Content uploaded by Jing Zhong
Author content
All content in this area was uploaded by Jing Zhong on Sep 29, 2019
Content may be subject to copyright.
1 / 5
Magnetic field orientation dependent dynamic susceptibility and Brownian
relaxation time of magnetic nanoparticles
Jing Zhong1,a), Niklas Lucht2, Birgit Hankiewicz2, Meinhard Schilling1, and Frank Ludwig1
1Institut für Elektrische Messtechnik und Grundlagen der Elektrotechnik, TU Braunschweig D-38106 Braunschweig, Germany
2Institut für Physikalische Chemie, Universität Hamburg D-20146 Hamburg, Germany
a)E-mail: j.zhong@tu-braunschweig.de
This paper investigates the dynamic ac susceptibility (ACS) and the Brownian
relaxation time of magnetic nanoparticles (MNPs) in dc magnetic fields with arbitrary
orientations with respect to the ac magnetic field. A CoFe2O4 MNP sample, dominated
by Brownian relaxation, is used to perform ACS measurements in an ac magnetic field
with constant amplitude of 0.2 mT (from 2 to 3000 Hz) and a superposed dc magnetic
field with amplitudes ranging from 0 to 5 mT. Experimental results indicate that the
ACS and Brownian relaxation time are significantly affected not only by the strength
but also by the orientation of the dc magnetic field. Moreover, a mathematical model is
proposed to analyze the ACS and Brownian relaxation time in dependence of the
orientation of the dc magnetic field, which extends the established models for parallel or
perpendicular to arbitrary-oriented dc magnetic fields. Experimental results indicate that
the good fitting between the experimental data (ACS and Brownian relaxation time) and
the proposed models demonstrates the feasibility of the proposed model for the
description of ACS and Brownian relaxation time in arbitrary-orientated ac and dc
magnetic fields.
Magnetic nanoparticles (MNPs) are of great promise
in biological and biomedical applications, such as
heaters in magnetic hyperthermia,1-4 markers in
biosensors,5 sensors in MNP thermometry6-8 and
contrast agents in magnetic particle imaging (MPI).9-14
In these applications, the dynamic MNP magnetization
and its spectra induced by ac and/or dc magnetic fields
significantly affect their performance. For instance, the
MNP dynamic magnetization is crucial to the heating
efficiency in magnetic hyperthermia15 while its spectra
play significant roles in temperature sensitivity when
using the spectra for thermometry16,17 and image
resolution/quality in MPI.12,18 In a sufficiently
high-frequency ac magnetic field, MNP relaxation may
worsen the systematic error19 in MNP thermometry
while it further blurs the MPI images.12,18 A
comprehensive study of the dynamic ac susceptibility
(ACS) and relaxation of MNPs in ac and/or dc
magnetic field is of great significance and interest to
biomedicine and biology.
Brownian relaxation depends not only on the
hydrodynamic size, viscosity and temperature but also
on external excitation magnetic fields. It is well-known
that an increase in the amplitude of an ac magnetic
field as well as of superposed parallel/perpendicular dc
magnetic fields decrease the Brownian relaxation
time.20-23 Physical and/or phenomenological models
were proposed to describe the dependence of MNP
Brownian relaxation on ac and dc magnetic fields, as
well as the ACS spectra.20-22 However, regarding
superposed dc magnetic fields, only the parallel or
perpendicular orientation with respect to the ac
magnetic field were taken into account. In some
specific applications, e.g. MPI, the ac and dc magnetic
fields have arbitrary orientations in the imaging field of
view. Understanding the MNP spectra and Brownian
relaxation in the arbitrary-orientated ac and dc
magnetic fields, as well as the underlying physical
mechanisms, is of great significance to viscosity and
molecule-binding imaging with MPI. To date, there
have not been any studies or physical models to
describe the MNP relaxation and ACS in
arbitrary-oriented dc magnetic fields.
In this paper, we investigate the ACS and the
Brownian relaxation time of a CoFe2O4 sample in
arbitrary-orientated ac and dc magnetic fields. ACS
spectra of a Brownian relaxation-dominated CoFe2O4
sample are measured whereas the Brownian relaxation
time is extracted from the ACS spectra. Moreover, a
mathematical model is proposed to describe the
experimental results.
