Synthesis and reactions of N-heterocycle functionalised variants of heterometallic {Cr 7 Ni} rings †

Here we present a series of linked cage complexes of functionalised variants of the octametallic ring {Cr 7 Ni} with the general formula [ n Pr 2 NH 2 ][Cr 7 NiF 8 (O 2 C t Bu) 15 (O 2 CR)], where HO 2 CR is a N-heterocycle containing carboxylic acid. These compounds are made by reacting [ n Pr 2 NH 2 ][Cr 7 NiF 8 (O 2 C t Bu) 15 (O 2 CR)] with a variety of simple metal salts and metal dimers. The carboxylic acids studied include iso-nicotinic acid, 3-(4-pyridyl)acrylic acid and 4-pyridazine carboxylic acid. These new linked cage complexes have been studied structurally and the study highlights the versatility of functionalised {Cr 7 Ni} as a Lewis base ligand. As {Cr 7 Ni} is a putative molecular electron spin qubit this work contributes to our understanding of the chemistry that might be required to assemble molecular spin qubits.


Introduction
Several strategies have been proposed to create qubits that could act as the fundamental unit of a quantum computer; these include quantum dots, 1 semiconductor nanowires, 2 large atomic ensembles, 3 ion traps, 4 defects in solids, 5 carbon nanotubes, 6 nuclear spin 7 and molecular nanomagnets. 8 Molecular nanomagnets have one major advantage over most other strategies, which is that through chemistry it is possible to engineer the interaction between individual qubits. We have been studying heterometallic rings as possible qubits and have shown they have reasonable coherence times, and that they can be linked so that spin on individual rings entangle with neighbours. 9 Here we report coordination chemistry to produce a series of functionalised rings.
Functionalisation of [ n Pr 2 NH 2 ][Cr 7 NiF 8 (O 2 C t Bu) 16 )] (hereafter {Cr 7 Ni}, Fig. 1) rings with N-heterocycle functional groups allows the ring to act as a Lewis base towards metal sites, essentially acting as a bulky N-donor ligand. 10 By synthesising the functionalised {Cr 7 Ni} rings prior to binding to another metal or simple dimer, we produce a controlled approach to synthesis of linked systems. This should allow the design of appropriate linkers with the capability of providing switchable communication between qubits. The functionalised rings have the general formula [ n Pr 2 NH 2 ][Cr 7 NiF 8 (O 2 C t Bu) 15 (O 2 CR)], herein {Cr 7 Ni(O 2 CR)}, where HO 2 CR is either iso-nicotinic acid (HO 2 C 5 H 4 N, herein HO 2 C-py), 3-(4-pyridyl)acrylic acid (HO 2 C 2 H 2 C 5 H 4 N, herein HO 2 C-4pyac) or 4-pyridazine carboxylic acid (HO 2 C 4 H 3 N 2 , herein HO 2 C-pyd). These complex N-donor ligands have remarkable solubility in non-polar solvents and are very sterically demanding. In the following we discuss the diversity of structures that can be formed from reactions with simple metal salts and metal dimers. The range of metal used represents a wide selection from across the periodic table including 3d, 4d and 5d transition metals with a range of anions, including chelating and bridging ligands. The resulting complexes are discussed in terms of their structure and crystal packing.

Experimental
Column chromatography was carried out using Silica 60A ( particle size 35-70 μm, Fisher, UK) as the stationary phase, and TLC was performed on pre-coated silica gel plates (0.25 mm thick, 60 F 254 , Merck, Germany). Elemental analyses were performed by departmental services at The University of Manchester. Carbon, nitrogen and hydrogen analysis was performed using a Flash 200 elemental analyser. Metal analysis was performed by Thermo iCap 6300 Inductively Coupled Plasma Optical Emission Spectroscopy (ICP-OES).

