Correlation between Supramolecular Connectivity and Magnetic Behaviour of [Fe III (5-X-qsal) 2 ] + -Based Salts Prone to Exhibit SCO Transition

: We present an extensive study to determine the relationship between structural features of spin crossover (SCO) systems based on N-(8-quinolyl)salicylaldimine (qsal) ligand derivatives and their magnetic properties. Thirteen new compounds with general formula [Fe III (5-X-qsal) 2 ] + (X = H, F, Cl, Br and I) coupled to Cl − , ClO 4 − , SCN − , PF 6 − , BF 4 − and BPh 4 − anions were prepared and magnetically characterized. The structure/properties correlations observed in these compounds were compared to those of salts with the same [Fe III (qsal-X) 2 ] + cations previously reported in the literature. These cations favour the LS conﬁguration in compounds with the weakest connectivity. As connectivity increases most of them present HS states at room temperature and structures may be described as arrangements of parallel layers of interacting cation dimers. All the compounds based on these cations undergoing complete SCO transitions within the 4–300 K temperature range have high intralayer connectivity. If, however, the interlayer connectivity becomes very strong they remain blocked in the HS or in the LS state. The SCO transition may be affected by the slightest change of solvent molecules content, disorder or even crystallinity of the sample and it remain difﬁcult to predict which kind of ligand substituent should be selected to obtain compounds with the desired connectivity.


Introduction
Spin crossover (SCO) is the ability that some complexes have to switch reversibly between two possible spin configurations, a low-spin (LS) and a high-spin (HS) state. This phenomenon can be observed in octahedral complexes with d 4 -d 7 electronic configuration [1]. It can be triggered by variation in temperature, pressure or by light irradiation and can be monitored by a wide variety of techniques [1][2][3]. SCO can be incomplete, complete, gradual or abrupt, hysteretic, in one-step or in multiple steps. These various behaviours are connected to the way the structural changes induced by the SCO transition are propagated through the entire solid structure, i.e., the cooperativity between the SCO centres. This cooperativity can be understood as the sum of the contributions of a series of parameters correlated with supramolecular interactions between the molecules that constitute each system, cations, anions and solvating molecules. To obtain compounds with different levels of cooperativity it is necessary to synthesize them considering several supramolecular contacts such as hydrogen [4,5] and halogen bonding [6,7], and π-π stacking [8,9]. However, it has proven difficult to understand how a specific combination of these interactions can result in structures that have a desired SCO behaviour. Often in the literature structures that have SCO behaviour are described but no information is given regarding the specific type of supramolecular structure and the interactions that correlate with the observed SCO behaviour. This kind of studies are however important since they provide useful design clues in the search for stable, hysteretic SCO complexes at room temperature.
In this work we present an extensive study to determine the relationship between structural features or patterns common in SCO systems based on N-(8-quinolyl)salicylaldimine (qsal, Figure 1) ligand derivatives and the resulting magnetic properties. We have prepared a series of Fe III SCO complexes with general formula [Fe(5-X-qsal) 2 ] + (X = H, F, Cl, Br and I) coupled to Cl − , ClO 4 − , SCN − , PF 6 − , BF 4 − and BPh 4 − anions in Table S1. The resulting compounds were magnetically characterized, and their structure analysed in order to detect structural similarities between those with similar magnetic behaviour. The structure/properties correlations observed in this work were also compared to those of salts with the same cations [Fe III (5-X-qsal) 2 ] + previously reported in the literature. Hereafter the reference to the compounds will have one of the following numerical prefixes depending on X, 1 (X = H), 2 (X = F), 3 (X = Cl), 4 (X = Br), 5 (X = I), 6 (X = OCH 3 ) and 7 (X = 5-tBu). Complexes coupled with transition metal anions such as dithiolates like [Ni(dmit) 2 ]− (dmit = 4,5-dithiolato-1,3-dithiole-2-thione) were excluded from this study due to their disparity of supramolecular arrangements arising from the ease with which they create segregated layers of highly connected anions. These materials are expected to follow different supramolecular arrangements than the compounds addressed in this work.
of these interactions can result in structures that have a desired SCO behaviour. Often in the literature structures that have SCO behaviour are described but no information is given regarding the specific type of supramolecular structure and the interactions that correlate with the observed SCO behaviour. This kind of studies are however important since they provide useful design clues in the search for stable, hysteretic SCO complexes at room temperature.
In this work we present an extensive study to determine the relationship between structural features or patterns common in SCO systems based on N-(8quinolyl)salicylaldimine (qsal, Figure 1) ligand derivatives and the resulting magnetic properties. We have prepared a series of Fe III SCO complexes with general formula [Fe(5-X-qsal)2] + (X = H, F, Cl, Br and I) coupled to Cl -, ClO4 -, SCN -, PF6 -, BF4 -and BPh4 -anions in Table S1. The resulting compounds were magnetically characterized, and their structure analysed in order to detect structural similarities between those with similar magnetic behaviour. The structure/properties correlations observed in this work were also compared to those of salts with the same cations [Fe III (5-X-qsal)2] + previously reported in the literature. Hereafter the reference to the compounds will have one of the following numerical prefixes depending on X, 1 (X = H), 2 (X = F), 3 (X = Cl), 4 (X = Br), 5 (X = I), 6 (X = OCH3) and 7 (X = 5-tBu). Complexes coupled with transition metal anions such as dithiolates like [Ni(dmit)2]− (dmit = 4,5-dithiolato-1,3-dithiole-2-thione) were excluded from this study due to their disparity of supramolecular arrangements arising from the ease with which they create segregated layers of highly connected anions. These materials are expected to follow different supramolecular arrangements than the compounds addressed in this work.   show the temperature dependence of χPT, product of the paramagnetic susceptibility, χ, by temperature for all the compounds synthesized. All compounds shown in Figure 2 are in the LS state up to approximately 200 K with low temperature χPT values (considering 20 K for all compounds) ranging from 0.40-0.46 emu K mol −1 . The magnetic behaviour of compound 3.PF6.MeCN is typical of a SCO compound with a sharp transition and transition temperature T1/2 ~ 256 K, calculated as the maximum of the first derivative / . At 315 K 3.PF6.MeCN has a χPT value of 3.72 emu K mol −1 which can be attributed to an incomplete spin transition. A far as the remaining compounds are concerned, gradual transitions to the HS state start to occur approximately at 250 K. At 320 K the values of χPT for 3.Cl.MeCN.H2O, 3.ClO4, 3.BF4 and 3.BPh4 are 1.53 emu K mol −1 , 0.60 emu mol K −1 , 0.93 emu K mol −1 and 1.96 emu K mol −1 , respectively. Magnetic susceptibility measurements were also obtained on cooling but no evidence of hysteresis was found for these compounds.

