Modulation of Mn3+ Spin State by Guest Molecule Inclusion

Spin state preferences for a cationic Mn3+ chelate complex in four different crystal lattices are investigated by crystallography and SQUID magnetometry. The [MnL1]+ complex cation was prepared by complexation of Mn3+ to the Schiff base chelate formed from condensation of 4-methoxysalicylaldehyde and 1,2-bis(3-aminopropylamino)ethane. The cation was crystallized separately with three polyatomic counterions and in one case was found to cocrystallize with a percentage of unreacted 4-methoxysalicylaldehyde starting material. The spin state preferences of the four resultant complexes [MnL1]CF3SO3·xH2O, (1), [MnL1]PF6·xH2O, (2), [MnL1]PF6·xsal·xH2O, (2b), and [MnL1]BPh4, (3), were dependent on their ability to form strong intermolecular interactions. Complexes (1) and (2), which formed hydrogen bonds between [MnL1]+, lattice water and in one case also with counterion, showed an incomplete thermal spin crossover over the temperature range 5–300 K. In contrast, complex (3) with the BPh4−, counterion and no lattice water, was locked into the high spin state over the same temperature range, as was complex (2b), where inclusion of the 4-methoxysalicylaldehyde guest blocked the H-bonding interaction.


Introduction
Crystalline forms of spin crossover (SCO) transition metal complexes are well known to be sensitive to lattice contents, and crystal engineering can be used to a good effect to modulate the thermal evolution pathway. This has been convincingly demonstrated in mononuclear complexes of Fe 2+ [1][2][3][4][5][6], Fe 3+ [7][8][9][10] and Co 2+ [11][12][13][14] and with SCO 1-D [15][16][17][18], 2-D [19][20][21][22][23] and 3-D [24][25][26][27] polymeric networks. Intermolecular interactions including hydrogen bonding [28][29][30] and π-π stacking [16,[31][32][33][34][35] can be used to connect SCO-active sites and to change internal lattice pressure. Hydrogen-bonding is a particularly effective tool in modulating SCO and this approach was successfully used to generate an extended homochiral 2D sheet of mixed valence Fe 2+ /Fe 3+ sites, which exhibited SCO and light induced excited spin state trapping (LIESST) [36]. In our own work with the [Mn(R-sal 2 323)] + series of Schiff base complexes, we have established that thermal SCO is possible in the less studied d 4 ion Mn 3+ [37][38][39][40][41][42][43][44][45][46]. The hexadentate ligand type used in this approach results from condensation of 1,2-bis(3-aminopropylamino)ethane with a substituted 2-hydroxybenzaldehyde, and the ligand series may be abbreviated as R-Sal 2 323 to indicate the 323 alkyl connectivity in the starting tetraamine and the substitution (R) on the phenolate ring. Changing lattice solvation and/or the identity of the polyatomic charge-balancing counterion X − in the [Mn(R-sal 2 323)]X complex type has enabled examination of the effect of crystal engineering on the choice of the spin state and has yielded a variety of thermal evolution profiles [38,41,42,44]. Recently we have determined that use of the more sterically demanding naphtholate donor in place of a substituted phenolate stabilizes the rare spin triplet form of Mn 3+ up to room temperature in crystalline samples, regardless of choice of counterion. [43] We also showed that in the naphtholate donor series, introduction of ethanol solvate guest molecules to the lattice changes the preferred electronic configuration of the Mn 3+ to the spin quintet form, and in neither case (spin triplet or spin quintet) did we observe thermal SCO [43]. This latter result nicely demonstrates the delicate interplay between lattice contents and choice of spin paired or unpaired arrangement in Mn 3+ and we build on this result now to demonstrate the effect of introducing a larger guest molecule. This was achieved by the serendipitous cocrystallization of some unreacted starting aldehyde during preparation of [Mn(4-methoxy-sal 2 323)] + complexes with a variety of counterions designed to help (CF 3

Synthetic Approach
The reaction of 4-methoxysalicylaldehyde in a 2:1 ratio with 1,2-bis(3-aminopropylamino)-ethane, led to the formation of the Schiff base ligand H 2 L 1 , which is a suitable hexadentate ligand to chelate a Mn 3+ centre. This resulted in the formation of dark red/black crystals of complexes (1)-(3), which were prepared in a one-pot synthesis, Scheme 1. The structures of all compounds were established by single crystal X-ray diffraction and the bulk samples were then fully characterized using elemental analysis, IR spectroscopy and magnetic measurements. Complexes (1)-(3) were formed by a salt metathesis procedure with the target counterions introduced as their group 1 salts, for example sodium as indicated in Scheme 1. This method led to the successful crystallization of [Mn (4-