Starting materials
All reagents and solvents used to synthesise the functionalised rings and linkers were commercially available and used as )], 3, were synthesised by an analogous procedure to 1, the details of which are in the ESI. †

Coordination compounds with rings as ligands
The linked ring systems are obtained from the reaction of the corresponding N-heterocycle functionalized {Cr 7 Ni} ring, in slight excess, and the link of choice in an appropriate solvent, usually hot acetone or toluene. The reaction mixture is then stirred for anywhere between 5 and 20 minutes (with the exception of ReCl(CO) 5 and Rh 2 (O 2 CCH 3 ) 4 due to their inertness) during which time a precipitate may have formed. If a precipitate forms, the solution is cooled, the precipitate collected, washed with the same solvent used in the reaction and recrystallized from an appropriate solvent mix. If no precipitate forms the reaction solution is left to cool and in most cases the desired product will crystallise in 24-48 hours. An example is given for compound 4 below and in detail for all compounds in the ESI. † The reagents involved, yield and elemental analysis are listed in Table 1 O (0.011 g, 0.049 mmol) was added to a solution of 2 (0.3 g, 0.13 mmol) in hot acetone (30 mL) with stirring. The solution was refluxed for 5 minutes before cooling and left undisturbed in a sealed flask at room temperature. Dark green crystals suitable for single crystal XRD formed after 1 week and were collected by filtration and washed with acetone.

X-ray crystallography
Structures 2, 3, 6, and 17 were collected on a Bruker X8 Prospector 3-circle diffractometer with a copper microfocus source and an APEX II CCD detector. Structure 21 were collected on an Enraf Nonius FR590 4-circle diffractometer with a molybdenum sealed-tube, fine focus molybdenum source, structures 1, 5, 10, 11, 12, 14, 16, 18, 19 and 20 were collected on Advanced Light Source station 11.3.1. Structures were solved and refined using SHELX97 and SHELX-2013. Structures 4,7,8,9,12 and 15 have already been communicated previously. 9c,10d Full crystallographic details for all crystal structures are included in Table S1. † CCDC numbers 1424323-1424336 contain the supplementary crystallographic data for the new structures reported in this paper.

Results and discussion
In all the structures discussed here the heterometallic ring is unchanged from the parent unsubstituted {Cr 7 Ni} ring. 11 The ring contains an octagon of metal sites, with each edge bridged internally by a fluoride and by two carboxylates. On each edge, one carboxylate lies close to the plane of the metal ring while the second carboxylate is perpendicular to the plane of eight metals; this site is labelled as the axial carboxylate, with the in-plane carboxylates described as equatorial.
In the functionalised derivatives the nickel(II) site is always found in the edge that is substituted and is disordered between the two sites in that edge. The incoming carboxylate (e.g. iso-nicotinate) is found in an axial site in the molecules described below and this is always the more common isomer.   nicotinic acid provides the ring with a very sterically congested pyridyl moiety, with the nitrogen of the pyridyl barely protruding beyond the neighbouring pivalate groups. This provides a pathway for spin propagation from the ring to whatever it may bind through the conjugated π-system.  (Fig. 3). The introduction of the 3-(4-pyridyl)acrylate group provides a longer, less sterically demanding linker whilst still maintaining the potential for spin propagation from the ring to whatever it may bind to through the conjugated π-system.  (Fig. 4). The 4-pyridazine carboxylate group provides a very sterically congested pyridazine moiety thorough which spin propagation can occur with the potential for different coordination modes and strength of coupling compared to 1.

Linked {Cr 7 Ni} systems
The following structures are described using three metric parameters; the torsion angle between the centre of the Cr-Ni edges and the centre of each ring, the N-M-N bond angle (in the case of dimetallic linkers the centroid between the metals is used) and the deviation from co-planar of the two rings. Schematics for the torsion and deviation from co-planar metrics are shown in Fig. 5. If the torsion angle is 180°, the rings are considered staggered and if the angle is 0°they are considered eclipsed and co-planar if there is 0°deviation between the planes.