Magnetic Characterisation
Figures 2-4 show the temperature dependence of χ P T, product of the paramagnetic susceptibility, χ, by temperature for all the compounds synthesized. All compounds shown in Figure 2 are in the LS state up to approximately 200 K with low temperature χ P T values (considering 20 K for all compounds) ranging from 0.40-0.46 emu K mol −1 . The magnetic behaviour of compound 3.PF 6 .MeCN is typical of a SCO compound with a sharp transition and transition temperature T 1/2~2 56 K, calculated as the maximum of the first derivative ∂χT/∂T. At 315 K 3.PF 6 .MeCN has a χ P T value of 3.72 emu K mol −1 which can be attributed to an incomplete spin transition. A far as the remaining compounds are concerned, gradual transitions to the HS state start to occur approximately at 250 K. At 320 K the values of χ P T for 3.Cl.MeCN.H 2 O, 3.ClO 4 , 3.BF 4 and 3.BPh 4 are 1.53 emu K mol −1 , 0.60 emu mol K −1 , 0.93 emu K mol −1 and 1.96 emu K mol −1 , respectively. Magnetic susceptibility measurements were also obtained on cooling but no evidence of hysteresis was found for these compounds.
Compounds in Figure 3 exhibit three distinct magnetic behaviours. 4.SCN.0.5H 2 O has a magnetic behaviour consistent with a HS species with an average value of χ P T~4.14 emu mol K −1 in the 50-320 K range and χ P T = 4.23 emu mol K −1 at 320 K. 4.PF 6 .H 2 O reveals a spin transition with T 1/2 ≈ 203 K. At low temperature χ P T = 0.30 emu K mol −1 , the usual value for a LS state in these compounds [1]. At 317 K χ P T = 3.90 emu K mol −1 indicating an almost complete spin transition. Compounds 4.ClO 4, 4.BF 4 Figure 3 exhibit three distinct magnetic behaviours. 4.SCN.0.5H2O has a magnetic behaviour consistent with a HS species with an average value of χPT ~ 4.14 emu mol K −1 in the 50-320 K range and χPT = 4.23 emu mol K −1 at 320 K. 4.PF6.H2O reveals a spin transition with T1/2 ≈ 203 K. At low temperature χPT = 0.30 emu K mol −1 , the usual value for a LS state in these compounds [1]. At 317 K χPT = 3.90 emu K mol −1 indicating an almost complete spin transition. The magnetic behaviour of the compound 5.Cl.MeOH.2.5H2O ( Figure 4) is consistent with HS behaviour with an average χPT ~ 3.96 emu K mol −1 slightly increasing with temperature up to χPT = 4.01 emu K mol −1 at 300 K. In the case of 5.PF6.1.5H2O it is consistent with SCO transition with T1/2 ≈ 270 K. At low temperatures (20 K) χPT = 0.47 emu K mol −1 , a typical value for a LS Fe III compound [1]. At 314 K χPT = 4.31 emu K mol −1 is indicative of a HS state. The χT curves obtained on cooling and warming overlap with no evidence of hysteresis.

