Dynamic Complex‐to‐Complex Transformations of Heterobimetallic Systems Influence the Cage Structure or Spin State of Iron(II) Ions

Abstract Two new heterobimetallic cages, a trigonal‐bipyramidal and a cubic one, were assembled from the same mononuclear metalloligand by adopting the molecular library approach, using iron(II) and palladium(II) building blocks. The ligand system was designed to readily assemble through subcomponent self‐assembly. It allowed the introduction of steric strain at the iron(II) centres, which stabilizes its paramagnetic high‐spin state. This steric strain was utilized to drive dynamic complex‐to‐complex transformations with both the metalloligand and heterobimetallic cages. Addition of sterically less crowded subcomponents as a chemical stimulus transformed all complexes to their previously reported low‐spin analogues. The metalloligand and bipyramid incorporated the new building block more readily than the cubic cage, probably because the geometric structure of the sterically crowded metalloligand favours the cube formation. Furthermore it was possible to provoke structural transformations upon addition of more favourable chelating ligands, converting the cubic structures into bipyramidal ones.


Syntheses of the complexes Metalloligand ML-1(BF4)2
A solution of 2-formyl-6-methyl-5-(4'-pyridyl) pyridine 1 (50.00 mg, 0.25 mmol, 3.00 equiv) and tris(2-aminoethyl)amine (2, TREN; 12.59 μL, 0.08 mmol, 1.00 equiv) in 8 mL of acetonitrile was degassed by applying a vacuum and flushing with argon three times and stirred for 30 min at room temperature under an argon atmosphere. Subsequently, iron(II) tetrafluoroborate hexahydrate (28.38 mg, 0.08 mmol, 1.00 equiv) was added. The red solution was degassed again and stirred at 50 °C for 16 h under an argon atmosphere. After cooling to room temperature, the solution was transferred into 100 mL of degassed diethyl ether. The red precipitant was collected and carefully washed with diethyl ether several times. After drying in air, the product was obtained as a bright red solid in 92% yield (73.12 mg, 0.08 mmol).

Trigonal bipyramid BP-1(OTf)6(BF4)4
A solution of ML-1(BF4)2 (10.00 mg, 10.90 μmol, 2.00 equiv) and 1,3-bis(diphenylphosphino)propanylpalladium(II) triflate ([Pd(dppp)(OTf)2]; 13.37 mg, 16.40 μmol, 3.00 equiv) in 1.4 mL of acetonitrile was degassed by applying a vacuum and flushing with argon three times and heated under an argon atmosphere at 50 °C for 16 h. The resulting solution was filtered, and the product was precipitated by the diffusion of diethyl ether vapor into an acetonitrile solution of the complex. The red solid was filtered off and carefully washed with diethyl ether several times. After the solid was dried in a stream of air, the product was obtained as a red solid in 94% yield (22.18 mg, 5.08 μmol).

Cube CU-1(BF4)28
A solution of ML-1(BF4)2 (5.00 mg, 5.46 μmol, 8.00 equiv) and tetrakis(acetonitrile)palladium(II) tetrafluoroborate ([Pd(CH3CN)4](BF4)2; 1.82 mg, 4.09 μmol, 6.00 equiv) in 0.7 mL of acetonitrile was degassed by applying a vacuum and flushing with argon three times and heated under an argon atmosphere at 50 °C for 16 hours. The resulting solution was filtered, and the product was precipitated by the diffusion of diethyl ether vapor into an acetonitrile solution of the complex. The red solid was filtered off and carefully washed with diethyl ether several times. After the solid was dried in a stream of air, the product was obtained as a bright red solid in 89% yield (5.49 mg, 0.61 μmol).
The 1 H-NMR spectrum of ML-1(BF4)2 shows only one set of signals, indicating the presence of C3-symmetric complexes. As dictated by the covalently bridged ligand system, only the fac-isomer of the mononuclear complex can be formed and observed. The chemical shift of δ = 237.0 ppm is consistent with the predicted chemical shift for imine protons in paramagnetic iron(II) compounds with S = 2. 3 The magnetic susceptibility of ML-1(BF4)2 in solution at room temperature was obtained by employing the Evans' method as described in the literature. 4 The molar magnetic susceptibility multiplied with temperature was determined to be 3.07 cm 3 K mol -1 (Table S1, Figure S6) and is consistent with an expected value for ΧmT of one uncoupled iron(II) cation in the high-spin state (S = 2) of 3.001 cm 3 K mol -1 . 5 H-NMR spectra of ML-1(BF4)2 ( Figure S9) show that the paramagnetic high spin state of the iron(II) cation is stabilized in the temperature range from 233 K to 298 K in solution. Please note that chemical shifts of paramagnetic samples are temperature-dependent and the following dependency is valid in these cases: δT ~ ΧT (δ = chemical shift, T = temperature, X = magnetic susceptibility). Therefore, the product δT is constant, if the magnetic susceptibility does not change (Table S2).  Figure S10. UV-vis spectrum of ML-1(BF4)2 (acetonitrile, 170.5 µM, 10 mm cuvette).
