Exploring the Influence of Counterions on a Hysteretic Spin-Transition in Isomorphous Iron(II) Complex Salts

[FeL2]X2·2H2O (L = 2,6-bis{5-methyl-1H-pyrazol-3-yl}pyridine; X‒ = BF4‒ or ClO4‒) are readily dehydrated upon mild heating. Anhydrous [FeL2][BF4]2 exhibits an abrupt spin-transition at T½ = 205 K, with a 65 K...


Experimental details S4
Table S1 Experimental data for the crystal structure determinations of 2b•2H2O and 2b. S5

Definitions of the structural parameters discussed in the paper S6
Scheme S1 The parameters θ and ϕ, used to discuss the structures of [ML2] 2+ (M = Fe, Zn). S6 The formula unit of 2b phase B, showing the hydrogen bonds in the lattice. S10 Table S2 Selected bond lengths and angles in the crystal structure of 2b•2H2O and 2b. S11

Figure S7
The full anion disorder in phase B of 2b at 300 K (top) and 250 K (bottom). S12

Table S3
Hydrogen bond parameters in the crystal structure of 2b•2H2O and 2b. S13

Figure S9
Packing diagram of 2b phase B at 250 K, viewed parallel to the terpyridine embrace layers. S15 Figure S10 Packing diagram of 2b phase C, viewed parallel to the terpyridine embrace layers. S16

Figure S11
Measured and simulated powder diffraction data for 2b·2H2O and the phases of 2b. S17

Figure S1
Variable temperature magnetic susceptibility data for 1a•2H2O (top) and 1b•2H2O (bottom), at a scan rate of 2 K min -1 . The data are taken from refs. 1 and 2. An initial 290→5→350 K scan of the asprepared samples shows their in situ dehydration (black), and a 350→5→350 K scan of the dehydrated materials is in red.
Freshly prepared 1a•2H2O and 1b•2H2O are crystallographically isomorphous at room temperature, while anhydrous 1a and 1b are also isomorphous by powder diffraction. However, their spin state properties are very different.
Crystals of 1a•2H2O contain a 1:1 high:low-spin population which is retained on cooling, but converts to a fully high-spin state on heating to 350 K following in situ loss of lattice water. 1 Dehydrated 1a (phase B) transforms to a new phase C near 300 K, then exhibits an abrupt spin-transition at T½ = 205 K which is associated with another crystallographic phase change. 3 The spin-transition has a 65 K thermal hysteresis loop in the first scan (shown), which narrows to 37 K hysteresis width upon multiple thermal scanning ( Figure 4, main article). 1 In contrast, 1b•2H2O exhibits a gradual, highly structured SCO on cooling which involves two crystallographic phase transitions: one just below room temperature, and the other associated with the final hysteretic step near 120 K. The annealed material (phase B) also transforms to phase C around 300 K, but then remains fully high-spin on further cooling. 2 We unravelled the structural chemistry of 1a through investigations of its zinc(II) analogue 2a, whose crystals are more robust during the dehydration process. 3 This study is a continuation of that work, to investigate why 1b is inactive to SCO despite being isomorphous with 1a.

Experimental
Synthetic and crystallisation procedures, and analytical characterisation data, for the compounds in this work are presented in the main article.

Single Crystal Structure Analyses
Experimental details of the structure determinations are given in Table S1. All the structures were solved by direct methods (SHELXS97 1 ), and developed by full least-squares refinement on F 2 (SHELXL2018 1 ). Crystallographic figures were prepared using XSEED. 2 Unless otherwise stated, the following procedures were applied to the refinements. All crystallographically occupied non-H atoms in the following refinements were refined anisotropically, while H atoms were placed in calculated positions and refined using a riding model. Disordered anions were treated with refined Cl-O and O···O distance restraints.

