Slow Phase Transition-Induced Scan Rate Dependence of Spin Crossover in a Two-Dimensional Supramolecular Fe(III) Complex

Abstract

Open AccessCCS ChemistryRESEARCH ARTICLE18 Mar 2022Slow Phase Transition-Induced Scan Rate Dependence of Spin Crossover in a Two-Dimensional Supramolecular Fe(III) Complex Zhao-Yang Li, Ying-Ying Wu, Yao Li, Jin-Hua Wang, Arshia Sulaiman, Mohammad Khurram Javed, Ya-Chao Zhang, Wei Li and Xian-He Bu Zhao-Yang Li School of Materials Science and Engineering, Nankai University, Tianjin 300350 Google Scholar More articles by this author , Ying-Ying Wu School of Materials Science and Engineering, Nankai University, Tianjin 300350 Google Scholar More articles by this author , Yao Li School of Materials Science and Engineering, Nankai University, Tianjin 300350 Google Scholar More articles by this author , Jin-Hua Wang School of Materials Science and Engineering, Nankai University, Tianjin 300350 Google Scholar More articles by this author , Arshia Sulaiman School of Materials Science and Engineering, Nankai University, Tianjin 300350 Google Scholar More articles by this author , Mohammad Khurram Javed School of Materials Science and Engineering, Nankai University, Tianjin 300350 Google Scholar More articles by this author , Ya-Chao Zhang *Corresponding authors: E-mail Address: [email protected] E-mail Address: [email protected] Guizhou Provincial Key Laboratory of Computational Nano-Material Science, Guizhou Education University, Guiyang 550018 Google Scholar More articles by this author , Wei Li School of Materials Science and Engineering, Nankai University, Tianjin 300350 Google Scholar More articles by this author and Xian-He Bu *Corresponding authors: E-mail Address: [email protected] E-mail Address: [email protected] School of Materials Science and Engineering, Nankai University, Tianjin 300350 College of Chemistry, Nankai University, Tianjin 300071 Google Scholar More articles by this author https://doi.org/10.31635/ccschem.022.202101721 SectionsSupplemental MaterialAboutAbstractPDF ToolsAdd to favoritesDownload CitationsTrack Citations ShareFacebookTwitterLinked InEmail Spin crossover (SCO) is commonly accompanied by a synchronous phase transition. A few phase transition-coupled SCO compounds have been reported, yet the synergy between SCO and phase transition on different time scales has not been explored. Herein, we report an [Fe(H-5-Cl-thsa-Et)(5-Cl-thsa-Et)]·H2O ( 1·H2O; H2-5-Cl-thsa-Et = 5-chloro-salicylaldehyde ethylthiosemicarbazone) Fe(III) complex that displays a two-dimensional supramolecular structure and SCO behavior above room temperature. Its dehydrated form 1 exhibits a two-step spin transition with a plateau in the temperature-dependent magnetization (M−T) curve at room temperature and a 51 K thermal hysteresis loop (Tc↑↓ = 299/248 K) at a rate of 5 K/min. The improved SCO performance in 1 could be attributed to the stronger intralayer but weaker interlayer interactions, which is supported by single-crystal structural analysis and density functional theory calculations. Remarkably, complex 1 displays an unusual scan rate-dependent SCO behavior at rates of 0.5−30 K/min, in which M−T curve plateaus appear at lower scan rates (<10 K/min) but vanish at faster scan rates (≥10 K/min). Scan rate-dependent differential scanning calorimetry, powder X-ray diffractometry, time-dependent magnetic moment decays, and infrared spectroscopy consistently reveal that the slow structural relaxation is coupled with a slow crystallographic phase transition, which is the mechanism for the unusual scan rate-dependent SCO. Download figure Download PowerPoint Introduction Spin crossover (SCO) complexes have been among the most intensively studied switchable molecular materials over the past few decades. Reversible switching between high-spin (HS) and low-spin (LS) states is generally accompanied by changes in magnetic, optical, dielectric, and other material properties. This feature has prompted efforts to fundamentally understand the SCO process and develop potential applications for information