A crystalline sample of ML-1(BF4)2 was used to determine the magnetic properties in the solid state ( Figure S11). During sample preparation we observed a very rapid loss of solvent after removing the crystals from the mother liquor, which hindered the exact determination of the mass of the sample. As a result we observed a molar magnetic susceptibility multiplied with temperature of 3.5 cm 3 K mol -1 in the temperature range from 100 to 300 K, which is slightly higher than the expected value for one uncoupled iron(II) cation in the high-spin state of XmT = 3.001 cm 3 K mol -1 . 5 However, consistent with the result of the determination of the magnetic susceptibility in solution (Table S1), the measurement shows the paramagnetic character of the complex.  The observed chemical shift of δ = 237.7 ppm is consistent with the predicted chemical shift for imine protons in paramagnetic iron(II) compounds with S = 2. 3 The magnetic susceptibility of BP-1(OTf)6(BF4)4 in solution at room temperature was obtained by employing the Evans' method as described in the literature. 4 The molar magnetic susceptibility multiplied with temperature was determined to be 5.95 cm 3 K mol -1 (Table S3, Figure S13) and is consistent with an expected value for ΧmT of two uncoupled iron(II) cations in the high-spin state (S = 2) of 6.002 cm 3 K mol -1 . 5
The temperature dependent 1 H-NMR spectra of BP-1(OTf)6(BF4)4 ( Figure S17) show that the paramagnetic high spin state of the iron(II) cations is stabilized in the temperature range from 233 K to 298 K in solution. Analogue to the findings with ML-1(BF4)2 ( Figure S9, Table S2) the product δT is constant, if the magnetic susceptibility does not change (Table S4). Table S4. Comparison of δT values for one selected 1 H-NMR signal of BP-1(OTf)6(BF4)4. The observed variation of the product δT is 1 %, proving that the paramagnetic high-spin state is stabilized in the given temperature range.    , most probably because this concentration is below the critical self-assembly concentration of the complex. A crystalline sample of BP-1(OTf)6(BF4)4 was used to determine the magnetic properties in the solid state ( Figure S20). During sample preparation we observed the rapid loss of solvent after removing the crystals from the mother liquor. However, the loss of solvent was less pronounced than with ML-1(BF4)2. We observed a molar magnetic susceptibility multiplied with temperature of 6.0 cm 3 K mol -1 in the temperature range from 200 to 300 K, which is in very good agreement with the expected value for two uncoupled iron(II) cations in the high-spin state of XmT = 6.002 cm 3 K mol -1 . 5 Figure S20. Temperature dependent VSM measurement (10000 Oe) of a crystalline sample of BP-1(OTf)6(BF4)4. Figure S21. Paramagnetic 1 H-NMR spectrum (300 MHz, acetonitrile-d3, 298 K) of CU-1(BF4)28.
The chemical shift of δ = 238.8 ppm is consistent with the predicted chemical shift for imine protons in paramagnetic iron(II) compounds with S = 2. 3 The magnetic susceptibility of CU-1(BF4)28 in solution at room temperature was obtained by employing the Evans' method as described in the literature. 4 The molar magnetic susceptibility multiplied with temperature was determined to be 24.10 cm 3 K mol -1 (Table S5, Figure S22) and is consistent with an expected value for ΧmT of eight uncoupled iron(II) cations in the high-spin state (S = 2) of 24.008 cm 3 K mol -1 . 5  The progress of the self-assembly to CU-1(BF4)28 was followed using NMR spectroscopy. Wide sweep 1 H-NMR spectra were measured after different times of heating the complex solution after the addition of [Pd(CH3CN)4](BF4)2 to a solution of ML-1(BF4)2 in acetonitrile-d3 ( Figure S23). The spectra show the impressively rapid formation of the heterobimetallic cubic cage. Already after combining all building blocks and leaving the solution at room temperature first signals assigned to the complex were detected. However, these signals are very broad and have a low intensity. After heating the solution to 50 °C for 3 hours the complex formation was finished and all signals referring to the cubic cage were detected. Further heating overnight did not lead to any changes in the 1 H-NMR spectra. This very fast complex formation impressively shows both, the kinetic lability of this complex, as well as the much faster self-assembly compared to the analogues diamagnetic system that took 5 days at 65 °C to assemble.
