New Silver(I) Coordination Polymer with Fe4 Single-Molecule Magnets as Long Spacer

In continuation of our work on supramolecular architectures of single-molecule magnets (SMMs) as a promising strategy in developing their magnetic performance, in this paper we report the synthesis and single crystal X-ray structure of the centered triangular tetrairon(III) SMM, [Fe4(PhpPy)2(dpm)6], Fe4 (Hdpm = dipivaloylmethane, H3PhpPy = 2-(hydroxymethyl)-2-(4-(pyridine-4-yl)phenyl)propane-1,3-diol), and its assembly in the coordination polymer {[Fe4(PhpPy)2(dpm)6Ag](ClO4)}n, Fe4Ag, upon reaction with silver(I) perchlorate. Thanks to the presence of the pyridyl rings on the two tripodal ligands, Fe4 behaves as divergent ditopic linker, and due to the Fe4:AgClO4 1:1 ratio, Fe4Ag probably possesses a linear arrangement in which silver(I) ions are linearly coordinated by two nitrogen atoms, forming 1D chains whose positive charge is balanced by the perchlorate anions. The stabilization of such a polymeric structure can be ascribed to the long distance between the two donor nitrogen atoms (23.4 Å) and their donor power. Fe4Ag shows slow relaxation of the magnetization which follows a thermally activated process with Ueff/kB = 11.17(18) K, τ0 = 2.24(17) 10−7 s in zero field, and Ueff/kB = 14.49(5) K, τ0 = 3.88(8) 10−7 s in 1-kOe applied field, in line with what reported for tetrairon(III) SMMs acting as building blocks in polymeric structures.


Introduction
Single-molecule magnets (SMMs) [1], thanks to their directionally bistable magnetic moment, have been investigated over the last two decades as nanoscale, chemically tunable entities for encoding binary information. Though SMMs developed so far show slow relaxation of the magnetization only at low temperatures (below liquid nitrogen), preventing their application in spintronics devices [2], a fast increase of their performance is expected, as recently shown through the synthesis of a dysprosocenium complex with opening of magnetic hysteresis at a temperature as high as 60 K [3,4]. Another promising strategy to develop SMM features consists of the assembly into extended coordination networks [5][6][7][8][9][10], which may allow enhanced magnetic remanence due to intramolecular magnetic coupling [11].
Complexes of the family of tetrairon(III) SMMs with general formula [Fe 4 (L) 2 (dpm) 6 ] [12][13][14] (H 3 L = tripodal ligand, Hdpm = dipivaloylmethane) include four coplanar high spin iron(III) ions (s = 5/2) forming a centered triangular geometry. The central metal atom (Fe c ) is antiferromagnetically coupled to the three peripheral ones (Fe p ) so as to give an S = 5 spin ground state. In general, clusters of this family feature easy axis anisotropy normal to the metal plane. Despite the low energy barrier of Fe 4 SMMs reaching up to 15-17 K, their chemical versatility has been exploited by creating clusters able to sublimate by fluorination of the peripheral β-diketonato ligands [15][16][17] or anchor on gold surfaces by sulfur-rich tripodal ligands [18,19], as well as modulating the spin ground state by selective replacement of the central metal ion with chromium(III) [20,21], vanadium(III) [22], or lanthanide ions [23,24]. This family of SMMs has been also characterized as single-molecule junction in scanning tunneling microscope (STM) experiments at 0.5 K [25] and through theoretical calculations for its behavior in spin-polarized transport [26], revealing how Fe 4 SMMs would be excellent candidates for molecular spintronics devices. Recently, the derivative [Fe 4 (pPy) 2 (dpm) 6 ] (A) (H 3 pPy = 2-(hydroxymethyl)-2-(pyridine-4-yl)propane-1,3-diol) [11] has also revealed to behave as efficient ditopic supramolecular synthon due to the presence of two divergent 4-pyridyl substituents on the tripodal ligands, as formerly found in the assembly of 1D chains by using ruthenium paddlewheel dimers as coordination nodes [11]. By modulating the redox state of the Ru 2 units (II,II vs. II,III), we could introduce a significant exchange bias with intra-chain coupling, and concomitant enhancement of the remnant magnetization. The ability of A as building block of supramolecular structures has been further confirmed by reaction with silver(I) perchlorate yielding {[Fe 4 (pPy) 2 (dpm) 6 ] 2 Ag}(ClO 4 ) (B), a tridimensional metal-organic framework (3D-MOF) [14]. In this compound, Ag + ions behave as tetrahedral nodes coordinated by four pyridyl nitrogen atoms of Fe 4 units, and the cubic crystals, although with no overall second-order anisotropy, can be persistently magnetized parallel or antiparallel to the four main diagonals of the unit cell.
Compound B also possesses accessible voids (14.3% of unit cell volume) able to trap small solvent molecules, which can be exploited as a means for modulating the magnetic properties. However, due to the presence of two interlocked dia networks (according to IUPAC nomenclature [27]), the space left inside the 3D-MOF is not available for bigger molecules [14]. In the attempt to build-up a new polymeric structure with larger cavities by elongating the SMM spacer, in this work we synthesized the new tripodal ligand H 3 PhpPy, 2-(hydroxymethyl)-2-(4-(pyridine-4-yl)phenyl) propane-1,3-diol (Scheme 1), where we added a phenyl ring between the triol unit and the pyridyl terminal fragment with respect to H 3 pPy. This ensures the same divergent ditopic linker behavior once incorporated into an Fe 4 SMM, constituting at the same time a longer spacer upon coordination of the pyridyl rings to silver(I) ions. Here we report the synthesis of H 3 PhpPy, its assembly into [Fe 4 (PhpPy) 2 (dpm) 6 ] (Fe 4 ), whose structure was confirmed by single crystal X-ray diffraction studies, and further reaction of Fe 4 with AgClO 4 . The behavior as SMM of the resulting coordination polymer {[Fe 4 (PhpPy) 2 (dpm) 6 Ag](ClO 4 )} n (Fe 4 Ag) will be then described, in comparison with the 3D-MOF B and the Fe 4 building block.

