Modulation of the CO2 fixation in dinickel azacryptands†

While bimetallic azacryptands are known to selectively coordinate CO2, there is little knowledge on how different substitution patterns of the azacryptand cage structure influence CO2 coordination. Stoppedflow UV-vis spectroscopy, electrochemical analysis and DFT calculations were performed on a series of dinickel azacryptands and showed different rates of CO2 coordination to the complexes. We herein present data showing that the different flexibility of the azacryptands is directly responsible for the difference in the CO2 uptake capability of dinickel azacryptand complexes.


Introduction
The fixation and utilization of CO 2 as a C1-building block is an important research field towards the recycling of the potent greenhouse gas CO 2 . 1 While enzymes like CO-dehydrogenases 9 and ureases 10 allow selective CO 2 fixation under mild and aqueous conditions, 7,8 they are not cost effective for industrial use.Therefore, it is vital to develop materials that are as selective and efficient as enzymes but at a much lower cost.Indeed a number of synthetic approaches for the fixation and transportation of CO 2 have been reported, such as metal organic frameworks (MOFs), 2 covalent organic frameworks (COFs) 3,4 or even inorganic carbonates. 5,6Although they show promising properties, most of them have low selectivity towards CO 2 in the presence of other atmospheric gases and also low stability in the presence of moisture. 7,8Cryptands, such as bis-Tren azacryptands (Tren = tris(2-aminoethyl)amine), have been shown to allow selective fixation and transportation of small molecules (e.g.][11][12][13] By using cryptands, nowadays frequently used for anion recognition as well as in metal chelation, an attempt for a comparable strategy for CO 2 fixation was made. 146][17][18] As a result, the uptake and binding properties of small molecules (e.g.halogenides and pseudohalogenides) can be selectively tuned by increasing the size of the cryptand and/or altering the binding motifs.The size of the binding cavity of cryptands for a potential small molecule to enter can be rationally designed by using different linker molecules connecting both Tren-moieties (Scheme 1). 19Along this line, Nelson et al. recently showed crystallographic evidence for different CO 2 coordination in dicobalt-azacryptand complexes. 20 ](ClO 4 ) 3 was provided by Chen as well as Mooney et al., where they highlighted the necessity of an additional hydroxyl group on one metal site for successful CO 2 uptake. 19,21 Notably, alteration of the linker not only had an effect on the substrate binding but also had a strong influence on the metal binding strength as was shown for L A Thio , L A Fur and L A

Py
. 22 We recently showed that [Ni 2 L A H ](Cl)(ClO 4 ) 3 is capable of performing rapid CO 2 uptake from air (k = 0.067 ± 0.005 M −1 s −1 ) and was able to reversibly bind CO 2 by substitution with azides. 23urthermore, we could show that the azide ligand could be replaced by atmospheric CO

Synthesis and characterization
4][25][26] In a first step, (ClO 4 ) 2 in 33% yield by a Ni-catalyzed imine hydrolysis in the presence of water of crystallization (Scheme 2). 27The molecular structure was unequivocally confirmed by X-ray crystallography (Fig. S3 †).Similar decomposition results were also observed for other hexa-imine azacryptands by ESI-MS.The only exception was L I OH,Me , which afforded a mononuclear complex when reacted with MnCl 2 ; similar results are reported in the literature. 28ue to the instability of the imine complexes in the presence of moisture, we did not further investigate the imines towards the possibility of CO  shared structural motif of the above-mentioned complexes. 23ikewise, ESI-MS experiments support a similar composition of the compounds by showing comparable mass patterns (Fig. S6 †).Structural analysis revealed that one of the two Ni(II)-centers is coordinated by a water molecule and an acetonitrile, while the other is coordinated to a chloride. 23

