Kenneth
Zhang‡
a,
Matthew J.
Wallis‡
a,
Alexander R.
Craze
b,
Shinya
Hayami
c,
Hyunsung
Min
ad,
Daniel J.
Fanna
d,
Mohan M.
Bhadbhade
e,
Ruoming
Tian
e,
Christopher E.
Marjo
e,
Leonard F.
Lindoy
f and
Feng
Li
*a
aSchool of Science, Western Sydney University, Locked Bag 1797, Penrith, NSW, Australia. E-mail: Feng.Li@westernsydney.edu.au
bDepartment of Chemistry, University of Oxford, Oxford OX1 3TA, UK
cDepartment of Chemistry, Graduate School of Science and Technology, Kumamoto University, 2-39-1 Kurokami, Chuo-ku, Japan
dAdvanced Materials Characterisation Facility, Western Sydney University, Locked Bag 1797, Penrith, NSW, Australia
eMark Wainwright Analytical Centre, University of New South Wales, Kensington, NSW, Australia
fSchool of Chemistry, The University of Sydney, NSW 2006, Australia
First published on 3rd May 2024
Two new salts of a mononuclear tripodal Fe(II) complex were prepared, using ClO4− and Cl−. The ClO4− sample (1) remained HS at low temperatures, similar to the previously reported BF4− analogue. Crystallising with the Cl− anion (2) led to a markedly different crystal packing arrangement, and engendered SCO activity. This has been correlated to the lower crystal packing density in 2 and the coordination complex conformational differences arising due to the packing motifs of 1 and 2. Further, solvent ordering effects have been proposed to facilitate spin transition behaviour in 2.
Contrasting with SCO behaviour in the solution state, the SCO transition characteristics are usually drastically influenced in the solid state by the supramolecular environment. Thus, by ‘tuning’ factors such as the counter-ion identity12–18 and/or the presence of lattice solvent molecules,16,19,20 potentially important properties such as the abruptness and completeness of a spin transition (ST), the transition temperature (T1/2) and the number of SCO steps can be modified.13,15,17–23 Further to this is the effect of co-crystallising other components on the material's crystal structure,13,15,17,24 as varied crystal packing motifs will often impose different steric constraints on the SCO complexes, which need to be afforded enough steric freedom if a spin transition is to occur. Previous studies have shown that the nature of a SCO material can be altered by the size and electronic characteristics of counter-ions present, while particular studies have demonstrated a change of counter-ion can completely change the crystal packing arrangement, thus drastically altering the associated magnetic behaviour.12,15,22–26 Further, the degree of solvation often plays a crucial role in determining the supramolecular environment, and thus also can be used to modify the magnetic behaviour of SCO materials.
The loss of solvent within a crystal lattice affects its constituent intermolecular interactions and may have drastic effects on the SCO behaviour of compounds, for example, due to the shrinkage of the lattice volume27,28 and in some cases a change of phase.23,29,30 The loss of solvation may influence the cooperativity of the crystal system in various ways, with reports of magnetic switching behaviour being either enhanced or quenched. Enhancements of magnetism from desolvation may occur from enhanced elastic cooperativity resulting in a denser crystal lattice,23,27,29,31 or by lowering of the steric hindrance between SCO complexes and solvent, enabling intramolecular adaptations typical of ST.4 On the other hand, quenching of magnetism may be caused by the formation of anti-cooperative interactions,19,30,32 or by an increase in steric hindrance in the crystal lattice between adjacent complexes and/or counter-ions.28,33–35 The effect of desolvation on a SCO system is heavily dependent on the contacts formed between the solvent and other units in the crystal and, in particular, the position of the solvent in the lattice relative to the SCO complexes.
Tripodal metalloligands incorporating Fe(II) and the triethylamine linker moiety have been widely investigated for their SCO capabilities. Often, these complexes bear coordinating domains such as imidazole-imine18,36–38 or thiazole-imine.39–41 We have recently reported the application of a similar semi-rigid tripodal metalloligand [FeL]2+ (see below) for the self-assembly of heteronuclear tetradecanuclear cubic cages.42,43 When [FeL]2+ was combined with a secondary metal ion, cubic cages were formed, whereby eight Fe(II) ions occupy the corners of the cube, and six secondary metal ions are arranged on the cubic faces. The [Fe8Pd6L8]28+ cubic cage synthesised from [FeL](BF4)2 did not exhibit SCO, (and neither did the [FeL](BF4)2 metalloligand). When the identity of the face-occupying Pd(II) was replaced with Ni(II), forming [Fe8Ni6L8(CH3CN)12]28+, the product was observed to undergo SCO.
