Vol. 97 (2000) ACTA PHYSICA POLONICA A No.

ULTRASONIC WAVES IN MAGNETIC LIQUIDS:

TWO-PHASE APPROACH
AND EXPERIMENTAL RESULTS
Μ. KACZMAREK α T. HORNOWSKIb, A. SKUMIELb AND M. ŁABOWSKIb
,

α'Department of Environmental Mechanics, Pedagogical University in Bydgoszcz

Chodkiewicza 30, 85-064 Bydgoszcz, Poland
b lnstitute of Acoustics, Adam Mickiewicz University
Umultowska 85, 61-614 Poznań, Poland

(Received September 6' 1999; revised version November 29' 1999;

in final form February 2, 2000)

A two-phase model of dynamical behavior of magnetic liquid influenced

by constant magnetic field is considered. The model takes into account
mutual interaction of clusters giving skeleton's stiffness and interaction of
clusters with free liquid producing attenuation and dispersion of waves for
transversally isotropic system. Rotational degrees of freedom are disregarded.
Predictions of the model are discussed for variable amount of clusters and
compared with experimental data for different magnetic fields.
PACS numbers: 47.35.-+i, 75.50.-y, 62.60.+v

1. Introduction
In the past few decades an increase in technological importance as well as sig-
nificant interest in theoretical and experimental investigations of magnetic liquids
are observed. Magnetic liquids when not subjected to magnetic field are homo-
geneous colloidal suspensions of ferromagnetic particles in carrier liquids, such as
water, oil or kerosene. Typically, the diameters of solid particles range from 5
to 10 nanometers. To prevent coagulation of the particles they are coated with
surface-active dispersive medium. However, experimental investigations of mag-
netic liquids in DC magnetic field show that a certain amount of colloidal particles
forms aggregates (clusters) which tend to join into a chain-like structure as long as
hundreds of nanometers [1]. This gel-like microstructure causes an increase in stiff-
ness of the liquid, induces anisotropy of its mechanical properties, and contributes
to the dissipation of energy of internal processes due to interaction of clusters with
the surrounding liquid.
A useful way to study properties of magnetic liquids is based on applica-
tion of ultrasonic methods. These methods proved to be very powerful to observe
anisotropy of mechanical properties of the liquids in variable magnetic field and

(639)
640 M. Kaczmarek et a1.

a wide range of temperature. They can be also used to study the influence of
an internal structure of magnetic liquid constituted by clusters on macroscopic
rheological characteristics of the liquids.
Numerous models of ultrasonic wave propagation in magnetic liquids have
been proposed, see e.g. [2-6]. They are based on magnetohydrodynamics, theory
of liquid crystals or theory of mixtures. However, most of them fail to predict
properly some of the observed features of the liquids such as anisotropy of attenu-
ation and phase velocity of ultrasonic waves, and dependence of wave propagation
parameters on strength of magnetic field and temperature. The promising model
which offers a description of characteristic features of attenuation of mechanical
waves is given by Taketomi [4]. This model takes into account the presence and
motion of clusters in magnetic liquid and in particular the mutual interaction be-
tween different clusters and the interaction between clusters and free fluid. Since
the Taketomi approach to treat the magnetic liquid as a mixture is based solely
on an analysis of dissipation of energy of the system it is limited to the study of
attenuation and in fact cannot predict dispersion of waves as well as other conse-
quences of the two-phase nature of magnetic liquid (e.g. the total number of wave
modes in the liquid that can propagate taking into account a given anisotropy and
stiffness of skeleton).
A more thorough two-phase approach founded on field equations for mass
and momentum of two-component mixture was applied for the description of ul-
trasonic waves in magnetic liquids by Lipkin [5], Gogosov et al. [6], and Bádescu
et al. [7]. This model, however, treats the magnetic liquid as a suspension in which
interactions between clusters and as a result the existence of a skeleton composed
of clusters having certain stiffness and anisotropy are neglected.
The purpose of the present paper is to develop a linear two-phase model of
wave propagation in magnetic liquid using the concept of superimposed continua.
The governing equations of the model are formulated using the theory of mul-
tiphase media, and particularly the theory of anisotropic fluid saturated porous
materials, e.g. [8, 9], and [10]. It is assumed that the two phases of the medium
are: the skeleton made of interacting chains of clusters and the liquid composed
of carrier liquid with free colloidal particles (particles not gathered into clusters).
The mutual interactions of clusters are represented through components of stress
tensor for the skeleton. The interactions of clusters with liquid are expressed as
the sum of viscous and inertial drag forces. Given the assumed transversal isotropy
of the medium induced by DC magnetic field the possible wave modes predicted
by the model in infinite medium are specified. Quantitative results obtained from
the model for a fast quasi-longitudinal wave and variable amount of interacting
clusters are discussed. The numerical predictions are compared with experimental
data.