storage, sensors, and visual displays, among others.1–5 Many such applications require materials that exhibit thermal hystereses in their spin transitions, ideally at room temperature and with 30–50 K hysteresis loops.6,7 Many SCO molecules and polymers have been discovered that are primarily based on Fe(II) and Fe(III) centers. However, only a few SCO compounds exhibit concurrent room temperature transitions with wide hysteresis loops,8 and many examples face problems associated with atmospheric oxidation of their Fe(II) forms and crystal degradation through solvent loss. Therefore, designing robust Fe(III) SCO complexes that exhibit strong cooperativity and understanding the origin of SCO at the molecular level remain challenging.9,10 Thermal hysteresis is a collective phenomenon driven by communication between spin centers. The physical origin of the process is complicated but can be speculated using physical models that involve elastic interactions, electrostatic contributions, coupling between spins, degrees of spin-lattice freedom, and electron-deformation interactions through phonon fields.9,11–14 Experiments have revealed that pronounced cooperative effects can be acquired by building covalent bridges between metal centers or by introducing H-bonding or π–π stacking moieties into the surrounding coordination environment.7,8,15–17 These interactions serve as propagators that transmit electronic and structural information between the building blocks. Owing to its collective nature, hysteresis is difficult to control; however, cooperativity can be enhanced by increasing the strengths of effective intermolecular interactions.18 We recently studied a family of mononuclear Fe(III) multistep, cooperative SCO compounds that exhibit the following characteristics19–21: (1) H2O-mediated hydrogen bonds and other classical H-bonds, such as N−H···O and O−H···N linkages, play critical roles in enhancing system cooperativity through strong intermolecular interactions. (2) Symmetry breaking accompanying three-, five-, and six-step transitions associated with H2O-formed hydrogen bonds. (3) Hysteresis loops, but at temperatures <15 K. We concluded that water molecules in the lattice are essential for multistability and cooperativity. The importance of solvent molecules in the lattice is also apparent in Hofmann-type Fe(II) SCO systems.22–24 Guest molecules intercalated between layers or located in pores play vital roles via host–host and host–guest interactions that determine the strength of the cooperativity.25–29 To the best of our knowledge, solvent molecules are crucial for intermolecular interaction regulation and hence affect the entire SCO behavior. In addition to pursuing high-performance SCO compounds, that is, those exhibiting a large hysteresis loop, abrupt spin transition, and room temperature Tc, researchers have more recently focused on rare examples that exhibit scan rate dependence.30–33 The cooling branch of a hysteresis loop is commonly believed to depend more on the scan rate than the heating branch, as one generally starts with an HS lattice, and an LS center occupies less volume than an HS center. This causes a change in the width of the hysteresis loop but barely affects its shape. Herein, we report an unusual temperature-scan rate-dependent SCO complex 1 that exhibits significant shape and positional M−T curve changes at various scan rates. 1 was obtained by the in situ dehydration of [Fe(H-5-Cl-thsa-Et)(5-Cl-thsa-Et)]·H2O (H2-5-Cl-thsa-Et = 5-chlorosalicylaldehyde ethylthiosemicarbazone) ( 1·H2O). 