Simple metal salt linked structures
Reactions with group 11 metal nitrates.  (Fig. 6). The central copper(II) adopts a 7-coordinated geometry that could be described like a distorted pentagonal bipyramid, with two pyridine nitrogen atoms from two different iso-nicotinic groups of two 1 in a trans arrangement and two bidentate nitrate counter ions in a trans disposition and a water molecule completing the pentagonal basal plane. Similar coordination geometries have been seen previously, chiefly in coordination polymers. 14 The rings themselves adopt a staggered geometry with a torsion angle of 176.7(2)°. This is most likely due to the short distance between the rings binding through a single metal ion with an eclipsed geometry i.e. with 0°rotation resulting in a steric clash. The rings lie almost parallel with a deviation of 7.2(2)°from co-planarity, with the N-Cu-N angle being 176.3(10)°.
Reacting 1 (Fig. 7). 15 Unlike 4, the rings adopt an eclipsed geometry with a torsion angle of 8.3(2)°. However, the rings as are not parallel with an angle of 49.8(2)°from planar between the plane of the rings. The iso-nicotinates bind to the silver(I) with an N-Ag-N angle of 145.1 (6)°and N-Ag distances of N-Ag 2.246(15) Å and 2.284(15) Å.
The reaction of 2 with hydrated copper(II) nitrate gives [{Cr 7 Ni(O2C-4pyac)} 2 Cu(NO 3 ) 2 (OC 3 H 6 )] 6 (Fig. 8). The ethenelink to the pyridyl group in 2 leads to different crystal packing in 6 compared with 4; the rings adopt an eclipsed, rather than staggered geometry with a torsion angle of 19.49(2)°and with a 34.65(2)°deviation from co-planarity. The Cu-N bond distances of 2.146(6) Å for 6 are slightly longer than literature examples of similar motifs (the CSD average is 1.978(4) Å for   3-(4-pyridyl) acrylic moieties bound to {Cu(NO 3 ) 2 }). 14 As with 4, the central copper adopts a 7-coordinate geometry, with the rings binding in a trans disposition with two nitrates and a water molecule in the pentagonal basal plane. The reaction of 3 with copper(II) nitrate proved unsuccessful, which was unexpected as 4-pyradizine has previously been shown to coordinate to copper(II) nitrate. 16 Reactions of with M(II) acetylacetonate derivatives.  Fig. 9). In all cases the central metal lies on an inversion centre with a trans disposition of the ligands. The presence of electron withdrawing F atoms in the acac ligand leads to a shortening of the N-M bond distances from 2.104(7) and 2.122(7) in 7 to 2.075(9) in 8 and 2.074(9) Å in 8 and 9 respectively. In 7, the rings are not    The other examples of M(Hfac) 2 linked rings (10 and 11) crystallise in a triclinic P1 rather than monoclinic C2/c. They have the same disposition of rings as 9, with a torsion angle of 180°with 0°deviation from co-planar, and a N-M-N bond angle of 180. Interestingly, 10 does not appear to display a significant Jahn-Teller elongation, as might be expected from copper(II). However, this could be due to its position on an inversion centre averaging the Cu-O bond lengths. In any case, the Cu-N bond distance from the centre to the ring is shorter than the two Cu-O bond lengths, which could imply that the rings coordinate in the x-y plane of the central copper (Cu-N = 1.993(3) Å, Cu-O1 = 2.065(4) Å and Cu-O2 = 2.178(3) Å). In the case of 11, the central manganese adopts a slightly distorted octahedral geometry with the coordinating nitrogen atoms exhibiting a slightly longer bond length to the oxygen atoms of the Hfac (2.150 (3)  shows the Ni(Hfac) 2 adopting the cis conformation (Fig. 10), contrasting with the trans-geometry of 9. The rings are eclipsed, with a torsion angle of 26.93(3)°and the rings lie with a 53.73(3)°deviation from co-planar. The Ni-N distance is 2.073(6) Å and the N-Ni-N bond angle is 90.0(3)°. It is unclear why the 4-pyridazinecarboxylate favours a cis-geometry, which seems more sterically demanding. It is also intriguing that 3 seems to bind to Ni(II) but not Cu(II).
Reactions with 5d metals. We have also fully characterised one example of {Cr 7 Ni} bound to rhenium by reacting 2 with ReCl(CO) 5  forming in good yields (see Table 2). The water is displaced from the two water-bridged dimers. In all these compounds the N-donor from iso-nicotinate binds to the apical sites of the metal dimers. In the five crystal structures, three crystallise in monoclinic settings (15, 18 and 19) with similar unit cells and with half the assembly in the asymmetric unit. Compound 16, has the highest molecular symmetry with 1/4 of the assembly present in the asymmetric unit in the orthorhombic Pnnm space group. Compound 17, crystallises in tetragonal I4 1/a and features half the assembly in the asymmetric unit. The crystal     4 ] as linker (15)(16)(17)(18), the rings are perfectly staggered with respect to one another and are also completely co-planar (Fig. 12a). The only differences from 15 to 18 are the M-M bond distances, the M-N bond distances, the ring centroid distances and the ring torsion angle ( Table 2). For compound 19, where the link is [Rh 2 (O 2 CMe) 4 ], the rings are eclipsed with a torsion angle of 47.55°and a deviation of 50.48°from co-planarity (Fig. 12b). The N-M bond length is 2.223(7) Å, slightly shorter than the average found in the CSD of 2.231(15) Å for similar complexes. The N-M 2 -N bond angle is 177.26(4)°, measured through the dimer centroid and the M-M bond distance is 2.4038(10) Å, slightly longer than the CSD average of 2.398(4) Å found for related compounds. 18 Whilst 15, 18 and 19 exhibit similar M-M distances to other similar complexes found in the CSD, 16 and 17 exhibit a significant shortening of the M-M distances. In the case of 16, only one similar compound from the CSD has a bond length significantly less than 2.6 Å; one with a {Cr 7 Ni} bound through a pyridyl group on the thread. 10b It would appear that the bulky nature of the coordinating ligand forces the Ni⋯Ni and Co⋯Co bonds to shorten. For 15, favourable 5-coordinate geometry may explain a lack of shortening, while in the case of 18 and 19 the bond order of the metal-metal bond is presumably the controlling factor. For 16, 17 and 19 the central dimetallic link has all carboxylate ligands around 45°from the plane of the iso-nicotinate ligands, while in 15 and 18 two of the carboxylates are almost co-planar with the iso-nicotinates (Fig. 13 (Fig. 15). In 20, the rings bind to the apical positions of the copper sites and the packing is similar to that of 15, with the rings in a perfectly staggered geometry. However, the central bridging unit in 20 adopts the same as that in 16 and 17, with the bridging pivalates out of the plane of the pyridyl groups. The Cu-N bond length is 2.146(6) Å, which is similar to that in 15, but the Cu-Cu distance is 2.5865(14) Å, slightly shorter than 15 and the CSD average. 19   The structure of 21 is similar to that of 15, with the carboxylates of the central copper in plane with the pyridyl moieties and similar Cu-N bond lengths (2.15 (2) and 2.160(12) Å). The Cu-N bond distances are shorter than that for pyridazine bound to copper(II) acetate (2.211(3) Å), implying that 3 is a better donor than plain pyridazine. This could either be due to the presence of the ring or the carboxylate group in the paraposition. Unfortunately, no structure is known for 4-pyridazinecarboxylic acid bound to a dimetalic copper pivalate complex, so a direct comparison with the free carboxylic acid is not possible.

