María José Heras
Ojea
a,
Giulia
Lorusso
b,
Gavin A.
Craig‡
a,
Claire
Wilson
a,
Marco
Evangelisti
*b and
Mark
Murrie
*a
aWestCHEM, School of Chemistry, University of Glasgow, University Avenue, Glasgow, G12 8QQ, UK. E-mail: mark.murrie@glasgow.ac.uk
bInstituto de Ciencia de Materiales de Aragón (ICMA), CSIC and Universidad de Zaragoza, 50009 Zaragoza, Spain. E-mail: evange@unizar.es
First published on 13th April 2017
The adiabatic temperature change of the star-shaped {CoIII3GdIII3} magnetocaloric ring is enhanced via topological control over the assembly process, by using a pre-formed {CoII(H6L)} building block that undergoes oxidation to CoIII, successfully separating the GdIII ions.
We recently reported large heterometallic {Mn18Cu6} complexes, obtained by following a directed synthesis approach, based on the use of the metallo–organic precursor [Cu(H6L)Cl]Cl (H6L = bis–tris propane).6 This prompted us to investigate the reactivity of bis–tris propane with 4f ions in the presence of 3d ions.7 Similar aminopolyol-type ligands seem to promote the oxidation of different Co(II) starting materials.8 Therefore, the exploration of Co(II) precursors containing H6L in the design of new magnetic refrigerants becomes highly attractive. Our approach is to use {CoII(H6L)} precursors that can undergo facile oxidation to diamagnetic Co(III), whilst encapsulating the cobalt centres and directing/separating the Ln(III) ions. Using this strategy, we present the magnetocaloric properties of a new {CoIII3GdIII3} star-shaped ring, showing that the Co(III) ions have a significant impact on the adiabatic temperature change in this system, by separating the Gd(III) ions and weakening the Gd(III)⋯Gd(III) interaction. In terms of ΔTad, this complex is among the best gadolinium-based molecular refrigerants reported so far (vide infra).
By combining the metalloligand [Co(H6L)(CH3COO)2] (1) and Gd(acac)3·H2O we are able to obtain a new hexametallic complex [CoIII3GdIII3(H2L)3(acac)2(CH3COO)4(H2O)2] (2) with a unique alternating wheel-like structure (Fig. 1 and Fig. S4, S12, Table S1, ESI†). The pre-formation of the metalloligand (see Fig. S1–S3 and S10, ESI†) seems to be essential for the assembly of 2, as previously seen for the {Mn18Cu6} complexes. During the reaction, under aerobic conditions in the presence of bis–tris propane, Co(II) is oxidised to Co(III) and hence, the magnetic properties of 2 are defined exclusively by the paramagnetic Gd(III) ions. The structure of 2 contains three octahedral Co(III) ions, each one encapsulated by one tetra-deprotonated H2L4− ligand through four O and two N atoms. The two remaining ligand arms are uncoordinated. Each {Co(H2L)} unit is linked to two octa-coordinated Gd(III) ions through four μ-O bridging atoms forming a ring-like structure, in which Co(III) and Gd(III) centres alternately occupy the corners of a six-pointed star.
