Guest, Light and Thermally Modulated Spin Crossover in [Fe(II) 2 ] Supramolecular Helicates

: A new bis–pyrazolylpyridine ligand (H 2 L) has been prepared to form functional [Fe 2 (H 2 L) 3 ] 4+ metallohelicates. Changes to the synthesis yield six derivatives; X@[Fe 2 (H 2 L) 3 ]Cl(PF 6 ) 2 · x CH 3 OH ( 1 , x =5.7 and X=Cl; 2 , x =4 and X=Br), X@[Fe 2 (H 2 L) 3 ]Cl(PF 6 ) 2 · y CH 3 OH·H 2 O ( 1a , y =3 and X=Cl; 2 , y =1 and X=Br) and X@[Fe 2 (H 2 L) 3 ](I 3 ) 2 ·3Et 2 O ( 1b , X=Cl; 2b , X=Br). Their structure and functional properties are described in detail via single crystal X-ray diffraction (SCXRD) experiments at several temperatures. 1a and 2a are obtained from 1 and 2 , respectively, via single-crystal-to-single-crystal (SCSC) mechanisms. The three possible magnetic states, [LS–LS], [LS–HS] and [HS–HS] can be accessed over large temperature ranges, thanks to the structural non-equivalence of the Fe(II) centres. The nature of the guest (Cl – vs Br – ) shifts the SCO temperature by ca. 40 K. Also, metastable [LS–HS] or [HS–HS] states are generated through irradiation. Helicates (X@[Fe 2 (H 2 L) 3 ]) 3+ persist in solution. (A5), model Flash 1112 at the Servei de Microanàlisi of CSIC, Barcelona, Spain. IR spectra were recorded as KBr pellet samples on a Nicolet AVATAR 330 FTIR spectrometer. Positive ion ESI TOF mass spectrometry experiments were performed on a LC/MSD-TOF (Agilent Technologies) at the Unitat d’Espectrometria de Masses de Caracterització Molecular (CCiT) of the University of Barcelona. The experimental parameters were: capillary voltage 4 kV, gas temperature 325ºC, nebulizing gas pressure 15 psi, drying gas flow 7.0 L min -1 , and fragmentor voltage ranging from 175 to 300 V. Samples (μL) were introduced into the source by a HPLC system (Agilent 1100), using a mixture of H 2 O/MeCN (1/1) as eluent (200 μL min –1 ).


Introduction
The phenomenon of spin crossover (SCO) may be encountered for transition metals that exhibit two possible distributions of the electrons among their crystal field split d orbitals, if both configurations are sufficiently close in energy. [1] By far, the most commonly studied case is that of octahedral Fe(II) ions, which, may switch between a diamagnetic (S = 0, low spin, LS) and a paramagnetic state (S = 2, high spin, HS). [2] The switching comes about by means of external stimuli, [3] such as changes to the temperature [4] or pressure, [5][6] light irradiation, [7] or modulation of the crystal field near the metal by secondary components not directly bound to it. [8][9][10] The spin transition (ST) is accompanied by drastic changes to the physical properties of the material (optical, magnetic, electrical, etc.) also affecting the structure, which converts these systems into very promising candidates for the implementation of functional/switchable nanoscopic devices. [11] The tools and concepts developed in coordination supramolecular chemistry [12][13] could be very beneficial for exploiting the SCO at the molecular scale. One particular challenge is designing molecular systems featuring more than one spin active center with spin states controllable by means of external stimuli. A pioneering example was a molecular grid of four Fe(II) centers made with a specially designed ligand. Thise assembly couldan be brought to three different HS/LS state combinations of its metals by using light irradiation or by controlling the temperature. [14] In this context, a well-known family of coordination architectures that can be very useful are the so called metallohelicates. [15] These species are amenable to rational design; the right choice of metals and polydentate ligands allows the prediction and formation of helicates with different numbers of metals and strands. [16][17] In addition, with the right choice of ligand donors, it is possible to chemically tune the crystal field around the metals (ege.g. Fe(II)) of the supramolecular assembly in order to facilitate the occurrence of SCO. [18][19][20][21][22] Some of the reported examples show evidence that the spin active centers can be brought from the LS to a metastable HS state in response to light irradiation, via the LIESST (light induced excited state trapping) effect. [23] With a more sophisticated design of ligands, metallohelicates are also amenable to selectively capture guest species inside them, [24][25][26][27] which offers a valuable opportunity for modulating the functional properties of their components, such as the switching behavior of potential SCO metals. In fact, such a tuning of the SCO behavior through encapsulation of guests is extremely rare in supramolecular chemistry, and when encountered, it has indeed led to only very minor effects. [28]