2. Model description

Following the introductory assumptions the magnetic liquid is modeled as

consisting of two phases: solid skeleton (s) made of interacting clusters and fluid (f)
composed of carrier liquid with free magnetic particles. The motion of the phases
 Ultrasonic Waves in Magnetic Liquids ... 641

is governed by the macroscopic linearized balances of linear momentum

where ρf and ps stand for the average mass densities of fluid and solid phase,
respectively, νf and i's are macroscopic velocities of the fluid and solid skeleton,
respectively, Tf and I5 denote stress tensors, and R f = — Rs are interaction forces
between the phases. All the above quantities are understood as averages over
volume regimes which are much smaller than the wavelength of the considered
waves.
The proposed model neglects rotational degrees of freedom of magnetic par-
ticles, the momentum interaction of the particles with carrier liquid and external
magnetic field and the mutual momentum interaction of the particles (antisymme-
try of the stress tensor). This assumption is justified if the direction of magnetic
field is parallel to the propagation vector of ultrasonic wave.
The stress tensors and interaction forces are assumed to be constitutive func-
tions that express material properties of the medium. The stress tensor in fluid
phase is split into two parts: hydrostatic and viscous components

where p is the average of microscopic hydrostatic pressure, n is the volume fraction

of clusters which constitute the skeleton and πij is the macroscopic viscous stress
tensor. Assuming that axis x 3 is perpendicular to the isotropy plane which includes
axes χ 1 and x 2 , the hydrostatic component of the stress tensor in fluid phase for
the transversally isotropic medium reads [9]

where ej denotes the strain tensor for the skeleton and e is the relative volume
change (dilatation) of fluid that can be expressed by displacements of skeleton, ui,
and fluid, of as following:
,

and

The coefficients M and Q relate deformations of solid to that of fluid while R is a

measure of the hydrostatic stress required to change the volume of fluid phase if the
volume of skeleton remains constant. The components of the macroscopic viscous
stress tensor in fluid phase are assumed to be dependent on rate of deformation of
fluid

and for a given symmetry are the following:

642 M. Kaczmarek et al.

where μ, α, γ, β, and δ are the viscosity constants.

The stress tensor for the skeleton expressing interaction between clusters
depends in general on deformation of skeleton and dilatation of fluid and its com-
ponents for transversally isotropic case take on the following form [9]:

where N, A, F, C, and L are the constants characterizing elastic response of the

skeleton.
The interaction force between carrier liquid and the skeleton is assumed
to be a sum of viscous (frictional) force and inertial force and the appropriate
components are the following:

where b t b n , c t , and c n are the coefficients characterizing viscous and inertial

,

couplings in a transversally isotropic medium. Taking into account the fact that
in this paper the waves in ultrasonic frequency range are considered, the above
specified interaction force does not include the component which represents the
history dependence of the interaction between phases (Basset force) [6, 10].
.

