A ferromagnetically coupled Fe42 cyanide-bridged nanocage

Self-assembly of artificial nanoscale units into superstructures is a prevalent topic in science. In biomimicry, scientists attempt to develop artificial self-assembled nanoarchitectures. However, despite extensive efforts, the preparation of nanoarchitectures with superior physical properties remains a challenge. For example, one of the major topics in the field of molecular magnetism is the development of high-spin (HS) molecules. Here, we report a cyanide-bridged magnetic nanocage composed of 18 HS iron(III) ions and 24 low-spin iron(II) ions. The magnetic iron(III) centres are ferromagnetically coupled, yielding the highest ground-state spin number (S=45) of any molecule reported to date.

One of the challenges in the field of molecular magnetism is to synthetically prepare new nanoarchitectures with high-ground state spin numbers. To date, numerous high-spin (HS) molecules have been reported 25,26 . Here, we report a giant-spin nanocage that contains 18 HS ferromagnetically coupled Fe III (S ¼ 5/2) ions resulting in a molecular ground state spin of S ¼ 45, the largest value known to date 27 . This nanoarchitecture is a mixed valent HS and low-spin (LS) cyano-bridged Fe III-HS 18 Fe II-LS 24 compound, structured as a supramolecular cage with a nanometre-sized inner cavity space.

Results
Preparation of Fe 42 cyanide-bridged nanocage. The strategy used to construct the magnetic nanocage is based on the preparation of metal-organic polyhedra 28 . We used metal-organic complexes as building blocks, which not only act as caps but also contain metal centres and cyano groups that enable the introduction of magnetic interactions. Namely, instead of organic tridentate pyridyl ligands, which can provide a large hollow polyhedral structure, we used monoanionic complex ligand {Fe(Tp)(CN) 3 } À (Tp ¼ hydrotris(pyrazolyl)borate) units [29][30][31] as a trinucleating ligand of the metal ions for constructing magnetic nanocage. In addition, the choice of counter metal ions in these structures importantly enables the adjustment of their magnetic properties, facilitating the creation of HS ground states. We employed iron ions in consideration of the magnetic interaction and redox activity of the ferromagnetic metal-cyanide compound: Prussian blue Fe III 4 Fig. 3, with further details of this measurement).
The most remarkable structural feature of 1 Á 18H 2 O is that the 18 Fe III-HS ions can be identified as the vertices of a highly symmetric entity shown in Fig. 2 Magnetic properties of Fe 42 cyanide-bridged nanocage. Figure 3a shows the magnetic properties of a polycrystalline sample of 1 Á 18H 2 O under a direct current field of 10 kOe from 300 to 30 K and 100 Oe from 30 to 2 K. It indicates the existence of predominantly ferromagnetic interactions and a resulting giant ground-state spin for 1 Á 18H 2 O. At 300 K, the w m T product is 85.5 cm 3 mol À 1 K, and the data in the range 300-30 K can be fitted to the Curie-Weiss law, yielding C ¼ 83.2 cm 3 mol À 1 K and y ¼ þ 6.7 K. This C value is consistent with the expectations (78.8 cm 3 mol À 1 K with g ¼ 2.0) for 18 uncoupled Fe III centres (S ¼ 5/2). On cooling, the w m T value becomes slightly larger with temperature, abruptly increasing to 863 cm 3 mol À 1 K at 2 K ( Supplementary Fig. 4). This magnetic behaviour and the positive Weiss constant suggest the existence of dominant ferromagnetic exchange interactions in 1 Á 18H 2 O. Figure 3b shows w À 1 versus the T-plot for various applied fields, which displays an inflection in w À 1 between 10 and 5 K as excited states depopulate, on further cooling below 5 K the slope starts to become linear again, suggesting that it is just the ground state that is mainly populated, ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6955 following the dependence of a paramagnetic S ¼ 45, g ¼ 2.0, spin unit (dot line).
Moreover, the magnetization (M) at 2 K (Fig. 3c) rapidly increases at low fields, and then steadily increases with H415 kOe to reach a near saturation value of 88.4 m B at 50 kOe, which is in good agreement with the expected value of 90 m B (with g ¼ 2.0) for a ground state of S T ¼ 90/2 ( Supplementary Figs 5 and 6). This magnetization behaviour is significantly higher than the Brillouin curve corresponding to 18 non-interacting S Fe spins (S ¼ 5/2, green line), fitting more closely the Brillouin curve for one S ¼ 45 centre (with g ¼ 2.0, blue line, see Methods). These data support the maximum possible spin state, S ¼ 45, which is the largest spin ground state number of any molecule ever prepared 27 .
To rule out the possibility of intermolecular interaction or a magnetic ordering between the molecule-based giant spins, electron paramagnetic resonance (EPR) spectra of 1 Á 18H 2 O have been examined. Figure 3d shows no evidence of intermolecular interaction is present within EPR on decrease of temperature to 1.6 K; the g ¼ 2.0 absorption increases in intensity with no significant variation of line width or shift in resonance field position to evidence intermolecular interactions. No evidence of anisotropy is observed in EPR measurements, it is reasoned that the geometry of the molecular structure of 1 Á 18H 2 O causes the cancelling of single ion anisotropic contributions. In addition, the temperature dependence of magnetization for 1 Á 18H 2 O under various applied magnetic fields does not show evidence of spontaneous magnetization down to 0.5 K (see Supplementary Fig. 7). These results indicate that the cyanide-bridged magnetic Fe 42 nanocage exhibits the maximum spin ground state with isolated molecule S ¼ 45.