In a small-amplitude ac magnetic field Hac without a
superposed dc magnetic field Hdc, the Debye model is
used to describe the dynamic ACS24
B
i
+
=1
0
, (1)
where
0 is the static value calculated from the static
Langevin function,
is the angular frequency and
B is
the Brownian relaxation time. Note that at infinite
frequencies some MNPs with relaxation times below 1
s and/or intra-potential-well contributions cause the
dynamic ACS not drop to zero but have a constant
value.25
This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.
PLEASE CITE THIS ARTICLE AS DOI: 10.1063/1.5120609
2 / 5
A superposed Hdc significantly affects the ac
susceptibility and the effective Brownian relaxation
time. A physical model for the description of the
dynamic ACS
para (
perp) in a parallel (perpendicular)
Hdc is given by20
( ) ( ) ( )
para
para i
L
+
=1
1
30
, (2a)
( ) ( ) ( )
perp
perp id
dL
+
=1
1
30
, (2b)
where
para (
perp) is the ac susceptibility in a parallel
(perpendicular) bias dc magnetic field.
para (
perp) is the
effective Brownian relaxation time in a superposed Hdc
parallel (perpendicular) to the small-amplitude ac
magnetic field. L(
) is the static Langevin function,
=
0mHdc/kBT, m is the MNP magnetic moment, kB is
Boltzmann constant and T is the absolute temperature.
Herein,
para and
perp are given by19
( ) ( )
( )
Bpara d
Ld
ln
ln
=
, (3a)
( ) ( )
( )
Bperp L
L
−
=2
, (3b)
where
B = 3
Vh/kBT,
is the medium viscosity and Vh
is the hydrodynamic volume of the MNPs.
To investigate the dependence of ac susceptibility
and effective Brownian relaxation time of MNPs on
the magnetic field orientation, a CoFe2O4 MNP sample
suspended in water and glycerin (glycerin weight
percentage: 80%) is used for experiments. The
CoFe2O4 MNPs have a nominal core diameter of 15.5
nm and a hydrodynamic diameter of 38.6 nm.26 The
viscosity of the MNP suspension amounts to about
34.8 mPa· s from theoretical calculation.27 The MNPs
are dominated by Brownian relaxation due to the very
high anisotropy energy barrier of CoFe2O4. A
rotating-magnetic-field (RMF) system with
two-dimension Helmholtz coils and a fluxgate-based
measurement system is used to generate
arbitrary-orientation Hac and Hdc with different
frequencies and measure the ACS spectra. The details
of the RMF system can be found in Ref. [28]. A
2-dimensional Helmholtz coil system is used to
generate the arbitrary-orientated ac and dc magnetic
fields in one plane while a fluxgate based measurement
system measures the MNP response in the direction
perpendicular to the plane of the applied magnetic
fields. The ac magnetic field with constant amplitude
of 0.2 mT is applied with a frequency range from 2 Hz
to 3000 Hz.
Fig. 1 shows the measured ACS spectra (real
’ and
imaginary
’’ parts) at room temperature without a
superposed Hdc.
’’ shows a peak frequency at about
128 Hz, which can be used to determine the effective
Brownian relaxation time. To accurately determine the
Brownian relaxation time from the spectra, the
measured ACS spectra are fitted with the
phenomenological Havriliak-Negami model22,29
( )
( )
−
+
+= 1
inf
1B
amp
i
, (4)
where
inf is the measured susceptibility at a very high
frequency,
amp is the amplitude of the
frequency-dependent ACS,
inf +
amp is the DC
susceptibility.
(
) affects the width (asymmetry) of
the imaginary parts of the ac susceptibility. In this
paper, Eq. (4) is only used to fit the ACS spectra at
different dc fields and angles to accurately obtain the
characteristic Brownian relaxation time. The fits of the
experimental spectra with Eq. (4) are presented as
solid/dashed lines in Fig. 1. The effective Brownian
relaxation time
B (about 1.32 ms for the given case) is
determined from the peak frequency of
’’.
Fig. 1. Ac susceptibility spectra of the CoFe2O4 MNP sample.
Symbols are experimental data whereas solid/dashed lines are fits
with Eq. (4).
Fig. 2. Real and imaginary parts vs. angle
at different frequencies.
Symbols are experimental data whereas solid/dashed lines are guides
to the eye.