The temperature dependent 1 H-NMR spectra of CU-1(BF4)28 ( Figure S26) show that the paramagnetic high spin state of the iron(II) cations is stabilized in the temperature range from 233 K to 298 K in solution. Analogue to the findings with ML-1(BF4)2 ( Figure S9, Table S2) the product δT is constant, if the magnetic susceptibility does not change (Table S6). Table S6. Comparison of δT values for one selected 1 H-NMR signal of CU-1(BF4)28. The observed variation of the product δT is below 1 %, proving that the paramagnetic high-spin state is stabilized in the given temperature range.   The superimposed UV-vis spectra of CU-1(BF4)28 with different concentrations ( Figure S28) show that the UV-vis spectrum of this heterometallic cage is independent from the complex concentration in the investigated concentration range. This indicates that the critical self-assembly concentration is below 17 µmol L -1 and may demonstrate the higher thermodynamic stability of CU-1 over BP-1, which already decomposed at a concentration of 79 µmol L -1 ( Figure S19). This finding might also be responsible for the more readily incorporation of the less bulky subcomponent 3 by BP-1(OTf)6(BF4)4 compared to CU-1(BF4)28 (vide infra). The slow diffusion of diethyl ether vapor into a concentrated solution of ML-1(BF4)2 in acetonitrile gave X-ray quality single crystals, suitable to determine the solid state structure of the metalloligand as the tetrafluoroborate salt (figure S29 and figure  S30). The mononuclear complex crystallizes in the achiral triclinic space group P-1. The coordination sphere around the iron cation is best described as octahedral and arises from six nitrogen atoms of the tris(pyridylimin) binding site with an average Fe-N bond length of 2.168 Å, being consistent with a paramagnetic high-spin iron(II) complex. 7 As dictated by the ligand system only the fac-isomer of ML-1(BF4)2 is observed. Figure S29. Solid state structure of ML-1(BF4)2 as determined by single crystal X-ray diffraction. Hydrogen atoms, solvent molecules and counter anions are omitted for clarity.
[S8] The structures were solved by direct methods and refined anisotropically by the leastsquares procedure implemented in the ShelX program system.
[S9] The hydrogen atoms were included isotropically using the riding model on the bound carbon atoms. The contribution of the electron density from disordered counterions in BP-1(OTf)6(BF4)4, which could not be modelled with discrete atomic positions were handled using the SQUEEZE 10 routine in

Crystal structure of CU-1(BF4)28
Single crystals suitable for X-ray diffraction were obtained by very slow evaporation of solvent from an acetonitrile solution of CU-1(BF4)28 over 3 months. Single crystals of CU-1(BF4)28 was transferred onto a glass slide covered by NVH oil. Seven crystals were quickly mounted onto 40 to 200 µm nylon loops and immediately flash cooled in liquid nitrogen to avoid collapse of the crystal lattice. Crystals were stored at cryogenic temperature in dry shippers, in which they were safely transported to macromolecular beamline P11, 12 Petra III, DESY, Hamburg, Germany. Samples were mounted using the StäubliTX60L robotic arm. A wavelength of  = 0.6889 Å was chosen using a liquid N2 cooled double crystal monochromator. Single crystal X-ray diffraction data was collected at 80(2) K on a single axis goniometer, equipped with an Oxford Cryostream 800 and a Pilatus 6M fast. 1800 diffraction images were collected in a 360° φ sweep at a detector distance of 250 mm, 100% filter transmission, 0.2° step width and 0.2 second exposure time per image. Due to radiation damage indicated by drastically reduced resolution in the second half of the collection, only the first 900 frames were used in final integration with XDS. 13 The data was cut at 1.18 Å, as the signal to noise ratio has dropped below I/σ(I) < 2.0. Due to high mosaicity and disorder in the solvent region a higher resolution could not be achieved. Nevertheless the resolution achieved was sufficient to solve the structure by intrinsic phasing/direct methods using SHELXT 14 . SHELXL 15 (version 2014/7) was used for refinement and ShelXle 16 as a graphical user interface. The DSR program plugin was employed for modelling. 17-18 All cycles were refined against F 2 until convergence using the conjugate-gradient algorithm (CGLS). Only for computing the crystallographic information file (CIF) the full-matrix least-squares routine was employed. Hydrogen atoms were included as invariants at geometrically estimated.

