Synthesis, structure, magnetic and magnetocaloric properties of a series of {CrIII4LnIII} complexes

The hydrothermal reaction of CrCl3 6H2O and Ln(NO3)3 xH2O in the presence of di(2-pyridil)ketone, and sodium acetate in water leads to the formation of a series of [Cr4Ln(CH3COO)(pyCOO)((py)2COO)4] (NO3)2 nH2O complexes (with Ln = Gd(III), Tb(III), Dy(III), and n = 19 or 20). The four newly synthesized complexes were characterized by single crystal X-ray diffraction technique and their magnetic properties were fully investigated by dc (direct current) and ac (alternating current) magnetic susceptibility measurements. Single crystal X-ray studies revealed that the dysprosium derivative crystallizes as two different structural forms: a triclinic P% 1 form and a monoclinic P21/n form. The gadolinium and the terbium derivatives were isostructural to the dysprosium monoclinic form. Surprisingly, full static and dynamic magnetic studies revealed that none of the synthesized complexes exhibit single molecule magnet behaviour, while the gadolinium derivative displays a magnetocaloric effect (MCE).


Introduction
Heterometallic 3d-4f coordination complexes have gained substantial interest over the last decade as potential candidates for the design of multifunctional materials, with emphasis on their peculiar magnetic behaviour, such as single molecule magnet, SMM, behaviour 1 and molecular refrigerants. 2 For molecular refrigerant properties, the use of weakly exchange coupled 3d-4f systems, based on isotropic ion such as Gd(III) 3 and featuring small molecular weight ligands has been proposed. Magnetic refrigeration is based on the magnetocaloric effect (MCE), an isothermal magnetization process, leading to a decrease of magnetic entropy of the material, followed by an adiabatic demagnetization, resulting in a temperature decrease. 4 Recent studies demonstrated that molecular clusters show a magnetocaloric effect much larger than that found in the best intermetallic and lanthanide alloys at low temperatures, and magnetic nanoparticles employed commercially, due to the increase of magnetic entropy at low temperature when these materials are subjected to an adiabatic demagnetization. In the last few years, it has been demonstrated that useful chemical approach consists in combining the essentially isotropic Gd(III) ion 3 (4f 7 ) with transition metal ions, such as Mn(III), 5 Co(II), 6 Ni(II), 7 Zn(II), 8 Cu(II), 9 as well as the isotropic Cr(III) ion, 2a or the use of high-spin ions such as Fe(III) and Mn(II). 2b, 10 Particularly, Cr(III)-Ln(III) complexes, highlighted by a series of papers by Pedersen and co-workers, 2b,11,17b,c have recently demonstrated promising magnetocaloric effect through the use of fluoride bridged-clusters, leading to weak exchange interactions and showing relevant magnetic entropy changes.
For the study of single molecule magnets, the use of exchangecoupled systems, 12 has been proposed as a viable method to reduce the very efficient quantum tunnelling (QT) processes in zero field often observed in mononuclear 4f systems. This may hamper the observation of magnetic bistability despite very large anisotropy barriers often reported for Tb(III) and Dy(III) derivatives. 13,14 Despite encouraging results 15,16 there are however two main issues to be solved in the use of 3d-4f systems for efficient SMMs: (i) the relatively weak exchange coupling provided by lanthanide ions, due to the inner nature of the 4f orbitals compared to the 3d ones; (ii) the relative orientation of the magnetic anisotropy of the 3d and 4f metal ions must be controlled to obtain large global anisotropy.
In our group, we are interested, in particular, in complexes where the core is comprised of lanthanides and chromium: the number of Cr(III)-Ln(III) coordination complexes is indeed relatively small, rendering this heterometallic combination a challenging area of chemistry. To the best of our knowledge, the design of new molecular Cr(III)-Ln(III) materials is based on four main chemical approaches. The first one involves the use of {Cr-F-Ln} fluoride bridges, 17 while the second one is built around cyano-bridged ligands for {Cr-CN-Ln} derivatives. 18 The third and the fourth ones incorporate oxo-bridged and oxalato-bridged ligands respectively for {Cr-O-Ln} 19,20 and {Cr-OOCCOO-Ln} derivatives. 21 While promising results have been  obtained for fluoride bridged-ligands as well as on cyano-bridged  ligands, only very few example of Cr(III)-Ln(III) complexes  The new series of complexes has been investigated by single crystal X-ray diffraction techniques, revealing that the Dy(III) derivative crystallizes into two structural forms, triclinic P% 1 and monoclinic P2 1 /n. The magnetic characterization showed that none of the complexes reported in the present manuscript exhibit SMM behaviour, while MCE of reasonable magnitude is observed for the Gd(III) complex.