Synthesis
The bis-β-diketone H2L1 (Scheme I) was prepared through a common Claisen condensation between a ketone and an ester (Fig. S1), as previously reported for other related ligands. [31] This species served as precursor of the bis-pyrazolylpyridine ligand H2L (Scheme I), accessed through a double ring closure following its reaction with hydrazine (Fig. S1). This bis-chelating ligand could potentially form helicates, furnish an appropriate environment to Fe(II) for the SCO and encapsulate anions via N−H···X hydrogen bonds. The reaction in methanol of FeX2 (X=Cl, Br) salts with H2L in the presence of Bu4NPF6 (the latter as a source of a counterion) leads indeed to crystallization of the assemblies X@[Fe2(H2L)3]X·(PF6)2 (de-solvated 1 and 2 for Cl and Br, respectively) consisting of cationic triple stranded Fe(II) dinuclear helicates, encapsulating one X − anion. Upon prolonged exposure to air, crystals of 1 and 2 experience the exchange of guest lattice molecules producing 1a and 2a. The latter are thus solvathomorphs of 1 and 2 after losing 2.7 or 3 molecules of MeOH, respectively, and absorbing one equivalent of H2O. Transformations 1 → 1a and 2 → 2a remarkably occur in a single-crystal-to-single-crystal (SCSC) manner, thus allowing the full structure determination of both resulting products. Attempts to encapsulate the larger halide I − by employing the salt FeI2 in the original reaction led to crystallization from Et2O of the novel compound Cl@[Fe2(H2L)3](I3)3·3Et2O (1b) in very low yield, with Cl − arising from a trace impurity. The anion I3 − was the result of aerobic oxidation of I − , which is not rare in inorganic chemistry. [32] The reaction was then optimized by using the FeCl2 salt and introducing excess iodide as Bu4NI, although the yield of the crystallized compound remained quite small, most likely due to low concentrations of the in situ formed I3 − anions. The same procedure could be then replicated with Br -, with the convenient formation of Br@[Fe2(H2L)3](I3)3·3Et2O (2b), although again in moderate yields. The inability to encapsulate Iis thought to be due to the large volume of this anion, exceeding the space available inside the [Fe2(H2L)3] 4+ host, as could be ascertained by analyzing the molecular structure of the helicates (see below). A summary of the six derivatives prepared with their main features is in Table 1, to facilitate the reading.

Description of Structures
Cl@[Fe2(H2L)3]Cl·(PF6)2·5.7CH3OH (1). Compound 1 is found at 100 K in the tetragonal space group I41cd (Tables S1 and S6). The asymmetric unit consists of one helical [Fe2(H2L)3] 4+ complex cation with an encapsulated Clanion ( Fig. 1), together with an external Cland two PF6ions, in addition to five full solvent MeOH molecules, and one with 70% occupancy. Of the solvent molecules, three are disordered over two positions, as well as one of the PF6anions. The unit cell encloses a total of sixteen such ensembles. The cationic helicate ( Fig. 1) is formed by two Fe(II) metal centers defining the central axis and three H2L ligands acting as strands. The latter chelate both metals through their pyrazolyl-pyridine moieties (each approximately confined in one plane) completing distorted chiral six-coordination around them. In this manner, each helicate in the lattice displays either ΔΔ or ΛΛ metal configuration sets, leading to enantiomeric species, present as racemic mixtures in the crystal by virtue of its group symmetry.  The chloride encapsulated by the helicate is held within the cavity through six hydrogen bonding interactions with the N-H groups of the pyrazolyl moieties ( Fig S2). Of these, two are clearly stronger than the rest (Tables 2 and S11), which causes the Clanion to be closer to one iron center (Fe1) than to the other by 0.33 Å. 3.567(7)/3.59   The external Clgroup is located next to the complex cage, establishing one hydrogen bond with an N-H group, also near Fe1, and one molecule of methanol (Fig. S2). Thus the Fe atoms in 1 differ on the nature of the species forming hydrogen bonds with their β-N-H groups. The N-H groups adjacent to Fe1 act as donors to Clions, whereas the N-H moieties near Fe2 interact with the oxygen atom of MeOH molecules. These differences translate into two distinct magnetic responses (see below), which at 100 K causes Fe1 to be in the high spin (HS) state while Fe2 is low spin (LS). This is reflected in the metric parameters around these ions, such as the average Fe-N bond distances, <d(Fe-N)>, of 2.190 and 1.980 Å for Fe1 and Fe2, characteristic