Synthesis of the Tripodal Ligand H 3 PhpPy
Two possible synthetic paths for H 3 PhpPy were identified, both starting from commercial 2-(4-bromophenyl)ethanol and involving three synthetic steps. The most convenient method proved to be the one featuring a first Suzuki cross-coupling with pyridin-4-ylboronic acid, followed by oxidation of the homobenzylic alcohol and completed by a Tollens addition with paraformaldehyde (Scheme 1). Another variant was also investigated, featuring the oxidation, Tollens and Suzuki sequence, but its efficacy was lower (Scheme S1 in Supplementary Materials).
Following Scheme 1, the cross-coupling reaction between 2-(4-bromophenyl)ethanol and pyridin-4-ylboronic acid was performed on the basis of a procedure reported in the literature [28] by means of Pd(PPh 3 ) 4 as catalyst. The presence of the pyridyl group gave us the opportunity to separate the product 1 from N,N-dimethylformamide (DMF) by precipitating it as hydrochloride salt, avoiding the use of column chromatography and still guaranteeing good yield and optimal purity as detectable with GC-MS and 1 H NMR analyses. The oxidation of 1 was secured with a radical method, employing trichloroisocyanuric acid (TICA) and (2,2,6,6-tetramethylpiperidin-1-yl)oxyl (TEMPO) radical [29], while other tested methods caused the partial conversion to the more stable benzaldehyde derivative. The following reaction of 2 with paraformaldehyde and Ca(OH) 2 in tetrahydrofuran (THF) [29][30][31] allowed the double aldolic addition and the Cannizzaro reaction to the final tripodal ligand H 3 PhpPy with a reasonable yield of 41%. The addition of Al(OiPr) 3 as additive had not a clear scope in this reaction, but we could observe easier purification of the crude product. The overall yield of the tripodal ligand starting from commercial 2-(4-bromophenyl)ethanol is then 23%. The alternative pathway reported in Scheme S1 in Supplementary Materials was also attempted in order to move the triol-formation step, characterized by the lowest yield, before the cross-coupling. In this way it would have been possible to save on the amount of used palladium catalyst, which is the most expensive reagent in this synthesis. The first oxidation step of 2-(4-bromophenyl)ethanol with TICA and TEMPO led to 3 in very high yield, and the following reaction with paraformaldehyde allowed the isolation of 4 with the formed triol fragment. Unfortunately, the cross-coupling of 4 with pyridin-4-ylboronic acid in the presence of Pd(PPh3)4 did not yield the desired final tripodal ligand even after several attempts. This might possibly due to an interference of the triol unit that can coordinate the palladium ion, sequestering it from the catalytic cycle, or an interaction between the triol unit and the boronic acid.

Assembly of Fe4 and Its Molecular Structure
The assembly of the Fe4 cluster with H3PhpPy was performed as previously established [11,12] by reacting a solution of [Fe4(OMe)6(dpm)6] [32] in diethyl ether (Et2O) with the tripodal ligand suspended in ethanol (EtOH). Upon addition of H3PhpPy a darkening of the orange solution could be observed, hint of the exchange of the six bridging methoxido ligands with two tripodal ones. Nevertheless, by vapor diffusion of EtOH with the reaction mixture we could almost always obtain crystals of the dimeric cluster [Fe2(OEt)2(dpm)4] (thermodynamic product), whose formation is known to be favored in basic conditions [33]. Differently from H3pPy, the inductive effect of the added phenyl ring might pushes electron density toward the pyridyl moiety [34] in the tripodal ligand H3PhpPy, with resulting pronounced basicity.
Nevertheless, we could isolate single crystals of Fe4 suitable for X-ray diffraction from the concentrated mother liquor of one synthetic attempt. They belong to the monoclinic space group C2/c as most of the reported homologous tetrairon(III) clusters [12,13,29], and in particular its shorter analogue A⋅2EtOH [11]. Selected geometrical parameters at 293(2) K are reported in Table 1, and a view of the molecular structure is shown in Figure 1.
The four iron(III) ions define a metal-centered triangle that develops around the crystallographic two-fold axis passing through Fe1 and Fe2. The molecular structure is held together by two tri-deprotonated tripodal ligands that bridge the central ion (Fe1) to the peripheral ones (Fe2, Fe3, Fe3'; primed atom is generated by the two-fold rotation). The coordination sphere of each peripheral metal ion is completed by two bidentate dpm − ligands to form the classical propeller-like The alternative pathway reported in Scheme S1 in Supplementary Materials was also attempted in order to move the triol-formation step, characterized by the lowest yield, before the cross-coupling. In this way it would have been possible to save on the amount of used palladium catalyst, which is the most expensive reagent in this synthesis. The first oxidation step of 2-(4-bromophenyl)ethanol with TICA and TEMPO led to 3 in very high yield, and the following reaction with paraformaldehyde allowed the isolation of 4 with the formed triol fragment. Unfortunately, the cross-coupling of 4 with pyridin-4-ylboronic acid in the presence of Pd(PPh 3 ) 4 did not yield the desired final tripodal ligand even after several attempts. This might possibly due to an interference of the triol unit that can coordinate the palladium ion, sequestering it from the catalytic cycle, or an interaction between the triol unit and the boronic acid.