Syntheses
All chemicals were commercially purchased from Sigma-Aldrich, and used without further purification.
Cr 4 Dy (1a) and (1b). CrCl 3 Á6H 2 O (26.6 mg, 0.10 mmol), Dy(NO 3 ) 3 Á6H 2 O (137 mg, 0.30 mmol), (py) 2 CO (110.5 mg, 0.60 mmol) and CH 3 COONaÁ3H 2 O (81.6 mg, 0.60 mmol) were placed in a 23 mL Teflon-lined steel autoclave stirred all together in 5 mL of distilled water. After 10 min, the autoclave was closed and heated at 140 1C for 2 days, followed by 12 hours of cooling to room temperature. The solution was then filtrated to remove all impurities. The resulting clear solution was kept undisturbed at room temperature. Crystals suitable for X-ray studies were collected after several days. Compounds 1a and 1b crystallize simultaneously as a triclinic form and as a monoclinic form. Crystals were manually selected and collected under a stereo microscope exclusively for magnetic measurements and ICP analysis. Quantity of collected crystals for 1a, m = 7. 36

X-ray crystallography
Single crystal X-ray diffraction data were collected at 183(1) K on an Oxford Instruments Cryojet XL cooler and on an Agilent Technologies Xcalibur Ruby area detector diffractometer using a single wavelength enhance X-ray source with Mo K a radiation (l = 0.71073 Å) 25 from a micro-focus X-ray source for complexes 1a and 1b, and on a Agilent Technologies SuperNova Atlas area detector diffractometer using a single wavelength enhance X-ray source with Cu K a radiation (l = 1.54184 Å), from a microfocus X-ray source, for 2 and 3.
The selected suitable single crystals were mounted using polybutene oil on a flexible loop fixed on a goniometer head and immediately transferred to the diffractometer. Pre-experiment, data collection, data reduction and analytical absorption correction 26 were performed with the program suite CrysAlisPro. 27 Using Olex2 28 the structures were solved by the SHELXS97 29 (1a, 1b, 2) or SHELXT 30 (3) and then refined with the SHELXL 31 program package by full-matrix least-squares minimization on F 2 . Platon 32 was used to check the results of the X-ray analyses. For more details about the data collections and refinements parameters, see the crystallographic information files and the ESI. † Crystal data and refinement parameters of 1a, 1b, 2 and 3 are given in Table 1.

Physical measurements
Manually collected crystals were analysed by AES-ICP method. The direct current (dc) magnetic susceptibility measurements as well as the field dependence of the magnetization (M vs. H) were carried out on a Quantum Design MPMS SQUID magnetometer on microcrystalline samples pressed in a pellet to avoid field induced orientation of the crystallites. The ac magnetic susceptibility measurements were carried out on a Quantum Design PPMS in ac mode at both zero and applied external dc field in the presence of 5 Oe oscillating magnetic field. Isothermal magnetization curves for 3 were performed using the VSM option of the Quantum Design PPMS. Raw data were reduced to paramagnetic molar susceptibilities using molecular formula obtained by X-ray data, including all water molecules. This also allowed to evaluate intrinsic diamagnetism of the sample by use of corresponding Pascal's constants.