In this work, in order to gain further insight into the emergence of SCO in this system, we focus on the metalloligand material [FeL]X2 (where X = ClO4− or Cl−), and how SCO can be elicited in this building block, rather than the more complex cage architecture. The material [FeL](ClO4)2 (1) portrayed similar magnetic and structural behaviour to the [FeL](BF4)2 metalloligand species described above, not exhibiting SCO. Several key similarities in regard to important intermolecular parameters and the crystallographic packing motifs were observed between these two HS species. On the other hand, [FeL]Cl2 (2) exhibited a spin transition that was solvent dependant. The transition was two-step in the solvated and desolvated samples. Additionally, the residual HS fraction at low temperatures was higher in the desolvated sample. Magnetostructural correlations are presented, to give possible explanations as to how the [FeL]X2 material becomes SCO active when the identity of the counter-ion X is changed to Cl−. These materials were characterised by CHN, SEM-EDS, HR ESI-MS, TGA-DSC, PXRD, magnetometry and variable temperature single crystal X-ray diffraction (VT-SCXRD).
Caution! Perchlorate salts are highly explosive and should be handled with care and in small amounts.
Simultaneous thermal analysis (STA) experiments were conducted on a Netzsch STA-449C Jupiter instrument. STA measurements were acquired using argon for both the protective and purge gases at a flow rate of 25 mL min−1. The sample was weighed into a pierced aluminium crucible and measured between 303–500 K at a heating rate of 10 K min−1.
Powder X-ray diffraction measurements were conducted on a Bruker D8 ADVANCE X-ray diffractometer fitted with a Cu source (Cu Kα1 at 1.54 Å). The tube accelerating voltage and current were set to 40 kV and 40 mA, respectively. A LYNXEYE XE-T position sensitive detector (PSD) with energy discriminator settings optimised to reject Fe fluorescence was employed. All scans were collected at room temperature in Bragg–Brentano geometry. The scan for 1 was conducted with a fixed divergent slit opening of 0.16°, a 5–90° 2θ scan range with a step size of 0.005° and a dwell time of 0.5 seconds per step. The scan for air dried sample of 2 (2·6.5H2O) was conducted with motorised slits on primary optics set to fixed illumination mode with sample beam illumination set to 10 mm. Samples were scanned at room temperature across a 2.5–50°2θ scan range with a step size of 0.02° and a dwell time of 0.25 seconds per step. These scan parameters were optimised for this sample to achieve better intensity and lower amount of sample degradation. Details of sample preparation and data processing are available in the ESI (S1†).
Magnetic susceptibility measurements were performed using a Quantum Design SQUID magnetometer calibrated against a standard palladium sample under an applied field of 0.5 T. Air dried sample (2·6.5H2O) was used for measurement. Magnetic susceptibility measurements were conducted between 5 and 400 K at scan rates of 1 and 4 K min−1 in heating and cooling modes. Desolvation of the air-dried sample (2·6.5H2O) was achieved by holding the samples at 400 K for 60 minutes in situ, after which additional measurements were performed in heating and cooling modes at 1 and 4 K min−1. Light-induced excited spin state trapping (LIESST) studies were performed with red (800 nm) and green (532 nm) laser irradiation on the air dried sample. The sample was irradiated with one excitation wavelength at 5 K until the magnetic signal saturation and then heated at 1 K min−1.
Single crystal data (SCXRD) for all complexes were collected at the Australian Synchrotron from the MX1 beamline using silicon double crystal monochromated radiation (λ = 0.71073 Å). Variable temperature SCXRD experiments were conducted on crystals (2-CH3OH) from 100 K to 300 K at 50 K intervals. All data sets consisted of two collections consisting of a sweep through θ of 360° but differing by the setting of κ either 0° or 180°. XDS44 software at the Australian Synchrotron was utilised for merging, data integration, processing, scaling and the merging of raw datasets. Absorption corrections were applied using SADABS.45 The structures were solved and refined using a suite of ShelX programs46,47via the Olex248 interface. Octahedral distortion parameters were calculated using OctaDist.49 The crystallographic data in CIF format has been deposited at the Cambridge Crystallographic Data Centre with CCDC numbers 2338622–2338627.†
At a scan rate of 4 K min−1, the air-dried sample (2·6.5H2O) exhibits an incomplete two-step ST (Fig. 1) that occurs between 362 to 90 K. The first step occurs with a TSCO value of 292 K (2.61 cm3 K mol−1) (Fig. S8†). Between 250 and 190 K the profile adopts a shallow gradient between the two steps. The second spin transition event occurs with a TSCO value at 133 K (1.88 cm3 K mol−1). At temperatures under 100 K, the curve flattens out with an approximate residual χMT of 1.57 cm3 K mol−1, indicating a residual HS fraction of ∼53%. Desolvation (desolv. 2) further reduced the completeness of the transition. Polynomial fitting suggests two-step dynamics are maintained, with TSCO temperatures of 272 K (2.68 cm3 K mol−1) and 150 K (2.39 cm3 K mol−1) (Fig. S9†) at a scan rate of 4 K min−1. The higher residual magnetic susceptibility of ∼2.34 cm3 K mol−1 at low temperatures, corresponds to a residual HS fraction of ∼80% in desolv. 2. Additionally, LIESST studies were performed, indicating that 2·6.5H2O experiences photomagnetic behaviour in response to irradiation with red (800 nm) and green (532 nm) light (Fig. S10†).