Structure refinements of [ZnL2][ClO4]2•2H2O (2b•2H2O) and [ZnL2]
[ClO4]2 (2b). The asymmetric unit of 2b•2H2O contains one-quarter of a complex dication, with Zn(1) on the 4 site ¾, ¾, ¼ and N(2) and C(5) on the C2 axis ¾, ¾, z. The unique ClO4half-anion is disordered over two sites spanning the C2 axis ¾, ¼, z. These were modelled with occupancies of 0.35 and 0.15. The isotropic displacement ellipsoids of the Cl and O atoms in each partial anion were also constrained to be approximately equal (SHELXL SIMU restraint). The unique water half-molecule is crystallographically ordered in this refinement. All non-H atoms except the disordered anion were refined anisotropically, and C-and N-bound H atoms were placed in calculated positions and refined using a riding model. The disordered water H atoms were not located in the Fourier map and are not included in the model, but are accounted for in the molecular weight and density calculations.
Annealing at 350 K for 30 mins on the diffractometer led to dehydration of the crystal. Datasets collected from the 2b crystal at 350, 300, 250 and 200 K all retained the P42/n space group (phase B). The refinement protocol is similar to 2b•2H2O above, except the unique half-anion was refined over two 0.25-occupied disorder sites. The 200 K structure determination is noisier and less precise than the others, which may reflect its proximity to the phase B→C transition temperature.
At 120 K the crystal has transformed to triclinic phase C. This phase change leads to twinning of the crystals; four twin domains were resolved in the hkl file, related by ca 90° rotations about the crystallographic c axis. Phase C contains just one formula unit of the compound, with all residues on general crystallographic positions. No disorder is present in the phase C refinement.
CCDC-2195223-2195228 contain the supplementary crystallographic data for this paper (Tables S1 and S2). These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif.

Other measurements
Elemental microanalyses were performed by the microanalytical services at the University of Leeds and London Metropolitan University. 1 H NMR spectra employed a Bruker DPX300 spectrometer operating at 300.1 MHz. Magnetic susceptibility measurements were performed on a Quantum Design SQUID or SQUID/VSM magnetometer, with an applied field of 5000 G and a scan rate of 5 Kmin -1 . A diamagnetic correction for the sample was estimated from Pascal's constants; 3 a previously measured diamagnetic correction for the sample holder was also applied.
Thermogravimetric analyses (TGAs) were obtained with a TA Instruments TGA Q50 analyser heating at a rate of 10 K min -1 under a stream of nitrogen gas, while differential scanning calorimetry (DSC) used a TA Instruments DSC Q20 calorimeter, also with a temperature ramp of 10 Kmin -1 . X-ray powder diffraction data were obtained with a Bruker D8 Advance A25 diffractometer using Cu-Ka radiation (l = 1.5418 Å) .  (3) 9.5249(9) 9.5028

Figure S4
The complete formula unit of 2b•2H2O at 120 K, showing the hydrogen bonds in the lattice. Only the major orientation of the disordered half-anion is shown, and moieties that are not directly hydrogen bonded to the cation are de-emphasised for clarity. Displacement ellipsoids are at the 50 % probability level, where shown, and C-bound H atoms are omitted for clarity. Hydrogen atoms for the half-water molecule O(17) were not included in the refinement.
Symmetry codes: The half-anion and half-water molecule are disordered about the same lattice cavity, so the hydrogen bond network is in fact disordered within the crystal.

Figure S5
The complete formula unit of phase B of 2b at 250 K, showing the hydrogen bonds in the lattice. Only one orientation of the disordered half-anion is shown. Displacement ellipsoids are at the 50 % probability level, and C-bound H atoms are omitted for clarity.
Symmetry codes: (iii) y, 3 The disordered (half)-anions are themselves disordered across a crystallographic C2 axis, which randomly distributes them between hydrogen-bond donor groups on two different cations ( Figure S7).

Figure S6
The asymmetric unit of phase C of 2b at 120 K. Displacement ellipsoids are at the 50 % probability level, and C-bound H atoms are omitted for clarity.

Figure S7
The full anion disorder in phase B of 2b at 300 K (top) and 250 K (bottom). Displacement ellipsoids are at the 50 % probability level except for the anion, which is de-emphasised for clarity. C-bound H atoms are not included.
Colour code: C, white; H, pale grey; Cl, yellow; N, blue; O, red; Zn, green. The two disorder orientations associated with the symmetry-related half-anion sites, related by the operation 3 /2−x, 1 /2−y, z, are shown with dark and pale colouration.
The anion disorder is the same at 350 and 300 K; and, at 250 and 200 K. However the refinements imply the disordered anions reorient themselves between 300 and 250 K, which changes the pattern of hydrogen bonding. § At 350 and 300 K, all four partial anions form hydrogen bonds to both cations. § At 250 and 200 K, one partial anion (and its symmetry equivalent) hydrogen bond to both cations, but the other partial anion hydrogen bonds to just one cation site.
While the level of disorder means it must be interpreted with caution, such a rearrangement of the anions could be a precursor to the phase B→C transformation which occurs below 200 K. There is a rearrangement of the disordered anion sites between 300 and 250 K, which changes the geometry of the N−H···O hydrogen bonds ( Figure S7).