Results and discussion
Syntheses and crystal growth All compounds were obtained by a fast hydrothermal treatment of a mixture containing all starting materials and followed by the slow evaporation of the filtrates. Interestingly, the hydrothermal treatment at 140 1C during 2 days and 12 hours of cooling to room temperature lead to the formation of the gem-diol dianion form, (py) 2 COO 2 À , by reduction of the di(2-pyridil)ketone ligand 33 and to the decomposition of this ligand in a simplified ion form (py)COO À . Suitable crystals for single crystal X-ray diffraction analysis were collected after few weeks, by slow evaporation of the solutions obtained from the filtration of the reaction mixture after 2 day of heating and 12 hours of cooling to room temperature. The dysprosium derivative crystallizes simultaneously as two structural forms that are the triclinic and the monoclinic forms respectively for 1a and 1b. The two forms were easily separated manually under a stereomicroscope, due to their different shape forms. As for compounds 1a and 1b, compounds 2 and 3 were also collected manually, due to the low quantity of crystals present in solution, and in order to get samples in a pure form.
Compounds 1a and 1b crystallize simultaneously as two structural forms, respectively in the triclinic P% 1 and monoclinic P2 1 /n space groups and differ only by their water molecule content.
The Dy(III) ion is in a 8-fold coordination geometry of very low symmetry (see Fig. 2, left). Indeed, analysis of the coordination polyhedron via SHAPE software 34 did not provide any reasonable agreement with regular 8-vertex polyhedral (see ESI, † Table S1). The distortion with respect to the square-antiprismatic geometry is highlighted in Fig. 2. The eight oxygen atoms belong to one chelating acetate ligand on the top and four gem-diol ligands to which two are chelating the lanthanide (see Fig. 2, right). DyÁ Á ÁO bond lengths range from 2.365(2) to 2.431(2) Å, for the oxygen atoms of the gem-diol ligands, and 2.348(2) Å and 2.396(2) Å for the oxygen atoms of the acetate ligand.
The diamond-shape type structure is formed by the heterometallic core composed by the Dy(III) and the four Cr(III) ions. The structural study shows (see Fig. 4) that two opposite  Nonetheless, complex 1a crystallizing in the P% 1 triclinic space group, shows a slightly more extended pentanuclear core, compared to complexes 1b, 2 and 3. Indeed, the Cr1Dy1Cr3 angle of 110.041 in 1a is slightly bigger than the same angles in 1b, 2 and 3, with values of 107.201, 106.991 (for Cr1Tb1Cr3) and 106.971 respectively (for Cr2Gd1Cr3, see Fig. 6).

Magnetic characterization
Results of dc magnetic characterization are reported in Fig. 5 in the form of wT vs. T plot for 1a, 2 and 3, (complex 1b providing superimposable curves to those of 1a). At room temperature, the observed wT values of 21.6 and 19.21 cm 3 K mol À1 for Dy(III) and Tb(III) derivatives, respectively, are in good agreement with 21.65 and 19.3 cm 3 K mol À1 , expected for uncorrelated spins of the aforementioned lanthanide ions (Dy(III): J = 15/2, g J = 4/3; Tb(III): J = 6, g J = 3/2) and four Cr(III) (S Cr = 3/2, g = 1.98) ions. As expected for lanthanide containing complexes, the wT product slowly decreases on lowering temperature from 300 K to 50 K to thermal depopulation of the excited states of the ground multiplets of the lanthanide(III) ions, split by ligand field effects. The more rapid decrease observed below 50 K may be attributed to the effect of both intramolecular exchange coupling, either    Ln-Cr or Cr-Cr one, and crystal field effects leading to a minimum value of 16.57 and 12.7 cm 3 K mol À1 for Dy(III) and Tb(III) at 1.9 K, respectively. Due to the largely anisotropic nature of both Dy(III) and Tb(III) magnetic moments, a modelling of the data using the standard HDvV Hamiltonian was not possible. However, the isothermal M vs. H curves are clearly far from saturation even at 5 T, indicating either large anisotropy or the presence of low-lying excited states close to ground one. Since ac susceptibility measurements as a function of frequency at zero and applied external magnetic field (2 KOe) and variable temperature did not show any out-of-phase signal for either 1 or 2, indicating very fast relaxation process and a lack of SMM behaviour, the latter explanation appears more reasonable. A more detailed analysis of the magnetic behaviour was on the contrary, possible for 3, which contains the orbitally non degenerated Gd(III) ion with S = 7/2. For this complex, the magnetic data shows a room temperature value consistent with independent spins (expected spin-only value for g = 2.00: 15.38 cm 3 K mol À1 ; measured: 15.3 cm 3 K mol À1 ). The wT product starts to increase on lowering temperature witnessing intramolecular ferromagnetic exchange coupling interactions. However, the maximum value of 17.5 cm 3 K mol À1 reached at 5 K, is much lower than the one of 49.88 cm 3 K mol À1 expected for a full ferromagnetic state (S = 19/2), indicating that antiferromagnetic interactions are also active. Accordingly the value decrease to 15.2 cm 3 K mol À1 on further lowering temperature to 1.9 K. This qualitative picture is confirmed by the isothermal field dependent curves (see Fig. 5, inset), which are still far from saturation at 5 T, showing magnetization value which are below the maximum expected value of 19 m B .
Following these considerations and based on the molecular structure derived by X-rays, we attempted a fit of wT data using the following isotropic Hamiltonian: 35 where S are the spin operator, Js are the isotropic exchange coupling constants. To avoid over parametrization, we considered a single, averaged J Gd-Cr interaction value, despite the quite different geometrical parameters characterizing the exchange coupling paths of Gd(III) ion with Cr1 and Cr4, compared to that with Cr2 and Cr3. Best fit of the wT vs. T data was obtained 36 fixing g Cr = 1.98 and g Gd = 1.99 and provided J Gd-Cr = 0.25 AE 0.05 cm À1 and J Cr-Cr = À0.55 AE 0.05 cm À1 . Confirmation of the occurrence of ferromagnetic Gd-Cr interaction and of antiferromagnetic Cr-Cr ones was obtained by the best fit of the isothermal magnetization curves, which yielded a closely related set of parameters ( J Gd-Cr = 0.17 AE 0.04 cm À1 and J Cr-Cr = À0.35 AE 0.05 cm À1 ), the small discrepancy being attributable to the neglecting of zero field splitting contributions and to the simplified model of exchange pattern we adopted. We note that no magnetostructural correlation for the Gd(III)-Cr(III) exchange coupling with similar complexes is now possible, since to the best of our knowledge this is the first example of carboxylate-bridged Gd-Cr for which exchange coupling values, even if averaged, are reported. 17a Nonetheless, if we refer to fluoride bridged Gd(III)-Cr(III) complexes, for which a large deal of work has been performed in past years, 2,11,37 antiferromagnetic coupling is expected for large bridging angles (41401) whereas at smaller angles ferromagnetic coupling should occur. 17c This is in reasonable qualitative agreement with the results here reported, considering that all the Gd-O-Cr angles in  our structure are smaller than 1321, with considerably smaller angles bridges to Cr2 and Cr3 (103.86(11)1-106.87(12)1), see Fig. 6 (and ESI, † Table S4). The ground state resulting from best fit parameters is a S = 9/2. However, the small value of the coupling and the presence of competing ferromagnetic and antiferromagnetic interactions, yield a large number of very low lying excited states, all the 2048 levels lying within an energy range of 40 cm À1 . This suggests us that this system might be appealing for the investigation of potential MCE effect, which is favoured by this condition.