Structural Characterization
In the [Fe(X-qsal)2] + cations two X-qsalligands are coordinated to the central Fe(III) atom in a nearly perpendicular manner, yielding distorted octahedral FeN4O2

Structural Characterization
In the [Fe(X-qsal)2] + cations two X-qsalligands are coordinated to the central Fe(III) atom in a nearly perpendicular manner, yielding distorted octahedral FeN4O2 The magnetic behaviour of the compound 5.Cl.MeOH.2.5H 2 O (Figure 4) is consistent with HS behaviour with an average χ P T~3.96 emu K mol −1 slightly increasing with temperature up to χ P T = 4.01 emu K mol −1 at 300 K. In the case of 5.PF 6 .1.5H 2 O it is consistent with SCO transition with T 1/2 ≈ 270 K. At low temperatures (20 K) χ P T = 0.47 emu K mol −1 , a typical value for a LS Fe III compound [1]. At 314 K χ P T = 4.31 emu K mol −1 is indicative of a HS state. The χT curves obtained on cooling and warming overlap with no evidence of hysteresis.

Structural Characterization
In the [Fe(X-qsal) 2 ] + cations two X-qsal − ligands are coordinated to the central Fe(III) atom in a nearly perpendicular manner, yielding distorted octahedral FeN 4 O 2 coordination environments.
Hereafter cation-cation and cation-anion interactions will be referred to as DD and DA interactions, respectively. The extended aromatic systems of the X-qsal ligands give rise to quite strong ππ cation-cation, DD interactions, which play a dominant role in the crystal packing of these compounds. Strong DD pairwise ππ interaction due to the extensive overlap of ligands from neighbouring cations, often reinforced by hydrogen bonds, CH . . . O, and type II halogen bonds, CX . . . π, result in the appearance of two distinct basic structural motives, either dimers or chains of cations.
In the crystal structures of the salts with anions such as Cl − , SCN − , BF 4 − and ClO 4 − , weaker interdimer ππ interactions between the cations give rise to 2D cationic layers, while in the salts of the B(Ph) 4 − anions, strong cation-anion, DA, interactions, mostly concerning ππ and CH . . . π contacts, are responsible for the existence of chains of weakly interacting dimers isolated by the bulky B(Ph) 4 − anions. In the PF 6 − salts weaker DD ππ interactions give rise to crystal structures based on layers of parallel chains of cations.
The studied compounds may thus be sorted in four sets with different supramolecular configurations: Dimer Chains, Type I and Type II Dimer Layers and Chain Layers.  Table S2. Figure 5 shows a projection of the structure of 4.BPh 4 along the direction of the chains of cation dimers. The intrachain arrangements of the cations and the DD contacts in the crystal structure of 4.BPh 4 are shown in Figure S1.
The BPh 4 − anions display strong interactions with the cations mostly through a large number of DA ππ and CH . . . π contacts, as illustrated in Figure S2, where the arrangement of the most relevant DA contacts around one anion with cations from three neighbouring cationic chains, is shown. Some of the DA interactions mediated by the BPh 4 − anions appear to give rise to relatively rigid DAD interchain arrangements such as those involving the cations in orange, connected by DA ππ contacts through a single Ph group of the anion. Despite the strong DD intradimer and DA interactions, the weak DD interdimer and the softness of the BPh 4 − anion is expected to lead to a quite flexible crystal lattice from the point of view of the structural rearrangements associated with the SCO processes.
observed. The cations display strong interactions with the phenyl rings of the BPh4 anions through ππ and C-H…π contacts. The chains of cation dimers are isolated from each other by the BPh4 -anions and no interchain DD contacts are observed. The more relevant parameters describing the intra and interdimer DD interactions of 3.BPh4 and 4.BPh4 are summarized in Table S2. Figure 5 shows a projection of the structure of 4.BPh4 along the direction of the chains of cation dimers. The intrachain arrangements of the cations and the DD contacts in the crystal structure of 4.BPh4 are shown in Figure S1. The BPh4 -anions display strong interactions with the cations mostly through a large number of DA ππ and CH…π contacts, as illustrated in Figure S2, where the arrangement of the most relevant DA contacts around one anion with cations from three neighbouring cationic chains, is shown. Some of the DA interactions mediated by the BPh4 -anions Both compounds with this Dimer Chains supramolecular configuration are in the LS state at low temperatures with incomplete transitions up to 320 K (Figures 2 and 3).