Kinetic analysis
To further evaluate the differences of the azacryptand platform we performed UV-vis stopped-flow investigations.We expected a significant alteration of the CO 2 uptake kinetics with different substitution patterns groups according to L H > L F > L OMe > L Me ≫≫ L tBu .In addition, the application of the furan linker molecules results in a significant decrease of the rate of CO 2 uptake.This kinetic trend is valid for all temperatures measured (Tables S2 and  S3 †).The small ΔH ‡ and ΔS ‡ values obtained from the Eyringplots indicate that the coordination of CO 2 in all substrates proceeds in more than one step. 23Therefore these small values reflect a more complex association of CO 2 and thus preequilibrium and activation enthalpies compose the apparent ΔG ‡ value.Comparable changes of the reaction rates upon alteration of the reactive site environment were previously reported by Holm and co-workers on [Ni II ( pyN 2 R2 )(OH)] − . 31In contrast, we present an example that exhibits no obvious alteration of the steric bulk on the metal center.

Azide fixation
A likely explanation for the alteration of the CO 2 uptake kinetics is a decisive change of the Ni-Ni distance and cavity size due to the influence of the linker.A similar hypothesis was reported by Nelson and co-workers. 32Likewise, the particular shape of the formed HCO 3 − anion can play a significant part in the destabilization within the dinickel complex as one C-O bond is directed towards the opening of the cavity and can interfere with the ligand periphery.As such, the uptake of linear molecules, e.g.azides, should not be dramatically influ-enced by the alteration of the substitution pattern.11]23 Even when no CO 2 binding was observed, such complexes, e.g.(ClO 4 ) 3 (6.129Å) and [Ni 2 L A tBu (N 3 )](ClO 4 ) 3 (6.119Å). 23,33 The general coordination of azides within the cavity and the alteration of the Ni-Ni distances within structurally comparable metal complexes underline the influence of the substitution pattern.It also shows that CO 2 is a key component in the different uptake kinetics.Furthermore, the successful incorporation of negatively charged azides additionally shows that the lone pairs of the furan, pyridine or thiophene linker cannot be a major reason for the weak or no CO 2 binding in ](Cl)(ClO 4 ) 3 , respectively.In light of the acidic properties of CO 2 in an aqueous environment, changes of the substitution pattern might also alter the basicity of the coordinating N-donors and thus the nucleophilicity of the metal atoms.This hypothesis, however, has to be ruled out since the redox potentials of the complexes did not show a trend when electron withdrawing groups (e.g.F) or electron donating groups (e.g.t Bu, Me) were installed.For all complexes, multiple irregular electron transfer steps can be observed at ∼1.5 V vs. Fc/Fc + , which we were not able to assign (Fig. S13 †).

Theoretical analysis
In order to rationalize the experimentally observed differences in CO 2 binding of [Ni 2 L A R ](Cl)(ClO 4 ) 3 complexes, DFT calcu-    lations were performed for complexes with R = H, F, Me and t Bu (Fig. 5a).Both Ni 2+ centers were found to be in the highspin state (S = 1) consistent with the observed octahedral and trigonal-bipyramidal coordinations of the Ni 2+ centers and this finding is in line with previous SQUID measurements. 23The two triplet states were found to be exchange-uncoupled by broken-symmetry calculations.Though the substituents differ in their electron-donating and withdrawing capacities, no electronic effect was observed at the nickel ions as well as at all amines, as became apparent from unchanged Mulliken charges, bond distances and orbital compositions.The latter is in-line with the experimental observation that no clear correlation between the CO 2 uptake kinetics and electron donating capacity of the substituent could be found.
However, a noticeable steric effect was observed in the calculations in that rotation of the phenyl groups leads to a steric clash with the bulky t Bu-substituent (Fig. 5b) while the smaller CH 3 groups allow larger rotational flexibility of the ligand.For R = H and F, the phenyl rotation is essentially unhindered (Fig. 5c).
The bulkiness and flexibility of the substituent correlates with the observed CO 2 uptake kinetics, suggesting that these two factors help in tuning the kinetics and that it is most likely the rate-determining step in the reaction mechanism.Moreover, since the electronic structure at the nickel centers and the amines are the same for all of the complexes, every complex should in principle be able to take up and convert CO 2 .This was tested by performing an additional calculation in which Cl − and MeCN solvent molecules were removed, hydroxide was inserted at the position where the crystal structure contains a H 2 O molecule and CO 2 was introduced near the Ni-ions.Geometry optimization indeed leads to a barrier-less formation of HCO 3 − with the driving force being C-O bond formation.
An intermediate structure of this mechanism is shown in Fig. 5d, which is in agreement with the mechanism proposed for copper cryptates. 19Of note here is that contrary to the experimental findings, the Ni-Ni distance is largely independent of the substituent and ranges from 6.09 to 6.

