1

General introduction on magnetic fluids

Chapter 1

1.1
1.1.a

Magnetism
Atomic magnetism, paramagnets and diamagnets [1-3]

Wherever electrical currents flow, magnetic fields are generated. Within atoms, the orbital motions of electrons around the nucleus and protons within the nucleus produce magnetic fields. In addition, the spinning motion of electrons, proton, and neutrons around their axes generate magnetic fields. However, because the contribution to the induced magnetic field of nucleons is typically two thousand times smaller than that of electrons, it is usually neglected. In case of current loops, the generated magnetic field strength is conveniently written as being proportional to the magnitude (current )(loop area) of the magnetic moment m, a vectorial quantity. Many atoms have a permanent non-zero magnetic moment, in the order of 10-23 A m2. The net magnetic moment of an assembly of atoms, however, is generally zero due the random orientation of atomic moments within the assembly (see figure 1.1a). The net magnetic moment can be changed by bringing the assembly into an external magnetic field H. This effect of H can usually be expressed as M =H, where M = i m i / V is the magnetization (net dipole moment per unit volume), V the volume of the system, and the susceptibility, describing the strength of the response to an external magnetic field. Microscopically, the effect of the external field is twofold. First, the external field exerts a torque =mH on the atomic moments that forces them to align with H. Since Brownian rotation counteracts the alignment, the degree of alignment depends on the magnetic field strength and temperature. The alignment of dipoles with H gives a positive contribution to , called the paramagnetic susceptibility. Second, the external field induces a change in the orbital motion of electrons, thereby producing a magnetic field that opposes the external field. Accordingly, this effect gives a negative contribution to , called the diamagnetic susceptibility. The diamagnetic susceptibility is independent of temperature, and is present in all atoms, regardless of the presence of a permanent magnetic moment. Depending on the sign of the total susceptibility, materials are classified as paramagnets (>0, and of the order of 10-5 to 10-3) or diamagnets (<0, typically -10-5). 1.1.b Ferromagnets, antiferromagnets and ferrimagnets [1,2]

In the previous section, magnetic materials were treated as systems containing atoms with noninteracting dipoles. In some materials, however, strong dipole-dipole interaction of quantummechanical origin causes long range orientational correlations of

General introduction

Figure 1.1. Dipole orientations in four types of magnetic materials: (a) paramagnet; (b) ferromagnet; (c) antiferromagnet; (d) ferrimagnet.

the permanent atomic dipoles. In metallic iron, for example, dipolar interaction favors parallel alignment of the dipoles (figure 1.1b), hence this material has a net magnetization even in the absence of an external field. Materials with this kind of ordering of dipoles are called ferromagnets, and are characterized by a high susceptibility, typically between 10-2 and 106, which, moreover, depends on the field strength. Above the so-called Curie temperature, which is material-specific, the dipolar ordering inside a ferromagnet is lost and the material becomes a paramagnet. Dipolar interaction does not always lead to net magnetization. In some materials, named antiferromagnets, neighboring dipoles are aligned antiparallel (figure 1.1c). Consequently, such materials do not have a net magnetization in the absence of an external field, and have a low susceptibility, generally between 0 and 10-2. At the so-called Nel temperature, a transition from antiferromagnetic ordering to paramagnetic occurs. Ferrimagnets form a third class of magnetic materials with ordered dipoles. On the microscopic scale, ferrimagnets are similar to antiferromagnets, with antiparallel arrangement of neighboring dipoles. However, because the number (or magnitude) of dipoles pointing in one direction differs from the number pointing in the opposite direction, ferrimagnets have a nonzero magnetization in zero field (figure 1.1d). Thus, ferrimagnets resemble ferromagnets on the macroscopic scale. Due to partial cancellation of dipole moments, the magnetization of ferrimagnets is generally lower than that of ferromagnets. The most well-known ferrimagnetic material is magnetite (Fe3O4). 1.1.c Magnetic domains [1,3]

The dipole ordering described in the previous section is long ranged, but usually does not extend over the entire volume of a sample. Rather, a sample of magnetic material is split up into domains in which all dipoles are ordered along a preferential direction (figure 1.2). This direction changes from domain to domain, and it is for this reason that bulk magnetic materials may as a whole be unmagnetized, even though they are 7

Chapter 1

Figure 1.2. In bulk ferromagnetic materials, dipole alignment is split up into domains.

magnetized on the length scale of the domains. In an external magnetic field, the magnetization directions of the domains are forced to align with the field, or domains with favorable magnetization direction will grow at the cost of domains with unfavorable directions. Both mechanisms increase the net magnetization of the bulk material. At infinite magnetic field strength, all dipoles are aligned with the field, and the system has reached its saturation magnetization. After the field is removed, the magnetization tends to relax into its original state with randomly oriented domains. However, in order to reach that equilibrium state, the system may have to pass unfavorable states that can keep it from actually reaching equilibrium. Materials for which the relaxation to the unmagnetized state is prevented, are called hard magnetic, as opposed to soft magnetic materials, which demagnetize quickly upon removal of the external field or application of a small field opposing the magnetization.