Synthetic Approach
The reaction of 4-methoxysalicylaldehyde in a 2:1 ratio with 1,2-bis(3-aminopropylamino)ethane, led to the formation of the Schiff base ligand H2L1, which is a suitable hexadentate ligand to chelate a Mn 3+ centre. This resulted in the formation of dark red/black crystals of complexes (1)-(3), which were prepared in a one-pot synthesis, Scheme 1. The structures of all compounds were established by single crystal X-ray diffraction and the bulk samples were then fully characterized using elemental analysis, IR spectroscopy and magnetic measurements. Complexes

Synthesis and Characterization of Compounds (1)-(3)
In the case of the single crystals used in the diffraction experiments the solvent/guest could not be determined crystallographically in terms of atomic sites. Therefore, Platon SQUEEZE [47] was used to compensate for the spread electron density, leading to the respective molecular formulae listed in Table A1.

Magnetic Characterisation
The magnetic susceptibility of the bulk samples of compounds (1)-(3) was measured using a SQUID magnetometer and the data were collected from 300 K down to 5 K under an applied dc field of 1000 Oe, Figure 1. No thermal hysteresis was detected on warming back to room temperature and plots of χ M T versus T are shown in the cooling mode only, Figure 1. In the case of the single crystals used in the diffraction experiments the solvent/guest could not be determined crystallographically in terms of atomic sites. Therefore, Platon SQUEEZE [47] was used to compensate for the spread electron density, leading to the respective molecular formulae listed in Table A1.

Magnetic Characterisation
The magnetic susceptibility of the bulk samples of compounds (1)-(3) was measured using a SQUID magnetometer and the data were collected from 300 K down to 5 K under an applied dc field of 1000 Oe, Figure 1. No thermal hysteresis was detected on warming back to room temperature and plots of χMT versus T are shown in the cooling mode only, Figure 1. Both hydrated compounds (1) and (2) exhibited an incomplete thermal spin transition. Below 100 K both existed in the low spin state with χMT values close to the expected spin-only value of 1.0 cm 3 ·K/mol for S = 1, assuming g = 2. Above 100 K both show SCO upon warming, with a gentle sigmoidal pathway, although neither reached the pure high-spin state by 300 K. T1/2 values were determined to be 217 K for (1) and 242 K for (2). In sharp contrast both complexes (2b) and (3) were in the spin quintet form over the same temperature range with χMT values close to the expected spin only value of 3.0 cm 3 K/mol for a monomeric Mn 3+ complex with S = 2 and g = 2.