Investigation of magnetocaloric effect
In order to explore the possibility that the changes in the magnetocaloric effect in complex 3 derivative is large enough to be considered for application purposes, the isothermal magnetic entropy changes DS m were derived for several applied magnetic field changes DH = H f À H i . From the isothermal magnetization data at different temperatures the DS m values were obtained using the Maxwell equation: and are shown in Fig. 7. As expected an increase in DH corresponds to an increase in ÀDS m reaching a maximum value of 18.31 J K À1 kg À1 at 3.5 K, at the highest experimentally achievable DH = 8 T.
This value lies comfortably within the range reported in the literature for gadolinium based systems, 38 confirming the potentialities of this approach for magnetorefrigeration.
It is worth noting that the maximum observed entropy change is about two thirds of the full available magnetic entropy due to four Cr(III) and one Gd(III): R ln(2S Gd + 1) + R ln(2S Cr + 1) = 28.41 J K À1 kg À1 , while it is much higher than that pertaining to the ground state only (S = 9/2, DS m calc = 8.71 J K À1 kg À1 ), as well as of the one calculated assuming a pure ferromagnetic state (S = 19/2, DS m calc = 11.16 J K À1 kg À1 ). In this respect, the existence of competing ferro-and antiferromagnetic interactions of comparable energy scales, resulting in a non-isolated ground spin state, is confirmed to be an essential ingredient for observing MCE of reasonable magnitudes.

Conclusion
In conclusion, three novel pentanuclear Ln-Cr(III) derivatives (Ln = Dy(III), Tb(III) and Gd(III)) have been synthesized, and were structurally and magnetically characterized. Crystal structure studies revealed that the dysprosium derivative crystallizes simultaneously as two structural forms, respectively in the triclinic P% 1 and monoclinic P2 1 /n space groups. Terbium and gadolinium derivatives are isostructural to the monoclinic dysprosium complex. The magnetic characterization of the three novel Cr(III)-Ln complexes (1a, 2 and 3) revealed that while no ac susceptibility signal could be observed for the three derivatives, the gadolinium one showed interesting magnetocaloric properties. In this complex, the weak coupling between the ions in the complex have lead to the population of multiple paramagnetic states at low temperature, thus leading to a magnetocaloric effect magnitude somewhat in-between the two maxima of the calculated entropy. New experiments on the hydrothermal synthesis approach are currently carried out to improve the yield of all reactions, in order to plan further investigations on the luminescent properties of the Cr(III)-Ln(III) combination. Moreover, additional experimental procedures are also currently performed to replace the Cr(III) ion by others transition metals, in order to study the influence of the metal ions on the magnetic behaviour.