Assembly of Fe 4 and Its Molecular Structure
The assembly of the Fe 4 cluster with H 3 PhpPy was performed as previously established [11,12] by reacting a solution of [Fe 4 (OMe) 6 (dpm) 6 ] [32] in diethyl ether (Et 2 O) with the tripodal ligand suspended in ethanol (EtOH). Upon addition of H 3 PhpPy a darkening of the orange solution could be observed, hint of the exchange of the six bridging methoxido ligands with two tripodal ones. Nevertheless, by vapor diffusion of EtOH with the reaction mixture we could almost always obtain crystals of the dimeric cluster [Fe 2 (OEt) 2 (dpm) 4 ] (thermodynamic product), whose formation is known to be favored in basic conditions [33]. Differently from H 3 pPy, the inductive effect of the added phenyl ring might pushes electron density toward the pyridyl moiety [34] in the tripodal ligand H 3 PhpPy, with resulting pronounced basicity.
Nevertheless, we could isolate single crystals of Fe 4 suitable for X-ray diffraction from the concentrated mother liquor of one synthetic attempt. They belong to the monoclinic space group C 2 /c as most of the reported homologous tetrairon(III) clusters [12,13,29], and in particular its shorter analogue A·2EtOH [11]. Selected geometrical parameters at 293(2) K are reported in Table 1, and a view of the molecular structure is shown in Figure 1.
The four iron(III) ions define a metal-centered triangle that develops around the crystallographic two-fold axis passing through Fe1 and Fe2. The molecular structure is held together by two tri-deprotonated tripodal ligands that bridge the central ion (Fe1) to the peripheral ones (Fe2, Fe3, Fe3'; primed atom is generated by the two-fold rotation). The coordination sphere of each peripheral metal ion is completed by two bidentate dpm − ligands to form the classical propeller-like structure with helical pitch, calculated as dihedral angle between Fe 4 and FeO 2 Fe planes, of about 67-69 • . Thus, all metal centers achieve a distorted octahedral environment formed by six oxygen atoms with Fe-O distances falling in the 1.97-2.03 Å narrow range. The isosceles triangle given by the peripheral iron(III) ions possesses two long sides (5.3903(11) Å) and a short one (5.2063(15) Å), and no ethanol molecules are involved in hydrogen bonds with the nitrogen atom of the pyridyl ring in the solid state, differently from A·2EtOH [11]. Phenyl and pyridyl rings of the tripodal ligand are not co-planar, being rotated each other of 6.47(5) • . The last ones are involved in weak π···π stacking interactions (the shortest C-C distance is 3.67 Å) that, together with weak hydrogen bonds, contribute to stabilize the overall crystal structure ( Figure S1 in Supplementary Materials). , and no ethanol molecules are involved in hydrogen bonds with the nitrogen atom of the pyridyl ring in the solid state, differently from A⋅2EtOH [11]. Phenyl and pyridyl rings of the tripodal ligand are not co-planar, being rotated each other of 6.47(5)°. The last ones are involved in weak π⋅⋅⋅π stacking interactions (the shortest C-C distance is 3.67 Å) that, together with weak hydrogen bonds, contribute to stabilize the overall crystal structure ( Figure S1 in Supplementary Materials).  Table 1. Color code: Fe = orange, O = red, N = violet, C = grey, H = black. Table 1. Selected interatomic distances (Å) and angles (°) in Fe4 1 (see Figure 1 for atom labelling).  (11) Due to the difficulty to isolate the desired cluster as pure crystalline phase, we decided to investigate whether or not Fe4 was already formed in solution after 5 h of stirring between [Fe4(OMe)6(dpm)6] and H3PypPy. The reaction mixture was analyzed via ESI-MS spectrometry and we could positively identify the main peaks at m/z = 1874.7 and 1836.8, belonging to the +K and +H positive ions of Fe4 ( Figure S2 in Supplementary Materials). The only other detected peak appears at m/z = 644.4, which corresponds to [Fe(dpm)3 + K] + , a decomposition product of [Fe4(OMe)6(dpm)6] upon standing in solution [12]. Proved that the desired cluster is formed in solution (kinetic product), the reaction mixture was then taken to dryness under reduced pressure without heating and the infrared spectrum of the obtained orange solid was compared with the ones of the dimeric species [Fe2(OEt)2(dpm)4] and the starting cluster [Fe4(OMe)6(dpm)6] ( Figure S3 in Supplementary  Table 1. Color code: Fe = orange, O = red, N = violet, C = grey, H = black. Table 1. Selected interatomic distances (Å) and angles ( • ) in Fe 4 1 (see Figure 1 for atom labelling). Due to the difficulty to isolate the desired cluster as pure crystalline phase, we decided to investigate whether or not Fe 4 was already formed in solution after 5 h of stirring between [Fe 4 (OMe) 6 (dpm) 6 ] and H 3 PypPy. The reaction mixture was analyzed via ESI-MS spectrometry and we could positively identify the main peaks at m/z = 1874.7 and 1836.8, belonging to the +K and +H positive ions of Fe 4 ( Figure S2 in Supplementary Materials). The only other detected peak appears at m/z = 644.4, which corresponds to [Fe(dpm) 3 + K] + , a decomposition product of [Fe 4 (OMe) 6 (dpm) 6 ] upon standing in solution [12]. Proved that the desired cluster is formed in solution (kinetic product), the reaction mixture was then taken to dryness under reduced pressure without heating and the infrared spectrum of the obtained orange solid was compared with the ones of the dimeric species [Fe 2 (OEt) 2 (dpm) 4 ] and the starting cluster [Fe 4 (OMe) 6 (dpm) 6 ] ( Figure S3 in Supplementary Materials). Even if very similar, the three spectra contain substantial differences in the position and relative intensities of several bands, which all hint to confirm the formation of Fe 4 . In particular, the coordination of the tripodal ligand is supported by the alkoxido C-O stretching band shifted to 1096 from 1050 cm −1 in the progenitor tetranuclear cluster [Fe 4 (OMe) 6 (dpm) 6 ].