Dimer Layers
The crystal structures of the salts with the Cl − , SCN − , BF 4 − and ClO 4 − anions are also based on arrangements of dimers of cations; however, distinctively from those observed in the BPh 4 − based salts, these salts exhibit significant interdimer DD interactions, giving rise to crystal structures that are based on arrangements of parallel layers of dimers. The intradimer arrangements are similar to those described for 3.BPh 4 and 4.BPh 4 , with an overlap of the 5-X-qsal ligands from each cation extended through the entire length of the ligands. In the Dimer Layers structures, two distinct types of layers were observed.  4 salts, displaying "type I" layer arrangements, each of the two 5-X-qsal ligands dimers that is not involved in the intradimer contacts, shows sizable ππ interactions with the ligands from two neighbouring dimers, through Ph-Qn partial overlaps. In 5.Cl.MeOH.2.5H 2 O and 4.SCN.0.5H 2 O, exhibiting "type II" layer arrangements, each of the four 5-X-qsal ligands displays quite strong ππ interactions to a ligand from a neighbouring dimer through the overlap of both rings from the Qn fragments of two ligands. These two interdimer arrangements are shown in Figure 6a Tables S3 and S4. interdimer arrangements are shown in Figure 6a,b for 3.Cl.MeCN.H2O (type I) and 5.Cl.MeOH.2.5H2O (type II), respectively. The crystal packing of both these salts based on both types of arrangements will be described below, in detail. The crystal packings of the other salts displaying these types of the Dimer Layers arrangements will be compared with those of 3.Cl.MeCN.H2O and 5Cl.MeOH.2.5H2O. The main parameters describing the intradimer interactions (DD d ) as well as the interdimer intra (DD L ) and interlayer (DD iL ) DD interactions are summarized in Tables S3 and S4.