of their respective spin states. [33] These observations confirm the established fact that out-of-sphere intermolecular interactions are crucial to the magnetic state of SCO centers. [34][35] Here, hydrogen bonds of the N-H groups towards Clstabilize the HS state more than the interactions with MeOH molecules. These results are to be compared with a previously reported [Fe(II)2] helicate, with a bis-(imidazolimine) ligand, [22] which shows the same magnetic dissimilarity between both metals of the molecule. In that case, the differing behavior is explained by the presence of a strong π···π contact, near one of the Fe(II) centers, perhaps restraining the structural changes related to a process of SCO. The supramolecular (Cl@[Fe2(H2L)3]) 3+ species are organized in sheets parallel to the crystallographic ab plane (Fig. S3), mutually connected through π···π and C-H···π interactions involving some of their numerous aromatic rings, where each helicate sees five related first neighbors (Fig.  2). These sheets, described as hydrophobic because of the organic ligands, intercalate with hydrophilic layers composed of the external Cland PF6ions and MeOH molecules (Fig. S3).
The space occupied by the layers of anions and solvents seems appropriate for polar molecules to diffuse and exchange with other such molecules from the environment. This seems to be the space used by the ambient H2O molecules to diffuse inside the crystal and occupy the space left by the MeOH species leaving the structure upon formation of 1a while maintaining the crystallinity. The evacuation of guest MeOH molecules occurs probably through these layers as well. The asymmetric structure of 1 causes its Fe(II) centers to be in two different spin states at 100 K. A variable temperature crystallographic study was performed in order to visualize the thermal evolution of the magnetic states of both distinct Fe centers in this molecular assembly.
The asymmetric unit also differs on the number of solvate molecules, here with four solvent MeOH molecules present. The structural and electronic dissimilarity between the Fe atoms ( Fig.  S5) of the assembly is here slightly less pronounced than in 1 (Tables 2 and S11) while still causing a different magnetic behavior. The metals at 100 K are thus in the HS (<d(Fe-N)> = 2.193 Å) and LS (<d(Fe-N)> = 1.980 Å), respectively. The variable temperature structural study of 2 (Table S2) shows that the gradual SCO of Fe1 is now shifted approximately 40 K towards lower temperatures, as is the evolution of the cell parameters (Figs. 3 and S4). In this evolution, a change of tendency is seen at 280 K and above probably related to the onset of MeOH extrusion.

Cl@[Fe2(H2L)3]Cl(PF6)2·3CH3OH·H2O (1a).
Compound 1a is organized in the tetragonal space group I41acd (Tables S3 and  S8) and incorporates sixteen asymmetric units into the unit cell. The composition of the asymmetric unit of 1a differs from that of 1 in that 2.7 molecules of MeOH have now been replaced by one molecule of H2O. The latter and one molecule of MeOH are disordered over two positions. The exchange of MeOH by H2O, which occurs in a SCSC manner, leads to other important changes. The (Cl@[Fe2(H2L)3]) 3+ assembly becomes symmetric by virtue of a binary axis, thus the encapsulated Clion is located at the center of the host helicate at the same distances from both Fe atoms, now crystallographically equivalent (Tables  2 and S8). The increased symmetry is achieved because the external Clion is now disordered over two equivalent positions, near either one or the other Fe center (Fig. S6). This disorder shows that the exchange of guest molecules causes important relocations of atoms (besides these from the migrating species). Thus 50% of the Clions have experienced, either a displacement of approximately 9 Å (Fig. S7) or a partial substitution with the encapsulated anion in going through the helicate, from one side to the other. The crystallographic equivalence of the Fe ions in the lattice results from the averaging of all the disordered components, thus, is not mirrored by an equivalence of both atoms within individual molecules, which are clearly different (each external Clion can lie close to only one of both Fe centers at a time). This translates into a slightly different magnetic behavior for each metal (see below) that cannot however be put into evidence crystallographically, contrary to the case of 1. Here both Fe centers are not so different from each other because the distribution of N−H···Cl − vs N−H···O interactions (Tables 2 and S11) near one or the other metal is more even (Fig. S6) (Tables S5 and S10).