1.2
1.2.a

Magnetic fluids
Introduction

Unlike paramagnets and diamagnets, which can be gaseous, liquid or solid, ferromagnets are almost exclusively solid1. There exist, however, liquids with a susceptibility of roughly 1, not as high as many ferromagnets but still orders of magnitudes higher than that of paramagnets. Moreover, like ferromagnets these liquids can easily approach their saturation magnetization. These so-called magnetic fluids or ferrofluids are actually two-phase systems, comprised of small solid ferro- or ferrimagnetic particles dispersed in a liquid [6-8]. Because the size of the particles is in the colloidal range, i.e. between about 1 and 1000 nm, these particles are often referred to as magnetic colloids. Although intensive research on magnetic fluids did not start until the 1960s, the preparation of

Only very few exceptions are known, such as liquid 3He [4] and undercooled,

metastable melts of Co80Pd20 [5]. 8

General introduction

(water-based)

magnetic

fluids

had

already

been

described

in

1938

by

Figure 1.3. Schematic drawing of a ferrofluid on three length scales. On the macroscopic length scale (left), it resembles an ordinary liquid. On the colloidal length scale (middle), the fluid is appears to consist of small solid particles dispersed in a liquid. Each particle consists of a single domain magnetic iron oxide core, and a surface grafted with polymer chains (right).

Elmore [9]. The first ferrofluids were primarily used as a means to study magnetic domain structures in solids. Three decades later, Papell [10] and Rosensweig [11] developed the synthesis of oil-based ferrofluids, which are still used today. The most commonly used ferrofluid contains spherical magnetite particles with a typical size of 10 nm, dispersed in an apolar solvent. Sedimentation of these particles is sufficiently counteracted by their Brownian motion to keep them dispersed for years. A prerequisite for such long-term stability is that particles do not aggregate, since aggregates sediment faster and have slower Brownian motion to compensate for sedimentation. Unfortunately, without special measures magnetite particles will form large clumps that settle quickly. To prevent aggregation, the colloids can be covered with a thin layer of polymer, commonly a monolayer of oleic acid, which makes the particles soluble in many organic solvents. A schematic drawing of a ferrofluid, seen on three length scales, is shown in figure 1.3. Because of their small size, these magnetic colloids contain a single magnetic domain, and therefore have a permanent magnetic moment proportional to their volume. Although magnetic colloids are ferromagnetic on the molecular scale, they resemble a paramagnet on the colloidal scale, with the major difference that the magnetic moments of magnetic colloids are much larger than the moments in a paramagnet (typical values are 10-19 A m2 for magnetic colloids and 10-23 A m2 for paramagnets). It is for this reason that ferrofluids are sometimes called superparamagnetic. In order to be (super)paramagnetic, the dipole moment of each particle must be free to rotate on the time scale of experiments. Two modes of rotation are operative in magnetic colloids. One is Brownian rotation, with a relaxation time

Chapter 1

B =

3v0 kT

(1.1)

where v the volume of the particle, 0 the viscosity of the solvent, k the Boltzmann constant and T the temperature. For 10-nm colloids in a solvent with 0=10-3 Pa s, B is 4 10-7 s. The other mode of rotation is Nel rotation, which involves rotation of the magnetization with respect to the crystal lattice of the magnetic colloid. The relaxation time for this process is Kv N = f 0 1 exp , kT (1.2)

where K is the (material-dependent) anisotropy constant, and f 0 is the Larmor frequency, about 109 s-1. Clearly, the Nel relaxation time stronly depends on the particle volume. For example, the N of magnetite colloids (K=1.1 104 Jm-3 [7]) increases from 4 10-9 s to 7 10-5 s upon increasing the particle diameter from 10 nm to 20 nm. 1.2.b Applications of magnetic fluids

