Cryogenic Magnetocaloric Effect in a Ferromagnetic Molecular Dimer

Zuschriften DOI: 10.1002/ange.201102640 Magnetocaloric Effect Cryogenic Magnetocaloric Effect in a Ferromagnetic Molecular Dimer** Marco Evangelisti,* Olivier Roubeau, Elias Palacios, Agustn Camn, Thomas N. Hooper, Euan K. Brechin, and Juan J. Alonso Over the last few years, great interest has emerged in the synthesis and magnetothermal studies of molecular clusters based on paramagnetic ions, often referred to as molecular nanomagnets, in view of their potential application as lowtemperature magnetic refrigerants.[1, 2] What makes them promising is that their cryogenic magnetocaloric effect (MCE) can be considerably larger than that of any other magnetic refrigerant, for example, lanthanide alloys and magnetic nanoparticles.[3] The MCE is the change of magnetic entropy (DSm) and related adiabatic temperature (DTad) in response to the change of applied magnetic field, and it can be exploited for cooling applications via a field-removal process called adiabatic demagnetization. Although the MCE is intrinsic to any magnetic material, in only a few cases are the changes sufficiently large to make them suitable for applications. The ideal molecular refrigerant comprises the following key characteristics:[1] 1) a large spin ground state S, since the magnetic entropy amounts to R ln(2S+1); 2) a negligible magnetic anisotropy, which permits easy polarization of the net molecular spins in magnetic fields of weak or moderate strength; 3) the presence of low-lying excited spin states, which enhances the field dependence of the MCE owing to the increased number of populated spin states; 4) dominant ferromagnetic exchange,[3c] favoring a large S and hence a large field dependence of the MCE; 5) a relatively low molecular mass (or a large metal/ligand mass ratio), since the nonmagnetic ligands contribute passively to [*] Dr. M. Evangelisti, Dr. O. Roubeau, Dr. E. Palacios, Dr. A. Camn Instituto de Ciencia de Materiales de Aragn (ICMA) CSIC-Universidad de Zaragoza Departamento de Fsica de la Materia Condensada 50009 Zaragoza (Spain) Fax: (+ 34) 976-761-229 E-mail: evange@unizar.es Homepage: http://molchip.unizar.es/ T. N. Hooper, Dr. E. K. Brechin EaStCHEM School of Chemistry, The University of Edinburgh West Mains Road, Edinburgh, EH9 3JJ, Scotland (UK) Dr. J. J. Alonso Departamento de Fsica Aplicada I, Universidad de Mlaga 29071 Mlaga (Spain) [**] This work was supported by the Spanish MICINN through grants MAT2009-13977-C03, MAT2007-61621 and CSD2007-00010. E.K.B. thanks the EPSRC. Supporting information for this article, including experimental methods, the structure of 1 and views of the hydrogen-bonding networks, Monte Carlo numerical simulations of the magnetic ordering, and representative direct measurement of the temperature evolution of 1 under quasi-adiabatic conditions, is available on the WWW under http://dx.doi.org/10.1002/anie.201102640. 6736 the MCE. Although this last point is crucial for obtaining an enhanced effect, it has been mostly ignored to date. Molecular cluster compounds tend to have a very low magnetic density because of the large complex structural frameworks required to encase the multinuclear magnetic core. Herein we propose a drastically different approach by focusing on the simple and well-known ferromagnetic molecular dimer gadolinium acetate tetrahydrate, [{Gd(OAc)3(H2O)2}2]·4 H2O (1).[4a,b] The structure of 1 (Figure 1) com- Figure 1. The molecular structure of 1. Gd = black, O = dark gray, C = light gray, H = small bullets. H atoms of the methyl groups are omitted for clarity. Intramolecular hydrogen bonds are depicted as thin lines. prises a dimer of Gd3+ ions bridged through two of the six carboxylate groups, which bond in an h2 :h1:m2-fashion.