples. Both of them look quite similar but do not match the phase characterized by single crystal X-ray diffraction. The loss of solvent molecules is also reflected in the elemental analysis which matches more closely the expected values of the desolvated polymer with small amounts of residual solvent (Table 4). The best match for the experimentally determined values is achieved for two dichloromethane molecules per unit cell.
The elemental analysis matches the results obtained in the thermogravimetric analysis (Fig. 10). {[NiPt 2 Cl 6 (TRIP-Py) 4 ]·-2DCM} n loses weight in two well-separated steps. First, a gradual loss of 6.7% of mass until 160 � C is observed, which agrees with the desolvation of two dichloromethane molecules. The second step begins at 350 � C and ends at 520 � C, after which 37.5% of the original sample weight is left. The identity of the remaining black powder could not be identified unambiguously. Its diffraction pattern displays merely reflections for elemental Pt. These are very broad, indicating a small average particle size. Using energy-dispersive X-ray spectroscopy (EDX), the elements Pt, Ni, Si and P were detected in a ratio of 2.0 (3):1.3 (4):4.1 (5):4.0 (5), in acceptable match with the composition in the original CP. From this we propose that Ni and Si stay in their oxidation states Ni II and Si IV , that phosphorus is oxidized to phosphate anions and oxide anions balance the remaining positive charge. The total sum formula of this mixture has a molecular weight of 38.1% of the original CP and two molecules of dichloromethane, matching the experimental weight loss from the TGA measurement. The EDX measurement does however reveal a higher than expected value for oxygen. Our suggested composition would require a Pt:O ratio of 2:19; the EDX analysis yields 2.0 (3):34 (3). This discrepancy may be caused by a contribution to the oxygen signal from the material used for fixing the sample.

Conclusion
The structural characterization of 1 proved challenging but also rewarding. The PtCl 2 moieties in the heterometallic polymer are exposed towards the periphery and therefore potentially useful for follow-up reactions. They might, for

Figure 10
Thermogravimetric analysis of [NiPt 2 Cl 6 (TRIP-Py) 4 ] n , with a heating rate of 5 K min À 1 in a stream of air.

Table 3
Void volume (V) and electron content (e À ) according to the program SQUEEZE (Spek, 2015) as implemented in PLATON (Spek, 2020), and the average electron count per volume in the void � for the pore treated with the bypass algorithm.