Conclusion
The coordination of CO 2 in dinickel azacryptands can be manipulated through the presence of different linker molecules comprising Tren cages.UV-vis spectroscopic analyses, as well as ESI-MS analyses clearly show an influence of different functional groups on the CO 2 uptake.Functional groups pointing into the cryptand cavity, as in L A Fur , L A Py or L A Thio , and L A OH,Me significantly slow down or even prohibit a coordination of CO 2 .In contrast to this, functional groups pointing out of the cavity show an increasing CO 2 -fixation rate with decreasing steric demand ( Both DFT calculations and cyclic voltammetry demonstrate that there are no electronic effects at the nickel centers as a result of the different substituents.Therefore, we attribute the observed changes in reactivity to structural changes.Furthermore, the DFT calculations performed herein show that with increasing steric demand of the linker, the flexibility of the azacryptand core is decreased, providing a kinetic barrier to the initial coordination of CO 2 .In contrast to the binding of CO 2 , all dinickel complexes show fixation of azides.The results clearly show that controlling the flexibility of the cryptand can regulate binding of different substrates.With this in hand, new applications might be accessible for azacryptands, e.g.within catalysis or gas separation utilizing cryptands as the ligand platform.

General techniques
All reactions were performed under either a dry N 2 atmosphere using standard Schlenk techniques or in a glovebox.All solvents were dried according to standard methods. 1 H, 13 C NMR spectra were recorded on a Bruker DPX-200 NMR, Bruker DPX-250 NMR or a DPX-400 NMR spectrometer at room temperature.Peaks were referenced to residual 1 H signals from the deuterated solvent and are reported in parts per million ( ppm).IR spectra were measured with a Bruker Tensor 27 FT-IR spectrometer as a KBr pellet and are reported in cm −1 .Mass spectra were measured with a Shimadzu QP-2010 instrument.The dialdehydes 17,[34][35][36][37] as well as the azacryptands , 26 L A Py , 20 and L I OH,Me 31 were synthesized according to literature procedures.All other chemicals were used as received from commercial vendors.Caution!Perchlorate salts of metal complexes with organic ligands are potentially explosive.They should be handled with care, and prepared only in small quantities.

X-ray data collection and structure solution refinement
Single crystals suitable for X-ray analysis were coated with Paratone-N oil, mounted on a fiber loop, and placed in a cold, gaseous N 2 stream on the diffractometer.Diffraction intensities were measured using graphite-monochromatic Mo Kα radiation (λ = 0.71073 Å).Data collection, indexing, initial cell refinements, frame integration, final cell refinements, and absorption corrections were accomplished with the program CrysAlis Pro (Agilent Technologies, Version 1.171.37.34, 2014) and X-Area, respectively.Space groups were assigned by analysis of the metric symmetry and systematic absences (determined by XPREP) and were further checked by PLATON 38,39 for additional symmetry.Structures were solved by direct methods and refined against all data in the reported 2Θ ranges by full-matrix least squares on F 2 with the SHELXL program suite 40,41 using the OLEX2 interface. 42The program PLATON SQUEEZE was used for the structures L A Py and [Ni 2 L A Thio ](ClO 4 ) 4 to eliminate non-refinable solvent molecules. 43Crystallographic data as well as refinement parameters are presented in Tables S4-S7 in the ESI.†

Stopped-flow measurements
Time-dependent spectrophotometry was measured with a UV-Vis spectrophotometer S600 from Analytik Jena and a SFA-20 Rapid Kinetics Accessory from Hi-Tech Scientific.
Temperature control was obtained with an attached cryostat and a cuvette-holder with a temperature-unit.The used MeCNsolutions were prepared from a stock-solution of MeCN saturated with CO 2 ([CO 2 ] 298K = 0.28 mol L −1 ) 44 and degassed MeCN.The complex was synthesized in situ in degassed MeCN under an N 2 atmosphere.