Structural Characterisation of Compound (1)-(3)
Complexes (1) and (2) crystallize isostructurally in monoclinic space group P21/c with Z = 4 where the asymmetric unit contains a full [MnL1] + cation and one full triflate or one full hexafluorophosphate anion, respectively, Figure 2. The hexadentate Schiff base ligand chelates the Mn 3+ centre in pseudo octahedral geometry with two trans-phenolate donors, O1 and O2, two cisamine and two cis-imine donor atoms, in the same way as previously observed for manganese complexes with this ligand type [37][38][39][40][41][42][43][44][45][46]48]. Residual electron density in both data sets was modelled as part occupancy water molecules using Platon SQUEEZE [47]. Data for complex (2) was collected at 100 K and 293 K on two different crystals and the water content was modelled as half-occupancy and full occupancy respectively (Table A1).
Compound 2b, with partial occupancy of guest molecule 4-methoxysalicylaldehyde, crystallizes in the monoclinic space group P21/n with Z = 4. Occupancy of the lattice contents was refined with Both hydrated compounds (1) and (2) exhibited an incomplete thermal spin transition. Below 100 K both exist in the low spin state with χ M T values close to the expected spin-only value of 1.0 cm 3 ·K/mol for S = 1, assuming g = 2. Above 100 K both show SCO upon warming, with a gentle sigmoidal pathway, although neither reached the pure high-spin state by 300 K. T 1/2 values were determined to be 217 K for (1) and 242 K for (2). In sharp contrast both complexes (2b) and (3) were in the spin quintet form over the same temperature range with χ M T values close to the expected spin only value of 3.0 cm 3 K/mol for a monomeric Mn 3+ complex with S = 2 and g = 2.
Compound 2b, with partial occupancy of guest molecule 4-methoxysalicylaldehyde, crystallizes in the monoclinic space group P2 1 /n with Z = 4. Occupancy of the lattice contents was refined with 0.3 molecules of water and 0.3 molecules of unreacted 4-methoxysalicylaldehyde, yielding a formula of [MnL 1 ]PF 6 ·0.3H 2 O·0.3sal for the crystalline sample. The benzene ring of the disordered salicylaldehyde was restrained to be regular using SADI and FLAT. Rigid Bond (RIGU) restraints were applied to all non-hydrogen atoms of this molecule. Note that elemental analysis of the bulk sample used in the SQUID measurements showed a better fit to [MnL 1 ]PF 6 ·0.4H 2 O·0.1sal so the latter formula was used in the calculation of molar magnetic susceptibility. The cell parameters of a = 17.2 Å, b = 9.6 Å and c = 19.0 Å for complex (2b) suggest similar dimensions to complex (2), (a = 8.0 Å, b = 20.8 Å and c = 16.2 Å), which was modelled as having 0.5 molecules of water in the crystal lattice. An overlay of the complex cations in (2) and (2b) is shown in Figure 2b.
Complex (3)  In Mn 3+ SCO compounds of the [Mn(R-sal2323)] + type, the average bond lengths change upon spin transition, but only significantly for the amine and imine bonds in the equatorial positions. The Mn-Nimine bond lengths are typically 1.95-2.00 Å in the spin triplet form, increasing to 2.10-2.15 Å in the spin quintet state, while the Mn-Namine bond lengths change from 2.00-2.10 to 2.20-2.30 Å during a spin transition [41]. The bond lengths of complexes (1)-(3) ( Table 1) indicate that the SCO complexes (1) and (2) had typical low spin bond lengths at 100 K while high spin complexes (2b) and (3) had those typical for the S = 2 spin state. Upon warming, the bond lengths of complexes (1) and (2) show the expected equatorial elongation (Table 1) indicative of the transition towards the high spin state. The room temperature bond lengths of complexes (1) and (2) show that both compounds have not yet fully reached the pure spin quintet state, which is in good agreement with the magnetic data. This behaviour has also been observed in other Mn 3+ SCO compounds, which show a gradual and incomplete thermal crossover [37,44], whereas the full transition to high-spin almost always leads to Mn-Namine bond lengths above 2.2 Å [40,42].
In complexes (1) and (2) the increase in bond lengths in the equatorial plane led to a displacement of the benzene ring, which can be seen in the view of the overlaid structures in Figure 2b where the structural differences between the low spin compound (2) and the high-spin compound (2b) are highlighted. While most of the flexible backbone part of the Schiff base ligand overlapped almost  Table 1) indicate that the SCO complexes (1) and (2) had typical low spin bond lengths at 100 K while high spin complexes (2b) and (3) had those typical for the S = 2 spin state. Upon warming, the bond lengths of complexes (1) and (2) show the expected equatorial elongation (Table 1) indicative of the transition towards the high spin state. The room temperature bond lengths of complexes (1) and (2) show that both compounds have not yet fully reached the pure spin quintet state, which is in good agreement with the magnetic data. This behaviour has also been observed in other Mn 3+ SCO compounds, which show a gradual and incomplete thermal crossover [37,44], whereas the full transition to high-spin almost always leads to Mn-N amine bond lengths above 2.2 Å [40,42].
In complexes (1) and (2) the increase in bond lengths in the equatorial plane led to a displacement of the benzene ring, which can be seen in the view of the overlaid structures in Figure 2b where the structural differences between the low spin compound (2) and the high-spin compound (2b) are highlighted. While most of the flexible backbone part of the Schiff base ligand overlapped almost perfectly, there were discrepancies visible in the benzene rings and the peripheral methoxy substituents. Hexacoordinated Mn 3+ compounds exhibit a stronger distortion of the octahedral environment in the S = 2 state than in the almost perfect octahedral geometry observed in the S = 1 state due to the Jahn-Teller effect in the high spin form. The degree of distortion, analysed by the distortion parameters Σ and Θ as defined by McKee et al. [49] highlights the local angular deviation from the cis octahedral angles of 90 • , while Θ measures the trigonal torsion, which is defined as the degree of twist from a perfect octahedron towards trigonal prismatic geometry. Both values are zero in the case of a perfect octahedron, and the reported literature values for Mn 3+ compounds are shown in Table 2. [41,44].  [50] and the observed parameters are summarized in Table 3, which reflect the structural distortion due to the different spin states of the molecules. At 100 K, compounds (1) and (2) clearly exhibit Σ and Θ values which confirm that both are in the low spin state at this temperature. Upon warming, the values of both complexes increase slightly to Σ = 45.3 • for (1) (46.3 • for (2)) and Θ = 140.3 • for 1 (145.7 • for (2)), which reflected the gradual spin state change within these compounds. In contrast, analysis of the distortion in compounds (2b) and (3), which were determined by magnetometry to be high-spin over the whole measured temperature range, had high Σ and Θ values, confirming the S = 2 assignment. Even though the angular distortion parameter, Σ, is high for these latter two, the values are still within the range that was reported for other SCO Mn 3+ compounds [41,44], whereas the Θ values of both are higher than what has been reported before. Olguin has recently summarized the Σ values of Mn 3+ complexes of the [Mn(R-Sal-323] + type, and shows that Σ values above 70 • often lead to complexes that are locked in the high-spin state [48]. The fact that this is not necessarily always the case can be seen in the [Mn(3-MeO-Sal-323-]NO 3 complex [37]. This complex has a Σ value of 70.7 • in the high spin state, with Θ values of 129.8 • in the low spin state and 222.3 • in the high spin state, while still being able to undergo spin transition, in contrast to related complexes with lower Σ values, which are locked in the high spin state [48]. It appears that there is a threshold for the Σ values at around 70 • but it is not exact, as SCO is observed in some Mn 3+ complexes with Σ values higher than 70 • . On the other hand, the Θ values may be a more reliable indicator of SCO in Mn 3+ , as it seems that for Θ values higher than 250 • , the compounds are not able to undergo a spin transition and will remain in the high spin state. This demonstrates the considerable elastic flexibility of the [MnL 1 ] + complex type, which can accommodate bulky anions and neutral molecules in the crystal lattice, resulting in deformation of the complex cation with the possible loss of SCO behaviour. Examination of the intermolecular arrangements in complexes (1)-(3) revealed that only the two complexes that exhibited thermal SCO, complexes (1) and (2), make strong intermolecular contacts (See Supplementary Materials Figures S1-S4). These were by way of hydrogen bonding, which in the case of triflate complex (1) connected the complex cation to the counterion via the guest water molecule at both 100 K and 293 K (Figures S1 and S2). In the case of the hexafluorophosphate complex, a hydrogen bond was confirmed between the complex cation and partial occupancy water molecule at both measured temperatures ( Figures S3 and S4), but the nature of the interaction with the counterion was less clear, especially given the significant distortion of the PF 6 − counterion in the 100 K structure.
In marked contrast both high spin complexes (2b) and (3) had no intermolecular interactions with the complex cation and this likely contributed to the locking-in of the spin quintet state over the thermal range. In complex (2b) the partial occupancy guest water and 4-methoxysalicylaldehyde molecules were highly disordered but were distinct from the [MnL 1 ] + cation and PF 6 − counterion, the latter two forming neither hydrogen bonds nor π-π interactions, Figure S5. Thus the [MnL 1 ] + spin carrier was isolated in this lattice with the large guest molecule, Figure S6, and persisted in a fixed spin state while the temperature was reduced. A similar fixed spin-state response was observed in complex (3) with the largest counterion, BPh 4 − , which was not disposed to form hydrogen bonds, and which also therefore tended to isolate the complex cation, thereby preventing SCO [41].