Due to the symmetry of the Hac and Hdc orientations
in the four quadrants, the ACS of the MNPs in the first
quadrant should be the same as those in the other three
quadrants. To verify this, the ACS, in dependence of
the angle
between Hac and Hdc, are measured at three
different frequencies in all four quadrants with Hdc = 3
mT, as shown in Fig. 2. It demonstrates the symmetry
of the ACS across the four quadrants. In addition, the
values of the ACS (
’ and
’’) in perpendicular Hdc are
greater than those in parallel Hdc. We have performed
simulations with Eqs. (2) and (3) to calculate the ACS
in parallel and perpendicular Hdc, which shows the
same behaviour (simulation results are not presented
here). Therefore, due to the symmetry in four
quadrants, the experiments on the ACS measurements
in this paper are only performed for
arbitrary-orientation Hac and Hdc in the first quadrant.
The superposed Hdc was varied from 1 mT to 5 mT
with a step of 1 mT while the angle
between Hac and
Hdc varies from 0° to 90° with a step of 10°. Fig. 3
shows the experimental results of parallel and
perpendicular imaginary parts
’’ in different-strength
Hdc. The maximum value of the parallel and
perpendicular
’’ decreases while the peak frequency
shifts to higher frequencies (the Brownian relaxation
time decreases) with increasing Hdc. Qualitatively, the
experimental results agree with Eqs. (2a) and (2b), as
well as Eqs. (3a) and (3b). We have also tried to fit the
ACS spectra in different Hdc with Eqs. (2a) and (2b) by
considering a lognormal size distribution of the core
This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.
PLEASE CITE THIS ARTICLE AS DOI: 10.1063/1.5120609
3 / 5
and hydrodynamic MNP diameters. However, the
fitting does not work very well, which might be caused
by some clusters in the MNP sample. Previous studies
have already demonstrated that the used MNP sample
contains some trimers, causing the fitting of the ACS
spectra to be more complicated. In addition, the trimers
may lead to a change in the effective magnetic moment
in different applied magnetic fields,26,30 which might
also be responsible for the comparably poor quality of
the fits. Nevertheless, it does not affect the study of the
dependent ACS and Brownian relaxation time.
Figure 4 shows the experimental curves of
’’
versus frequency at Hdc = 1 mT (Fig. 4a), 3 mT (Fig.
4b) and 5 mT (Fig. 4c). Fig. 4a indicates that – at Hdc =
1 mT – a decrease in
in the first quadrant shifts the
peak frequency of
’’ to a higher frequency and
decreases the amplitude of
’’. In addition, with
increasing Hdc, the peak frequency shifts to an even
higher frequency and the amplitude further decreases,
which qualitatively fits with Eqs. (2) and (3). Herein,
we propose a mathematical model to describe the
dependent susceptibility:
( )
+= 22 sincos, perppara
. (5)
The measured
para and
perp obtained from Fig. 3 are
used to calculate the ACS in arbitrary-oriented Hdc
with Eq. (5), the imaginary parts
’’ of which are
presented as dashed lines in Fig. 4. The good
agreement between the calculated and measured
’’
(coefficient of determination better than 0.997)
demonstrates the feasibility of Eq. (5) for the
description of
dependent
(
,
). The squares of
cos(
) and sin(
) in Eq. (5) serve to make
(
= 0,
)
equal to the Debye model – Eq. (1) with Hdc = 0 mT. In
addition, cos2(
) and sin2(
) can be converted to
cos(2
), meaning that the ACS vs.
curve has two
cycles in a
range from 0° to 360°, which reflects the
180° periodicity. It fits very well with our measured
data as shown in Fig. 2. Note that in Eq. (5) the
parameter
is calculated from the absolute value of the
applied dc magnetic field, i.e., not from the projection
of the dc magnetic field in either parallel or
perpendicular direction.
Fig. 3. Imaginary part versus frequency in parallel (a) and
perpendicular (b) dc magnetic fields. Symbols are experimental data
whereas solid lines are fits with Eq. (4).
To quantitatively investigate the dependence of the
effective Brownian relaxation time
B(
) on
,
’’
are fitted with Eq. (4) while the peak frequency is used
to calculate
B(
). Fig. 5 shows
B(
) versus
in
superposed Hdc with different strengths and
orientations. With
= 0,
para =
B(
= ) decreases
from 1.32 ms to 0.23 ms while
perp =
B(
= °)
decreases from 1.32 ms to 0.38 ms by increasing Hdc
from 0 mT to 5 mT. With Hdc ≠ 0 mT,
B(
)
decreases from
perp to
para with decreasing
in the
first quadrant. Herein, we propose a mathematical
model to describe the
dependent effective Brownian
relaxation time
B(
)
( )
+= 22 sincos, perpparaB
. (6)
With the measured
parp and
perp, one can calculate the
B(
) in arbitrary-oriented Hdc with Eq. (6), shown
as dashed lines in Fig. 5. It indicates that the calculated
Brownian relaxation time fits very well (the coefficient
of determination better than 0.984) with the
experimental results, which demonstrates the
feasibility of Eq. (6) for the description of
dependent
B(
).