Electrochemical analysis
The electrochemical studies were performed on a Gamry Reference 600 in 100 mM tetrabutylammonium hexafluorophosphate (TBAPF 6 ) as a supporting electrolyte, 20 mM Ni(ClO 4 ) 6 •6H 2 O and 10 mM L A R in degassed MeCN.Glassy carbon, Pt wire and Ag|AgNO 3 (10 mM) in MeCN were used as working, counter and reference electrodes respectively.Cyclic voltammograms (CV) were recorded between −2.0 and +1.5 V at 100 mV s −1 in degassed MeCN and after CO 2 purging.The working electrode was polished with alumina paste 0.3 µm (Buehler) before each measurement.The solutions were purged for 10, 20, 60 and 120 seconds with CO 2 .The results are reported versus Fc/Fc + .

DFT calculations
All calculations have been performed with the ORCA program. 45The BP86 functional 46 was used along with the Def2-svp basis set. 47The resolution of the identity (RI) approximation has been employed to speed up the calculation time. 48,49Scalar relativistic effects are included in zero order regular approach (ZORA). 50,51Solvent effects were taken into account by using the COSMO solvation model. 52neral synthetic procedure for L I R In a typical experiment, the respective dialdehyde (3. Scheme 1 Azacryptands L A R and imines L I R with different linker molecules L R .The linkers are arranged according to their (i) different steric bulks on the central benzyl unit, (ii) capability to directly alter the electron density within the azacryptand cavity and (iii) different cage sizes. 5,14,15,17,20† Electronic supplementary information (ESI) available: Synthesis and characterization of compounds, X-ray crystallographic analysis, UV-Vis spectra, SQUID and kinetic data.CCDC 1517780-1517787 and 1517950.For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c6dt04527h [2 + 3]-Schiff-base condensation of the respective dialdehydes and Tren afforded the imines L I R in good yields (57-77%).The reaction of L I R with KBH 4 yielded the azacryptands L A R in good to excellent yields (66-99%).The molecular structures of the hexa-imines L I F , L I OMe and L I Me , and the hexa-amines L A Py and L A OH are presented in Fig. S1 and S2.† A simple way to investigate the influence of the linker molecule on the metal cryptand properties during CO 2 uptake is the application of the imine species L I R since they are structurally more rigid than their amine counterparts.We therefore attempted complex formation of the hexa-imines with Ni(ClO 4 ) 2 •6H 2 O.The reaction solely afforded [(Tren)Ni(CH 3 CN) 2 ]
revealed a different coordination behavior in the solid state where both Ni-atoms are coordinated on the outside of the cryptand cavity, as was reported for the structurally related complexes [Cu 2 L A Thio ] (O 3 SOCF 3 ) 2 and [Ag 2 L A Thio ](ClO 4 ) 2 . 29Each Ni(II)-center is octahedrally coordinated to only three N-donor atoms of the Trenmoiety as well as two additional acetonitriles and one water ligand.We assume that the different structure of [Ni 2 L A Thio ] (ClO 4 ) 4 compared to e.g.[Ni 2 L A H ](Cl)(ClO 4 ) 3 can be explained by a different metal binding affinity and therefore an altered complex stabilization, as was reported by Nelson and coworkers. 22We next investigated the CO 2 fixation behavior of all [Ni 2 L A R ](Cl)(ClO 4 ) 3 complexes by UV-vis spectroscopy and ESI-MS.A clear change in the UV-vis spectra upon purging the [Ni 2 L A R ](Cl)(ClO 4 ) 3 (R = H, F, OMe, Me and Fur) solutions with CO 2 is observed, showing a decrease in absorption intensity of the bands between 430-580 nm with the formation of an intense absorption band at about 610 nm (Fig. 2 and S4 †).The amplitude of the absorption band as well as the disappearance of the original bands between 430-580 nm depends on the substitution pattern at the linker unit.In analogy to our recent finding for [Ni 2 L A H ](Cl)(ClO 4 ) 3 , such changes can be attributed to the coordination of CO 2 within the cavity of the cryptand to afford a bicarbonate dinickel complex. 23ESI-MS analysis further supports the fixation of either 12 CO 2 or 13 CO 2 for the reported complexes by the appearance of the [Ni 2 L A R (HCO 3 )] mass-peak (Fig. S7 and S8 †). 23Notably, while the color changes of the complexes upon reaction with CO 2 are usually unclear from dark to light blue, [Ni 2 L A F ](Cl)(ClO 4 ) 3 reveals a distinct color change upon CO 2 fixation from blue to red (Fig. S9 and S10 †).In contrast, no apparent changes could be observed for the complexes comprising the L A Thio , L A OH , L A Py , or L A OH,Me moiety.Likewise, ESI-MS analysis solely revealed the mass peaks of the starting complexes.It can thus be assumed that these complexes do not possess the capability to fixate CO 2 under the described reaction conditions, although a small shift of the main band from 564 nm to 571 nm was observed in the UV-vis spectrum upon CO 2 addition to [Ni 2 L A H,para ](Cl)(ClO 4 ) 3 .ESI-MS showed a new mass-peak at m/z = 756, which clearly indicates a reaction of [Ni 2 L A H,para ](Cl)(ClO 4 ) 3 to afford a new complex.This behavior can most likely be attributed to the formation of [Ni 2 L A H,para (CN)](ClO 4 ) 3 comprising a bridging CN − -ligand but no coordinated bicarbonate.Further evidence for the presence of a CN − ligand was provided by IR spectroscopy showing a signal at 2022 cm Fig. 2 UV-vis spectra (MeCN/MeOH 4 : 1, RT) of the reaction of [Ni 2 L A R ] (Cl)(ClO 4 ) 3 with CO 2 : (a) [Ni 2 L A F ], (b) [Ni 2 L A OMe ], (c) [Ni 2 L A H,para ] and (d)