Specific refinement details
In order to generate a molecular model and increase robustness of the refinement we have adapted and exploited techniques commonly applied in macromolecular structure refinement. The obtained geometry from the ML-1(BF4)2 structure was used as input for stereochemical restraints generation by the GRADE program using the GRADE Web Server (http://grade.globalphasing.org). The restraint dictionary was applied to the ligand in the refinement using residue name MPY. A GRADE dictionary for SHELXL contains target values and standard deviations for 1.2-distances (DFIX) and 1.3-distances (DANG), as well as restraints for planar groups (FLAT). The refinement of ADP's for non-hydrogen atoms was enabled by using the rigid bond restraint (RIGU) 19 in the SHELXL program in combination with SIMU restraints. The contribution of the electron density from disordered counterions, and solvent molecules, which could not be modelled with discrete atomic positions were handled using the SQUEEZE 7 routine in PLATON. 8 The solvent mask file (.fab) computed by PLATON were included in the SHELXL refinement via the ABIN instruction leaving the measured intensities untouched.
ML-1(BF4)2 (10.00 mg, 10.90 µmol, 1.00 equiv) together with 2-formyl-5-(4'-pyridyl) pyridine 3 (12.06 mg, 65.50 µmol, 6.00 equiv) was dissolved in 1 mL acetonitrile-d3. The solution was degassed by applying a vacuum and flushing with argon three times. The reaction was indicated by a change of colour from bright red to dark purple only few minutes after starting the reaction. To reach equilibrium the reaction mixture was heated to 40 °C for 16 hours. Then, a 1 H-NMR spectrum was measured to evaluate the reaction progress without any work up. In order to isolate ML-2(BF4)2 the reaction mixture was added drop wise to 25 mL of diethyl ether. The resulting suspension was stirred for 30 minutes at room temperature and then the purple precipitant was collected and washed with generous amounts of diethyl ether. Drying in a stream of air gave ML-2(BF4)2 as a purple solid in 99% yield (9.48 mg, 10.8 µmol).   Comparison of the 1 H-NMR spectra of ML-2(BF4)2 and the spectrum received from the isolated complex of the complex-tocomplex transformation according to scheme 2 indicates the successful formation of ML-2(BF4)2 ( Figure S35). An Evans' experiment with the isolated complex also revealed, that ML-2(BF4)2 from the complex-to-complex transformation shows no more magnetic moment, since the same signal for t-BuOH from the internal and external reference was found ( Figure S36). This complex-to-complex transformation is indicated by a significant change in colour. Starting from the bright red ML-1(BF4)2 complex solution, the transformation yields a dark purple solution of the analogues diamagnetic complex ML-2(BF4)2 ( Figure  S37). Figure S37. Colour change during the complex-to-complex transformation of ML-1(BF4)2 to ML-2(BF4)2. The reaction was carried out with 6 equivalents of aldehyde 3 at room temperature and the pictures show the reaction progress over 3 hours.
The reaction progress can be tracked by measuring a UV-vis spectrum of the reaction mixture every 15 minutes ( Figure S38). The increasing signal at 580 nm can be assigned to the 1 A1  1 T1 transition of iron(II) cations 20 of the diamagnetic metalloligand ML-2(BF4)2 and therefore can be used to show the reaction progress. Scheme S3. Complex-to-complex transformation from ML-1 to ML-2 using 3 equiv of 3.
ML-1(BF4)2 (5.00 mg, 5.46 µmol, 1.00 equiv) together with 2-formyl-5-(4'-pyridyl) pyridine 3 (3.03 mg, 16.40 µmol, 3.00 equiv) was dissolved in 0.7 mL acetonitrile-d3. The solution was degassed by applying a vacuum and flushing with argon three times. The reaction was indicated by a change of colour from bright red to dark purple only few minutes after starting the reaction. To reach equilibrium the reaction mixture was heated to 40 °C for 36 hours. Then, a 1 H-NMR spectrum was measured to evaluate the reaction progress without any work up. In order to isolate ML-2(BF4)2 the reaction mixture was added drop wise to 25 mL of diethyl ether. The resulting suspension was stirred for 30 minutes at room temperature and then the purple precipitant was collected and washed with generous amounts of diethyl ether. Drying in a stream of air gave ML-2(BF4)2 as a purple solid in 95% yield (4.53 mg, 5.18 µmol).   (400 MHz, acetonitrile-d3, 298 K) from the complex-to-complex transformation from ML-1 to ML-2 (Scheme S3).
