398Rev.Adv.Mater.Sci. 10 (2005) 398-402 Srikanth Hariharan and James Gass

SUPERPARAMAGNETISM AND MAGNETO-CALORIC EFFECT (MCE) IN FUNCTIONAL MAGNETIC NANOSTRUCTURES

Srikanth Hariharan and James Gass

Materials Physics Laboratory, Department of Physics, University of South Florida, Tampa, FL 33620, USA Received: May 25, 2005

Abstract. Magnetic exhibit superparamagnetic and ferromagnetic behaviors con- trolled by the particle size and interactions. While monodisperse nanoparticles are of interest from fundamental science perspective, functional nanostructures involve dispersion of particles in host matrices such as . We discuss synthesis and of nanocomposites based on PMMA with embedded Fe O nanoparticles. An interesting applica- 3 4 tion of assemblies is in magneto-caloric effect (MCE) that has the potential for local cooling/heating of NEMS and MEMS devices. We have studied MCE in CoFe O nanoparticles by 2 4 measuring family of M-H curves at set temperature intervals and calculated the entropy change (∆S) for this system.

1. INTRODUCTION ogy, electromagnetic devices [2,3]. For example, the lack of magnetic hysteresis coupled with high behave very differently from saturation magnetization is highly desirable in RF bulk or thin film systems. When the size of the and microwave magnetic applications as the mag- particles is reduced below the limit netic losses are considerably reduced. In addition, (~15 to 20 nm for iron oxide), they exhibit the granular and disconnected nature of nanoparticle superparamagnetism at room temperature followed assemblies prevents eddy current losses too. by a spin-glass like transition at low temperature For device applications, since particles are gen- [1]. Superparamagnetism is essentially a result of erally dispersed in a non-magnetic (conducting or non-interacting, thermally fluctuating nanoparticle non-conducting) matrix, it is important to investigate moments where the particle volume dependent ef- the collective magnetic response and possible influ- fective anisotropy energy (KV) of each particle is ence of the matrix itself. It is well known that in easily overcome by the thermal energy (k T). The B polymer host matrices, nanoparticle agglomeration effect of this is that the nanoparticles have a large and magnetic interactions can affect the overall prop- moment with high saturation magnetization (as erties to a great extent [4-6]. Many attractive prop- expected in ferromagnets) but a non-hysteretic M- erties of polymers like non-corrosiveness, light H curve with zero remanence and coercivity. weight, mechanical strength and dielectric tunability Superparamagnetic particles have a number of ap- can be utilized along with novel magnetic and opti- plications in fields as varied as biomedical technol-

Corresponding author: Srikanth Hariharan, e-mail: [email protected]