The asymmetric unit corresponds to one third of the empirical formula while the unit cell encloses six times this formula. The main component of 1b is a (Cl@[Fe2(H2L)3]) 3+ helical unit (Fig. 4) very similar to that seen in 1 and 1a, present in the lattice as a racemic mixture of both possible enantiomers. The encapsulated Clfeatures also varying N-H···Clinteractions (Tables 2 and S11) rendering the two metal ions of the helicate inequivalent. . This is reflected in the <d(Fe-N)> values observed at 100 K (of 2.185 and 2.188 Å for Fe1 and Fe2, respectively). Thus, in this compound, the crystallographic differences between both Fe centres do not reflect on disparate magnetic properties. The interaction of the I3ions with the (Cl@[Fe2(H2L)3]) 3+ fragment of 1b occurs via a succession of "lone pair -π" interactions with various aromatic rings from the H2L ligands. In fact, a total of six contacts fulfil the criteria to consider such interaction [36] (Fig. 5), in a rare example where the three atoms of the anion interact with one or more aromatic rings. These interactions keep the central phenylene spacer and the concerned pyrazolylpyridine group of H2L almost within the same plane (mutual angle of 10.3º) whereas the other pyrazolylpyridine moiety is twisted by 29.96º with respect to the central phenylene. Each (Cl@[Fe2(H2L)3]) 3+ assembly is surrounded also by six first neighbouring molecules of Et2O (Fig.  4). Thus, the organization of this compound in the lattice does not produce alternating layers of respective hydrophilic and hydrophobic character as in 1 and 2. Instead, the helical ensembles of 3 are disposed as infinite rods running along their axial direction, parallel to the crystallographic c axis (Fig. S10) and mutually shifted. Within the rods, pairs of enantiomeric (Cl@[Fe2(H2L)3]) 3+ groups are interlocked pairwise face to face, connected through six complementary and identical C-H···π contacts (Fig. S11). The mutual shift between rods cause each helicate to be surrounded laterally by six other equivalent neighbours connected to it by pairs of π···π interactions (Fig.  S12). Br@[Fe2(H2L)3](I3)3·3Et2O (2b). Compound 2b is isostructural with 1b, with the only significant difference that the anion encapsulated is now a Brspecies instead of Cl -. This does not lead to significant differences to the lattice organization (Table  S5, Figs. 4, 5 and S9 to S12), or to the metric parameters (Tables 2, S10, S11 and caption of Fig 5), thus, the description made for the Clanalogue is valid for the Brderivative. In addition, the Fe centers exhibit the same spin state as in 1b, ie HS as observed crystallographically at 100 K (<d (Fe-N) (1b), resulting from the presence of Cltraces in the system. The socalled packing coefficient, PC, has been previously employed to evaluate the possibility of encapsulating guests within a host cavity and its efficiency. [37] The PC of a host/guest system is the ratio of the volume of the guest over that of the host cavity (PC=Vguest/Vcav,). The ideal PC for the case of encapsulation of liquids was shown to be 0.55 ± 0.09. [37] Higher values (0.60 -0.79) have been found with host/guest systems involving strong intermolecular interactions. [38] The volume of the cavity inside the [Fe2(H2L4)3] 4+ host was calculated from the molecular structure of compounds 1 and 2 as 28 and 33 Å 3 , respectively, using Swiss-Pdb Viewer 4.1 (Fig. S13). The difference suggests that this host has a certain degree of flexibility and is capable to adjust its size depending on the nature of the guest. On the other hand, the volume of halide ions was calculated from their ionic radii as 19.51, 25.52 and 36.62 Å 3 , for Cl − , Br − and I − , respectively. [39] The calculated PC values for Cland Brin 1 and 2 are 0.697 and 0.773, respectively. These higher than the ideal value (0.55) numbers are expected, considering the strong N-H···X -H-bonding interactions involved in these host/guest systems. In addition, it is possible that for the case of monoatomic anions (such as halides) the ideal PC value is larger than for liquids. For the case of I -, it appears indeed that the volume of the anion seems excessive to accommodate within this host.