Comparison of the 1 H-NMR spectra of ML-2(BF4)2 and the spectrum received from the isolated complex of the complex-tocomplex transformation according to Scheme 3 indicates the successful formation of ML-2(BF4)2 ( Figure S41). An Evans' experiment with the isolated complex also revealed, that ML-2(BF4)2 from the complex-to-complex transformation shows no more magnetic moment, since the same signal for t-BuOH from the internal and external reference was found ( Figure S42). Figure S42. 1 H-NMR spectrum (400 MHz, acetonitrile-d3, 298 K) of the Evans' experiment with isolated ML-2(BF4)2 from the complex-to-complex transformation from ML-1 to ML-2 (Scheme S3).
BP-1(OTf)6(BF4)4 (11.69 mg, 2.73 µmol, 1.00 equiv) together with 2-formyl-5-(4'-pyridyl) pyridine 3 (6.03 mg, 32.70 µmol, 12.00 equiv) was dissolved in 0.7 mL acetonitrile-d3. The solution was degassed by applying a vacuum and flushing with argon three times. The reaction was indicated by a change of colour from bright red to dark blue only few minutes after starting the reaction. To reach equilibrium the reaction mixture was heated to 40 °C for 16 hours. Then, a 1 H-NMR spectrum was measured to evaluate the reaction progress without any work up. In order to isolate BP-2(OTf)6(BF4)4 the reaction mixture was added drop wise to 25 mL of diethyl ether. The resulting suspension was stirred for 60 minutes at room temperature and then the blue precipitant was collected and washed with generous amounts of diethyl ether. Drying in a stream of air gave BP-2(OTf)6(BF4)4 as a blue solid in 97% yield (11.13 mg, 2.64 µmol).  Comparison of the 1 H-NMR spectra of BP-2(OTf)6(BF4)4 and the spectrum received from the isolated complex of the complexto-complex transformation according to scheme 4 indicates the successful formation of BP-2(OTf)6(BF4)4 ( Figure S45). An Evans' experiment with the isolated complex also revealed, that BP-2(OTf)6(BF4)4 from the complex-to-complex transformation shows no more magnetic moment, since the same signal for t-BuOH from the internal and external reference was found ( Figure S46). from the complex-to-complex transformation from BP-1 to BP-2 (Scheme S4).
This complex-to-complex transformation is indicated by a significant change in colour. Starting from the bright red paramagnetic complex solution, the transformation yields a dark blue solution of the analogues diamagnetic complex ( Figure  S47). Figure S47. Colour change during the complex-to-complex transformation of BP-1(OTf)6(BF4)4 to BP-2(OTf)6(BF4)4. The reaction was carried out with 12 equivalents of aldehyde 3 at room temperature and the pictures show the reaction progress over 11.5 hours.
The reaction progress can be tracked by measuring a UV-vis spectrum of the reaction mixture after different reaction times ( Figure S48). The increasing signal at 590 nm can be assigned to the 1 A1  1 T1 transition of iron(II) cations of the diamagnetic bipyramid BP-2(OTf)6(BF4)4 and therefore can be used to show the reaction progress. Scheme S5. Complex-to-complex transformation from BP-1 to BP-2 using 6 equiv of 3.
BP-1(OTf)6(BF4)4 (11.69 mg, 2.73 µmol, 1.00 equiv) together with 2-formyl-5-(4'-pyridyl) pyridine 3 (3.02 mg, 16.40 µmol, 6.00 equiv) was dissolved in 0.7 mL acetonitrile-d3. The solution was degassed by applying a vacuum and flushing with argon three times. The reaction was indicated by a change of colour from bright red to dark blue only few minutes after starting the reaction. To reach equilibrium the reaction mixture was heated to 40 °C for 36 hours. Then, a 1 H-NMR spectrum was measured to evaluate the reaction progress without any work up. In order to isolate metal-organic compounds the reaction mixture was added drop wise to 25 mL of diethyl ether. The resulting suspension was stirred for 60 minutes at room temperature and then the bluish precipitant was collected and washed with generous amounts of diethyl ether. Drying in a stream of air gave 11.18 mg of a mixture of BP-2(OTf)6(BF4)4 together with side products of unknown incorporated proportions of 1 and 3 as a blueish solid.  Using only six equivalents of 3 to transform the paramagnetic bipyramid BP-1 into the diamagnetic bipyramid BP-2 does not lead to a quantitative reaction. About 20% of subcomponent 3 is still not incorporated into the aggregate after 36 hours, leading to some background signals of undefined structures. However, the equilibrium of the reaction is on the side of BP-2.  acetonitrile-d3, 298 K) isolated product mixture from the complex-to-complex transformation from BP-1 to BP-2 (Scheme S5).