Type I Dimer Layers
The crystal structure of 3.Cl.MeCN.H 2 O is based on an arrangement of layers of [Fe(5-X-qsal) 2 ] + dimers parallel to the ac plane. Figure S3 shows a view of the structure of 3.Cl.MeCN.H 2 O along b, where the atoms belonging to one of the cationic layers are depicted in dark blue. Figure  In the case of salts displaying the Type I Dimer Layers configuration, minor variations were also observed in intralayer DD interactions, resulting from the overlap between the Qn and Ph-X fragments of the ligands. The variation in the separation of the two aromatic systems suggests that those interactions appear to be slightly stronger The connectivity resulting from the contacts mediated by the anions and solvent molecules is illustrated in Figure S6 (Figures 2-4). The slower progress of the SCO processes that is clearly noticed in particular for the 3.BF 4 , 3.ClO 4 and 4.ClO 4 salts can be related to the larger rigidity of the lattice of these salts resulting from the stronger interlayer DD interactions, thus requiring more energy in order to distort the crystal lattice.  Table S4.
Besides the strong intradimer DD d and interdimer DD L 1 interactions, each [Fe(5-I-qsal) 2 ] + cation displays a fairly strong interaction to one cation located in a neighbouring layer, through a ππ contact resulting from the overlap between two Ph fragments of neighbouring ligands. This interaction DD iL is further assisted by two-fold type I halogen contacts C-I . . . O in 5.Cl.MeOH.2.5H 2 O or C-Br . . . O in 4.SCN.0.5H 2 O. The arrangement of the various DD contacts relative to a central cation is illustrated in Figure S9. The cation bonded to the central one within the dimer is depicted in light grey. The cations displayed in orange are located in the same layer as the central cation. In the case of the dark orange cation the ππ interactions to the central cation are reinforced by twofold CH . . . π contacts and this DD L 1 interaction is expected to be slightly stronger than the DD L 2 interaction with the light orange cation. Finally, the cation located in a neighbouring layer, displayed in violet in Figure S9, interacts with the central cation through the combination of ππ overlap between two Ph groups and two-fold C-I . . . O interactions DD iL .
The anions and solvent molecules are confined within hollow channels formed by the cationic sub-lattice, and establish very strong OH . . . O, CH . . . O and OH . . . Cl hydrogen bond type contacts that propagate along the channels. The Cl − anions as well as the MeOH and H 2 O molecules are involved in various contacts with the cations: CH . . . O, CH . . . π, CH . . . Cl hydrogen bonds. These contacts are expected to reinforce the connectivity between the cations located in neighbouring layers. However, considering that the stronger contacts involve the MeOH molecules, which have half occupancy, the reinforcement of the connectivity is expected to be relatively modest. Figure S10 illustrates a detail of the interlayer coupling mediated by strongly coupled aggregates containing two Cl − anions, six H 2 O molecules and one methanol molecule with two possible locations. For clarity the contacts mediated by the H 2 O molecules were omitted in Figure S10. For each molecule those contacts involve only cations located in one of the layers. In 4.SCN.0.5H 2 O, the H 2 O molecules display quite strong interactions with two cations located in the neighbouring layers, and the SCN − anions display relatively strong interactions to cations in distinct layers, which anticipates a much more effective contribution of these interactions to the overall lattice connectivity than in 5.Cl.MeOH.2.5H 2 O.
The strong DD interactions found in the salts displaying the Type II Dimer Layers arrangements are expected to lead to more rigid crystal packings than in type I. The  Besides the intrachain DD interactions, DD Ch 1 and DD Ch 2, each cation shows a strong ππ interaction to another cation located in a neighbouring chain due to an overlap of the Qn fragments, reinforced by twofold CH…π contacts. These strong interchain DD interactions, DD L , give rise to strongly coupled layers of cations that are parallel to the bc plane in 3.PF6.MeCN and 4.PF6.H2O or to the ac plane in 5.PF6.1.5H2O. Figure S13 illustrates the cation arrangement and interchain DD L contacts of 5.PF6.1.5H2O, where the cations that interact with those from the central chain are depicted in orange.
The combination of strong intra (DD Ch i) and interchain intralayer (DD L ) interactions in the 3.PF6.MeCN, 4.PF6.H2O and 5.PF6.1.5H2O salts is expected to lead to relatively rigid layers of cations, although less rigid in comparison with the Type II Dimer Layers due to differences in interlayer DD interactions. In the Chain Layers structure each of the [Fe(5- Besides the intrachain DD interactions, DD Ch 1 and DD Ch 2 , each cation shows a strong ππ interaction to another cation located in a neighbouring chain due to an overlap of the Qn fragments, reinforced by twofold CH . . . π contacts. These strong interchain DD interactions, DD L , give rise to strongly coupled layers of cations that are parallel to the bc plane in 3.PF 6   The combination of strong intra (DD Ch i ) and interchain intralayer (DD L ) interactions in the 3.PF 6 .MeCN, 4.PF 6 .H 2 O and 5.PF 6 .1.5H 2 O salts is expected to lead to relatively rigid layers of cations, although less rigid in comparison with the Type II Dimer Layers due to differences in interlayer DD interactions. In the Chain Layers structure each of the [Fe(5-X-qsal) 2 ] + cations show only weak interactions to three cations located in neighbouring layers. Figure S14 shows in the default colour scheme the cations that interact with a central one in 5.PF 6 .1.5H 2 O. Those belonging to the same chain are depicted in light grey and the cation in orange corresponds to the one located in a neighbouring chain within the layer (contacts with these cations are omitted). The cation in violet and those in light blue are those that interact with the central cation and are located in the neighbouring layers. The stronger interlayer DD iL 1 interactions concern the violet cation and consist of a two-fold CH . . . I contact, along with a weaker CH . . . π contact. The weaker, DD iL 2 , concerns halogen bond CI . . . π contacts to two cations that are located in both neighbouring layers.
The  Figure S15, which portrays the five cations (D1-D5) that exhibit short contacts to one PF 6 − anion. Four of the cations are located in one layer, belonging to two neighbouring chains (D1and D2 in one chain and D3 and D4 in another) and a fifth, D5, is located in a distinct layer. The DA interactions are particularly strong to D1 and D2, slightly weaker to D3 and relatively weak to D4 and D5. The solvent molecules, H 2 O in 4.PF 6 Figure S16. It is possible to observe that the indirect connectivity essentially reinforces the already strong intrachain and intralayer interactions, involving mostly cations that show already quite strong direct DD interactions between themselves, such as DD Ch 1 , DD Ch 2 and DD L , while the reinforcement of the intralayer direct DD interactions is relatively modest in case of DD iL 1 and non-existent for DD iL 2 . In 3.PF 6 .MeCN and 4.PF 6 .H 2 O the indirect DAD connectivity is similar to that observed for 5.PF 6 .1.5H 2 O; however, the effect of DA interactions seems to be stronger in 3.PF 6 .MeCN and 4.PF 6 .H 2 O, particularly concerning the interlayer connectivity. Unlike in 5.PF 6 .1.5H 2 O, in these salts the interlayer DD iL 2 interactions, corresponding to the interactions involving the blue cations in Figure S14, are also assisted by anion mediated interactions.
Furthermore, in 4.PF 6 .H 2 O, the effect of the DA interactions is expected to be slightly enhanced, as in this salt there are two anions contributing to the reinforcement of DD iL 1 and the anions establish an additional DAD interaction to a second cation located in the neighbouring layer, reinforcing the overall interlayer connectivity. The indirect connectivity assistances to the direct DD intralayer, DD L , and interlayer, DD iL 1 and DD iL 2 , interactions regarding 3.PF 6 .MeCN, 4.PF 6 .H 2 O and 5.PF 6 .1.5H 2 O are compared in Figure S17.
A slight intrachain DD alternation is observed in the distances between the average planes of the interacting 5-X-qsal ligands (DD Ch 1 and DD Ch 2 , in Table S5). This seems to be mostly caused by the alternation of the interactions that are mediated either by the solvent molecules in 2 ) lead to a larger separation in the intrachain. This effect appears to be slightly larger in 3.PF 6 .MeCN that shows stronger DA interactions. The small differences observed on the intralayer DD contacts (DD L , in Table S5), in particular for 3.PF 6 .MeCN, seems also to result from the stronger indirect connectivity provided by the PF 6 − anions and the solvent molecules in Figure S18.
The crystal packing and in particular the direct DD contacts are very similar and the strong intrachain and interchain intralayer DD interactions are expected to lead to rigid layer of cations, which seem to be responsible for the easy propagation of the lattice distortions and the SCO processes.