Thermal Spin Crossover Properties
The extensive crystallographic studies on the supramolecular assemblies 1, 1a, 1b, 2, 2a and 2b, revealed a fascinating variety of magnetic states and STs that were investigated in detail by means of bulk magnetic and calorimetric measurements.
Magnetic susceptibility data were collected on microcrystalline samples of all compounds described above, under a constant magnetic field of 0.5 T, from 1.9 to either 350 or 380 K. The results are represented on Figs. 6 and S14 in form of χT vs T plots. For compound 1, the value of χT at 380 K is 7.15 cm 3 Kmol -1 , which shows that both Fe(II) ions of the supramolecular helicate are in the HS state ([HS-HS], S = 2, g = 2.38), as also indicated by the single crystal X-ray diffraction experiments (see above). A clear decline is observed immediately upon cooling, reaching a plateau near 250 K, at 3.9-3.7 cm 3 Kmol -1 down to 50 K, showing that half of the Fe(II) centers of the system experience an almost complete SCO, centered near room temperature (T1/2 = 305 K). This is consistent with the crystallographic data at 100 K, which shows that each of the two crystallographically distinct Fe centers of the helicate of 1 is at this temperature in a different spin state ([LS-HS]). Near 25 K, the χT curve exhibits a new sharp decline that is attributed to the zero field splitting (ZFS) effect of the remaining HS metal centers. The Branalogue (compound 2) exhibits an almost identical behavior to 1, with the only difference that the SCO is shifted 40 K to lower temperatures (T1/2 = 265 K).
The above observations are mirrored by broad anomalies observed at the corresponding temperatures in the Cp(T) curves (Fig. S15, Cp is the molar heat capacity at constant pressure), and thus ascribed to the SCO processes of compounds 1 and 2.
From the excess heat capacity associated with these, obtained by subtracting the estimated lattice contribution (Fig. 7), the excess enthalpy HSCO and entropy SSCO due to the SCO are determined to be 4.79/3.85 kJmol -1 and 15.95/15.01 Jmol -1 K -1 , respectively, for 1/2. In both compounds, the excess entropy is only slightly higher than the purely electronic component of the ST of one Fe(II) center, Rln5, indicating a very weak coupling of the SCO with lattice phonons, in agreement with the gradual nature of the transition. A more quantitative measure of the cooperative character of the SCO is derived by modeling the excess heat capacity with Sorai's model, that considers domains with interacting n like-spin centers. [40][41] Here, the data for 1/2 are nicely reproduced with TSCO = 302.1(3)/258.2(3) K and n = 15.8(5)/14.6(4) (Fig. 7). The derived TSCO values are in excellent agreement with the magnetic susceptibility data, while the values of n are intermediate between weekly cooperative SCO compounds [42][43] and highly cooperative ones. [44][45]   The only chemical difference between compounds 1 and 2 is the nature of the encapsulated and external Xions (Clvs Br -), which influence the crystal field around Fe1 via N-H···Xhydrogen bonds. The presence of Clstabilizes the LS state with respect to Br -, in contradiction with previous observations made for other SCO systems exhibiting similar N-H···Xout-ofcoordination sphere interactions. [34] The discrepancy could be due to the fact that in the current systems the N-H groups are in α with respect to the Fe-N bonds, whereas in the reported compounds the N-H functionalities are two bonds away from the Fe-N moieties. Calorimetry studies support these observations. First, two consecutive anomalies associated with two SCO steps are clearly distinguished for 1a (Fig. S15). Indeed, the corresponding excess heat capacity is nicely reproduced by the domain model using two transitions with TSCO = 185.0(3)/258.9(3) K and n = 7.1(3)/15.5(1) (Fig. 7), thus in excellent agreement with the magnetic data. The lower temperature step is thus markedly less cooperative, with a domain size n about half that for the higher temperature step or these for the SCO in 1 and 2. The total excess enthalpy HSCO and entropy SSCO amount to 6.03 kJmol -1 and 26.60 Jmol -1 K -1 , characteristic of a weakly-cooperative system, involving a very weak coupling of the SCO with lattice phonons. On the other hand, the Cp(T) of compound 2a hardly exhibits any anomaly, as a consequence of the extremely gradual and broad nature of the SCO process, in addition to the fact that it occurs down to the lowest temperature accessible with the DSC set-up, 100 K.