1·H2O undergoes SCO above room temperature, whereas 1 exhibits room temperature SCO (Tc↑↓ = 299/248 K) and a fixed hysteresis loop of 51 K at a scan rate of 5 K/min. Single-crystal structure analyses reveal that both 1·H2O and 1 have clear two-dimensional (2D) layer structures. Dehydration expands the interlayer distances and strengthens intermolecular interactions within 2D layers. Density functional theory (DFT) calculations support a molecular mechanism in which dehydration relieves the internal chemical pressure of the lattice by destroying water-mediated hydrogen bonds. Both experimental and theoretical results consistently indicate that (1) scan rate-dependent SCO involves a slow crystallographic phase transition and (2) dehydration results in intensifying the intralayer intermolecular interactions, which effectively enhances SCO cooperativity. To the best of our knowledge, 1 is the first example of Fe(III) SCO compounds that exhibit scan rate dependence. Experimental Methods Measurements Magnetic susceptibility data were collected using a Quantum Design MPMS3 superconducting quantum interference device (SQUID) magnetometer with an applied magnetic field of 0.1 T in sweep mode. Data were first collected for 1·H2O between 5 and 300 K and then in situ collected between 5 and 400 K at a scan rate of 5 K/min for four continuous cycles for 1. The scan rate-dependent behavior of 1 was then studied at 0.5, 1, 2, 5, 10, 20, and 30 K/min. The data were corrected for contributions from the sample holder and intrinsic diamagnetism. Mössbauer experiments were performed using a 57Co/Rh (925 MBq) source in a constant-acceleration transmission spectrometer (Ulvac Cryogenics Inc., Japan) equipped with a closed-cycle helium refrigerator (Ulvac Cryogenics Inc.). The spectrometer was calibrated using standard Fe foil. The driving speed of the radiation source was ±6.0 mm/s. A fresh sample of 1·H2O was analyzed at 5 K. The sample was cooled to 80 K after being heated in situ to 400 K (identical process to SQUID) and then examined at 320 K. The data collected at 320 K correspond to complex 1. High-resolution powder X-ray diffractometry (PXRD) profiles were obtained using a Rigaku SmartLab instrument that features a PhotonMax high-flux 9 kW rotating anode X-ray source coupled with a D/tex Ultra 250 detector and attached to an Anton Paar TTK 600 Low Temperature Chamber. The detailed conditions were a reflection geometry between 100 and 400 K at 2, 5, and 10 K/min (2θ scan range: 5−35°; increment: 0.02°; 2θ scan rate: 10°/min; sample atmosphere: vacuum). Thermogravimetric analysis (TGA) was conducted on a Rigaku standard thermogravimetric-differential thermal analysis (TG-DTA) analyzer from room temperature to 800 °C at 10 K/min in a flowing argon atmosphere. 1·H2O was subjected to differential scanning calorimetry (DSC) between 170 and 410 K on a TA (United States) DSC-25 instrument in a helium atmosphere in an aluminum crucible at 15 K/min for the first cycle (in situ dehydration); 1 was then subjected to multicycle DSC between 170 and 370 K at 2, 5, 10, 15, 20, 25, and 30 K/min. Scanning electron microscopy (SEM) images were acquired using a JSM-7800F instrument. X-ray photoelectron spectroscopy (XPS) was performed using a Thermo Scientific ESCALAB 250Xi instrument. Time- and temperature-dependent infrared (IR) reflectivity spectra for 1 were acquired using a Bruker Tensor 27 instrument with an in situ cell (OPERANDO-DRCRO-R). 1·H2O was cooled to 100 K after being heated to 440 K and then heated to 320 K at 20 K/min. Single-crystal X-ray diffractometry (SC-XRD) data for 1·H2O were collected on an ADSC Quantum-210 detector with synchrotron radiation (λ = 0.700 Å) at the two-dimensional supramolecular crystallography (2D-SMC) beamline of the Pohang Accelerator Laboratory, South Korea. The structures were solved using direct methods (SHELXS-97/SHELXS-2014) and refined using full-matrix least-squares calculations on F2 (SHELXL-97) included in the SHELX-TL program package.34 SC-XRD data for 1 were collected on a Rigaku XtaLAB MM007 CCD diffractometer with Cu Kα radiation (λ = 1.5418 Å). Diffraction profiles were integrated using the CrystalAlice PRO program. Structures were solved and refined by full-matrix least-square methods using Olex2 software35 within the SHELXT and SHELXL programs, respectively. All nonhydrogen atoms were refined with anisotropic displacement parameters. X-ray crystallographic data have been deposited at the Cambridge Crystallographic Data Centre (CCDC): CCDC number 2035662 for 1·H2O (100 K), 2145471 for 1·H2O (300 K), 2035660 for 1 (100 K), and 2035661 for 1 (310 K). These data can be obtained free of charge from the Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif Synthesis of H2-5-Cl-thsa-Et H2-5-Cl-thsa-Et was synthesized in a similar manner to that reported.36,37 5-Chloro-2-hydroxybenzaldehyde (156.0 mg, 1.0 mmol) and 4-ethyl-3-thiosemicarbazide (119.0 mg, 1.0 mmol) were refluxed in dried ethanol (10.0 mL) for 3 h. Colorless crystals of H2-5-Cl-thsa-Et were obtained in 78% yield upon slow evaporation of the filtrate. Synthesis of 1·H2O A 14.0-mg quantity (0.4 mmol) of dimethylamine was added to a solution of H2-5-Cl-thsa-Et (51.6 mg, 0.2 mmol) in MeOH (8.0 mL), and the resulting mixture was stirred for 20 min at room temperature. A 1.0-mL portion of an aqueous solution containing 0.1 mmol (40.4 mg) [Fe(NO3)3]·9H2O was added to the mixture, which was heated at 80 °C for 8 h. Black microcrystals of 1·H2O were obtained in 79% yield (based on [Fe(NO3)3]·9H2O) after cooling to room temperature and washing with cold methanol/water. Single crystals of 1·H2O suitable for XRD were obtained by recrystallization from 1:1 MeOH:CH3CN. Elemental analysis results for 1·H2O (C20H23Cl2FeN6O3S2): calculated, H, 3.95; C, 40.97; N, 14.33%. Found: H, 4.052; C, 40.827; N, 14.385%. Acquisition of 1 Dehydrated single crystals of 1 were collected after the in situ dehydration of 1·H2O at 380 K for 2 h in the SQUID instrument (pressure: 3–8 Torr; state: purged). Single crystals of 1 suitable for SC-XRD were carefully picked out from the vacuum grease. Finding high-quality single crystals of 1 after annealing is extremely difficult, as reported for most cases. After dozens of trials, we determined the structure of 1 at 100 K, which corresponds to the LS state. Single-crystal experiments did not provide a reasonable structure for the HS state. As an alternative, we used DFT calculations to optimize the HS state structure based on the experimental structure of the LS state (see Computational section). Elemental analysis for 1 (C20H21Cl2FeN6O2S2): calculated, H, 3.72; C, 42.27; N, 14.79%. Found: H, 4.079; C, 42.180; N, 14.780%. Computational Section DFT calculations were performed with the SIESTA package.38 Full structure relaxations, including the unit cell and atomic positions, were performed based on the conjugate gradient algorithm. The X-ray crystal structure determined at 100 K was chosen as the starting geometry. We set the parameter controlling the forced convergence of the geometry optimization to 0.04 eV/Å. Our calculations used the Perdew–Burke–Ernzerhof (PBE)39 exchange-correlation functional based on the generalized gradient approximation. We applied a Hubbard-U correction40 at the Fe site to account for the strong correlation effects of the 3d electrons. The Ueff parameter was set to 1.55 eV, which is optimal for PBE in studying spin-crossover properties.41 The DFT-D2 correction developed by Grimme42 was included to consider noncovalent interactions involved in the secondary coordination environment in the molecular crystals. We employed Troullier–Martins norm-conserving pseudopotentials,43 and nonlinear core correction was included for the magnetic Fe atom. We used a triple-ζ plus polarization numerical atomic basis for the basis sets. The finite range of the atomic orbitals was controlled using an energy shift of 50 meV. The fineness of the real-space grids for the numerical integrals was controlled with a mesh cutoff of 200 Ry. Only the Γ point was used for Brillouin zone sampling. We simulated the smearing of electron occupations based on the Methfessel–Paxton function at an electronic temperature of 1000 K. The electronic energies in different spin states were obtained by standard spin-polarized DFT calculations under a fixed spin-moment scheme. Results and Discussion Design and synthesis Complex 1·H2O was prepared by refluxing H2-5-Cl-thsa-Et, Fe(NO3)3·9H2O, and diethylamine in methanol/water (8:1 v/v). The single-crystal structure of 1·H2O confirms that it is a neutral complex (see “Single-crystal structures” section). While 1·H2O was prepared in a