Fig. 1 Synthetic approach and structure of 2. C, grey; Co, fuchsia; Gd, green; N, blue; O, red; hydrogen atoms and solvent molecules are omitted for clarity. |
The oxidation state of each cobalt centre has been confirmed by bond valence sum calculations (BVS).9 Two different triangular dodecahedral Gd ions can be distinguished based on the co-ligands that complete their coordination sphere: two bidentate acetates for Gd1, or one acac−, one monodentate acetate and one water ligand for Gd2 and Gd2′ (see Table S2, ESI†). The average intramolecular Gd⋯M distances (M = Co, Gd) are d(Gd⋯Co) = 3.389(1) Å, and d(Gd⋯Gd′) = 5.802(9) Å. Note that the {CoIII3GdIII3} wheel is not perfectly planar (see Fig. S5, ESI†), with a dihedral angle between the {Co3} and {Gd3} planes of 9°. To the best of our knowledge, the structure shown by {CoIII3GdIII3} is unprecedented for Co/4f complexes (CSD 5.37, November 2016, where 3d = Sc → Zn, 4f = La → Lu) and remarkably only larger alternating 3d/4f rings (where 3d = Mn or Fe only), such as {Mn4Ln4}, {Mn8Ln8} or {Fe10Yb10} have been reported to date, where the structures are all more puckered than in 2.10
The experimental value of the static magnetic susceptibility temperature product χMT at 290 K (23.37 cm3 mol−1 K) for 2 is consistent with that expected for three uncoupled Gd(III) centres (23.63 cm3 mol−1 K, 8S7/2, s = 7/2, g = 2). χMT displays an almost imperceptible decrease between 290 and 26 K and then drops to 21.33 cm3 mol−1 K at 2 K (Fig. 2), consistent with very weak spin ordering promoted either by antiferromagnetic correlations or zero-field splitting (ZFS) of the ground multiplet. Exchange coupling through the diamagnetic Co(III) ions is a potential pathway for very weak Gd⋯Gd′ intramolecular interactions, similar to previous Cu⋯Cu′ coupling through Zn(II) in heterometallic Cu/Zn bis–tris propane complexes.11 Within the exchange coupling model, the simultaneous fit of the susceptibility and magnetisation data at T = 2 to 10 K, step 1 K (see spin Hamiltonian below, and magnetic model in Fig. S11, ESI†) gives , and g = 1.988 ± 0.002.12
(1) |
In the mean-field approximation, we can interpret as an effective interaction constant resulting from dipole–dipole magnetic interactions, which are well documented for molecular materials.13 The g value obtained from the fit is reasonable for Gd(III) ions considering similar complexes.14 The small deviation from the expected spin-only g value could be a consequence of a small ZFS, induced by crystal-field effects combined with spin–orbit coupling.2b,15 Considering the relatively large average Gd⋯Gd′ distance (5.802(9) Å), one could argue that exchange coupling is likely to be less effective for spin ordering. All sources, though, are extremely weak and we cannot discriminate between them, on the basis of the experimental data. The DC experiments suggest, therefore, that 2 should display a relatively large MCE, arising from the weakly-interacting Gd(III) ions.1b,16 The MCE is best evaluated from heat capacity cp experiments (Fig. 3, top). As is typical for molecular magnetic materials,1a lattice vibrations contribute predominantly to cp as a rapid increase above ca. 5–10 K. The non-magnetic lattice contribution can be described by the Debye model (dotted line in Fig. 3), which simplifies to a clatt/R = aT3 dependence at the lowest temperatures, where a = 6.7 × 10−3 K−3 for 2. At such low temperatures, cp is mainly determined by a Schottky-like anomaly, which is strongly dependent on B and can be well modelled by Hamiltonian (1), using the same parameters obtained from fitting the susceptibility and magnetisation data. For B = 0, the system becomes sensitive to any perturbation, hence the need to add an effective internal field Beff ≈ 0.3 T to our model, in order to simulate the zero-applied-field cp. For B ≥ 1 T, such a correction is not necessary. We ascribe Beff to the dipole–dipole magnetic interactions, although to a minor extent, it could also be associated with a small ZFS at the Gd(III) sites. From the experimental heat capacity data, we derive the entropy S of the system, according to . Similarly, we derive the lattice entropy Slatt from clatt and the magnetic entropy Sm from cm, that is, the magnetic contribution to cp, viz., the aforementioned Schottky-like anomaly, calculated on basis of Hamiltonian (1). The bottom panel of Fig. 3 shows the resulting temperature dependence of the experimental entropy S, which thus corresponds to the sum of Slatt (dotted line) and Sm (solid lines), for any applied magnetic field employed. Because of the very weak ZFS and interactions present, temperatures as low as ca. 3–4 K are already sufficient for fully decoupling the Gd(III) spins. Therefore, within the investigated temperature range, the zero-applied-field S reaches the maximum entropy value per mole involved, which corresponds to three non-interacting Gd(III) spins s = 7/2 and is calculated as 3 × Rln(2s + 1) ≈ 6.2R (see Fig. 3).