Parameter
Furthermore, the orange solid was analyzed via 1 H NMR spectroscopy dissolving the product in toluene-d 8 : the spectrum ( Figure S4 in Supplementary Materials) clearly entails the intense broad signal at 10.9 ppm characteristic of the paramagnetically shifted tBu substituents in this kind of Fe 4 complexes [12,15,16]. Other four broad signals at 12.0 (shoulder) 8.8, 8.5 and 6.5 ppm can be observed, ascribable to the corresponding four different aromatic protons of the tripodal ligand; in particular the shoulder at 12.0 ppm can be attributed to H5 (Scheme 1 for atom numbering), in analogy to the known Fe 4 complexes with phenyl substituent [12,16], while the narrower peak at 8.8 ppm should be generated by H9, close to the nitrogen atom of the pyridyl ring and farther from the paramagnetic core. An additional shoulder at 13.1 ppm due to decomposition to [Fe(dpm) 3 ] is present [12], in agreement with what observed in the ESI spectrum; anyway, this signal is weaker than the ones of the desired main product Fe 4 . Most important, the 1 H NMR spectrum remained almost unchanged after 16 h, which allow us to conclude that the desired cluster is stable in toluene and suitable for its subsequent use.

Assembly of Fe 4 Ag
The reaction of Fe 4 with AgClO 4 was performed by liquid/liquid diffusion of a toluene solution containing the tetrairon(III) cluster into a THF solution of AgClO 4 . Regardless of the relative concentration, we could always isolate needle-like microcrystals of Fe 4 Ag with a 1:1 ratio between the two components. This was confirmed by comparing the experimental percentages with the calculated elemental analysis of different stoichiometric combinations, always fitting with the [Fe 4 AgClO 4 ] minimal formula. When using a Fe 4 :AgClO 4 2:1 ratio, the formation of microcrystalline Fe 4 Ag resulted in a darker orange solution containing the excess of unreacted Fe 4 .
The infrared spectrum of Fe 4 Ag fully resembles the one of Fe 4 ( Figure S5 in Supplementary Materials), which proves the intactness of the starting tetrairon(III) cluster. The presence of AgClO 4 is confirmed by the appearance of a unique perchlorate stretching band at 1122 cm −1 , distinctive of an anion in tetrahedral symmetry and purely ionic character [35], as previously observed for B [14]. It is also possible to detect a weak band at 479 cm −1 tentatively ascribable to the Ag-N bond stretching [36], confirming the coordination of silver(I) by the nitrogen atoms of the pyridyl rings. Another proof can be found in the ESI-MS spectrum of Fe 4 Ag in THF/CH 3 CN with the peak at m/z = 2014.7 for the [Fe 4 Ag(THF)] + ion, in addition to the peaks at m/z = 1836.4 and 1874.8, as previously observed for Fe 4 ( Figure S6 in Supplementary Materials). Further confirmation is given by the X-ray fluorescence (XRF) spectrum, where Fe, Ag, and Cl are clearly detected ( Figure S7 in Supplementary Materials).
The orange microcrystals of Fe 4 Ag are practically insoluble in all common solvents, clearly suggesting the polymeric nature of the product. It revealed to be soluble only in pyridine, which coordinates silver(I) ions replacing Fe 4 . The Fe 4 :AgClO 4 1:1 ratio, the needle-like crystal habit which resemble the 1D chains with ruthenium paddlewheel dimers [11], the insolubility of the product and the ionic T d symmetry of the perchlorate anions as given by the infrared spectrum led to identify Fe 4 Ag as a probable linear coordination polymer of silver(I) ions linked together by intact Fe 4 units acting as long spacer (N1···N1' = 23.413(11) Å), whose positive charge is balanced by ClO 4 − ions occupying voids in between adjacent 1D chains (Scheme 2). The propensity of silver(I) to give linear coordination environment with pyridine [37][38][39] or nitrogen ligands [40][41][42][43] is reported, and in this case the high basicity of the pyridyl ring given by the releasing electronic effect of the attached phenyl ring [34] might convey enough donor power to only two nitrogen atoms in order to satisfy the coordinative electronic demand of silver(I).

Magnetic Characterization
Thanks to the intactness of Fe4 units within the polymer and the purity of the crystalline phase as given by elemental analysis and optical microscope inspection, we performed direct current (dc) magnetic measurements on Fe4Ag. The temperature dependence of the molar magnetic susceptibility, χM, in low fields (1-10 kOe) was measured between 1.9 and 300 K, and the data are reported in Figure 2 as χMT vs. T plot. The shape is characteristic for triangular-centered Fe4 clusters, where the dominant antiferromagnetic interactions between Fe c and Fe p centers lead to an S = 5 ground state [12,14,15,32]. The χMT value at room temperature is about 13.0 emu K mol −1 , hence lower than the Curie constant for four s = 5/2 spins (17.51 emu K mol −1 with g = 2.00); upon cooling, it first decreases towards a minimum at 95 K (ca. 10.2 emu K mol −1 ) and then increases again reaching 12.9 emu K mol −1 at 8 K, before a further drop at the lowest temperatures. The χMT value at low T is then rising toward the Curie constant for the expected S = 5 ground state (15.00 emu K mol −1 with g = 2.00). Magnetic exchange interactions among the four iron(III) centers are usually obtained by fitting the χMT vs. T curve to a Heisenberg plus Zeeman spin Hamiltonian that assumes threefold molecular symmetry to avoid over-parametrization [15]: where S1 is the spin vector for Fe c ; S2, S3, and S4 are the spin vectors for Fe p ions; S is the total spin vector; H is the applied magnetic field; and J and J' coupling constants are the nearest-neighbor (n.n.) and next-nearest-neighbor (n.n.n.) interactions, respectively. Due to the lower increase of the χMT value at low T with respect to the expected Curie constant for the S = 5 ground state reported above and the fact that n.n. interactions are most effective in the high-temperature regime, the best-fit procedure was conducted in the 40-300 K range with data obtained with a 10-kOe applied field (filled circles in Figure 2) using PHI program [44], yielding a satisfactorily reproduction of the experimental data with g = 1.9840 (12), J = 13.98(8) cm −1 and J' = 0.86(2) cm −1 (black curve in Figure 2). These parameters are in line with what reported for the Fe4 SMM family with isotropic g close to 2.00, as expected for high spin iron(III) ions, and the presence of dominant n.n. antiferromagnetic coupling and J' of one order of magnitude smaller than J [11,12,14,15]. This last coupling constant is slightly lower than expected (15-19 cm −1 [11,13]), but this can be easily envisaged by the lower value