Combining the above introduced constitutive functions and equations of lin-

ear momentum for solid and fluid phase one can obtain the equations for macro-
scopic displacements of phases. In particular, the equations govern the propagation
of plane harmonic waves in an infinite medium for which the solutions take on the
following form: ui = ŭ-fi exp [iω(t — qjxj)], and usi = 1i exp [iω(t — qjxj)], where
,su,={fuai4r}etnmdp3l-s?f}o2acemntluid
and solid, ω is the angular frequency, and q = {q1, 0, q3} is the slowness vector
which is related to the complex wave vector, k, by the relation k = ωq. The com-
ponents of the slowness vector are chosen in such way that the plane of constant
phase is parallel to the axis x2. Plugging the solutions into the governing equations
 Ultrasonic Waves in Magnetic Liquids ... 643

of the two-phase model of magnetic liquid one obtains six homogeneous equations
with unknown amplitudes ί4 and u-si

where the coefficients Dij are the combinations of the material constants, angular
frequency, and components of the slowness vector for which the explicit forms
are given in Appendix. Since only the last two equations contain amplitudes i
andu-s2,the quationsareindepnde tof heother quations,andonecan otice
that the waves which are related to the motion of particles in χ 2 -directiοn are not
coupled with the waves with the motion of particles in the plane determined by
x1 and x3. The set of four homogeneous Eqs. (22-25) has a non-trivial solution
if the appropriate determinant of the matrix of coefficients, D I , is equal to zero,
D I = 0. This condition yields a dispersion equation corresponding to the waves
which are disturbances of displacement in the plane of x1 and x3. Similarly, the
uniqueness of the solution of Eqs. (26) and (27) demands that the corresponding
determinant is equal to zero, det D II = 0. This condition yields a dispersion
equation corresponding to the propagation of the waves with a motion of particles
in direction of x2.
Introducing into the dispersion equations a complex slowness parameter ά
such that q1 = ά sin θ, and q3 = d cos θ, where θ is the angle between the magnetic
field and the direction of wave propagation, one obtains the dispersion equations in
forms of polynomials in ά. The roots of the polynomials with positive values of real
components of ά enable us to determine the number of the wave modes predicted
by the model as well as the velocities and the coefficients of attenuation of the
waves. The model predicts four types of waves which are related to disturbances in
plane x1 , x3. Two of the modes are quasi-longitudinal waves for which the motion
of particles of fluid and solid phase take place in or out of phase. The remaining
waves are the shear-like waves resulting from the existence of the stiffness of solid
phase and viscosity of carrier liquid.

3. Experimental method
Measurements of absorption and velocity were carried out using the Matec
pulse echo-overlap technique (Fig. 1). The radio-frequency gated amplifier model
755 and gating modulator model 7700 were used to drive the piezo-ceramic trans-
ducer. The ultrasonic pulse, on traversal of the sample, was detected by the receiver
transducer and amplified in a wide-band amplifier. The resulting pulse-echo train
was observed on a CRT display. When the system was properly synchronized by
644 M. Kaczmarek et al.

the CW signal from the model 110 high resolution frequency source, the two cho-
sen echoes overlapped on the oscilloscope screen. By tuning the frequency of the
CW source, the correct, "cycle-to-cycle", match between the two echoes can be
established. Then the velocity of the ultrasonic wave can be evaluated from the
expression

where 1is the distance between the transducers and f — the frequency of the CW
source equal to the reciprocal of the round trip time in the sample.

The model 2470B automatic attenuation recorder measures the logarithmic

difference (in dB) between two selected echoes, say Α 1 and Α2, if the time gates
correctly cover the main portion of the echoes. When Α1 and Α2 are two consec-
utive echoes the log(Α1/Α2 ) output is proportional to the absorption coefficient
of the sample. The variations of log(Α 1/Α2 ), e.g. as a function of the magnetic
field strength, give direct changes in the absorption coefficient. The accuracy of
our determination of the velocity was of the order of 0.05% whereas changes in
the absorption coefficient were measured with an error less than 1%. The absolute
values of the absorption coefficient were accurate within ±5%. Α more detailed
description of the experimental setup can be found elsewhere [11].
 Ultrasonic Waves in Magnetic Liquids ... 645

In order to measure the effect of the magnetic field on the velocity and
attenuation of the ultrasonic wave, the measuring cell (made out of brass) with
ferrofluid inside was placed between pole pieces of electromagnet which yielded
a maximum field of l T at a spacing of 5 cm. The magnetic field induction was
measured to within 0.5% with a F.W. Bell gaussmeter model 9200.