[Fe(5-X-qsal) 2 ] + Based Salts Reported in the Literature
Most of the salts based on [Fe(5-X-qsal) 2 ] + cations and small anions reported so far have crystal structures displaying arrangements of parallel chains of cations, resulting from an extensive overlap of the 5-X-qsal ligands. In these salts, as in the case of 3.PF 6 .MeCN, 4.PF 6 .H 2 O and 5.PF 6 .1.5H 2 O, the cation chains are organized as strongly coupled layers, through relatively strong interchain DD interactions, mostly associated with close contacts between the Qn fragments of cations located in neighbouring chains, typically with relatively modest interlayer DD interactions.

1.[Fe(CN) 5 NO].MeOH
In spite of the crystal packing similarities of these salts, the progress of the SCO transitions is quite diverse, ranging from single step gradual SCO processes, typical of weak cooperativity, to relatively sharp SCO processes related to strong cooperativity, either in a single step or in multistep SCO processes (Table S6). In a few cases, the SCO processes were irreversible, possibly due to solvent loss or other severe structural modification as in 1.SCN.DCM [15], 1.SeCN.DCM [25], 5.CF 3 SO 3 .EtOH [18] and 6.SCN.DCM [23]. In 1.[Fe(CN) 5 [11], upon cooling, half of the [Fe(5-X-qsal) 2 ] + cations remain blocked in the HS state and in 7.Cl.4MeOH.H2O, 7.ClO 4 .MeOH and 7.NO 3 [21] all the cations remain blocked in the HS state and no SCO transition is observed.
Gradual processes are only observed for three salts, 1.[Fe(CN) 5 NO].2MeCN [11], 5.CF 3 SO 3 .nPrOH [18] and 5.CF 3 SO 3 .iPOH [18]. In 1.[Fe(CN) 5 NO].2MeCN [11], the crystal structure obtained at 100 K, consists of layers based on alternated chains of HS and LS cations, with an overall arrangement similar to that described for 5.PF 6  . . π short contacts between the Ph fragments of the ligands, while in the case of the highly distorted HS cations involve a variety of ππ and CH . . . π contacts between a Ph fragment of one cation and Ph and Qn fragments from two neighbouring cations. In 1.[Fe(CN) 5 NO].2MeCN the strengthening of intra and interlayer DD coupling seems to induce both the blocking of HS configuration in one of the cation sites and the gradual nature of the SCO process in the second cation site. 5.CF 3 SO 3 .nPrOH and 5.CF 3 SO 3 .iPOH [18] exhibit crystal structures also quite similar to the one previously described for 5.PF 6 .1.5H 2 O. However, in those salts the interchain intralayer interactions between the two Qn fragments are expected to be considerably weaker due to a much poorer overlap between the aromatic systems of the interacting ligands, which is expected to lead to softer cationic layers and less cooperative SCO behaviours. The [Fe(5-tBu-qsal) 2 ] + cations were found to remain blocked in the HS state in 7.Cl.4MeOH.H 2 O, 7.ClO 4 .MeOH and 7.NO 3 [21]. These compounds display crystal structures similar to those observed in 3.PF 6 .MeCN, 4.PF 6 .H 2 O and 5.PF 6 .1.5H 2 O, also based on arrangements of layers composed by strongly coupled parallel chains of cations. However, the bulky t-Bu groups located in the periphery of the layers leave them in close proximity to those located in neighbouring layers. The blocking of the HS state in these salts most probably results from the strong repulsive interactions between the t-Bu groups, which prevent the structural rearrangements associated with the SCO processes.
The slight reduction in the sharpness of the SCO process observed for 1.I 3 [13], 3.SCN.MeOH [16], 3.PF 6 .MeCN and 4.PF 6 .H 2 O appears to correlate with two different factors. First, the lower rigidity in the cationic layer observed for 3.SCN.MeOH, where the distance between the interacting Qn fragments is considerably larger than for the other salts. Second, the stronger interlayer interactions, either due to the direct DD interactions in 3.SCN.MeOH and 3.PF 6 .MeCN or to the DAD assisted connectivity that seems to be more effective in 1.I 3 , 3.SCN.MeOH and 4.PF 6