Magnetic Characterization
Thanks to the intactness of Fe 4 units within the polymer and the purity of the crystalline phase as given by elemental analysis and optical microscope inspection, we performed direct current (dc) magnetic measurements on Fe 4 Ag. The temperature dependence of the molar magnetic susceptibility, χ M , in low fields (1-10 kOe) was measured between 1.9 and 300 K, and the data are reported in Figure 2 as χ M T vs. T plot. The shape is characteristic for triangular-centered Fe 4 clusters, where the dominant antiferromagnetic interactions between Fe c and Fe p centers lead to an S = 5 ground state [12,14,15,32]. The χ M T value at room temperature is about 13.0 emu K mol −1 , hence lower than the Curie constant for four s = 5/2 spins (17.51 emu K mol −1 with g = 2.00); upon cooling, it first decreases towards a minimum at 95 K (ca. 10.2 emu K mol −1 ) and then increases again reaching 12.9 emu K mol −1 at 8 K, before a further drop at the lowest temperatures. The χ M T value at low T is then rising toward the Curie constant for the expected S = 5 ground state (15.00 emu K mol −1 with g = 2.00). Magnetic exchange interactions among the four iron(III) centers are usually obtained by fitting the χ M T vs. T curve to a Heisenberg plus Zeeman spin Hamiltonian that assumes threefold molecular symmetry to avoid over-parametrization [15]: where S 1 is the spin vector for Fe c ; S 2 , S 3 , and S 4 are the spin vectors for Fe p ions; S is the total spin vector; H is the applied magnetic field; and J and J' coupling constants are the nearest-neighbor (n.n.) and next-nearest-neighbor (n.n.n.) interactions, respectively. Due to the lower increase of the χ M T value at low T with respect to the expected Curie constant for the S = 5 ground state reported above and the fact that n.n. interactions are most effective in the high-temperature regime, the best-fit procedure was conducted in the 40-300 K range with data obtained with a 10-kOe applied field (filled circles in Figure 2) using PHI program [44], yielding a satisfactorily reproduction of the experimental data with g = 1.9840 (12), J = 13.98(8) cm −1 and J' = 0.86(2) cm −1 (black curve in Figure 2).