An important property of concentrated ferrofluids is that they are strongly attracted by permanent magnets, while their liquid character is preserved. The attraction can be strong enough to overcome the force of gravity. Many applications of ferrofluids are based on this property [7]. For example, ferrofluids are widely used as lubricating, airtight seals in rotary shafts. A magnetic field gradient keeps the ferrofluid in place, even in case of pressure differences between the two separated compartments. Today, many computer hard disk drives contain a ferrofluid-sealed shaft. Ferrofluids are also used to improve heat dissipation in loudspeaker coils, enabling higher output power. When non-magnetic objects are immersed in a ferrofluid and subjected to the field gradient of a permanent magnet, the objects will be effectively repelled by the magnet (actually, the ferrofluid is attracted and drives away the object). When combined with a gravitational or centrifugational force opposing the effective magnetic force, this effective repulsion has the same effect as a density gradient in the solvent. This principle is used to separate materials into density fractions, for instance in the mining industry or waste processing [12]. Because the effective 'density' of ferrofluids can be much higher than that of ordinary liquids, density-based separation with ferrofluids is also suitable for high density materials such as metals. In the aforementioned applications, the ferrofluid was considered to be homogeneous. However, structural changes can occur on a microscopic level when magnetic fluids are subjected to a magnetic field. Because an external field aligns the dipole moments of magnetic colloids, it can increase the average interaction strength between magnetic colloids sufficiently 10

General introduction

to induce aggregation of the colloids into concentrated, micron-sized droplets [13]. Because the size of such droplets is comparable to the wavelength of light, the optical properties of ferrofluids depend on the direction and strength of the external field. Optical devices employing the strong magneto-optical effects of magnetic fluids are still in development [7,14,15]. 1.2.c Phase behavior of magnetic fluids

The macroscopic properties of ferrofluids not only depend on the properties of single particles, but also on their mutual interaction. As can be inferred from the previous section, some applications depend on the stability of ferrofluids, others depend on their instability. In addition to the irreversible aggregation mentioned in section 1.2.a, aggregation may also be reversible. Because of the colloid's Brownian motion, the behavior of a magnetic fluid is dictated by thermodynamics, so given the characteristics of the magnetic colloids, the temperature and concentration, the system adapts a state or phase that minimizes its free energy. The anisotropic nature of magnetic interaction leads to a rich phase behavior of magnetic fluids. For example, because magnetic interaction favors head-to-tail configurations of magnetic colloids, worm-like structures can be expected in dilute solutions of strongly interacting magnetic colloids [16]. In concentrated magnetic fluids, calculations suggest that strong interaction may lead to macroscopic parallel alignment of dipole moments, yielding a liquid permanent magnet [17,18]. These two phases have not yet been found experimentally, and some even question if the spontaneously magnetized liquid exists at all [19]. Another unresolved and actively debated issue is whether magnetic attraction can give rise to gas-liquid separation [20], i.e. spontaneous separation of an initially homogeneous magnetic fluid into two fluids with different concentrations. Interactions between magnetic colloids can be modified by applying an external field, thereby enriching the phase behavior. The external field will align magnetic moments of colloids along the field, thereby increasing their average strength of interaction. This can also lead to gasliquid phase separation [17,18], though in this case both fluid phases are magnetized. Fieldinduced phase separation has been experimentally demonstrated [13], but the factors influencing this type of phase instability of magnetic fluids are only partly understood, both theoretically and experimentally.

1.3

Outline of this thesis

In this thesis, the relation between the microscopic and macroscopic properties of magnetic

11

Chapter 1

fluids is investigated. The subject is dealt with experimentally as well as theoretically. This research on dipolar colloids is an extention of the research previously done at the Van 't Hoff Laboratory on other model colloids, such as spherical [21], rod-like [22] and plate-like [23] colloids. Chapter 2 describes the synthesis of several types of magnetic colloids, all based on 10-nm magnetite particles. The interaction strength between these colloids ranges from very weak to very strong. Weakly interacting colloids are prepared by covering the magnetic core by a nonmagnetic shell. Magnetizable colloids that interact strongly in a magnetic field are obtained by making composite colloids, consisting of large silica spheres covered by many small magnetite particles. In chapter 3, several theories describing the magnetic susceptibility of dipolar hard spheres, a model for magnetic colloids, are tested against measurements on oleic acid stabilized magnetite colloids in cyclohexane. The polydisperse ferrofluid is fractionated by means of partial precipitation, yielding a fraction of large particles with relatively strong dipolar interaction. The concentration dependence of the susceptibility of both the unfractionated and the fractionated ferrofluid deviates from all prevailing theories, though for some of them deviations are not large. Apart from magnetic interaction, the stability of ferrofluids is also influenced by non-magnetic constituents, such as an excess polymeric stabilizer. Chapters 4 to 6 deal with the phase behavior of magnetic fluids-polymer mixtures, where the non-adsorbing polymer acts as a destabilizing agent. In chapter 4, the phase diagram of magnetic colloid-polymer mixtures is treated theoretically by incorporating magnetic interactions into a well-known mean field theory for non-magnetic colloids-polymer mixtures. Calculations with this theory show that the interactions in common ferrofluids are too weak to explain the observed field-induced phase separation. The presence of free polymer, however, may account for the experimentally observed instabilities. Chapters 5 and 6 describe experiments on phase behavior of magnetic colloid-polymer mixtures. Phase instabilities are detected and characterized with a Colpitts oscillator, which is used as a local concentration meter. The principle of operation of this device and its application to the study of the phase behavior of magnetic colloids and polymers in absence of an external magnetic field is described in chapter 5. In chapter 6, the destabilizing influence of an external field is investigated. It is shown that, contrary to many experimental observations but in line with theoretical prediction, the ferrofluid investigated here is stable at any concentration and field strength. This may be due to the high quality (absence of clusters) of the investigated ferrofluid. In addition, the influence of an applied field on the stability of ferrofluid-polymer mixtures agrees quite nicely with the theory described in chapter 4. 12

