Iron in a Cage: Fixation of a Fe(II)tpy2 Complex by Fourfold Interlinking

Abstract The coordination sphere of the Fe(II) terpyridine complex 1 is rigidified by fourfold interlinking of both terpyridine ligands. Profiting from an octa‐aldehyde precursor complex, the ideal dimensions of the interlinking structures are determined by reversible Schiff‐base formation, before irreversible Wittig olefination provided the rigidified complex. Reversed‐phase HPLC enables the isolation of the all‐trans isomer of the Fe(II) terpyridine complex 1, which is fully characterized. While temperature independent low‐spin states were recorded with superconducting quantum interference device (SQUID) measurements for both, the open precursor 8 and the interlinked complex 1, evidence of the increased rigidity of the ligand sphere in 1 was provided by proton T2 relaxation NMR experiments. The ligand sphere fixation in the macrocyclized complex 1 even reaches a level resisting substantial deformation upon deposition on an Au(111) surface, as demonstrated by its pristine form in a low temperature ultra‐high vacuum scanning tunneling microscope experiment.

The reaction mixture was cooled to room temperature, was washed with aqueous saturated NaHCO3 solution and was extracted with DCM. The combined organic layers were washed with brine, dried over MgSO4, filtered and concentrated under reduced pressure to yield the product as a white solid (711 mg, 2.36 mmol, 100%). The product was sufficiently pure and was used for the next reaction without further purification.

UV-Vis Spectra of Compound 8 and Compound 1
The UV-Vis spectrum of compound 8 was recorded on a Shimadzu UV spectrometer UV-1800. The wavelength was measured in nm. The solution was measured under air saturated conditions in acetonitrile. The spectrum was recorded using optical 1115F-QS Hellma cuvettes (10 mm light path).

S38
The UV-Vis spectrum of compound 1 was recorded on a Jasco V-770 spectrophotometer. The wavelength was measured in nm. The solution was measured under air saturated conditions in acetonitrile. The spectrum was recorded using optical 1115F-QS Hellma cuvettes (10 mm light path).  The combined UV-Vis spectra of compound 1 and 8 can be found in figure SI6. Furthermore, the absorption maxima for both compounds are summarized in table SI3. It can be seen that the metal-to-ligand charge transfer (MLCT) absorption is in the same range for both complexes. Only a small bathochromic shift for the cage compound 1 of 3 nm is observed. The intensity of the MLCT band is slightly decreased upon macrocyclization. These minuscule spectral changes indicate that the surrounding ring system does not distort the coordination geometry at the metal center and is only weakly coupled to the complex' tpy ligand system. A bigger difference can be observed for the ligand-centered (LC) transitions. For the aldehyde complex 8 a weak transition at 332 nm is observed, whereas we observe a much more intense transition for the complex 1 at 302 nm. This intense transition is attributed to the macrocycle formed upon interlinking the ligands in 1 resulting in a large delocalized π-system.  Fluorescence spectra of the complexes 1 and 8 were recorded in degassed acetonitrile at room temperature using quartz fluorescence cuvettes with a 1 cm optical path length on a Horiba Jobin-Yvon FluoroMax 4 fluorimeter. However, no emission of the complexes was observed under these conditions.

Crystal Data of 8 and Unsuccessful Crystallization Attempts of 1
Single crystals of compound 8 were grown by vapor diffusion technique with acetonitrile as solvent and diethyl ether as anti-solvent. Solid state structure of complex 8 with rotation ellipsoids at 50% probability are displayed. Hydrogen atoms, solvent molecules and the PF6counter ions were omitted for clarity. Color code: N: blue, Fe: orange, O: red and C: gray.

Crystallization Attempts of 1:
In spite of numerous attempts, crystals suitable for x-ray diffraction analysis of complex 1 could not be obtained. Based on our experience with Fe(II)tpy2 model complexes, crystallization attempts of 1 were based on both, vapor diffusion and the solvent layering techniques. In particular the following combinations were investigated repeatedly without success. Solutions of 1 in acetonitrile, acetone, methanol, and dichloromethane were exposed to the "anti-solvents" diethyl ether, diisopropyl ether, heptane, hexane, petroleum ether, water, toluene, tetrahydrofuran, cyclohexane, and isopropanol. Also to vary the initial concentration of 1 in the solvents did not result in crystals of the required quality.