The blue-green lines are guides to the eye.
Comparison of the 1 H-NMR spectra of BP-1, BP-2 and the product mixture received from the complex-to-complex transformation from BP-1 to BP-2 according to scheme S5 shows, that after the transformation no more signals of paramagnetic species were recorded ( Figure S51) and BP-2 was formed as a product in the reaction ( Figure S52). However, the spectrum of the product mixture clearly shows that there is a significant amount of background signals, most likely resulting from cages with unknown proportions of 1 and 3 incorporated. This is also corroborated by the finding that 20% of 3 was not incorporated into the cages after 36 hours ( Figure S50). The Evans' experiment with the isolated product mixture revealed a molar magnetic susceptibility multiplied with temperature of 1.38 cm 3 K mol -1 (Table S7, Figure S53) after the transformation, proving the existence of a small amount of paramagnetic species. Please note that for this calculation the molar mass of BP-2(OTf)6(BF4)4 was used, since the composition of the side products in unknown. Therefore the calculated value for XmT is not an exact value for the magnetic susceptibility. Table S7: Magnetic susceptibility of the isolated product mixture from the transformation from BP-1 to BP-2 according to Scheme S5 in acetonitrile-d3 as determined by the Evans' method. *The concentration was calculated using the molar mass of BP-2(OTf)6(BF4)4.  Figure S53. Evans' experiment (300 MHz, acetonitrile-d3, 298 K) of the isolated product mixture from the complex-to-complex transformation according to Scheme 5.
CU-1(BF4)28 (6.82 mg, 0.68 µmol, 1.00 equiv) together with 2-formyl-5-(4'-pyridyl) pyridine 3 (6.03 mg, 32.70 µmol, 48.00 equiv) was dissolved in 0.7 mL acetonitrile-d3. The solution was degassed by applying a vacuum and flushing with argon three times. The reaction was indicated by a change of colour from bright red to dark blue only few minutes after starting the reaction. To reach equilibrium the reaction mixture was heated to 50 °C for 3 days. Then, a 1 H-NMR spectrum was measured to evaluate the reaction progress without any work up. In order to isolate metal-organic compounds the reaction mixture was added drop wise to 25 mL of diethyl ether. The resulting suspension was stirred for 30 minutes at room temperature and then the bluish precipitant was collected and washed with generous amounts of diethyl ether. Drying in a stream of air gave 6.18 mg of a mixture of CU-2(BF4)28 together with side products of unknown incorporated proportions of 1 and 3 as a bluish solid. The product was hardly soluble in acetonitrile. The spectrum in Figure S55 shows that the complex-to-complex transformation from CU-1 to CU-2 did not lead to a quantitative subcomponent exchange, although 48 equivalents of aldehyde 3 were used. The equilibrium ratio between 1 and 3 was determined to be 40:60. As consequents some background signals that do not refer to 1, 3 nor CU-2 are observed and probably belong to heterometallic species with unknown proportions of 1 and 3. Please note, that one would expect a 2:1 ratio of 3:1 if the exchange would happen in a statistical manner without any driving force. Therefore, this experiment shows that incorporation of 3 into the cubic assembly is slightly more favoured than incorporation of 1, but less pronounced than in the case of the bipyramidal assemblies. Comparison of the 1 H-NMR spectrum from the isolated product (Scheme S6) with the spectrum from CU-2 ( Figure S56) shows that CU-2 was formed as the main product in this transformation. However, there is a significant amount of background signals, most probably from species with unknown proportions of 1 and 3 incorporated. Unfortunately, we could not perform an Evans' experiment using the isolated complex mixture, due to the very low solubility of the complex in acetonitrile. Figure S57 shows that the complex concentration is not sufficient to observe shifted t-BuOH signals. Figure S57. Evans' experiment (300 MHz, acetonitrile-d3, 298 K) of the isolated product mixture from the complex-to-complex transformation according to Scheme 6.