© 2005 Advanced Study Center Co. Ltd. Superparamagnetism and magneto-caloric effect (MCE) in functional magnetic nanostructures 399 cal properties of nanoparticles to make multifunc- ing their attraction to a bar , the particles in tional materials. Inclusion of ferromagnetic or solution were coated with oleic acid. The surfac- superparamagnetic nanoparticles in polymers is tant-coated nanoparticles were suspended in hex- especially important as magnetic nanoparticles have ane and further centrifuged to separate any agglom- shown promise in various potential applications like erated clusters and/or particles of larger size out. spin-polarized devices, carriers for drug delivery, To prepare polymer nanocomposite thin films, 5% magnetic recording media, high-frequency applica- poly-methyl-methacrylate (PMMA) by weight was tions, etc. dissolved in chlorobenzene and vigorously stirred Another area where there is relatively little work for a few hours, until it is all dissolved. The Fe O 3 4 is in exploring the possibility of refrigeration using nanoparticles coated with oleic acid were added and the magneto-caloric effect (MCE) in nanoparticle sonicated to facilitate uniform dispersion. The solu- clusters or assemblies [7]. While the concept of tion was spun coat onto substrates at 4000 rpm magnetic refrigeration has been around for a long which yielded films of about 2 microns thickness. time, lack of easily processable materials have lim- One of our goals was to spin coat a layer of con- ited the effectivness of this approach to very low ducting polymer (polypyrrole) on top to form a bi- temperatures. MCE is an inherent property of all layer. The reason for a conducting polymer sheath magnetic materials and is due to the coupling of was to develop all-polymer layered structures that the magnetic sublattice with the external applied integrate electromagnetic interference (EMI) sup- field. The discovery of giant MCE in Gd (Si Ge ) pression capabilities for RF applications. The 5 x 1-x 4 alloys [8] with promising refrigeration characteris- polypyrrole layer is non-magnetic and does not in- tics in the intermediate temperature (40K to 200K) teract with the PMMA based nanocomposite film in range has revived interest in the search for new MCE any way interfering with its magnetic response. The materials. Recently, the importance of nanoscale top panel of Fig. 1 shows a cross-sectional TEM inclusions of Fe in this alloy has been discussed image of the bi-layer. The uniform dispersion of Fe O 3 4 by Provenzano et al. [9]. Their results indicate that nanoparticles is clearly evident in the bottom PMMA the magnetic entropy change is enhanced and also layer and it is important to note that individual par- pushed to near room temperature in Fe containing ticles are clearly resolved with very little clustering systems in addition to reduction in hysteretic observed. This is quite significant as it is a chal- losses. There have also been investigations of MCE lenging task to control the nanoparticle diffusion and in nanoparticle-based systems as it has been theo- avoid agglomeration during the polymerization pro- retically predicted that the entropy change could be cess [11]. Our contention is that the solvent-based enhanced in these systems [10]. technique that we employed with individual In this paper we report on two functional aspects nanoparticles coated by fatty acid and the overall of magnetic nanostructures currently being explored polymerization process assisted by vigorous soni- in our laboratory. The first is a demonstration of con- cation may be responsible for the nearly cluster- trolled nanoparticle dispersion in polymer free dispersion inside the polymer matrix. Molecu- nanocomposites exhibiting superparamagnetism lar Dynamics (MD) simulations have been particu- and the second is an analysis of MCE in the vicinity larly successful in investigating the dispersion of of the blocking temperature for chemically synthe- nanoparticles in polymers [4,12]. When the poly- sized Co-ferrite nanoparticles containing small and mer-polymer interaction strength is stronger than large size distribution. the polymer-particle interaction strength, the sys- tem potential energy is lowered by clumping the 2. SUPERPARAMAGNETIC POLYMER particles. If the interaction between the particles and NANOCOMOSITES the polymer can be increased, the particles will then disperse throughout the polymer. In our case, the Nanoparticles of Fe O were synthesized using a 3 4 oleic acid surfactant coating on the particles in- standard chemical co-precipitation method and char- creases the interaction with the polymer and thus acterized using X-ray diffraction (XRD) and trans- favors the process of uniform dispersion. mission electron microscopy (TEM). Analysis of the Fig. 1 also shows the magnetization vs field (M- particle size distribution showed a mean size of 11 H) measurements on the sample done using a com- nm with a standard deviation of 4 nm. We have also mercial Physical Property Measurement System further improved the size distribution using an ultra- (PPMS) from Quantum Design. M-H curves taken centrifuge and reduced the standard deviation to 2 at two temperatures (300K and 10K) are shown and nm. After confirming their magnetic nature by test- 400 Srikanth Hariharan and James Gass

3. MAGNETO-CALORIC EFFECT (MCE) IN FERRITE NANOPARTICLES MCE in conventional ferromagnetic materials is lim- ited to the entropy change associated with the spin disorder-order states achieved across the paramag- netic to ferromagnetic transition as one cools the sample below the (T ). The en- c tropy change in a magnetic system can be calcu- lated from the thermodynamic Maxwell relation:

F ∂S I F ∂M I = (1) H H K H T K ∂ TH∂ or integrating over the field,

H F ∂M I ∆S = z dH . H T K 0 ∂ H From measurements of M(H,T), the magnetic en- tropy change can be calculated. Typical values of the entropy change range in the 0.1 to 1 J/kg-K over a field of 1 or 2 Tesla. In systems exhibiting metamagnetic transitions and close coupling be- tween the magnetic and structural degrees of free- dom (such as the manganites), giant MCE with one or two orders of magnitude bigger entropy change have been reported in the vicinity of Curie tempera- Fig. 1. (Top) Cross-sectional TEM image of spun ture. However these changes do not persist over a coat PMMA/Polypyrrole bilayer films showing the broad temperature range. As indicated before, it is dispersion of Fe O nanoparticles in the PMMA somewhat cumbersome to systematically tune the 3 4 layer. (Bottom) Magnetization vs Field at room tem- ferromagnetic Curie temperature in materials through perature and 10K. Inset shows the low-field region state chemistry and doping. It is precisely for indicating the presence of hysteresis at low tem- this reason that magnetic nanoparticles have a perature. unique advantage. In nanoparticles there is a versa- tile order-disorder transition associated with the superparamagnetic-blocked states referred to as the blocking transition (T ). Unlike the Curie tempera- B ture, T can be varied by adjusting the particle size, B shape, dispersion and inter-particle interactions. While it is unlikely that simple elemental and oxide both of them display a strong non-linear behavior nanoparticles will exhibit giant magnetic entropy with saturation magnetization. The inset shows the changes, the advantages lie in the ease of synthe- expanded view of the low-field region. The M-H curve sis, tunability of T and also the dispersion of B at 300K does not show any hysteresis whereas the nanoparticles in other host matrices that themselves loop at 10K shows a clear hysteresis with a coer- can be good MCE materials. MCE in several civity of around 150 to 200 Oe. This is characteris- nanoparticle systems have been investigated by a tic of the sample being superparamagnetic at room number of groups over the years. It has been theo- temperature and entering a blocked state at low retically predicted [10] that nanocomposite temperature which results in opening up of the hys- superparamagnetic nanoparticles have the poten- teresis loop. The cross-sectional TEM image along tial to exceed the performance of garnets and other with this magnetic response provides evidence that commonly used magnetic refrigerants particularly we have indeed obtained superparamagnetic poly- at low fields generated by simple electromagnets mer nanocomposites at room temperature. Superparamagnetism and magneto-caloric effect (MCE) in functional magnetic nanostructures 401

Fig. 2. (Top) Zero-field cooled (ZFC) and Field-cooled (FC) magnetization data with 100 Oe applied field Fig. 3. Magnetic entropy change for CoFe O for CoFe O nanoparticles. (Bottom) Family of M-H 2 4 2 4 nanoparticles of mean size 5 nm (top) and 8 nm curves at various temperatures from which the en- (bottom). Insets show TEM images of as-synthe- tropy change is extracted. sized particles.

(1 to 2 T range as opposed to 5 – 10 T or higher der form at this stage for magnetic measurements which are only attainable in superconducting or or suspended in hexane for further size selection. pulsed ). We were able isolate two sets of nanoparticles with In this work, we have measured and calculated mean size distribution centered around 5 nm and 8 the magnetic entropy change using the standard nm. The TEM images for these two sets are shown Maxwell method in CoFe O nanoparticles prepared in the insets of Fig. 3 and the magnetic character- 2 4 by chemical methods. izations for the 5 nm particles are plotted in Fig. 2. To make CoFe O nanoparticles, a mixture of The top panel in Fig. 2 shows the zero-field cooled 2 4 cobalt(II) acetylacetonate, phenyl ether,1,2- and field-cooled magnetization with a broad peak hexadecanediol, oleic acid, and oleylamine are seen in ZFC curve characteristic of the blocking tran- mixed and heated to 156 °C. At this point iron(III) sition followed by a slow decrease in magnetization acetylacetonate mixed in phenyl ether was added at low temperature. Our goal was not to have a sharp and heated to reflux (260 °C) for 1 hour. The par- blocking transition (expected for a monodisperse ticles precipitated upon addition of excess ethanol system with a very tight size distribution) but rather and centrifuging. They were then washed several have a broad transition which is realized in polydis- times using deionized water and magnetic decan- perse nanoparticles with a large size distribution. tation. After that, they were dried and used in pow- The reasoning behind this is based on the simple 402 Srikanth Hariharan and James Gass