Dalton Transactions Paper
It adopts the same perfectly staggered geometry as 15 with the central dimer also adopting the same geometry with respect to the iso-nicotinate groups with a similar ring centroiddimer centroid angle of 77.13(3)°and a slightly longer Cu-Cu distance of 2.612(4) Å.
The focus of this paper is the structural chemistry of these materials. Magnetic studies of the substituted rings and linked arrays show variable temperature susceptibility behaviour that matches that of the sum of components. As we have reported previously 9c,10 interactions between such components tend to too weak to measure using conventional magnetometers and very low temperatures are required to measure the exchange interactions directly. Such measurements are only useful if new physics is to be uncovered. We have recently reported the use of double electron-electron resonance spectroscopy 20,21 to study such interactions in other supramolecular species; such studies will be undertaken for some of the compounds reported here in the future.

Conclusions and future perspectives
In summary, we have shown that N-heterocycle functionalised {Cr 7 Ni} rings can act as Lewis bases towards simple metal salts and metal carboxylate dimers to form linked ring systems. The simplicity of the method, using some of the most basic principles of simple coordination chemistry, coupled with a diverse range of metal salts and carboxylate dimers, means we can produce a range of coordination complexes where functionalised {Cr 7 Ni} act as bulky ligands. We have established that {Cr 7 Ni} rings functionalised with different N-heterocycles can all act as bulky Lewis bases, with some subtle variations based on the heterocycle present. This allows a degree of control over the synthesis of linked {Cr 7 Ni} systems and allows us to work towards the design and realisation of a quantum gate.
This approach to linking {Cr 7 Ni} furthers its potential for use as a qubit in quantum information processing and quantum simulation applications by addressing one DiVincenzo criteria that {Cr 7 Ni} does not fulfil. 22 Future work will look to the introduction of fully switchable linkers with a look to entanglement of linked {Cr 7 Ni} in order to implement quantum gate operations. We will also look at introducing hetero-functionality to the {Cr 7 Ni} in order to fulfil the schematic for a quantum simulator put forward by Santini et al. 23