General introduction

Most interactions between colloids are understood, although not always quantitatively known. For example, in addition to magnetic interaction magnetic colloids experience Van der Waals attraction, the strength of which is often uncertain. Sometimes, however, colloids exhibit unexpectedly strong interaction which can not be explained by the usual theories. In chapter 7, experimental evidence is presented for the presence of a strong, anomalous attraction between small magnetite colloids and large silica spheres, leading to irreversible adsorption of magnetite on silica. Textbook theories fail to describe the strength of the interaction and the very slow kinetics of the adsorption process. A proposed proton transfer between the magnetite and silica colloids is consistent with the experiments.

References
1 2 3 4 5 6 7 8 9 D. Jiles, Introduction to magnetism and magnetic materials, Chapman & Hall, London (1991). D. Craik, Magnetism, principles and applications, John Wiley & Sons, Chichester (1995). E. M. Purcell, Electricity and magnetism, McGraw-Hill, New York (1963). T. Lang, P. L. Moyland, D. A. Sergatskov, E. D. Adams, and Y. Takano, Phys. Rev. Lett. 77, 322 (1996). D. Platzek, C. Notthoff, D. M. Herlach, G. Jacobs, D. Herlach, and K. Maier, Appl. Phys. Lett. 65, 1723 (1994). R. E. Rosensweig, Ferrohydrodynamics, Cambridge University Press, Cambridge (1985). Magnetic fluids and applications handbook, ed. B. Berkovski and V. Bashtovoy, Begel House Inc., New York (1996). E. Blums, A. Cebers, and M. M. Maiorov, Magnetic fluids, Walter de Gruyter, Berlin (1997). W. C. Elmore, Phys. Rev. 54, 309 (1938). 10 S. S. Papell, U.S. Patent 3,215,572 (1965). 11 R. E. Rosensweig, J. W. Nestor, and R. S. Timmins, in Mater. Assoc. Direct Energy Convers. Proc. Symp. AIChE-I. Chem. Eng. Ser. 5 (1965), p. 104. 12 J. Shimoiizaka, K. Nakatsuka, T. Fujita, and A. Kounosu, IEEE Trans. Magn. MAG16, 368 (1980). 13 R. E. Rosensweig and J. Popplewell, in Electromagnetic Forces and Applications, ed. J. Tani and T. Takagi, Elsevier Science Publishers, Amsterdam (1992), p. 83.

13

Chapter 1

14 S. Taketomi, U.S. Patent 4,812,767 (1989). 15 H.-E. Horng, C.-Y. Hong, W. B. Yeung, and H.-C. Yang, Applied Optics 37, 2674 (1998). 16 P. G. De Gennes and P. A. Pincus, Phys. Cond. Matter 11, 189 (1970). 17 A. O. Tsebers, Magnetohydrodynamics 18, 137 (1982). 18 K. Sano and M. Doi, J. Phys. Soc. Jpn. 52, 2810 (1983). 19 L. Onsager, J. Am. Chem. Soc. 58, 1486 (1936). 20 P. I. C. Teixeira, J. M. Tavares, and M. M. Telo da Gama, J. Phys. 12, R411 (2000). 21 A. K. van Helden and A. Vrij, J. Colloid Interface Sci. 78, 312 (1980). 22 P. A. Buining, C. Pathmamanoharan, J. B. H. Jansen, and H. N. W. Lekkerkerker, J. Am. Ceram. Soc. 74, 1303 (1991). 23 F. M. van der Kooij, E. Kassapidou, and H. N. W. Lekkerkerker, Nature 406, 868 (2000).

14