Magnetic Characterization
Thanks to the intactness of Fe4 units within the polymer and the purity of the crystalline phase as given by elemental analysis and optical microscope inspection, we performed direct current (dc) magnetic measurements on Fe4Ag. The temperature dependence of the molar magnetic susceptibility, χM, in low fields (1-10 kOe) was measured between 1.9 and 300 K, and the data are reported in Figure 2 as χMT vs. T plot. The shape is characteristic for triangular-centered Fe4 clusters, where the dominant antiferromagnetic interactions between Fe c and Fe p centers lead to an S = 5 ground state [12,14,15,32]. The χMT value at room temperature is about 13.0 emu K mol −1 , hence lower than the Curie constant for four s = 5/2 spins (17.51 emu K mol −1 with g = 2.00); upon cooling, it first decreases towards a minimum at 95 K (ca. 10.2 emu K mol −1 ) and then increases again reaching 12.9 emu K mol −1 at 8 K, before a further drop at the lowest temperatures. The χMT value at low T is then rising toward the Curie constant for the expected S = 5 ground state (15.00 emu K mol −1 with g = 2.00). Magnetic exchange interactions among the four iron(III) centers are usually obtained by fitting the χMT vs. T curve to a Heisenberg plus Zeeman spin Hamiltonian that assumes threefold molecular symmetry to avoid over-parametrization [15]: where S1 is the spin vector for Fe c ; S2, S3, and S4 are the spin vectors for Fe p ions; S is the total spin vector; H is the applied magnetic field; and J and J' coupling constants are the nearest-neighbor (n.n.) and next-nearest-neighbor (n.n.n.) interactions, respectively. Due to the lower increase of the χMT value at low T with respect to the expected Curie constant for the S = 5 ground state reported above and the fact that n.n. interactions are most effective in the high-temperature regime, the best-fit procedure was conducted in the 40-300 K range with data obtained with a 10-kOe applied field (filled circles in Figure 2) using PHI program [44], yielding a satisfactorily reproduction of the experimental data with g = 1.9840 (12), J = 13.98(8) cm −1 and J' = 0.86(2) cm −1 (black curve in Figure 2). These parameters are in line with what reported for the Fe4 SMM family with isotropic g close to 2.00, as expected for high spin iron(III) ions, and the presence of dominant n.n. antiferromagnetic coupling and J' of one order of magnitude smaller than J [11,12,14,15]. This last coupling constant is slightly lower than expected (15-19 cm −1 [11,13]), but this can be easily envisaged by the lower value These parameters are in line with what reported for the Fe 4 SMM family with isotropic g close to 2.00, as expected for high spin iron(III) ions, and the presence of dominant n.n. antiferromagnetic coupling and J' of one order of magnitude smaller than J [11,12,14,15]. This last coupling constant is slightly lower than expected (15-19 cm −1 [11,13]), but this can be easily envisaged by the lower value of the minimum of the χ M T curve, which usually falls between 110 and 120 K in other Fe 4 derivatives [12][13][14]. Furthermore, the shift to about 90 K of the reproduced minimum from the experimental value of 95 K suggests that the real J might be slightly higher. Correction of this shift was carried out by simulating different curves with J ranging from 12 to 16 cm −1 , leaving g and J' unaltered. The best prevision of the minimum in the χ M T curve is obtained with J = 14.6 cm −1 (dashed blue curve in Figure 2), still maintaining a satisfactory reproduction of the experimental data at T > 40 K. Upon further cooling below 40 K, the χ M T product (empty circles in Figure 2 obtained with a 1-kOe applied field) follows the expected growing trend toward the Curie constant of the S = 5 ground state, but it is lower than the expected values from the fitting curve, and this can be probably ascribed to the presence of anisotropic effects or weak intermolecular interactions, which are more effective at low temperature.
Even if not isolated as crystalline phase, dc magnetic data were also collected for a powder sample of Fe 4 , whose elemental analysis closely reproduced the one of the pure compound, in order to perform a direct comparison with Fe 4 Ag. The temperature dependence of χ M in low fields (1-10 kOe) was measured between 5 and 250 K, and the data are reported in Figure S8 in Supplementary Materials as χ M T vs. T plot. As before, data were well-fitted by means of Equation (1) in the whole temperature range obtaining g = 1.9832(9), J = 15.10(6) cm −1 , J' = 0.80(2) cm −1 . Differently from Fe 4 Ag, in the case of Fe 4 the anisotropic effects at low temperature seem to be negligible; nevertheless, both n.n. and n.n.n. interactions in the polymeric 1D chain are fully comparable with the ones in the progenitor cluster.
The magnetization dynamics of both Fe 4 Ag and Fe 4 was then probed with alternating current (ac) susceptibility measurements in zero static field as a function of both temperature (down to 1.9 K) and frequency of the oscillating field (ν = 10-10,000 Hz). Characteristic frequency-dependent maxima in χ" M vs. T plots were detected, which indicate that Fe 4 Ag and Fe 4 behave as SMMs (Figure 3a and Figure S9a in Supplementary Materials). Measurements were then repeated at 1.9 K by applying a static field H in the 0-50 kOe range for Fe 4 Ag in order to evaluate the most effective field in slowing down relaxation by cutting quantum tunneling (QT) pathways (Figure 3b), and the static field H = 1 kOe was found to be the best option to repeat ac measurements at variable T (Figure 3c for Fe 4 Ag and Figure S9b in Supplementary Materials for Fe 4 ). Quantitative data treatment was based on a generalized Debye model [45,46] to extract the temperature-dependent relaxation time τ and the width of its distribution α. The model was found to be fully satisfactory and afforded α values that approach 0.20 and 0.23 at H = 0 and 1 kOe, respectively, at the lowest T, while they decrease toward zero at high T for both compounds.
By plotting ln(τ) vs. 1/T, linear trends can be clearly recognized at both static fields (Figure 3d for Fe 4 Ag and Figure S9c These thermally activated relaxation parameters fully resemble the ones found for the 3D-MOF B (τ 0 = 1.65(8) 10 −7 s and U eff /k B = 11.46(10) K in zero static field, and τ 0 = 3.21(11) 10 −7 s and U eff /k B = 14.25(8) K at 1 kOe) [14], which are similar but lower compared to Fe 4 clusters not involved in coordination networks [12], and especially with A·2EtOH [11] as building block of B [14]. This seems to confirm the different spin-phonon interactions, which affect the magnetic relaxation [47], between an organized 1D/3D framework and a molecular crystal held together only by van der Waals forces. The values for Fe 4 , similar to the ones of Fe 4 Ag and B and lower to the ones of A·2EtOH [11] and the tetrairon SMM with phenyl rings on the tripodal ligands [12], seems also to suggest that the weak but extended π···π stacking interactions between the phenylpyridine fragments of adjacent molecules in the crystal packing of Fe 4 are sufficient to mimic the spin-phonon effect in a coordination polymer.

General Information
Reagent-grade 2-(4-bromophenyl)ethanol, pyridin-4-ylboronic acid, TICA, TEMPO radical, paraformaldehyde, Pd(PPh3)4, solvents and inorganic reagents for organic synthesis, AgClO4, anhydrous toluene (VWR, max. 20 ppm H2O) and anhydrous THF (VWR, max. 30 ppm H2O) were used as received. Petroleum ether (PE) used was the 40-60 °C boiling fraction. EtOH was distilled over magnesium ethoxide and stored over 3 Å molecular sieves, while Et2O was pre-dried over CaCl2 overnight and distilled from sodium and benzophenone under N2 before use. Elemental analyses were performed with a Carlo Erba EA1110 CHNS-O automatic analyzer. IR spectra were recorded as KBr discs by using a Jasco FTIR-4700 LE spectrophotometer with 2 cm −1 resolution; bands are reported as wavenumbers (cm −1 ) together with the assignment and relative intensity (vs = very strong, s = strong, m = medium, w = weak, br = broad). Thin layer chromatography (TLC) was performed on silica gel and retention factors (r.f.) are reported for the given eluent. GC-MS analyses were performed with an HP GCD G1800C gas-chromatograph with injector and detector temperatures of 250 and 280 °C, respectively; the heating procedure started at 100 °C for 5 min, heated to 240 °C with a rate of 10 °C min −1 , and then kept at 240 °C for 16 min; retention times (r.t.) are given in minutes. 1 H and 13 C NMR spectra at 400 and 100 MHz, respectively, were recorded at room temperature with a Bruker Advance 400 spectrometer with chemical shifts given in parts per million (ppm) versus external TMS. The chemical shifts reported were determined by reference to the solvent residual signal (proton: 7.26, 2.50 and 2.08 ppm and carbon: 77.2, 39.5, and 20.4 ppm for CDCl3, DMSO-d6, and toluene-d8, respectively); coupling constants are given in Hz.  Ag measured at (a) zero applied static field in the 1.9 (red)-5.5 (blue) K temperature range, (b) 1.9 K and variable applied static field in the 0 (red)-50 (blue) kOe range, (c) 1 kOe applied static field in the 1.9 (red)-5.5 (blue) K temperature range; all measurements were performed in the 10-10,000 Hz frequency range, and grey lines are a guide for the eye. (d) Arrhenius plot for Fe 4 Ag both at zero and 1-kOe applied static fields; black lines are given by fitting data with the Arrhenius law.