4. DiscussiOn of results

Predictions of the proposed two-phase model of magnetic liquid with respect

to attenuation and wave velocity for variable volume fraction (stiffness) of skele-
ton are analyzed. The theoretical results are then compared with experimental
data for variable strength of magnetic field. The analysis is limited to the waves
which propagate along the magnetic field. In such case it is justified to disregard
rotational effects of clusters.
The measurements were carried out in a magnetic liquid denoted as ΕΜG-605
(produced by Ferrofluidics Inc.) consisting of magnetite particles Fe3O4 suspended
in water. The values of saturation magnetization, initial susceptibility, volume con-
centration, and viscosity were 20 mT, 0.5, 3.5%, and < 0.5 N s m -2 (at 25°C),
respectively. The particle distribution was a normal distribution with a maximum
at 100 Α. The measurements were carried out in 50°C at 1.18 MHz and the re-
sults are shown in Fig. 2. While the wave velocity increases monotonically with
strengthening of magnetic field the attenuation approaches a maximum value and
then drops to the level observed for the case of no magnetic field.

Fig. 2. Velocity and attenuation as a function of magnetic field strength H. Data for
velocity are denoted by ■ and for attenuation by •. The measurements were carried out
in the temperature of T = 50°C, and for the ultrasonic wave of frequency f = 1.18 MHz.
The direction of magnetic field was parallel, θ = 0°, to the ultrasonic wave propagation
vector.

In order to obtain numerical predictions of the proposed model a number of

assumptions for material parameters must be specified. It is assumed that the in-
crease in magnetic field causes an approximately proportional growth of a number
of clusters while the radius of clusters remains constant and equal to 1000 nm.
646 M. Kaczmarek et ad.

For the volume fraction of clusters lower than ~ 0.02 (this threshold value was
chosen arbitrarily) the interaction of existing clusters is assumed to be negligible
and there is no stiff or load bearing skeleton. When the volume fraction increases
above the threshold value, clusters start to interact stronger and the interaction is
modeled as proportional to the volume fraction of clusters. The interaction takes
place solely in direction χ 3 . There is no further increase in stiffness of skeleton when
all magnetic particles are gathered into clusters, which corresponds to magnetic
saturation (n approaches the maximum value). This pattern of cluster formation
is supported by experimental evidences (see e.g. [121). The assumed dependence of
the bulk modulus of skeleton Kb on the volume fraction of clusters with smoothed
breaking points is shown in Fig. 3.

Taking into account the fact that in the case of magnetic liquid the bulk
modulus of solid material, K s , is much greater than the bulk modulus of fluid
phase, Kf, and both parameters are a few orders of magnitude larger than the
bulk modulus of skeleton of magnetic liquid, Kb i.e. K s » Kf »Kb, and using the
,

relationship between macroscopic parameters of the two-phase model of saturated

porous media and the moduli of phases, see [13], for a given symmetry of the
material one can assume that elastic parameters of the model are as following:

The mass density and bulk modulus of fluid phase are calculated for the
series model taking into account the presence of free colloidal particles in fluid.
All the coefficients, which determine components of viscous stress tensor in fluid,
are assumed to have the same values irrespective to anisotropy and to get the
level of the observed attenuation they are taken to be thirteen times larger than
the viscosity of carrier fluid ζ. The inertial coupling in the model is disregarded
assuming that c t = c n = 0. The components of the viscous drag are calculated
 Ultrasonic Waves in Magnetic Liquids ... 647

assuming that the viscous interaction of a single spherical cluster of radius r,

moving with velocity vs in fluid which moves with the velocity of , is defined by
the Stokes formula