ML-1(BF4)2 (5.00 mg, 5.46 µmol, 8.00 equiv) and tetrakis(acetonitrile)palladium(II) tetrafluoroborate ([Pd(CH3CN)4](BF4)2; 1.82 mg, 4.09 μmol, 6.00 equiv) in 0.7 mL of acetonitrile-d3 was degassed by applying a vacuum and flushing with argon three times and heated under an argon atmosphere at 50 °C for 16 hours. Then, the reaction mixture was cooled down to room temperature and 2-formyl-5-(4'-pyridyl) pyridine 3 (12.66 mg, 68.20 µmol, 100.00 equiv) was added. The solution was degassed again. The reaction was indicated by a change of colour from bright red to dark blue only few minutes after starting the reaction. To reach equilibrium the reaction mixture was heated to 65 °C for 3 days. Then, a 1 H-NMR spectrum was measured to evaluate the reaction progress without any work up. In order to isolate metal-organic compounds the reaction mixture was added drop wise to 25 mL of diethyl ether. The resulting suspension was stirred for 30 minutes at room temperature and then the bluish precipitant was collected and washed with generous amounts of diethyl ether. Drying in a stream of air gave 6.03 mg of a mixture of CU-2(BF4)28 as main product together with side products of unknown incorporated proportions of 1 and 3 as a bluish solid. The product was hardly soluble in acetonitrile.  complex-to-complex transformation from CU-1 to CU-2 using 100 equiv of 3; Bottom (500 MHz, acetonitrile-d3, 298 K) CU-2 not measured as a wide sweep spectrum.
The complex-to-complex transformation of CU-1 to CU-2 using 100 equiv of aldehyde 3 yielded a conservatively estimated ratio of 1 to 3 of 30:70 ( Figure S59), corresponding to a nearly quantitative transformation with an expected ratio of 24:76. We note that we observed precipitation of a dark blue compound during the transformation experiment -this is most likely the diamagnetic CU-2. However, the precipitate might also contain small amounts of the less soluble aldehyde 3 which might lead to an overestimation of the relative amount of the better soluble aldehyde 1. Please note that we could not perform experiments with higher amounts of 3 due to its limited solubility. We do not observe any signals referring to CU-1 after the reaction ( Figures S60 and S61). Comparison of the 1 H-NMR spectrum from the isolated product (Scheme S7) with the spectrum from CU-2 ( Figure S62) shows that CU-2 was formed as the main product in this transformation. However, there is a minor amount of background signals that could be the result of the low solubility of CU-2 in acetonitrile and an overestimation of impurities. Unfortunately, we could not perform an Evans' experiment using the isolated complex mixture, due to the very low solubility of the complex in acetonitrile. Figure S63 shows that the complex concentration is hardly sufficient to observe shifted t-BuOH signals. However, from the aldehyde ratio observed in the previous 1 H-NMR spectrum ( Figure S59) we would not expect to observe a splitted t-BuOH signal, due to the nearly quantitative consumption of aldehyde 3. Figure S63. Evans' experiment (300 MHz, acetonitrile-d3, 298 K) of the isolated product mixture from the complex-to-complex transformation according to Scheme 7.
ML-1(BF4)2 (8.00 mg, 8.73 µmol, 8.00 equiv.) together with tetrakis(acetonitrile)palladium(II) tetrafluoroborate ([Pd(CH3CN)4](BF4)2; 2.91 mg, 6.55 μmol, 6.00 equiv) in 0.7 mL of acetonitrile-d3 was degassed by applying a vacuum and flushing with argon three times and heated under an argon atmosphere at 50 °C for 16 hours to assemble CU-1. After cooling to room temperature 1,3-bis(diphenylphosphino)propane (dppp; 2.70 mg, 6.55 µmol, 6.00 equiv) was added. The resulting solution was degassed again three times and heated under an argon atmosphere at 65 °C for 16 hours. Then, a 1 H-NMR spectrum was measured to evaluate the reaction progress without any work up. In order to isolate the product mixture, the solution was added to 25 mL of diethyl ether drop wise and was stirred for 30 minutes. Then, the red precipitant was collected, washed with generous amounts of diethyl ether and dried in air. The product mixture was obtained as a red solid (11.6 mg). We observed the formation of a considerable amount of insoluble red precipitate, when we heated the solution to 50 °C after addition of dppp, probably due to the formation of non-discrete oligomers. Subsequent heating to 65 °C led to even more precipitation and non-defined signals in the 1 H-NMR spectrum. Immediate heating of the solution to 65 °C instead, yielded a clear red solution ( Figure S64).  Figure S65 shows that the wide sweep 1 H-NMR spectrum of the reaction solution 16 hours after addition of 6 equivalents of dppp and after heating to 65 °C does not show any signals referring to the initial cage CU-1 anymore. Instead, the observed spectrum is an overlay of the 1 H-NMR spectra of the bipyramidal cage BP-1 and metalloligand ML-1. Please note that we did not observe any influences of the counter anions on the 1 H-NMR spectrum of BP-1. Therefore we used the 1 H-NMR spectrum of BP-1(OTf)6(BF4)4 for this stack.

Figure S64.