SQUID Measurements
The magnetic measurements were performed on a SQUID magnetometer (Quantum design, model MPMS-XL-5). For standard measurements, the temperature dependent magnetization was recorded at B=0.1 Tesla as external magnetic field. The temperature was sweeped in the range of 5 -365 K at the rate 3 K min -1 . Gelatine capsules were used as sample holders. The diamagnetic corrections of the molar magnetic susceptibilities were applied using Pascal's constants. From these measurements, we can infer that these compounds (9, 8 and 1) showed structurally low-spin state as the high-spin molecules show XT value around 3.5 emu mol -1 K. 5 Therefore, we concluded that all the compounds were diamagnetic and no SCO behavior was observed upon heating.

IR-Spectra
Infrared spectra were measured with a Bruker Platinum ATR Alpha in the region of 900-200 cm -1 . The spectra were recorded at room temperature with 150 scans and with a resolution of 2.0 cm -1 . The idea was to identify the Fe-N vibration to confirm the increasing stiffness upon interlinking of the structures. Unfortunately, the vibration could not be identified.

Determination of the Spin-Spin Relaxation Times T2 of Compound 8 and alltrans Isomer 1 by NMR Spectroscopy
The structures of 1 and 8 (illustrated in figure SI14) suggest an increased rigidity of the macrocyclized complex 1 compared to that of 8 as the isophthalaldehyde groups can freely rotate, which induces motion and flexibility all over the molecule. In order to gain insight into the dynamic behaviour of all-trans 1 and 8 we determined the spin-spin relaxation time T2 for all protons in 1 and 8 using a Carr-Purcell-Meiboom-Gill (CPMG) spin echo pulse sequence. An increase in T2 time at constant molecular weight is a very sensitive probe for the intramolecular mobility of a spin. The protons in the following figures and tables are assigned according to figure SI14.   To our delight, the analysis revealed highly significant differences between the central terpyridine units in 1 (0.496 to 0.752 s) and 8 (0.700 to 1.128 s, Table SI4)  show an up to three-fold T2 increase, as can be expected due to the higher degree of rotational freedom. Overall the T2 times of 1 are much more uniform throughout the molecule (max/min: 1.52), whereas the spread is 3.04 in 8.
The molecular weight increase of 20.9% between 1 and 8 is too small to explain the observed differences in T2 times, but aggregation of 1 to oligomeric assemblies would lead to effects in the observed order of magnitude. Therefore, also the hydrodynamic radii of both compounds in acetonitrile by diffusion ordered NMR spectroscopy were determined and a difference in diffusion coefficients of 5.2% corresponding to a mass increase for solvated 1 of 16.4% with respect to 8, thus corroborating the proposed rigidification by the four-fold interlinkage was found. Figure SI16: Diffusion ordered NMR spectroscopy of compound 8: Experimental intensity (blue circles) and 2parameter fit (red line). Figure SI17: Diffusion ordered NMR spectroscopy of compound 1: Experimental intensity (blue circles) and 2parameter fit (red line).

Monitoring the Imine Condensation towards the Fourfold Interlinked Complex 2 by DI-ESI-MS and NMR Spectroscopy
The course of the eightfold imine condensation forming the fourfold interlinked complex 2 was monitored by direct injection electrospray mass spectrometry (DI-ESI-MS). Representative spectra recorded of the crude reaction mixture are displayed in figure SI18 and figure SI19. In figure SI18 the course of the reaction after several minutes is depicted. The masses of the starting material and different intermediates, namely singly interlinked, twofold interlinked, threefold interlinked and the fourfold interlinked target structure 2 can be identified. Moreover, even an unwanted eightfold imine condensation product comprising five diamine subunits (three closed bridges and two additional bridging units masking both remaining aldehydes) can be observed. As described in the main text, performing the condensation reaction with an eightfold excess of meta-phenylenediamine in refluxing 1,2-dichloroethane (83.5 °C) for 10 days resulted in the DI-ESI-mass spectrum displayed in Figure SI19, with the mass of 670 m/z corresponding to the desired fourfold interlinked octaimine complex 2 as major peak. However, some of the signals like e.g. the threefold interlinked intermediate (633 m/z) or the overreacted side-product (723 m/z) are less pronounced, but still present. After 10 days, when the DI-ESI-MS displayed in Figure SI19 with the mass at 670 m/z assigned to the target complex 2 as dominant signal was recorded, the reaction was allowed to cool down. All attempts to isolate the pure imine interlinked complex failed. However, the crude reaction mixture enabled to record a high resolution mass spectrum of the complex 2.
The analysis of the crude reaction mixture by 1 H-NMR spectroscopy in acetone-d6 is displayed in Figure SI20. The signal at 9.87 ppm indicates the presence of benzaldehyde type structures in the reaction mixture, pointing either at partial incomplete closing of the cage or a reopening of the labile imine bonds. Attempts to increase the concentration of the caged target structure 2 failed. The decomposition of the complex during the purification attempts pointed at too labile imine bonds and triggered the here reported interlinking by Wittig-chemistry.