Discussion
A magnetostructural examination of a new series of SCO and high spin salts of the [MnL 1 ] + , complex cation, in which L 1 is the 4-methoxy-sal 2 323 hexadentate Schiff base ligand, revealed the delicate dependence of the complex spin state on lattice contents. This is because the geometry of the immediate coordination sphere of the Mn 3+ ion will be subject to different degrees of local distortion depending on the identity of the lattice partners in the space between spin-labile [MnL 1 ] + cations. This in turn impacts on the ability of the coordination sphere to undergo the significant rearrangement, which is required in spin state switching. The lattice partners, comprising counterions, and/or guest molecules such as solvent or unreacted starting materials, will have both a different volume and a different capacity for forming intermolecular interactions. A thermal spin transition typically necessitates a large change in volume and/or distortion as the antibonding orbitals are depopulated on cooling. This is especially relevant in ions which have a strong distortion in either spin state, as is the case with the Jahn-Teller ions Mn 3+ and Co 2+ , which have marked distortions in the high spin and low spin forms respectively [48]. The Jahn-Teller effect is also relevant in some Fe 2+ systems [5]. In the case of the complexes reported here two factors may be at work: Firstly, the inclusion of sterically demanding lattice partners such as a guest 4-methoxysalicylaldehyde molecule or large BPh 4 − counterion fills up the void space and causes a marked distortion of the coordination geometry around the Mn 3+ ion in the high spin state. Thus, the degree of distortion in the high spin forms, as revealed by the Θ angle (Table 3), is higher in those complexes which do not show SCO but which are locked into the high spin form over the temperature range. In contrast those complexes that can undergo SCO, complexes (1) and (2), show less distortion in their high spin forms, Table 3, suggesting that there is less strain with the less sterically demanding PF 6 − and CF 3 SO 3 − , counterions.
Secondly it appears that hydrogen bonding plays an important role in coupling the spin labile complex cation to the lattice, and in its absence the complex can get "stuck" in one spin state, in the case of [MnL 1 ] + this is the spin quintet form. The connectivity of the complex cation can be modulated by choice of counterion: use of counterions with oxygens and fluorines such as PF 6 − and CF 3 SO 3 − , enables efficient hydrogen bond formation with one of the phenoxy donors on the complex and SCO ensues. Substitution with BPh 4 − rules out hydrogen bonding and there is a concomitant loss of spin state lability for the cation in this lattice. We also established here that guest molecules other than the solvent could be accommodated, and inclusion of partial occupancy 4-methoxysalicylaldehyde in one of the synthetic attempts to make the PF 6 − complex, blocks the hydrogen bonding that was previously observed with this combination and the SCO is quenched. The importance of crystal engineering in modulating spin state preferences in such Mn 3+ chelates is clear and our work on related systems continues to try to establish the extent of the phenomenon in such materials.