.H 2 O.
It appears that the Chain Layers type of crystal packing consisting of an arrangement of relatively rigid 2D layers of HS [Fe(5-X-qsal) 2 ] + cations is quite promising in order to obtain cooperative SCO behaviours. However, the increase of the strength of interlayer interactions seem to constitute an effective perturbation, which leads to the loss of cooperativity or eventually to prevent the SCO processes in the case those interactions are too strong as in the Type II Dimer Layers. This seems to suggest that although these relatively rigid 2D cationic layers favour the propagation of the distortion of the cations associated with the SCO process throughout the lattice, the extremely strong interlayer interactions bring in a resistance to the propagation of that distortion.
It should also be noted that no correlation is observed between the type of supramolecular structures or the magnetic behaviour and the octahedral distortions of the coordination sphere of the Fe 3+ (Table S7).

Synthesis
Commercial solvents were used without further purification unless otherwise stated. The purity of the complexes and their solvation state were checked by determining the carbon, hydrogen, nitrogen and sulfur content at the C2TN Elemental Analysis Service.

3.Cl.MeCN.H 2 O-
In dry conditions a methanolic solution (30 mL) of 5-chlorosalicylaldehyde (2.0 mmol, 313.13 mg) was added to a methanolic solution (70 mL) of 8-aminoquinoline (2.0 mmol, 288.35 mg). The resulting mixture was stirred for two hours at room temperature and changed its colour from pale yellow to yellow. Afterwards a methanolic solution (10 mL) of sodium methoxide (2.5 mmol, 135.05 mg) was added to the yellow mixture followed by a methanolic solution (20 mL) of iron (III) chloride anhydrous (1.0 mmol, 162.20 mg). The solution turned black and the volume was reduced to 50 mL. The resulting mixture was then placed at −20 • C overnight. A dark solid was recovered by filtration

X-ray Crystallography
A summary of the crystal data, experimental details and refinement results are listed in Table S1. The X-ray diffraction, XRD, experiments were performed at 150 K with a Bruker AXS APEX CCD detector and a four-circle diffractometer using a graphite-monochromated Mo Kα radiation source (λ = 0.71073 Å) in the ψ and ω scans mode. A semiempirical absorption correction was carried out using SADABS [26]. Data collection, cell refinement, and data reduction were done with the SMART and SAINT programs [27]. Structures were solved by direct methods using SIR97 [28] and refined by full-matrix least-squares methods using the program SHELXL97 [29] and a WINGX software package [30]. Nonhydrogen atoms were refined with anisotropic thermal parameters, whereas H atoms were placed in idealized positions and allowed to refine riding on the parent C atom. Molecular graphics were prepared using MERCURY [31]. Data