The possibility of employing an additional means of manipulating the magnetic state of the Fe centers at low temperature in the present (X@[Fe2(H2L)3]) 3+ helicates by exploiting the LIESST effect [56] was investigated. Compounds 1/2 were brought to their [LS-HS] state by cooling them to 10 K and were then irradiated with light in the wavelength range of 500-650 nm. This caused a sudden increase of the χT value, reaching quasi saturation after <1000 s at 3.9/5.6 cm 3 Kmol -1 (Figs. 6 and S16). Considering that at this temperature the effect of Fe(II) ZFS is already very apparent, the jump in susceptibility in the case of 2 corresponds to a [LS-HS] → [HS-HS] transformation of the system of nearly 100% of the molecules, which reach in this manner a trapped metastable [HS-HS] state. Given that samples of very small and similar thickness were used in this study, the effect of light propagation should be minor. Therefore the incomplete transition in 1 can mostly be associated with the SCO occurring at higher temperatures. [57] Another contribution may also result from competition with the relaxation back to the [LS-HS] state, which appears to be slow, but active already at 10 K. Indeed, upon increasing the temperature, a decrease of the χT sets in almost immediately caused by the relaxation to the [LS-HS] state, which is completed at 80/75 K. As a consequence, a The process occurs with an ill-defined but low characteristic TLIESST can not be determined with confidence, although it is clearly rather low. [58] Irradiation of compounds 1a/2a at 10 K, lying in their stable [LS-LS] state, using the same quality of light, also causes a sudden increase of the χT product, reaching saturation at values of 3.1/4.0 cm 3 Kmol -1 after <2000 s (Figs. 6 and S16). Taking again into account the significant ZFS of the HS metastable state, these values point at a transformation from the LS to the HS state of at least 50% of the Fe centers of the sample. In the case that each metal of the molecule exhibits a different behavior, this would correspond to a 100% [LS-LS] → [LS-HS] transformation, possibly in addition to partial transformation to the [HS-HS] state. The aforementioned non-equivalence is plausible for 1a since the susceptibility measurements have served to demonstrate that there are two slightly magnetically different metals in this compound. The slight asymmetry would also be consistent with the interpretation of crystallographic data in both compounds. Upon increasing the temperature after turning off the light, both compounds exhibit a very similar behaviour, with the relaxation of the induced metastable state to the [LS-LS] state only occurring above ca. 25 K, with the same characteristic TLIESST of ca.

Stability in Solution
The stability of the (X@[Fe2(H2L)3]) 3+ assembly in solution was investigated by means of 1 H NMR and mass spectrometry. Such studies are dependent on the solubility of the compounds. DMSO was found to dissolve all complexes, however, it was established by NMR that the assemblies decompose in this medium. Compounds 1b and 2b are not soluble in any other common solvents, while MeCN is one of the few media where 1 and 2 can be dissolved. In this solvent, complex 1 exhibits a dominant set of nine broad, paramagnetically shifted peaks (between -6 and 60 ppm) with no hyperfine splitting of which, two integrate for half the intensity of the other seven (Fig S17). Two of the latter resonances are degenerate near 40 ppm, but comparison with the Branalogue (see below) unveils the existence of two peaks in that area. These features are consistent with the idealized symmetry shown by the helicate of 1 in the solid state (D3), suggesting that this is the major species in solution. The spectrum shows a smaller set of broad signals, spanning over a narrower range of chemical shifts (1 to 15 ppm). Their compared integrations are consistent with the ensemble arising from a multiple of sixteen protons (Fig S18). This is in agreement with a coordination complex with all identical H2L ligands but featuring lower symmetry than the helical assembly. In fact one species exhibiting H2L coordinated to only one Fe(II) with formula [Fe(H2L)3] 2+ can be isolated and characterized from this reaction system, which would explain this response. [59] Under this premise, comparison of total integration values indicates that the helicate in 1 and the mononuclear complex coexist in approximately 1:0.6 proportions. The 1 H NMR spectrum of 2 (Fig. 8) corroborates the observations made with its Clanalogue. The main differences are i) the set of peaks for the minor species is now much weaker (indicating now an approximate partition of 1:0.1, Fig. S19), ii) the signal that was degenerate in 1 is now resolved in two peaks, iii) a resonance that in 1 was located in between the signals of residual MeOH has now moved under one of the solvent peaks, iv) the most paramagnetically shifted peaks of the helicate exhibit now significantly larger chemical shifts. This experiment confirms that the (Br@[Fe2(H2L)3]) 3+ unit is stable in MeCN with clear dominance over the less symmetric species and that 2 in solution exhibits a larger fragment of Fe(II) centers in the HS state than 1 (consistent with the solid state behaviour) as indicated by much larger paramagnetic shifts.