Taking into account that a representation of the Stokes force in the continuum
model is obtained by referring the above force to a representative volume of the
fluid containing a single cluster V and that the volume fraction of clusters is given
by n = 4πr 3 /3V, the coefficients of viscous coupling are

where the negative sign results from the applied notation in Eqs. (1) and (19-20).
In Fig. 4 the predicted by the two-phase model values of attenuation and
velocity of the fast quasi-longitudinal wave propagating in n 3 -direction versus the
volume fraction of skeleton (clusters) are shown. The exact relation between the
volume fraction of skeleton and magnetic field is not yet known. However, there
are some evidences [14] that this relation is linear in some range of magnetic field
strength. Thus, it seems that the theoretical curve reproduces, at least qualita-
tively, the behavior of velocity and attenuation of ultrasonic waves in magnetic
liquids. The initial increase in attenuation due to a growing amount of suspended
in liquid freely moving clusters (clusters which do not interact between them-
selves) is stopped when clusters start to build up a stiff skeleton. The stiffness of
the skeleton limits a relative displacement of clusters with respect to surrounding
fluid and this is accompanied by a drop of attenuation with increasing volume
fraction of clusters (skeleton stiffness). The stiffness is responsible also for the
predicted increase in velocity of the ultrasonic wave.

Fig. 4. Predictions of two-phase model for velocity (solid line) and attenuation (dotted
line) of ultrasonic waves in magnetic liquid.

Although a further refinement of the model is necessary to include, e.g.

changes in viscosity of fluid phase and frequency dependence of interaction between
phases, one should notice from the comparison of the results given in Figs. 2 and 4
that there is good qualitative agreement of theoretical results and experimental
648 M. Kaczmarek et al.

data. It should be noted that the model predicts also the existence of the slow
quaSi-longitudinal wave. Its velocity — for the presumed material parameters —
is about ten times smaller, and the attenuation over thousand times larger in
comparison with the corresponding parameters for the fast wave. However, with
the current measuring setup we were unable to detect this wave experimentally.
An analysis of other predictions of the model in relation to the measurements to
evaluate, e.g. the effects of anisotropy of the material, and of the role of frequency
of ultrasonic waves is necessary and will be approached in the nearest future.
Appendix
The coefficients appearing in Eqs. (22-27) have the following form:
 Ultrasonic Waves in Magnetic Liquids ... 649

Acknowledgment
Supported by the Committee for Scientific Research grant No. 2Ρ03Β 07912.
References

[1] I. Nakatani, in: Proc. Int. Symp. on Hydrodynamics of Magnetic Liquids and its
Application' Sendai, 1997, p. 9.
[2] I.E. Tarapov, Magnetohydrodynamics 1, 1 (1972).
[3] J.D. Parsons, J. Phys. D 8, 1219 (1975).
[4] S. Taketomi, Jpn. J. Appl. Phys. 22, 1137 (1983).
[5] A.I. Lipkin, Magn. Gidrodin. No. 3, 25 (1985).
[6] W.W. Gogosow, W.W. Martynow, S.I. Curikow, G.A. Szaposznikowa, Magn.
Gidrodin. No. 2, 19 (1987).
[7] R. Bádescu, E. Luca, V. B^.descu, N. Rezlescu, J. Magn. Magn. Mater. 170, 235
(1997).
[8] R.M. Bowen, in: Theory of Mixtures, Ed. A.C. Erring, in series Continuum Physics,
Vol. 3, Academic Press, New York 1976, p. 1.
[9] M.D. Sharma, M.L. Gogna, J. Acoust. Soc. Am. 90, 1068 (1991).
[10] M. Kaczmarek, J. Kubik, Arch. Mech. 41, 769 (1989).
[11] A. Skumiel, M. Łabowski, T. Hornowski, J. Phys. D 30, 25 (1996).
[12] J. Cernák, P. Macko, J. Magn. Magn. Mater. 123, 107 (1983).
[13] M.A. Biot, D.G. Willis, J. Appl. Mech. 24, 594 (1957).
[14] A. Skumiel, T. Hornowski, Acoust. Lett. 20, 121 (1996).