Since the spectrum we received from this transformation is a fitting overlay of the 1 H-NMR spectra of ML-1 and BP-1, the magnetic susceptibility should be the same as found with the pure complexes. To further prove that we performed an Evans' experiment with the 1:2 mixture of BP-1 and ML-1 from the complex-to-complex transformation. The calculated value for XmT (per iron(II) cation) is 3.06 cm 3 K mol -1 , which fits the theoretical value of 3.001 cm 3 K mol -1 per uncoupled iron(II) cation 5 in the high-spin state, showing that the magnetic properties are maintained in this reaction (Table S8, Figure S66). Table S8: Magnetic susceptibility of the isolated product mixture from the transformation from CU-1 to BP-1 and ML-1 (in a 1:2 ratio) according to Scheme S8 in acetonitrile-d3 as determined by the Evans' method. *The concentration is given as the concentration of iron(II) cations.  Figure S66. Evans' experiment (300 MHz, acetonitrile-d3, 298 K) of the isolated product mixture from the complex-to-complex transformation from CU-1 to BP-1 and ML-1 according to Scheme S8.
ML-2(BF4)2 (8.00 mg, 9.15 µmol, 8.00 equiv.) together with tetrakis(acetonitrile)palladium(II) tetrafluoroborate ([Pd(CH3CN)4](BF4)2; 3.05 mg, 6.86 μmol, 6.00 equiv) in 0.7 mL of acetonitrile-d3 was degassed by applying a vacuum and flushing with argon three times and heated under an argon atmosphere at 65 °C for 5 days to assemble CU-2. After cooling to room temperature 1,3-bis(diphenylphosphino)propane (dppp; 2.83 mg, 6.86 µmol, 6.00 equiv) was added. The resulting solution was degassed again three times and heated under an argon atmosphere at 65 °C for 16 hours. Then, a 1 H-NMR spectrum was measured to evaluate the reaction progress without any work up. In order to isolate the product mixture, the solution was added to 25 mL of diethyl ether drop wise and was stirred for 30 minutes. Then, the bluish precipitant was collected, washed with generous amounts of diethyl ether and dried in air. The product mixture was obtained as a dark bluish solid 11.3 mg).

Figure S67.
1 H-NMR spectra of: Top (500 MHz, acetonitrile-d3, 298 K) ML-2; Second from top (700 MHz, acetonitrile-d3, 298 K) BP-2; Second from bottom (500 MHz, acetonitrile-d3, 298 K) CU-2 + 6 equiv. dppp after 16 hours at 65 °C; Bottom (500 MHz, acetonitrile-d3, 298 K) CU-2. The blue dotted lines are guides to the eye in order to show the absence of signals referring to CU-2 after the complex-to-complex transformation. Figure S67 shows that after 16 hours at 65 °C no signals referring to the cube can be observed anymore, proving that CU-2 was consumed quantitatively in the exchange reaction. Only the superimposed spectra of BP-2 and ML-2 were detected, showing the transformation to these complexes. The assembly process of CU-2 takes 5 days, whereas the formation of BP-2 is finished after only 16 hours, which shows that the metalloligand might be better preorganized to build up the bipyramidal cage, compared to cubic cages. This might be an additional driving force for this ligand exchange reaction. We also investigated the temperature dependent behaviour of the reaction mixture containing ML-2 and BP-2 ( Figures S68  and S69). Figure S68. Temperature-dependent 1 H-NMR spectra (500 MHz, acetonitrile-d3) of CU-2 + 6 equivalents dppp, showing signals of ML-2 and BP-2. Figure S69. Comparison of 1 H-NMR spectra of ML-2, BP-2 and CU-2 + 6 equivalents dppp (500 MHz, acetonitrile-d3) at selected temperatures. Figure S68 shows that upon heating the complex solution containing BP-2 and ML-2 the signals experience a significant lowfield shift above room temperature. We attribute this behaviour to a beginning spin crossover process, which was already reported with other electron deficient 2-formylpyridine derivatives. 21 The electron withdrawing 4-pyridyl substituent in this work is comparable to a fluoride substituent in the same position. The reduced electron density of the ligand system facilitates the stabilization of the paramagnetic high-spin state. However, in our case we only observe the very beginning of this spin crossover process with the imine signals shifted to only 12 ppm at 343 K ( Figure S70). As observed with the sterically strained complexes in this work one would expect this signal shifted to 250 ppm in case of a purely paramagnetic complex. Therefore the observations with ML-2 and BP-2 only show that a minor amount of iron(II) centres switched into the paramagnetic highspin state at 343 K. Also, figure S69 shows that the mixture of BP-2 and ML-2 shows the same behaviour as pure samples of these complexes.