Fig. 1 Magnetic susceptibility plot for air-dried (2·6.5H2O) and desolvated (desolv. 2) samples at a heating rate of 4 K min−1. |
All structures obtained for single crystal (2-CH3OH) were solved in the monoclinic C2/c space group. At 100 K, one metal complex and two chloride anions were found, as well as several solvent molecules. Two well-ordered methanols and one disordered methanol were found, as well as two water molecules in occupancies of 1 and 0.5. A solvent void occurring at a special position was masked, attributed to a water molecule with an occupancy of 0.8. This solvent composition was used to assign residual electron density in higher temperature structures, where thermal motion prohibited modelling of several solvent molecules (S6). In all structures, the metal complex and chloride anions were refined anisotropically and hydrogen atoms were fixed using a riding model. The restraints DFIX, DANG, FLAT and RIGU were used to model disordered pyridyl rings where appropriate. Solvent atoms were modelled anisotropically with hydrogen atoms fixed where possible, though in many cases were left isotropic and/or without hydrogen atoms fixed. DFIX and RIGU were applied to model disordered methanols.
Sample | [FeL](BF4)2a | 1 | 2·3CH3OH, 2.3H2O (2-CH3OH) | ||||
---|---|---|---|---|---|---|---|
Temperature (K) | 100 | 100 | 100 | 150 | 200 | 250 | 300 |
a [FeL](BF4)2 has been previously reported by our group, and is included here for structural comparison against the compounds studied in this work. | |||||||
Average Fe⋯N bond length (Å) | 2.20 | 2.2054(15) | 1.982(22) | 1.983(20) | 1.994(17) | 2.056(19) | 2.152(38) |
ζ (Å) | 0.16 | 0.155 | 0.031 | 0.027 | 0.027 | 0.043 | 0.025 |
Σ (°) | 105.0 | 104.0 | 53.6 | 53.5 | 54.6 | 64.1 | 77.4 |
Θ (°) | 280.4 | 280.3 | 174.3 | 174.4 | 179.0 | 208.4 | 246.4 |
Volume (Å3) | 3806.1(12) | 3880.2(12) | 8289(5) | 8313(5) | 8427(5) | 8600(5) | 8598(5) |
Packing coefficient (%) | 54.57 | 53.57 | 49.56 | 49.41 | 48.66 | 47.58 | 47.71 |
Previous investigations of the [Fe8Ni6L8(CH3CN)12]28+ coordination cage concluded that in order for the Fe(II) metalloligand to undergo SCO, the pyridyl termini must be able to spread apart from one another (see Fig. 3 for structural overlap). In the two structures that do not undergo SCO, 1 and its BF4− analogue, the packing arrangement is such that intermolecular contacts favour a more compact arrangement of the distal pyridyl units (specifically by N⋯HC and π⋯π contacts between B and C). The transition to LS may have a restrictive conformational barrier, or may be anticooperative with transitions in neighbouring units. Overall, the effect of these arrangements is stabilisation of the HS state in the P21/c lattices of 1 and [FeL](BF4)2, the ClO4− and BF4− analogues.