General Information
Reagent-grade 2-(4-bromophenyl)ethanol, pyridin-4-ylboronic acid, TICA, TEMPO radical, paraformaldehyde, Pd(PPh 3 ) 4 , solvents and inorganic reagents for organic synthesis, AgClO 4 , anhydrous toluene (VWR, max. 20 ppm H 2 O) and anhydrous THF (VWR, max. 30 ppm H 2 O) were used as received. Petroleum ether (PE) used was the 40-60 • C boiling fraction. EtOH was distilled over magnesium ethoxide and stored over 3 Å molecular sieves, while Et 2 O was pre-dried over CaCl 2 overnight and distilled from sodium and benzophenone under N 2 before use. Elemental analyses were performed with a Carlo Erba EA1110 CHNS-O automatic analyzer. IR spectra were recorded as KBr discs by using a Jasco FTIR-4700 LE spectrophotometer with 2 cm −1 resolution; bands are reported as wavenumbers (cm −1 ) together with the assignment and relative intensity (vs = very strong, s = strong, m = medium, w = weak, br = broad). Thin layer chromatography (TLC) was performed on silica gel and retention factors (r.f.) are reported for the given eluent. GC-MS analyses were performed with an HP GCD G1800C gas-chromatograph with injector and detector temperatures of 250 and 280 • C, respectively; the heating procedure started at 100 • C for 5 min, heated to 240 • C with a rate of 10 • C min −1 , and then kept at 240 • C for 16 min; retention times (r.t.) are given in minutes. 1 H and 13 C NMR spectra at 400 and 100 MHz, respectively, were recorded at room temperature with a Bruker Advance 400 spectrometer with chemical shifts given in parts per million (ppm) versus external TMS. The chemical shifts reported were determined by reference to the solvent residual signal (proton: 7.26, 2.50 and 2.08 ppm and carbon: 77.2, 39.5, and 20.4 ppm for CDCl 3 , DMSO-d 6 , and toluene-d 8 , respectively); coupling constants are given in Hz. [Fe 2 (OEt) 2 (dpm) 4 ] [33] and [Fe 4 (OMe) 6 (dpm) 6 ] [12,32] were synthesized as previously reported in the literature. Synthesis of 3 and 4 and attempted synthesis of H 3 PhpPy from 4 are reported in the Supplementary Materials.

Synthesis of 2-(4-(pyridin-4-yl)phenyl)ethanol (1)
This compound was obtained with a modification of a published procedure [28], especially in the final purification step. 2-(4-bromophenyl)ethanol (1.00 g, 4.97 mmol), pyridin-4-ylboronic acid (0.552 g, 4.49 mmol), and Na 2 CO 3 (0.526 g, 4.97 mmol) were introduced in a Schlenk tube under Ar. Then, Pd(PPh 3 ) 4 (0.262 g, 0.23 mmol) was added together with the solvent mixture (DMF:H 2 O 3:1, 24 mL overall). The yellow-green solution was kept under stirring at 110 • C for 24 h, and then it was cooled to room temperature and open to air to decompose any unreacted catalyst. The mixture was transferred in a round-bottom flask with the help of CH 2 Cl 2 (2 × 5 mL). The volatile organic phase was removed under reduced pressure yielding a yellow-green residue still containing DMF, H 2 O, or both. It was then transferred in a separating funnel and extracted with CH 2 Cl 2 (40 mL) and H 2 O (20 mL). The aqueous layer was further extracted with CH 2 Cl 2 (2 × 10 mL), and the organic phases were washed with H 2 O (10 mL). The aqueous phase was extracted with Et 2 O (3 × 10 mL) and all the organic phases were unified and concentrated under reduced pressure, yielding a yellow-orange concentrated solution still containing some DMF. Both TLC (PE:AcOEt 1:4) and GC-MS analysis showed the presence of the desired product as main component of the mixture. This was then diluted with Et 2 O (anhydrous over molecular sieves, 10 mL) and the resulting solution was treated with HCl (1.32 mol L −1 solution in Et 2 O, 5.0 mL, 6.6 mmol), obtaining the efficient precipitation of the product as light yellow hydrochloride salt. After 5 min under stirring the solid was filtered on a Gooch G3 and washed with Et 2 O (anhydrous over molecular sieves, 2 × 5 mL). The solid was then suspended in CH 2 Cl 2 (20 mL) and aqueous NaOH (5% w/w solution, 4.5 mL, 5.62 mmol) was slowly added under stirring. The mixture was transferred in a separating funnel, the aqueous solution was extracted with further CH 2 Cl 2 (5 mL) and the organic phases were unified, dried over MgSO 4 and taken to dryness under reduced pressure. The dense oil was dissolved in CH 2 Cl 2 (4 mL) and Et 2 O (15 mL) yielding a turbid yellow solution. This was refluxed for 5 min obtaining the precipitation of a brown solid, which was filtered off and discharged. The solution was then concentrated under reduced pressure till formation of a solid, then n-hexane (20 mL) was added and the mixture again concentrated under reduced pressure and cooled with an ice bath in order to favor the precipitation. The title compound as light yellow solid was filtered on a Gooch G3, washed with n-hexane (5 mL) and dried under vacuum