manner similar to our previously reported compounds, we emphasize that the use of diethylamine as a mild base is crucial in preparing this charge-neutral Fe(III) complex. The addition of an excess quantity of strong base leads to complete deprotonation and an anionic ferric compound, whereas the addition of one equivalent of a mild base leads to mono-deprotonation and a cationic ferric product. The use of a mild organic base in controlled amounts (1.5−2.0 equiv) is required to obtain the neutral ferric compound that displays SCO behavior (see “Magnetic properties” section). This crucial synthetic strategy is illustrated in Supporting Information Scheme S1. Basic characterizations The microcrystalline product was collected by filtration, washed with cold methanol/water and used for physical characterization studies (see Experimental Section). Single crystals of 1·H2O suitable for X-ray diffractometry were obtained by recrystallization from MeOH/CH3CN (1:1 v/v). The single-crystal structure of 1·H2O is shown in Supporting Information Figure S1 (also see the “Single-crystal structures” section). PXRD at room temperature confirmed the phase purity and crystallinity of 1·H2O ( Supporting Information Figure S2). TGA revealed that the water molecule in 1·H2O is gradually lost at T ≥ 340 K with a plateau appearing at 430 K, indicating the complete transformation of 1·H2O into 1 ( Supporting Information Figure S3a); 1 stabilizes at ∼500 K ( Supporting Information Figure S3b). Based on the TGA data, complex 1 can be obtained by in situ heating in a SQUID magnetometer. Single crystals of 1·H2O and 1 were subjected to SEM, which revealed that the crystal surface of 1 had slightly degraded, but its morphology was well retained upon thermal treatment, as shown in Supporting Information Figure S4, indicating that 1 is thermally robust, which is important for a utility material. XPS confirmed the exclusive presence of Fe(III) and the absence of Fe(II) following dehydration ( Supporting Information Figure S5). Magnetic properties and Mössbauer spectra The magnetic properties of 1·H2O were examined by temperature-dependent magnetic susceptibility (M−T) studies at 5−300 K using a SQUID magnetometer with a 0.1 T DC field. As shown in Supporting Information Figure S6, χMT at 5 K is 0.47 cm3 K mol−1, which is consistent with the expected value for LS Fe(III) ions (S = 1/2). χMT is ∼1.16 cm3 K mol−1 at 300 K, which indicates that an incomplete spin transition may occur above room temperature. Magnetic experiments were next performed at 5−400 K with a 5 K/min scan rate. Supporting Information Figure S7 shows the presence of a two-step SCO for the first cycle. Based on the sample environment (3−8 Torr, Tmax = 400 K) in the SQUID magnetometer and the TGA data, dehydration must have occurred during heating in the SQUID experiments, which indicates that 1· x H2O (x = 1–0) is present during the first heating process (5–400 K) and that 1 is present during the first cooling process (400–5 K). Thus, the first-cycle SCO behavior is irreversible and not reproducible. Further study of the magnetic properties of 1 at 5−400 K over three continuous cycles at 5 K/min reveals the stable and reproducible SCO behavior shown in Figure 1. The two-step transitions during heating and cooling display unusual shapes that are gradual on the HS side and more abrupt on the LS side near room temperature. The transition temperatures are Tc↑ = 299/344 K and Tc↓ = 290/248 K, yielding hysteresis loops of 51 K (symmetric on the LS side) and 54 K (nearly symmetric on the HS side). We observed that a plateau between 300 and 320 K appears only during heating and is potentially ascribable to an ordered HS/LS mixed state based on the χMT value. Figure 1 | M−T curve for 1 acquired during three continuous cycles at a scan rate of 5 K/min. The three curves are almost identical. Download figure Download PowerPoint Mössbauer spectra were acquired at 5 K for 1·H2O and at 320 K for 1 ( Supporting Information Figure S8 and Table S1). The data collected at 5 K show a single quadrupole doublet (isomer shift (IS) = 0.2654; quadrupole splitting (QS) = 2.6869 mm/s; LS area fraction = 100%), whereas the data collected at 320 K display a QS of 0.9281 mm/s for the HS fraction (54.1% area) and 0.2769 mm/s for the LS fraction (45.9% area). The results are consistent with the magnetic susceptibility data and suggest the presence of a mixed HS/LS state at 320 K that exhibits two-step spin transition behavior. The removal of molecular water from 1·H2O significantly improves its SCO behavior and results in a two-step spin transition (mixed HS/LS state appears), a room temperature Tc, and an ∼50 K hysteresis loop for 1. Any change in SCO behavior is highly likely to arise from structural modifications. Hence, we examined differences in the single-crystal structures of 1·H2O and 1 with the goal of finding clues about the molecular mechanism, specifically the structural change within and between 2D layers. Single-crystal structures 1·H2O was subjected to SC-XRD at 100 and 300 K, whereas 1 was measured at 100 and 310 K ( Supporting Information Table S2). 1·H2O crystallizes in the monoclinic P21/n space group, while 1 crystallizes in the C2 space group at 100 K but transforms into Pn at 310 K, each with Flack parameter values of 0.021(10) and 0.01(2), which confirms that the absolute structure of 1 is correct. Two inequivalent molecules are present in the unit cell of 1 at 310 K because of symmetry breaking, which enables the HS and LS Fe(III) centers to be distinguished based on the Fe−N bond lengths ( Supporting Information Figure S9). This is consistent with the observed magnetic quantity χMT = 2.71 cm3 K/mol, which corresponds to the sum of the magnetic susceptibilities of 1/2 HS FeIII and 1/2 LS FeIII. A single-crystal of 1·H2O was used to determine changes in the values of unit cell parameters upon in situ heating. The summary in Supporting Information Table S3 shows that although solvent loss occurs as the crystal is heated, the b axis elongates, and the cell volume expands ((VHS−VLS)/VLS = 5.7%), consistent with the involvement of SCO. In addition, we noted that the crystal volume of 1 (V = 2383 Å3) is almost the same as that of 1·H2O (V = 2385 Å3), which is ascribable to the increased interlayer spacing, consistent with the PXRD data. The asymmetric unit comprises an independent Fe(III) center, two ligands, and an uncoordinated water molecule. The metal center is hexacoordinate with a N2O2S2 donor set. 1·H2O exhibits Fe−N bond distances of 1.9493(16)/1.9501(16) at 100 K and 2.005(3)/2.011(3) at 300 K, which are characteristic of LS Fe(III) complexes, and the analogous distances in 1 are 2.021(12)/2.024(13) Å at 100 K, also characteristic of LS Fe(III) centers ( Supporting Information Table S4). The SC structure for the HS state of 1 was not obtained experimentally despite many efforts. Instead, we performed DFT calculations to optimize the HS structures of 1·H2O and 1, as shown in Supporting Information Figure S10. The Cartesian coordinates of the optimized HS state structures of 1·H2O and 1 are provided in Supporting Information Tables S5 and S6, respectively. The apparent differences between 1·H2O and 1 are derived from their different intermolecular interactions and packing structures. As shown in Supporting Information Figure S11, three types of classical hydrogen bonds are observed in the structure of 1·H2O ( Supporting Information Table S7). One class involves intermolecular hydrogen bonding to water molecules (O(water)−H···O(C−O)), with O···O distances of 3.171 and 3.189 Å. The second class involves N(2)−H···O(water) (2.904 Å) and O(water)−H···N(5) (3.147 Å) hydrogen bonding to H2O and results in the formation of a one-dimensional (1D) chain structure. The third involves intermolecular N(3)−H···O(1) and N(6)−H···O(2) hydrogen bonds with N···O distances of 2.758 and 2.826 Å, respectively. However, for 1, only one type of effective hydrogen bond (N−H···O(C−O)) exists ( Supporting Information Table S7), as hydrogen bonds involving H2O are absent. The intermolecular N−H···O(C−O) hydrogen bonds form a 1D chain