Scanning Tunneling Microscope (STM) Experiments of Complex 1
Clean and flat Au(111) surfaces were obtained by repeated cycles of Ar-ion sputtering and subsequent annealing to 500 °C.
The deposition of 1 was performed with a home-built electrospray setup 6 using a solution consisting of 1 dissolved in acetonitrile. The mass spectrum of the solution, acquired prior to deposition, exhibits a single peak with a mass to charge ratio at approximately 660 u/e, which is compatible with the expected mass to charge ratio of 665.7087 u/e within the accuracy of the mass spectrometer used. The measurements were performed with a scanning tunneling microscope operated at low temperature (4.4 K) in ultra-high vacuum. STM tips were electrochemically etched from tungsten wire and further prepared in situ by indentation in the Au(111) substrate.

Fe(II)tpy2 Model Complexes of Varying Rigidity in STM Experiments
The interdependence of the coordination sphere geometry and the spin-state of the central Fe(II) ion makes Fe(II)tpy2 complexes interesting model compounds for manipulating the molecular spin-state in STM experiments. To illustrate the unique stability of the coordination sphere of 1 on Au(111) in low-temperature STM experiments, its behavior is compared to less rigid Fe(II)tpy2 model complexes that were previously analyzed in similar experiments. Figure SI22 shows the structures together with STM measurements of three Fe(II)tpy2 model complexes with increasing numbers of linkers between the terpyridine ligands. As previously discussed in Ref.
[7], the compound without any linkers between both tpy-subunits ( Figure SI22 a) was found fragmented on Au(111). A majority of fragments assembles into irregular clusters with varying numbers of ligands and Fe atoms. Occasionally, flat dimers are observed (Figure SI22 d). The apparent height of the ligands is approximately 200 pm. We

S61
note that apparent height is related to the overlap of wave functions and may therefore differ from the geometric height.
Fragmentation on the surface is prevented by incorporating two linkers between the two tpy-ligands (Figure SI22 b). However, as displayed in Figure SI22 e, the complex is planarized on the surface because of the strong dispersion forces exerted by the metal substrate. The planarization is evident from the unchanged apparent height of the compound on Au(111) relative to that of the bare ligand.
A drastically different result is observed for the fourfold interlinked Fe(II)tpy2 complex 1 ( Figures SI22 c and f), where the apparent height is approximately 3 times larger than that of bare ligands. The large height indicates that the structure of 1 remains three-dimensional with limited surface-induced planarization.
In summary, the comparison of the STM results on the three Fe(II)tpy2 model complexes of increasing rigidity suggest that 1 (4 linkers) is least sensitive to dispersion force induced distortion. This corroborates the molecular design hypothesis of 1 being the most rigid compound.

HPLC Chromatograms of Complex 1
In a first HPLC run (for further details see experimental procedures) the threefold interlinked intermediate and the condenstation product with more than four diamines per complex were separated from the product. The all-trans isomer was isolated from other isomers by two different applied reversed-phase (C18) HPLC conditions (see figure SI22). Both times a 7:3 mixture of MeCN:H2O, but once with 10 mM NH4PF6 in both solvents and once with 10 mM NH4PF6 in water and 0.1% TFA in MeCN. The chromatograms are given at 360 nm and 371 nm, respectively. In both cases the almost perfectly baseline separation of the all-trans isomer could be achieved. In the chromatogram a shoulder for both marked peaks can be seen on the right side. Therefore, only the bluely marked part was isolated to prevent the impure isolation of the isomer. However, this and the fact that several different isomers are formed (theoretically 64 different isomer) led to the rather low yield of 4%.