Figure S8
Packing diagram of 2b·2H2O, viewed along the (010) crystal vector. Only the 'A' disorder orientation of the half-anion is included, but both symmetry equivalents for each anion and water site are shown. The ClO4 − ions are de-emphasised for clarity.
The half-water molecule H atoms were not included in the refinement, and so are not shown in the Figure  The small 2q offset between the measured and simulated peaks for 2b·2H2O can be attributed to the very different temperatures of the two measurements. Otherwise, the agreement between experiment and simulation is very good, with the two small caveats below.
The peak at 2q = 16.3° shown by 2b·2H2O, which has a weak counterpart in the simulation for that phase, is retained in the phase B powder pattern but not in its simulation. The reason for that small discrepancy is unclear.
The weak residual peak at 2q = 10.5° in the phase C powder pattern suggest a small fraction of the sample retains phase B below the nominal phase transition temperature. That was also observed in powder diffraction measurements of 1b (Figure 4 The initial warming ramps (grey trace) show strong endotherms arising from the dehydration of the samples ( Figure S2). Scans (ii) and (iii) (black) arise from the dehydrated materials.
The highlighted features in scans (ii) and (iii) of 2a are unambiguously assigned to the phase B→C transformation, as their temperature (average 290 K) coincides well with the phase transition measured crystallographically and by powder diffraction. 3 The average temperature of the corresponding peaks for 2b (268 K) is consistent with the onset of the phase B→C transition on cooling in the powder diffraction measurement (Figure 3, main article). However, there is no evidence in the DSC data for the broadening of the phase transition observed by powder diffraction, which occurs over a 25±5 K temperature window in that experiment. There is no evidence for broadening of the phase transition in the DSC trace.
The different characteristics for the phase B→C transition in 2b measured by crystallography, powder diffraction and DSC imply it depends on the measurement conditions and/or the history of the sample.

Figure S13
Measured room temperature powder diffraction data for 1c·2H2O and 1d·2H2O (black), and a simulation based on the structure of 1a·2H2O at 150 K (red). 1 Differences in peak intensities between the measurement and simulation arise from preferred orientation in the ground polycrystalline samples. However, there is good correspondence of the peak positions in the measurements and simulation, allowing for the different compositions of the samples and the low temperature of the simulation.
Notwithstanding the TIESST in 1d, in other respects scan 3 of 1c ( Figure S15) and scans 5 and 6 of 1d, closely resemble each other. Additional scanning of 1c causes its evolution to a symmetrical spin-transition hysteresis loop, whereas 1d remains trapped in that intermediate state under the same conditions. This is discussed in the main article.

Figure S17
Powder diffraction patterns on cooling and warming temperature ramps, for anhydrous 1d, produced by in situ dehydration of 1d·2H2O ( Figure S13).
The sample adopts phase B at 353 K, and phase C at the other temperatures (compare with Figure S11). There is no obvious temperature dependence of the phase C powder pattern, which resembles the behaviour shown by 1b and 2b (Figure 3, main article). That contrasts with 1a and 2a, whose phase C powder patterns are strongly temperature-dependent. 3 The sample has not transformed to its thermodynamic low-spin phase (phase E) at 149 K, the lowest temperature achieved on this diffractometer. 1,3 It is unclear how that correlates with the spin state properties of forms (i) and (ii) of 1d ( Figure S16 and Scheme 2, main article).
Multiple thermal scans of this material on the powder diffractometer were not achieved, because of instrument time constraints. However, on this evidence the complicated sequence of transformations in the magnetic data for 1d is not reflected in its powder pattern, which shows the same phase behaviour as the anion-pure salts 1a and 1b.