structure with N···O distances of 2.806/2.827 Å, as shown in Supporting Information Figure S12. The 2D layer structure involves a grid-shaped arrangement composed of crossed 1D chains. The nearest Fe(III)-ion distances in the 2D layer are 6.997/7.023/10.981 Å in 1, which are shorter than the values of 7.116/7.321/11.197 for 1·H2O, as shown in Figures 2a and 2b. These differences indicate that the removal of molecular water creates room within the 2D layer, which enables the Fe(III) centers to become more closely packed (0.12−0.30 Å shrinkage of Fe···Fe distances in 1). Figure 2 | 2D supramolecular structures of (a) 1·H2O and (b) 1, highlighting the shortest Fe···Fe distances. Note: red dashed lines represent hydrogen bonding; Fe: green ball atoms; O: red; N: blue; Cl: green; S: yellow; C: white. Download figure Download PowerPoint Another apparent difference between 1·H2O and 1 is the distance between the 2D layers, which is shorter in 1·H2O than in 1. As shown in Figures 3a and 3b, the distance between the virtual layers is 11.667/11.695 Å in 1·H2O but 12.462 Å in 1. The average Fe···Fe interlayer distance is 12.218 Å in 1·H2O, but in 1, it is increased by 10.4% (13.493 Å), as shown in Supporting Information Figure S13. There are no effective interlayer interactions in either 1·H2O or 1, except for the insignificant C−H···Cl and Cl···Cl contacts ( Supporting Information Figure S13), which indicates that the most effective intermolecular interactions in 1 originate from inside the 2D layer. Supporting Information Scheme S2 represents the structural modification caused by dehydration. Two factors can be taken into account in revealing the correlation of the 2D supramolecular structure and the improved SCO behavior: (1) the van der Waals forces are the dominant interactions between layers in both 1·H2O and 1, which makes it easier to change the spacing between layers compared with the three-dimensional (3D) covalent frameworks; (2) hydrogen bonding is the dominant interaction in the 2D supramolecular layers, which can be easily modulated or “tailored” to enhance the cooperativity compared with 2D/3D Hofmann-type SCO coordination polymers. Therefore, we conclude that cooperation between the interlayer van der Waals and intralayer hydrogen bonding interactions plays a crucial role in enhancing SCO cooperativity in supramolecular compound 1. Figure 3 | 3D structures of (a) 1·H2O and (b) 1, showing distances between virtual 2D layers (least-square planes generated by selecting the Fe atoms in Mercury software). Download figure Download PowerPoint DFT calculations We explored the electronic properties of molecular crystals of 1·H2O and 1 using periodic dispersion-corrected DFT calculations to understand the physical origin of the change in SCO behavior. The results show that the distances between adjacent magnetic centers in the cells of all spin states (S = 2, 6, and 10, Supporting Information Figure S14a) are shorter by 0.03–0.26 Å in the absence of molecular water ( Supporting Information Figure S14b). This finding is consistent with experimental X-ray structures. We first calculated the changes in electronic energies accompanying the spin transitions, which revealed spin-adiabatic energy differences (ΔE) for the mixed-spin−LS and HS−LS centers in 1 that are 1–5 kJ/mol less than those in 1·H2O due to the release of molecular water ( Supporting Information Figure S14c). In particular, the HS–LS energy splitting of 1·H2O is predicted to be 7.6 kJ/mol, which matches well with the measured enthalpy change (6.59 kJ/mol). In fact, if variations in the entropy differences caused by intermolecular interactions are neglected, a decrease in ΔE results in a lower Tc, which is consistent with the experimental results. Next, we monitored changes in the coordination environment upon water release by examining the partial density of states of the Fe-3d and ligand-p orbitals. In the HS state, the five 3d electrons fill the t2g- and eg-derived majority-spin orbitals, which leaves the minority-spin orbitals nearly unoccupied. This essential half-fi