In order to probe how the structure of 2 changed as a result of SCO, VT-SCXRD between 100–300 K was conducted on the crystals (2-CH3OH). Similar to 1, a distorted octahedral coordination sphere around the Fe(II) centre is formed by three imidazolylimine groups and both optical isomers are present within the crystal lattice. The compound is observed to be in the monoclinic C2/c space group with one complete Fe(II) metal complex, two chloride anions, three methanol molecules and three water positions, adding up to 2.3 waters per formula unit. A solvent mask was applied in all structures (S6). Interestingly, the distance between Fe(II) and the apical nitrogen atom (N1) shortens from 3.489 Å to 3.222 Å across the temperature series 100–300 K. A tightening of the average C–N1–C from 120° to 117° is also observed. The changing of these parameters is caused by the transition towards HS. The lone pair of the apical nitrogen is directed towards the t2g orbitals of the metal centre, which bear a lower occupancy in the HS state.39
To note, fewer direct interactions between adjacent [FeL]2+ cations were observed, in comparison to the tight N⋯HC and π⋯π interactions observed in 1. The crystal lattice packs as 1D chains through Cl⋯HC hydrogen bonding along the b-axis in an undulating sequence of ΔΛ complexes. Hydrogen bonding between the terminal pyridines of ligand arms B and C (N4B⋯C1C) forms voids, which are continuous along the c-axis (Fig. 4) and contain all solvent molecules. Along the c-axis, Cl− ions hydrogen bond with solvent ions in the pore. Further, hydrogen bonding between tripodal complexes and Cl− anions links CH groups of imidazole and pyridyl moieties of adjacent complexes (Fig. 4d). The Cl− ion designated Cl1 links the imidazole CH proximal to the coordination sphere of arm A with the imidazole CH groups distal to the coordination sphere on arms B and C. Meanwhile, Cl2 links the imidazole CH distal to the coordination sphere of arm A to the imidazole proximal to the coordination sphere of arm C.
Beginning at 100 K, the average Fe–N bond lengths (1.982 Å) and octahedral distortion parameters Σ and Θ49 of 53.6° and 174.3°, respectively, are consistent with the Fe(II) metal centre being in the LS state. The octahedral distortion values remain consistent up to 200 K, implying that the ST occurs primarily between 200 and 300 K, with the most significant change in octahedral distortions occurring between 250–300 K. At 300 K, the metalloligand demonstrated larger octahedral distortion parameter values (Fe–N = 2.15 Å, Σ = 77.4° and Θ = 246.4°, Table 1). Comparison of the HS values observed in 1 to the 300 K values of the single crystal (2-CH3OH), illustrates that the latter material has not reached a fully HS state at 300 K. This is in accord with the magnetic susceptibility data of the air dried sample (2·6.5H2O), at which point χMT = 2.70 cm3 K mol−1. This indicates a mixed spin fraction is present in single crystals of 2-CH3OH at 300 K. Based on the average coordinate bond lengths at this temperature of 2.152 Å, with respect to those observed in the fully HS (2.205 Å) and fully LS structures (1.982 Å), we estimate that at 300 K a HS fraction of ∼76% persists in the crystal, which is in agreement with the HS fraction in magnetic susceptibility measurements of the air-dried material at 300 K. This means that the higher temperature step of the HS > LS transition cannot be structurally characterised with the available crystallographic data. Below we show our findings which we attribute to the magnetostructural dynamics of the lower temperature step of spin transition in 2 (Fig. 1).
Of note in the manifestation of SCO in this material is the lower packing density of complex ions in the crystal lattice of 2-CH3OH, with respect to those found in the BF4− and ClO4− analogues (Table 1). This was gauged by measuring the metal complex crystal packing coefficient, as calculated in Olex248 using the ‘molinfo’ command. This provides a measure of the percentage of the unit cell volume which is occupied by coordination complexes. The lower crystal packing coefficient in 2-CH3OH correlates to metalloligand complexes occupying less of the total lattice volume, suggesting that the metal complexes have more space to undergo the conformational changes which must occur upon spin transition (as seen in comparing HS and LS structures). Interestingly, the evolution of the crystal lattice parameters in 2-CH3OH was non linear, with the volumes of the 250 K (8600 Å3) and 300 K (8598 Å3) lattices being within experimental error of one another. This occurs primarily due to growth of the ac plane at 250 K, with the a- and c-axis lengths at 300 K of 32.281(7) Å and 20.748(4) Å growing to 32.450(7) Å and 20.800(4) Å.
Inductive effects from intermolecular interactions between complexes and counter-ions may contribute to the ligand field of the Fe(II) centres impacting the overall magnetic profile. However, analysis of contacts between Cl− anions and metalloligands across the experimental temperature range has revealed that contact distances remain relatively stable (Tables S4–8†) with maximum differences between all temperatures ranging from 0.01–0.13 Å. The largest difference in hydrogen bonding distance of 0.13 Å (donor⋯acceptor) was ascribed to the interaction between a pyridyl CH (C10B) and Cl1. In fact, only three of these hydrogen bond distances in total exhibited changes greater than 0.1 Å, attributed in each case to pyridyl CH's. This suggests these contacts change primarily as a result of pyridyl groups twisting with respect to their respective imidazole counterparts, as opposed to a large-scale supramolecular reconfiguration.