Synthesis of 2-(4-(Pyridin-4-yl)phenyl)acetaldehyde (2)
This compound was obtained with a modification of a published procedure [29]. 1 (0.50 g, 2.51 mmol) was dissolved in CH 2 Cl 2 (13 mL, anhydrous over molecular sieves) and the resulting solution was cooled with an ice bath. After 10 min, TICA (0.50 g, 2.15 mmol) was added yielding a white suspension, which was further stirred for 20 min at low temperature. TEMPO (0.073 mol L −1 solution in CH 2 Cl 2 , 0.30 mL, 0.022 mmol) was added, the reaction mixture left at 0 • C for 15 min, and then diluted with CH 2 Cl 2 (7 mL). Aqueous Na 2 CO 3 (1.5 mL of a saturated solution) was added observing a color change from orange to yellow and formation of a white solid. The liquid phase was separated, the solid residue was washed with CH 2 Cl 2 (2 × 5 mL) and the organic phases taken to dryness under reduced pressure. The orange residue was dissolved in CH 2 Cl 2 (8 mL) and Et 2 O (10 mL), heated to reflux for 5 min, and the yellow liquid phase was filtered on a Gooch G3 to eliminate the solid and taken to dryness, yielding the title compound as yellow solid (0.47 g, 95%). TLC (PE:AcOEt 1:4) r.f.

X-ray Data Collection and Structure Determination
Crystals of Fe 4 suitable for X-ray diffraction experiments were selected from the mother liquor in EtOH/Et 2 O. Intensity data were collected on a Bruker Apex CCD area detector using graphite monochromatic Mo-Kα radiation. Crystal structure data for Fe 4  During data collection, no crystal decay was observed, so that no time-decay correction was needed. Data reductions were performed with SAINT, and absorption corrections based on multiscan were obtained with SADABS [48]. The structure was solved by direct methods and refined with SHELXL-2016/6 [49] implemented in WinGX-Version 2014.1 system [50]. The program ORTEPIII was used for graphics [51]. Three over six tBu groups were refined for their disorder by splitting them into two positions, A and B, with complementary occupancy factors of 0.624 (15) and 0.376 (15); restraint ISOR (sigma = 0.1) was also applied to the disordered terminal carbon atoms to ensure reasonable thermal ellipsoids. Anisotropic thermal parameters were used for all non-H atoms. The isotropic thermal parameters of H atoms were fixed at 1.2 (1.5 for methyl groups) times those of the atom to which they were attached. All H atoms were placed in calculated positions and refined by a riding model. CCDC 1842624 contains the supplementary crystallographic data for Fe 4 . These data can be obtained free of charge via http://www.ccdc.cam.ac.uk/conts/retrieving.html, or from the Cambridge Crystallographic Data Centre, 12 Union Road, Cambridge CB2 1EZ, UK; fax: (+44) 1223-336-033, or e-mail: deposit@ccdc.cam.ac.uk.

Magnetic Measurements
Dc magnetic data were recorded using a Quantum Design MPMS SQUID magnetometer. Magnetic susceptibilities were measured on 7.54 and 8.10 mg powder samples of Fe 4 and Fe 4 Ag, respectively, packed in a Teflon pellet with applied fields of 1 kOe from 1.9 (Fe 4 Ag) or 5 (Fe 4 ) to 40 K and of 10 kOe from 40 to 300 (Fe 4 Ag) or 250 (Fe 4 ) K. Data reduction was carried out by using 1835.54 and 2042.86 as molecular weights and −1054.9 and −1114.9 10 −6 emu mol −1 as diamagnetic contributions, estimated from the Pascal's constants [52], for Fe 4 and Fe 4 Ag, respectively. Data fitting was performed by using program PHI v3.1.1 [44]. Ac susceptibility was recorded on a Quantum Design PPMS susceptometer for Fe 4 Ag at zero static applied field and for both Fe 4 and Fe 4 Ag samples at 1 kOe static applied field, while the previously reported Quantum Design MPMS SQUID was employed for ac measurements on Fe 4 at zero static applied field.

Conclusions
As a promising strategy in developing the performance of SMMs by their assembly into coordination networks, in this paper we have described the behavior of a tetrairon(III) cluster, Fe 4 , as longer spacer (23.4 Å) when coordinating to silver(I) ions, in comparison with what previously observed with a shorter spacer (14.8 Å) of the same family, A [14]. As crucial point when engineering novel compounds with desired properties [53], the chemical design of the suitable tripodal ligand and its synthesis revealed fundamental for this project. 1D polymer chains were achieved in Fe 4 Ag with Fe 4 :AgClO 4 1:1 ratio, in contrast with the aimed 3D arrangement similar to B, which contains an A:AgClO 4 2:1 ratio [14]. The enormous cavities that would have formed with a 3D-MOF structure given by an Fe 4 :AgClO 4 2:1 ratio might probably suffer of high instability in the absence of any supplementary big molecules acting as sustain within the pores, and this point will be pursued in our future studies.
Despite the synthesized coordination polymer Fe 4 Ag does not possess the desired 3D structure, it still revealed invariant relaxation dynamics compared to B, in which Fe 4 SMMs behave as building blocks of polymeric structures. Further work to address 1D vs. 3D issue will be perceived also by testing other metal ions or coordination units like ruthenium dimers, which hold minor flexibility in their coordination behavior with respect to silver(I). In fact, this ion can assume different coordination environments only by subtle changes in the surrounding environment or the donor power of the coordinated atoms [43,54,55], as we could clearly observe here by exchanging H 3 pPy with H 3 PhpPy and that highlights how small molecular modifications can direct the fate of the final supramolecular assembly.