Fig. 4. Imaginary part vs. frequency in different-orientation ac and
dc magnetic fields with dc magnetic field strengths of 1 mT (a), 3
mT (b) and 5 mT (c). Symbols are experimental results whereas
dashed lines are the calculated values with Eq. (5).
Fig. 5. Effective Brownian relaxation time
B(
) versus
curves
at different Hdc. Symbols are experimental results whereas lines are
fits with Eq. (6).
With the proposed mathematical models for the
description of the
dependent ACS and Brownian
relaxation time – Eqs. (5) and (6) – the physical
models – Eqs. (2) and (3) – can be extended to a more
general case, including Hdc strength and orientation.
Inserting Eqs. (4a) and (4b) into Eq. (6), the effective
Brownian relaxation time
B(
) is given by
( ) ( )
( ) ( )
( )
BB L
L
d
Ld
−
+= 22 sin
2
cos
ln
ln
,
. (7)
Without a superposed Hdc (
= 0), Eq. (7) simplifies to
B(0,
) = (cos2
+ sin2
)·
B
=
B. With a parallel
superposed Hdc (
≠ 0), Eq. (3a) is obtained for the
case of a parallel Hdc (
= 0 °) while Eq. (3b) is
obtained from Eq. (7) for a perpendicular Hdc (
=
90°). Furthermore, by inserting Eqs. (2a) and (2b) into
Eq. (5),
(
,
) is given by
( ) ( )
( )
=+
+
=+
=
2
0
2
0
sin
)90,(1
1
3
cos
)0,(1
1
3,
B
B
i
L
i
d
dL
. (8)
Without a superposed Hdc (
= 0), Eq. (7) simplifies to
the Debye model – Eq. (1). With a superposed Hdc (
≠
0), Eq. (2a) is obtained for a parallel Hdc (
= 0 °)
while Eq. (2b) is obtained from Eq. (8) for a
perpendicular Hdc (
= 90 °). Therefore, the proposed
models – Eqs. (7) and (8) – are feasible to describe the
Hdc and
dependent Brownian relaxation time and the
regarding ACS spectra.
Note that Eqs. (7) and (8) are the combination of the
original model Eq. (2) and our proposed models Eqs.
This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.
PLEASE CITE THIS ARTICLE AS DOI: 10.1063/1.5120609
4 / 5
(5) and (6). The original model Eq. (2) for the
established cases of parallel and perpendicular dc
magnetic fields does not fit the experimental data very
well even if core and hydrodynamic size distributions
with a single set of parameters, e.g. mean and standard
deviation, are included. We speculate that it is caused
by the presence of small MNP cluster (mostly trimers),
leading to causing magnetic field dependent magnetic
moments. This results in the poor agreement of the
experimental data with Eqs. (7) and (8) although the
proposed models Eqs. (5) and (6) fit the experimental
data very well. Thus, the main contributions of this
paper – the proposed mathematical models for the
description of the magnetic field orientation dependent
ac susceptibility (Eq. 5) and Brownian relaxation time
(Eq. 6) – have been validated. We emphasize that the
proposed mathematical models are of great interest for
MPI, as well as corresponding biomedical applications.
With the extended model, one can simulate the ACS of
Brownian relaxation dominated MNPs and Brownian
relaxation time, in arbitrary-orientated Hdc superposed
to a small-amplitude Hac. In addition, taking into
account the phenomenological model for the
description of Hac dependent Brownian relaxation
time21,22 and the effective-field approaches,31,32 the
spectra of MNP magnetization induced in
arbitrary-orientated Hdc superposed to a
large-amplitude Hdc can also be simulated, which is
generally interesting to MPI related researches.