Next, we evaluate ΔSm and ΔTad, following a change of the applied magnetic field ΔB. The temperature dependencies of both ΔSm and ΔTad can be calculated straightforwardly from the experimental entropy (Fig. 3, bottom panel).1b Likewise, ΔSm can also be calculated from the magnetisation data (Fig. 2) by making use of the Maxwell relation . Fig. 4 shows the so-obtained dependence of ΔSm and ΔTad for 2versus temperature for selected ΔB values. Note the nice agreement between the ΔSm results obtained through both methods, thus validating the approaches employed. For the largest applied field change ΔB = (7–0) T, i.e., after a full demagnetisation from 7 T, the maximum value of −ΔSm is 5.6R at T = 1.9 K, which corresponds to 90% of the available entropy content and is equivalent to 23.6 J kg−1 K−1 per unit mass. Concomitantly, we obtain ΔTad = 10.7 K at T = 1.5 K for the same field change, that is, the temperature decreases down to a final temperature Tf = 1.5 K, on demagnetising adiabatically from B = 7 T and an initial temperature Ti = Tf + ΔTad = 12.2 K. Note that ΔTad is limited in Tf by sources of magnetic ordering (spin–spin interactions and magnetic anisotropies) and in Ti by the lattice entropy, which soon becomes the dominating contribution on increasing the temperature (see Fig. 3).
The precursor [Co(H6L)(CH3COO)2] (1) has successfully directed the molecular assembly in [CoIII3GdIII3(H2L)3(acac)2(CH3COO)4(H2O)2] (2), so that the Gd(III) ions are isolated, thus promoting the discussed magnetocaloric properties. Several Co(II)/Gd(III) molecular coolers have been studied so far.5a,17 The biggest hindrance to a large MCE in those compounds is the large magnetic anisotropy of the Co(II) ions.1b,2 In 2 we circumvent this impediment by oxidising Co(II) into Co(III) during the synthesis, thus leaving only the Gd(III) ions to contribute magnetically. In terms of the magnetocaloric properties of 2, the diamagnetic Co(III) ions still play a role, though passively. On the one hand, they influence negatively on the entropy change per unit mass. The lower the magnetic:non-magnetic ions ratio, the lower are the magnetic heat capacity and entropy per unit mass. The maximum value observed of −ΔSm = 23.6 J kg−1 K−1 at T = 1.9 K for 2 is large, though not outstanding. However, the key point is that the Co(III) centres impact positively on the adiabatic temperature change. In 2, the Co(III) and Gd(III) ions alternate with respect to one another. Therefore, the intermediate presence of the Co(III) ions weakens extremely the magnetic interaction between the Gd(III) ions, so the temperature-dependence of ΔTad has a maximum at a relatively lower temperature than usually found for pure-Gd molecular complexes. Among the few known systems that have a ΔTad maximum below e.g., 2 K for 7 T, complex 2 with ΔTad = 10.7 K at T = 1.5 K lags behind only the extraordinary {Gd2-ac} with ΔTad = 12.6 K at T = 1.4 K,3a while it outdoes {Zn2Gd2} with ΔTad = 9.6 K at T = 1.4 K,18 and {Gd7} with ΔTad = 9.4 K at T = 1.8 K.19
The UK Engineering and Physical Sciences Research Council are thanked for financial support (grant ref. EP/I027203/1). GL and ME thank the Spanish Ministry of Economy, Industry and Competitiveness for funding (MAT2015-68204-R) and a postdoctoral contract (to GL). The data which underpin this work are available at http://dx.doi.org/10.5525/gla.researchdata.223.
Footnotes |
† Electronic supplementary information (ESI) available: Experimental sections, spectroscopic studies, magnetic studies, crystallographic details. CCDC 1533722. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c7cc02243c |
‡ Current address: Institute for Integrated Cell-Material Science (WPI-iCeMS), Kyoto University, Yoshida, Sakyo-ku, Kyoto 606-8501, Japan. |
This journal is © The Royal Society of Chemistry 2017 |