As in the air-dried material (2·6.5H2O), solvent appears to play a key role in the magnetic behaviour of single crystals (2-CH3OH). In the 300 K structure, a high degree of thermal disorder prohibited the modelling of all but two solvent molecules – one methanol which is hydrogen bonded to the imidazole CH proximal to the coordination sphere of arm B, and a water molecule which is likely involved in hydrogen bonding interactions with that methanol and both Cl− anions. Both of these solvents are modelled with ¾ occupancy at 300 K, affording Ueq values approximately 2–2.5 times the Ueq values of main residue atoms. At lower temperatures, the occupancy of these two solvent positions increases to 1. Two additional methanol molecules are able to be modelled with partial occupancies at 250 K, which reach full occupancy in the 200 K structure. These latter methanol molecules occupy a pocket in the pore structure which contacts the triethylamine linker components of four Fe(II) complexes (Fig. 5b). Concomitant with the ordering of solvent molecules, the spacing of metal complexes aligned approximately with the a-axis direction increases (Fig. 5c). This effect pushes complexes outward from the solvent pocket, causing the increase in the a-axis as the temperature is lowered to 250 K. As temperature is lowered further, this spacing steadily decreases with reduced thermal motion. Meanwhile, the contact between the enantiomeric pair interacting via ligand arms B and C (running along the a-axis), remains largely unchanged (Tables S4–8†).
The effect of solvent ordering on the lattice positions of the SCO complexes affords more steric freedom to the terminal pyridyl groups on ligand arm A, providing more space to reconform as necessary for a transition to LS. This was quantified by calculating the mutual angles between the secondary bonding vectors of the terminal pyridyl units, defined as the interval between N and opposite C (S9). By measuring the angles between these intervals, we are able to quantify the degree of spread between the three different ligand arms, as we have done in our previous work.42,50
The mutual bond angles between the ligand arms in the HS trapped structure 1 have a much lower average value (55.8°) than the mixed spin structure of the single crystal (2-CH3OH) at 300 K (74.6°). As was found in our previous study36 of the coordination cage [Fe8Ni6L8(CH3CN)12]28+, a contributing factor to SCO behaviour in the [FeL]2+ metalloligand scaffold is the spread of the terminal pyridyl groups. The necessary distortions of the coordination sphere upon ST are more easily incorporated into the transitioning complexes with sufficient angular spread and steric freedom.
Regarding the SCO dynamics in the crystal (2-CH3OH), at 300 K the mutual bond angle (Fig. 6) between arms B and C is measured to be 74.2°, with slightly lower values at lower temperatures and a minimum value of 73.6° at 150 K. This implies that the terminal pyridyl groups of B and C slightly approach colinearity at lower temperatures. Meanwhile, the mutual bond angle values between A and B, as well as that between A and C both give their respective minimum values at 300 K, and both rise slightly to maximum values at 200 K, plateauing at lower temperatures (Fig. 6). Ultimately, these results indicate that the distortion of the octahedral coordination sphere on ST brings about a reconformation of ligand arms, and due to the contact strength between arms B and C, the relatively unhindered ligand arm A is more easily able to accommodate the change in coordination sphere size.
Fig. 6 Mutual bonding angles of the ligand arms in 2-CH3OH from 100 to 300 K. The notation I|J refers to the angle between the secondary bonding vectors of the Ith and Jth ligand arms. For full details and calculated values, see section S9.† |
In the air-dried material 2·6.5H2O, it is likely that solvent ordering/interactions may be a contributing factor to the second spin transition event. The degree of solvation would certainly affect the pore structure in this material, and whether the crystallographic location of solvent pockets remains equivalent between the crystal (2-CH3OH) and the air-dried sample (2·6.5H2O) is ambiguous at this stage, as PXRD results demonstrate that a phase change occurs upon air drying (Fig. S2†). In the fully desolvated material (desolv. 2), the loss of solvent molecules may result in the pore structure collapsing, forcing the material to adopt a higher crystal packing coefficient and thus a higher degree of steric crowding. This may also force molecules in the lattice to adopt intermolecular contacts which are anticooperative with a spin transition to LS, as was noted in the structures of the BF4− and ClO4− analogues.
Footnotes |
† Electronic supplementary information (ESI) available: HR ESI-MS, SEM-EDS, PXRD, STA, solvent masking details, intermolecular interaction tables, calculation of parameters and crystallographic data. CCDC 2338622–2338627. For ESI and crystallographic data in CIF or other electronic format see DOI: https://doi.org/10.1039/d4dt00706a |
‡ Equal contribution. |
This journal is © The Royal Society of Chemistry 2024 |