Positive electron spray ionization mass spectrometry (ESI + -MS) experiments were also performed for 1 and 2 in MeCN. The results for 1 confirm the existence of the helicate assembly in solution with identification of the fragments (Cl@[Fe2(H2L)2(HL)]) 2+ at m/z = 619.13 and [Fe2(H2L)(HL)2] 2+ at m/z = 601.15 (Fig. S20). The presence of the latter may reflect an equilibrium between the occupied and vacant host. In addition, peaks reflecting assemblies of the lower symmetry species [59] were also evident (associated as dimers, such as

X-ray crystallography
Data for compounds 1, 2, 1a and 2a were collected at various temperatures in the range 30-340 K on Beamline 11.3.1 at the Advanced Light Source, on a Bruker D8 diffractometer equipped with a PHOTON 100 CCD detector and using silicon 111 monochromated synchrotron radiation ( = 0.7749 Å). The crystals were mounted on a MiTegen kapton loop and placed in the N2 stream of an Oxford Cryosystems Cryostream Plus or for the lowest temperature (30 K) in the He stream from a Cryoindustries of America LT-HE Cool cryosystem. Data for compounds 1b and 2b were collected at 100 K on a Bruker APEXII QUAZAR diffractometer equipped with a microfocus multilayer monochromator with MoK radiation ( = 0.71073Å).

Physical Measurements
Variable-temperature magnetic susceptibility data were obtained with either MPMS5 or MPMS-XL SQUID magnetometers through the Physical Measurements unit of the Servicio de Apoyo a la Investigación-SAI, Universidad de Zaragoza. For the irradiation studies the commercial FOSH set-up was used in combination with a Xe arc lamp and short-pass and long-pass interference filters. The samples were in the form of small pieces of very thin pellets, to minimize the effect of the attenuation of the propagation of light through the sample. The data were corrected for the sample holder contributions, determined empirically as well as for the intrinsic diamagnetism of the samples, estimated using Pascal constants. Differential Scanning Calorimetry (DSC) measurements were done with a Q1000 calorimeter from TA Instruments equipped with the LNCS accessory, using aluminium pans crimped mechanically and an empty pan as reference. The temperature and enthalpy scales were calibrated with a standard sample of indium, using its melting transition (156.6 ºC, 3296 Jmol -1 ). The zero-heat-flow procedure described by TA Instruments was followed to derive heat capacities, using a synthetic sapphire as reference compound. An overall accuracy of about 0.2 K and up to 10% was estimated respectively for the temperature and heat capacity over the whole temperature range. The lattice contributions to the heat capacity were estimated from the data above and below the observed anomalies. Excess enthalpy and entropy were derived by integration of the excess heat capacity with respect to T and lnT, respectively. Elemental analyses were performed with an Elemental Microanalizer (A5), model Flash 1112 at the Servei de Microanàlisi of CSIC, Barcelona, Spain. IR spectra were recorded as KBr pellet samples on a Nicolet AVATAR 330 FTIR spectrometer. Positive ion ESI TOF mass spectrometry experiments were performed on a LC/MSD-TOF (Agilent Technologies) at the Unitat d'Espectrometria de Masses de Caracterització Molecular (CCiT) of the University of Barcelona. The experimental parameters were: capillary voltage 4 kV, gas temperature 325ºC, nebulizing gas pressure 15 psi, drying gas flow 7.0 L min -1 , and fragmentor voltage ranging from 175 to 300 V. Samples (μL) were introduced into the source by a HPLC system (Agilent 1100), using a mixture of H2O/MeCN (1/1) as eluent (200 μL min -1 ).