[4c] The remaining acetates are chelating, and the nine-coordinate (capped square-antiprismatic) geometry of the metal centers is completed by the presence of two terminally bound H2O molecules. These partake in intramolecular hydrogen bonding to the neighboring chelating acetate ligands and are responsible for both the direct intermolecular H-bonds in the ab plane and the interplane H-bonds mediated by the lattice H2O molecules (Figure S1 and Table S1 in the Supporting Information). Our theoretical and experimental investigations (see the Supporting Information for details) of the magnetothermal properties of 1 down to millikelvin temperatures reveal a truly enormous MCE. In addition to magnetization and heat capacity experiments, which we employ to indirectly estimate the MCE, we make use of a homemade experimental setup that allows us to measure both DSm and DTad,[5] thus directly probing the extraordinary cooling performance of 1. Figure 2 depicts the direct-current (dc) magnetic susceptibility (c) of 1 collected in an applied field B0 = 0.1 T over the 2–140 K temperature range. The room-temperature experimental value of c T agrees with that expected from two non-  2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. 2011, 123, 6736 –6739 Angewandte Chemie Figure 2. Temperature dependence (T > 2 K) of the dc susceptibility c T for 1 collected in an applied field of 0.1 T. Line is the fit, see text. Inset: magnetization of 1 versus applied field for several temperatures. Lines are guides to the eye. interacting Gd3+ ions with g = 2, that is, c T = 15.75 cm3 K mol1. On lowering T, c T stays nearly constant with decreasing temperature down to approximately 20 K, below which it increases significantly, reaching a value of 18.8 cm3 K mol1 at 2 K, thus corroborating the dominant ferromagnetic coupling between the two Gd3+ ions within each molecular unit. The isothermal molar magnetization (Mm, inset of Figure 2) shows a saturation value of 14.0 N mB (where N is the Avogadro constant and mB is the Bohr magneton) at the lowest investigated temperature of 2 K, in agreement with the ferromagnetic spin state S = 7. To estimate the intramolecular exchange constant, we fitted the experimental c T versus T curve (Figure 2) to a model based on the isotropic spin Hamiltonian H = J(SGd1 SGd2), obtaining J/kB = 0.068(2) K and g = 2.01(1), in agreement with previous studies.[4b] By Hall based micromagnetometry, we extended the magnetic measurements down to lower temperatures (Figure 3), with the isothermal Mm(B0) curve collected for T = 0.2 K (inset) supporting the dominant ferromagnetism of 1. The temperature dependence of the dc susceptibility deviates from the Curie law at the lowest temperatures, giving rise to an anomaly centered at approximately 0.3 K for B0 = 0.01 T. The slight decrease in c below this temperature and the disappearance of the anomaly for fields higher than 0.1 T (Figure 3) suggest that the complex undergoes a transition to a magnetically ordered phase in which ferromagnetic interactions are important. To further elucidate the mechanism of magnetic ordering, we have performed numerical simulations according to the standard Metropolis Monte Carlo algorithm.[6] We considered each Gd2 complex as an isotropic point dipole, that is, we assume a ferromagnetic J = 1 intramolecular Gd3+···Gd3+ exchange coupling and no intermolecular exchange paths. In satisfactory agreement with the experimental observations, we obtain a critical temperature of approximately 0.18 K for B0 = 0 and a magnetic structure formed by alternating ferromagnetic ab planes (see the Supporting Information). We next turn to the evaluation of the magnetothermal properties of 1 by presenting its experimental heat capacity (C). Figure 4 (top) depicts the experimental temperature Figure 4. Top: temperature dependencies of the heat capacity of 1 normalized to the gas constant R collected for B0 = 0, 1, 3, and 7 T. Bottom: temperature dependencies of the experimental magnetic entropy for several B0, as obtained from the respective heat capacity data after subtracting the lattice contribution (dashed line). Figure 3. Low-temperature (T  0.2 K) Hall micromagnetometry for 1. Temperature dependence of the dc-susceptibility c collected in applied fields of 0.01 and 0.1 T. Inset: experimental molar magnetization versus applied field for Hall micromagnetometry at 0.2 (c) and at 2.0 K (a) and for SQUID magnetometry at 2.0 K (&). Angew. Chem. 2011, 123, 6736 –6739 dependence of C for selected applied fields. At high temperatures the heat capacity is dominated by nonmagnetic contributions arising from thermal vibrations of the lattice, which can be modeled with the six-fold Debye function (dashed line in Figure 4), yielding a value of VD = 61.6 K for the Debye temperature, typical for this class of cluster compound.