Magnetic Characterization
Magnetic measurements on compounds were performed in a S700X SQUID magnetometer with a 7 T magnet (Cryogenic Ltd., London, UK) using polycrystalline samples inserted in gelatine capsules. The temperature dependences of the magnetic susceptibility in the temperature range 5 − 320 K, were measured in step mode under magnetic fields of 1 T for compounds 3 (X = Cl), and 4 (X = Br) and 0.05 T for compounds 5 (X = I), using the same rate on the heating/cooling cycles.
For each experimental point, the temperature of the sample was stabilized for at least 5 min. The paramagnetic susceptibility, χ P T, was deduced from the experimental magnetization data after correction of the diamagnetism of the sample holder and the diamagnetic susceptibility contribution of each compound, estimated from the tabulated Pascal constants. For the different set of compounds, the correspondent diamagnetic susceptibilities, χ D , were the following: 3

Conclusions
In this work we have characterized the structure and magnetic properties of 13 compounds based on the [Fe(5-X-qsal) 2 ] + cation (X = Cl, Br and I). Four structural motives emerged: Dimer Chains, Dimer Layers (Type I and Type II) and Chain Layers. The compounds that share the same structural motive reveal similar magnetic behaviours from room temperature down to 4K.
A Dimer Chains type of structure characterized the compounds with the weakest connectivity. They are organized as isolated chains with no interchain cation-cation (DD) connectivity and a weak intrachain or interdimer connectivity. These compounds are LS up to around 250 K, where a broad spin transition starts to occur. The lack of connectivity between the cationic centres should be the source for the lack of cooperativity of the cations and the softness of the structural network.
The compounds characterized with the Type I Dimer Layers structure, based on arrangements of parallel layers of interacting cation dimers, are also LS at around 250 K but display some diversity in the magnetic behaviour. Some show no change of the χ P T up to room temperature while others start to exhibit spin transition with various degrees of sharpness/broadening. These differences can be related to a degree of variability in the interlayer DD connectivity as the relatively weak intralayer connectivity of these compounds has few dissimilarities.
The Type II Dimer Layers compounds exhibit the highest degree of intra-and interlayer connectivity. The interlayer interactions are so strong that they intensify the rigidity of the structural network to a point where the transition from HS to LS states is blocked. In fact the compounds with this structure type remain in the HS state in the 295-4 K temperature range.
The Chain Layers structural motive shows significant intralayer interactions but weak interlayer interactions. These interactions allow for a relatively rigid layer of cations, which can induce easy propagation of the lattice distortions and more or less sharp SCO processes. The vast majority of the [Fe(5-X-qsal) 2 ] + based compounds reported in the literature show SCO transitions and their detailed structural analysis reveal that they are characterized as Chain Layers. The only Chain Layers compounds that remain blocked in the HS state at all temperatures are those based on the [Fe(5-tBu-qsal) 2 ] + cation. In these compounds the extreme steric hindrance provided by the bulky tert-butyl groups located in the periphery of the layers, may be the key factor for the HS trapping, as already theorized by the authors that first characterized these compounds.
The main structural features determining the SCO behaviour of [Fe(5-X-qsal) 2 ] + based salts were surveyed in the present study. Although it is still difficult to predict if a specific type of ligand substituent or counter ion leads to a desired supramolecular configuration, the categorization of these materials according to these supramolecular arrangements allows narrowing the pool of possible chemical combinations in order to achieve a desired arrangement. Further work is in progress, aiming to build such a library of compounds correlating supramolecular structure and magnetic behaviour, from which trends regarding the combination of ligands, counter ions and solvating molecules may emerge. The analysis of the data from this library will allow pre-selecting the best candidates for a desired magnetic behaviour, thus reducing laboratory time for the preparation and characterization of these materials, replacing the "bulk synthesis" approach by a more efficient "synthesis by design".  Table S1: Summary of the crystal data, experimental details and refinement results, Table S2: Key parameters describing the DD contacts in the crystal structures of 3.BPh 4 and 4.BPh 4 , with the Dimer Chains supramolecular configuration, Table S3: Key parameters describing the DD contacts in the crystal structures of the salts with type I Dimer Layers supramolecular configuration, Table S4: Key parameters describing the DD contacts in the crystal structures of the salts with type II Dimer Layers supramolecular configuration, Table S5: Key parameters describing the DD contacts in the crystal structures of 3.PF 6 .MeCN, 4.PF 6 .H 2 O and 5.PF 6 .2H 2 O, with the Chain Layers supramolecular configuration, Table S6: Summary of the parameters characterizing the SCO behaviour of the salts with the Chain Layers supramolecular configuration and with HS configurations at room temperature, Table S7 Summary of the parameters characterizing the octahedral distortion (Σ and θ).

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.