3 −
[Ni 2 L A tBu ](Cl)(ClO 4 ) 3 , allowed for rapid coordination of N between both nickel atoms.Correspondingly, we tested the capability of [Ni 2 L A R ](Cl)(ClO 4 ) 3 to allow azide coordination.All investigated azacryptands, except [Ni 2 L A OH,Me ](ClO 4 ) 2 , show fixation of N 3 − , which is obvious from the changes in their UVvis spectra by the formation of a new common absorption band at about 350 nm (Fig. S12 †). 23Additionally, ESI-MS analysis further confirms the formation of an azide complex and reveals the respective [Ni 2 L A R (N 3 )] mass peaks.Crystals suitable for X-ray crystallography were obtained for [Ni 2 L A F (N 3 )] (ClO 4 ) 3 (Fig. 4) and the results confirm the incorporation of N 3 − between the two Ni-centers.It is notable that the Ni-Ni distance (6.275 Å) is significantly larger than in [Ni 2 L A H (N 3 )] 10 −2 ± 2.7 × 10 −3

Fig. 3
Fig. 3 Plot of k obs vs. the CO 2 concentration of L A R in MeCN at 298.15 K for the reaction of [Ni 2 L A R ] with CO 2 .

Fig. 5
Fig. 5 (a) Geometry optimized structure of [Ni 2 L A F ] with Cl − , H 2 O and MeCN; (b) side view of [Ni 2 L A tBu ], showing the steric hindrance of t Bu with the neighboring phenyl group; (c) side view of [Ni 2 L A H ] where steric effects are absent; (d) intermediate structure with bent CO 2 where CO 2 has been introduced near Ni 2+ at 2.2 Å, leading to barrierless C-O bond formation towards HCO 3 − .