Materials and Physical Measurements
All chemicals and solvents if not otherwise mentioned were purchased from chemical companies and were reagent grade. They were used without further purification or drying. All reactions were carried out under ambient conditions. All measurements were carried out on powdered samples of the respective polycrystalline compound. Elemental analysis (C, H and N) were performed using a Perkin Elmer Vario EL (Waltham, MA, USA). A Bruker Alpha PlatinumATIR spectrometer (Billica, MA, USA), which was used to record the infrared spectra, and mass spectra were recorded on a Waters 2695 Separations Module Electrospray Spectrometer (Milford, MA, USA).

Single-Crystal X-ray Structure Determination
Suitable single crystals of complexes (1) to (3) were mounted on Oxford Diffraction Supernova E diffractometer (Oxford, UK) fitted with an Atlas detector; datasets were measured using monochromatic Cu-Kα radiation or Mo-Kα radiation and corrected for absorption [51]. The temperature (100 K) was controlled with an Oxford Cryosystem instrument. Structures were solved by dual-space direct methods (SHELXT) [52] and refined with full-matrix least-squared procedures based on F 2 , using SHELXL-2016. Non-hydrogen atoms were refined with independent anisotropic displacement parameters, organic H-atoms (i.e., bonded to C) were placed in idealized positions, while the coordinates of H-atoms bonded to O were generally refined with their O-H distance restrained to 0.88 (4) Å. Within complex (2b), the hydrogen atoms of the water molecules could not be detected and were therefore placed in the idealized position. The benzene ring of the disordered unreacted salicylaldehyde was restrained to be regular using SADI and FLAT. Rigid Bond (RIGU) restraints were applied to all non-hydrogen atoms of this molecule.
Selected crystallographic data and structure refinements are summarized in Table A1 and crystallographic data for the structures reported in this paper were deposited with the Cambridge Crystallographic Data Centre (CCDC) as supplementary publication numbers CCDC-2042004-2042009. Copies of the data can be obtained free of charge from https://www.ccdc.cam.ac.uk/structures/.

Magnetic Measurements
The magnetic susceptibility measurements were recorded on a Quantum Design SQUID magnetometer MPMS-XL (San Diego, CA, USA) operating between 1.8 and 300 K. DC measurements were performed on polycrystalline samples. Each sample was wrapped in a gelatine capsule and subjected to fields in the range from 0 to 7 T. The magnetization data was collected at 100 K in order to check for ferromagnetic impurities, which were found to be absent in the samples. Diamagnetic corrections were applied to correct for contribution from the sample holder, and the inherent diamagnetism of the sample was estimated with the use of Pascal's constants. AC measurements were carried out at frequencies between 1 and 1500 Hz.