conjecture that if the magnetic entropy change (DS) ACKNOWLEDGEMENTS is maximum in the vicinity of the blocking transi- The authors thank the US National Science Foun- tion, then having a broad transition would retain a dation for supporting this research through grant nearly constant DS over a broader temperature (CTS-0408933). Drew Rebar is acknowledged for his range. Such characteristics would be desirable for assistance with the experiments and the authors field-driven small scale refrigeration applications such gratefully acknowledge Dr. Pankaj Poddar for valu- as cooling MEMS devices. The bottom panel of Fig. able discussions and suggestions. 2 shows the M-H curves over a field range of 0.5 to 3 T taken at temperature intervals of 10K from 110 to 300K which spans a sizeable region covering the REFERENCES blocking peak. The DS were numerically calculated [1] Magnetic properties of fine particles, ed. by from these data sets using the Maxwell equation J. L. Dormann and D. Fiorani (North-Holland, and the results for the two CoFe O nanoparticle 2 4 Amsterdam, 1992). samples are plotted for three field values in Fig. 3. [2] T. Neuberger, B. Schopf, H. Hofmann, The values for the entropy change are quite small M. Hofmann and B. von Rechenburg // as reported in several other nanoparticle systems J. Magn. Magn. Mater. 293 (2005) 483. and comparatively slightly higher [13,14]. This is [3] J. I. Martin, J. Nogues, K. Liu, J. L. Vincent consistent with models that indicate that introduc- and I. K. Schuller // J. Magn. Magn. Mater. 256 ing some amount of inter-particle interactions could (2003) 449. lead to higher entropy change in comparison to [4] Polymer films with embedded metal strictly non-interacting superparamagnetic particles nanoparticles, ed. by A. Heilmann (Springer, [13]. Currently we are in the process of doing sys- New York, 2003). tematic MCE experiments on a single ferrite sys- [5] J. L. Wilson, P. Poddar, N. A. Frey, tem by varying parameters such as the overall par- H. Srikanth, K. Mohomed, J. P. Harmon, ticle size, dispersion and inter-particle interactions. S. Kotha and J. Wachsmuth // J. Appl. Phys. While the expected trend of large ∆S is observed 95 (2004) 1439. for higher applied fields, there is one aspect to the [6] D. Vollath, D. V. Szabo and S. Schlabach // data in Fig. 3 that needs to be mentioned. In both J. Nanoparticle Res. 6 (2004) 181. the systems, it is observed that the ∆S increases [7] V. Provenzano, J. Li, T. King, E. Canavan, uniformly as a function of temperature (up to 300 to P. Shirron, M. DiPirro and R. D. Shull // 350K which is the limit of our PPMS) and does not J. Magn. Magn. Mater. 266 (2003) 185. follow the shape of the blocking peak which is cen- [8] V. K. Pecharsky and K. A. Gschneider, Jr. // tered around 220K for the 5 nm particles and higher Appl. Phys. Lett. 78 (1997) 4494. for the 8 nm particles (not shown in this paper). A [9] V. Provenzano, A. J. Shapiro and R. D. Shull // similar trend was also observed in our studies on Nature 429 (2004) 853. other ferrite nanoparticles with much lower blocking [10] R. D. McMichael, R. D. Shull, L. J. peak (~40K) where a broad maximum in ∆S was Swartzendruber, L. H. Bennett and R. E. seen around a temperature about 4 to 5 times larger Watson // J. Magn. Magn. Mater. 111 (1992) than blocking temperature. This is quite promising 29. as it would imply that reasonably high ∆S can be [11] P. Poddar, J. L. Wilson, H. Srikanth, S. A. maintained at temperatures well above the blocking Morrison and E. E. Carpenter // transition which makes it different from conventional Nanotechnology 15 (2004) S570. ferromagnets where ∆S drops off sharply away from [12] T. Desai, P. Keblinski and S. K. Kumar // T . The origin of this behavior of S is unclear at c ∆ J. Chem. Phys. 122 (2005) 134910. this time. In nanoparticles, there is another charac- [13] M. S. Pedersen, S. Morup, S. Linderoth, teristic temperature above the blocking temperature C. Johansson and M. Hanson // J. Phys.: (T ) which is the irreversibility point (i.e. the tem- B Condens. Matter 9 (1997) 7173. perature below which the ZFC and FC curves devi- [14] T. Kinoshita et al. // J. Alloys and Com- ate and above which they lie on top of each other). pounds 365 (2004) 281. It remains to be seen if this temperature is relevant in any way for the ∆S dependence and more ex- periments are needed to further explore these is- sues.