[7] At low temperatures the applied field splits the molecular spin multiplet S = 7, gradually decoupling the intramolecular Gd3+···Gd3+ exchange coupling and giving rise to a broad (Schottky-type) feature, which shifts to higher T by increasing B0. Unfortunately, the experimentally accessible  2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.angewandte.de 6737 Zuschriften temperatures do not permit the observation of the phase transition. We note, however, that the zero-field C keeps increasing by decreasing T in the lowest temperature region, reaching values exceeding that expected for magnetically isolated Gd2 units (ffi1.9 R). From the experimental heat capacity, the temperature dependence of the magnetic entropy (Sm) is obtained by integration, using Sm(T) = sCm/ TdT, where the magnetic heat capacity Cm is obtained from C upon subtracting the lattice contribution. The thus obtained Sm(T) curves are shown in the bottom panel of Figure 4 for the corresponding applied fields. As expected, Sm for B0 > 0 tends to the maximum entropy value per mole involved at high temperatures, corresponding to two Gd3+ S = 7/2 spins (2 R ln(2 s+1)ffi4.16 R). In the case of B0 = 0, our experimental blindness for T lower than approximately 0.35 K forced us to add a constant value to the zero-field Sm(T) to match the limiting value at high T. As we shall see below, this crude procedure does not jeopardize our evaluation of the MCE of 1. From the Sm data, it is then straightforward to obtain the changes of magnetic entropy and adiabatic temperature, both indicated by arrows in Figure 4. The results are shown in Figure 5 as a function of T and for several field changes DB0 = Gd3+···Gd3+ exchange coupling, then the low-temperature value of DSm is much larger than can be produced in the absence of such a coupling. For instance, a relatively modest DB0 = 1 T is already sufficient to provide a DSm as large as 27 J kg1 K1 at T  0.5 K. This remarkable field dependence of the MCE is also observed in DTad. Figure 5 (bottom) shows that DB0 = 7 T provides a maximum DTad = 12.7 K for the same temperatures at which we observe the DSm maxima. By lowering DB0 to 3 and 1 T, DTad decreases to 9.0 and 3.5 K, respectively. Therefore, the field dependence of DTad increases from nearly 2 to well over 3 K T1, respectively, setting this material as the most efficient refrigerant for this low-temperature region.[8] In addition to heat capacity, magnetization data can also be employed for estimating DSm by making use of the Maxwell relation, DSm(T) = s[@Mm(T,B0)/@T]dB0. From the isothermal Mm(B0) curves of Figure 2, the so-obtained DSm(T) value for DB0 = 3 T is displayed in Figure 5 (top) and can be seen to be in perfect agreement with the data obtained from C, thus demonstrating that our experimental uncertainty in the zero-field Sm(T) does not affect the evaluation of DSm and DTad of 1. However, we note that both indirect procedures we followed for obtaining the MCE rely on numerical integrations that, by their nature, can be the source of large errors.[9] To overcome any possible shortfall inherent to these approaches, we have also measured DSm and DTad directly under quasi-adiabatic conditions (see Figure S3 in the Supporting Information).[5] By again considering DB0 = 3 T, we obtain the DSm and DTad values depicted in Figure 5, which rather beautifully corroborate our previous estimates. In conclusion, an unprecedentedly large magnetocaloric effect at extremely low temperature is reported to occur in the simple ferromagnetic molecular dimer gadolinium acetate tetrahydrate. The magnetic ordering originates from dipolar coupling, and its collective magnetic behavior is somewhat reminiscent of that reported for gadolinium sulfate octahydrate, a well-studied, purely dipolar system, which was the subject of the very first adiabatic demagnetization experiments.