Discussion
DFT calculations were carried out to estimate changes in the electronic structures of 1 Á 18H 2 O. Calculations were simplified to a cyano-bridged molecular square 38 Figure 4 shows HS ferromagnetic (HSFM) state was the ground state, and LS ferromagnetic (LSFM) and antiferromagnetic (LSAF) states were above 5.5 kcal mol À 1 and 1.5 kcal mol À 1 , respectively. DFT calculations indicate the ferromagnetic character of a magnetic coupling between the two diagonal Fe III-HS ions in the square framework. The calculated J value (35.5 cm À 1 at the B3LYP* level) for the tetranuclear cyanobridged square complex overestimates the magnitude of the exchange coupling parameters. The overestimation of coupling constants by a factor of 2-4 is not unusual in DFT calculations 40,41 . The calculation predicts the correct sign for J corresponding to the ferromagnetic nature of the ground state in the [Fe 42 ] nanocage.
We have presented a Fe 42 cyanide-bridged nanocage with a HS framework. Many metal-cyanide clusters have been synthesized since the discovery of Prussian blue. Our metal-cyanide polyhedron is the largest cyanide-bridged polynuclear cluster and exhibits a rare hollow structure. Among the various morphologies of nanoarchitectures, hollow spheres are of great interest because of their high surface to volume ratio and large pore volume, which could be exploited for promising applications in the controlled encapsulation and release of molecules.
In summary, we report a new high-nuclearity iron complex with a HS framework. In the Fe 42 cyanide-bridged nanocage,     3 ]. The crystals were washed with H 2 O and dried under reduced pressure for 12 h. As prepared compound, before vacuum drying, has approximately 50H 2 O molecules inside the cage (compound 1 Á ca 50H 2 O), which become partially desolvated when exposed to air at room temperature. Therefore, the physical measurements performed on compound 1 Á 18H 2 O were prepared carefully to prevent desolvation. The reported structures have been characterized by the single-crystal X-ray crystallography (Supplementary Tables 1-3 and Supplementary Data 1 and 2). X-ray structure determination. Single-crystal synchrotron radiation X-ray diffraction experiments were performed at 295 K for 1 Á 18H 2 O, and 100 K for 1 Á ca 50H 2 O using a Rigaku Mercury2 CCD detector at BL02B1/SPring-8 (Hyogo, Japan). The wavelength of the incident X-ray was 0.6186 or 0.6202 Å. We used anomalous dispersion coefficients for structure refinement, f 0 and f 00 , in dependence on X-ray energy calculated on the original FPRIME code of Cromer.
SEM measurements. Scanning electron microscopy studies were performed with a Hitachi Ultra-high Resolution Scanning Electron Microscope SU8000.
Physical measurements. Magnetic susceptibility measurements of samples were performed on a Quantum Design SQUID (MPMS XL-5 and MPMS XL-7) magnetometer. To prevent the loss of uncoordinated water molecules, the sample was introduced directly into the sample chamber at 100 K without purging, while flowing He gas. Keep the sample stay at 100 K for several minutes and then purge the chamber. At the end, vent the sample space with He gas and start the measurement. Data were corrected for the diamagnetic contribution calculated from Pascal constants. The Mössbauer spectra (isomer shift versus metallic iron at room temperature) were measured using a Wissel MVT-1000 Mössbauer spectrometer with a 57Co/Rh source in the transmission mode. All isomer shifts are given relative to a-Fe at room temperature. EPR measurements were performed with the Terahertz ESR Apparatus (TESRA-IMR) installed in the magnetism division of Institute of Materials Research (IMR), Tohoku University. A solenoid magnet is fed by a 90-kJ capacitor bank delivering a field pulse of 25 ms in width. Applied fields from 0 to 30 T were investigated for a range of EPR microwave frequencies from 90 to 405 GHz generated by backward-travelling wave oscillators. Measurements were performed in a He4 cryostat down to 1.6 K.
Simulation of magnetization process. The magnetization curve was compared with the mean-field simulation assuming an HDVV (Heisenberg-Dirac-Van Vleck)-type spin Hamiltonian: Ĥ ¼ À 2J P (i,j) Ŝ i Á Ŝ j , where a uniform isotropic exchange parameter J makes all the pair of S ¼ 5/2 spins coupled. This Hamiltonian gives the energy eigenvalues E(S total ) ¼ À J {S total (S total þ 1) À 18s(s þ 1)} with the resultant spin of the molecule S total , and the magnetic susceptibility is easily calculated following the Van Vleck formula. The simulation curve with a ferromagnetic coupling zJ/k B ¼ þ 0.5 K (z ¼ 17) and g factor of 2.0 agreed well with the experimental values.
XAS measurements. X-ray absorption spectra were measured on the soft X-ray undulator beam-line BL25SU at Spring 8, Japan. Soft X-rays circularly polarized from a twin helical undulator were monochromated and focused onto a thin polycrystalline layer of 1 Á 18H 2 O. The measured sample was fixed with carbon tape to a sapphire sample holder. X-ray absorption spectra were measured by the total electron yield method in which the sample current is directly measured while scanning the photon energy. Measurements were performed at zero applied magnetic field and hence both positive and negative X-ray helices resulted in equivalent absorption spectra. The sample chamber for the soft XAS keeps a high vacuum of 10 À 5 Pa or the better. Measurements of 1 Á 18H 2 O were repeated incrementally over several days and were found not to exhibit changes in spectral line shape with respect to the time spent under vacuum. Care was taken during XAS measurements to control the effect of photoreduction. The incident beam intensity was incrementally reduced until consistent multiple measurements at the same sample spot were obtained after an intensity reduction to 7%. To ensure damage was not encountered in short time periods, rapid measurements over defined features of the L 3 edge were measured repeatedly and ensured to be coincident.
DFT calculation. Full details of computational method are given in the Supplementary Methods.