[10] The enormous advantage of 1 over this prototype magnetic coolant is the intramolecular ferromagnetic exchange coupling, which favors the field-dependent enhancement of the magnetocaloric effect. Figure 5. Top: temperature dependencies of the magnetic entropy change DSm(T) as indirectly obtained from heat capacity and magnetization data, together with the direct measurements (gray dots) for the indicated applied-field changes DB0. Bottom: temperature dependencies of the adiabatic temperature change DTad(T) as indirectly obtained from heat capacity data, together with the direct measurements, for the indicated DB0. Received: April 16, 2011 Published online: June 7, 2011 BfBi, where f and i indicate final and initial states, respectively. A striking result is the DSm, which can be seen to reach values over 40 J kg1 K1 at T  1.8 K for DB0 = 7 T, which is much larger than any other value reported in the recent literature.[3] We note that DSm approaches the maximum entropy value per mole (4.16 Rffi42.5 J kg1 K1) for two fully decoupled Gd3+ ions. If DB0 is lower than or of the order of the strength of the ferromagnetic intramolecular 6738 www.angewandte.de . Keywords: gadolinium · low-temperature physics · magnetic properties · magnetocaloric effects · molecular refrigerants [1] For a recent review, see M. Evangelisti, E. K. Brechin, Dalton Trans. 2010, 39, 4672, and references therein. [2] a) Yu. I. Spichkin, A. K. Zvezdin, S. P. Gubin, A. S. Mischenko, A. M. Tishin, J. Phys. D 2001, 34, 1162; b) F. Torres, J. M. Hernndez, X. Bohigas, J. Tejada, Appl. Phys. Lett. 2000, 77, 3248; c) F. Torres, X. Bohigas, J. M. Hernndez, J. Tejada, J. Phys. Condens. Matter 2003, 15, L119; d) M. Affronte, A. Ghirri, S. Carretta, G. Amoretti, S. Piligkos, G. A. Timco, R. E. P.  2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. 2011, 123, 6736 –6739 Angewandte Chemie Winpenny, Appl. Phys. Lett. 2004, 84, 3468; e) J. Schnack, R. Schmidt, J. Richter, Phys. Rev. B 2007, 76, 054413. [3] a) M. Evangelisti, A. Candini, A. Ghirri, M. Affronte, E. K. Brechin, E. J. L. McInnes, Appl. Phys. Lett. 2005, 87, 072504; b) R. Shaw, R. H. Laye, L. F. Jones, D. M. Low, C. TalbotEeckelaers, Q. Wei, C. J. Milios, S. J. Teat, M. Helliwell, J. Raftery, M. Evangelisti, M. Affronte, D. Collison, E. K. Brechin, E. J. L. McInnes, Inorg. Chem. 2007, 46, 4968; c) M. Manoli, R. D. L. Johnstone, S. Parsons, M. Murrie, M. Affronte, M. Evangelisti, E. K. Brechin, Angew. Chem. 2007, 119, 4540; Angew. Chem. Int. Ed. 2007, 46, 4456; d) M. Manoli, A. Collins, S. Parsons, A. Candini, M. Evangelisti, E. K. Brechin, J. Am. Chem. Soc. 2008, 130, 11129; e) M. Evangelisti, A. Candini, M. Affronte, E. Pasca, L. J. de Jongh, R. T. W. Scott, E. K. Brechin, Phys. Rev. B 2009, 79, 104 414; f) G. Karotsis, M. Evangelisti, S. J. Dalgarno, E. K. Brechin, Angew. Chem. 2009, 121, 10 112; Angew. Chem. Int. Ed. 2009, 48, 9928; g) G. Karotsis, S. Kennedy, S. J. Teat, C. M. Beavers, D. A. Fowler, J. J. Morales, M. Evangelisti, S. J. Dalgarno, E. K. Brechin, J. Am. Chem. Soc. 2010, 132, 12983; h) S. Nayak, M. Evangelisti, A. K. Powell, J. Reedijk, Chem. Eur. J. 2010, 16, 12865; i) Y.-Z. Zheng, M. Evangelisti, R. E. P. Winpenny, Chem. Sci. 2011, 2, 99; j) Y.-Z. Zheng, M. Evangelisti, R. E. P. Winpenny, Angew. Chem. 2011, 123, 3776; Angew. Chem. Int. Ed. 2011, 50, 3692; k) S. K. Angew. Chem. 2011, 123, 6736 –6739 [4] [5] [6] [7] [8] [9] [10] Langley, N. F. Chilton, B. Moubaraki, T. Hooper, E. K. Brechin, M. Evangelisti, K. S. Murray, Chem. Sci. 2011, 2, 1166. a) M. C. Favas, D. L. Kepert, B. W. Skelton, A. H. White, J. Chem. Soc. Dalton Trans. 1980, 454; b) S. T. Hatscher, W. Urland, Angew. Chem. 2003, 115, 2969; Angew. Chem. Int. Ed. 2003, 42, 2862; c) CCDC 820956 (1) contains the supplementary crystallographic data for this paper. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif. a) L. Tocado, Ph. D. Thesis, Universidad de Zaragoza, 2008; b) L. Tocado, E. Palacios, R. Burriel, J. Magn. Magn. Mater. 2005, 290–291, 719. N. A. Metropolis, A. W. Rosenbluth, M. N. Rosenbluth, A. H. Teller, E. Teller, J. Chem. Phys. 1953, 21, 1087. M. Evangelisti, F. Luis, L. J. de Jongh, M. Affronte, J. Mater. Chem. 2006, 16, 2534. For an overview see, for example, V. K. Pecharsky, K. A. Gschneidner, Jr., J. Magn. Magn. Mater. 1999, 200, 44, and references therein. V. K. Pecharsky, K. A. Gschneidner, J. Appl. Phys. 1999, 86, 565. a) W. F. Giauque, D. P. MacDougall, Phys. Rev. B 1935, 43, 768; b) W. F. Giauque, D. P. MacDougall, J. Am. Chem. Soc. 1935, 57, 1175; c) R. F. Wielinga, J. Lubbers, W. J. Huiskamp, Physica 1967, 37, 375.  2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.angewandte.de 6739