In conclusion, we investigated the ac susceptibility
and Brownian relaxation time of MNPs in
different-orientated dc and ac magnetic fields. The ac
susceptibility of a CoFe2O4 MNP sample was
measured in arbitrary-orientated dc and ac magnetic
fields while the Brownian relaxation time was
calculated from the peak frequency of the imaginary
parts. Experimental results indicate that an increase in
the dc magnetic field strength decreases the Brownian
relaxation time while increasing the angle between the
dc and ac magnetic fields in the first quadrant increases
the Brownian relaxation time in the given range of dc
magnetic field strengths. Moreover, mathematical
models were proposed to describe the orientation of the
dc magnetic field dependent ac susceptibility and the
Brownian relaxation time. We envisage that the
proposed mathematical models are of great importance
and interest to MPI related research fields.
The financial support from the German Research
Foundation DFG (Project No.: ZH 782/1-1) and via
SPP1681 (LU 800/4-3 and FI 1235/2-2) is gratefully
acknowledged.
REFERENCES
1M. Johannsen, U. Gneveckow, L. Eckelt, A. Feussner,
N. Waldöfner, R. Scholz, S. Deger, P. Wust, S.
Loening, A. Jordan, Int. J. Hyperthermia 21, 637
(2005).
2Q. A. Pankhurst, J. Connolly, S. K. Jones and J.
Dobson, J. Phys. D-Appl. Phys. 36, R167 (2003).
3R. Hergt, S. Dutz, R. Muller and M. Zeisberger, J.
Phys.: Condens. Matter 18, S2919 (2006).
4E. A. Perigo, G. Hemery, O. Sandre, D. Ortega, E.
Garaio, F. Plazaola, and F. J. Teran, Appl. Phys. Rev.
2, 041302 (2015).
5S. Schrittwieser, B. Pelaz, W. J. Parak, S.
Lentijo-Mozo, K. Soulantica, J. Dieckhoff, F. Ludwig,
A. Guenther, A. Tschöpe, J. Schotter, Sensors 16, 828
(2016).
6L. He, W. Liu, Q. Xie, S. Pi, and P. Morais, Meas. Sci.
Technol. 27, 025901 (2015).
7M. Zhou, J. Zhong, W. Liu, Z. Du, Z. Huang, M.
Yang, and P. C. Morais, IEEE Trans. Magn. 51,
6101006 (2015).
8J. Zhong, M. Schilling, and F. Ludwig, Meas. Sci.
Technol. 29, 115903 (2018).
9B. Gleich and J. Weizenecker, Nature 435, 1214
(2005).
10T. Knopp, N. Gdaniec, and M. Moddel, Phys. Med.
Biol., 62, R124 (2017).
11M. Schilling, F. Ludwig, C. Kuhlmann and T.
Wawrzik, Biomed. Tech. 58, 557 (2013).
12S. Pi, W. Liu and T. Jiang, Meas. Sci. Technol. 29,
035402 (2018).
13T. Jiang, S. Pi, J. Shi, P. Zhang, W. Liu, and J. Cheng,
Meas. Sci. Technol. In press, doi:
10.1088/1361-6501/ab2d40.
14T. Knopp, S. Biederer, T. F. Sattel, M. Erbe, and T.
M. Buzug, IEEE Trans. Med. Imaging, 30(6),
1284-1292, 2011
15A. E. Deatsch, and B. A. Evans, J. Magn. Magn.
Mater. 354, 163 (2014).
16J. Zhong, J. Dieckhoff, M. Schilling and F. Ludwig, J.
Appl. Phys. 120, 143902 (2016).
17J. Zhong, M. Schilling and F. Ludwig, J. Magn.
Magn. Mater. 471, 340 (2019).
18L. R. Croft, P. W. Goodwill, and S. M. Conolly,
IEEE Trans. Med. Imag. 31, 2335 (2012).
19J. Zhong, M. Schilling and F. Ludwig, J. Phys. D:
Appl. Phys. 51, 015001 (2018).
20Y. L. Raikher and M. I. Shliomis, Adv. Chem. Phys.
87, 595 (2007).
21T. Yoshida and K. Enpuku, Jpn. J. Appl. Phys. 48,
127002 (2009).
22J. Dieckhoff, D. Eberbeck, M. Schilling, and F.
Ludwig, J. Appl. Phys. 119, 043903 (2016).
23M. A. Martens, R. J. Deissler, Y. Wu, L. Bauer, Z.
Yao, and R. Brown, Med. Phys. 40, 022303 (2013).
24R. E. Rosensweig, J. Magn. Magn. Mater. 252, 370
(2002).
25F. Ludwig, C. Balceris, C. Jonasson, and C.
This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.
PLEASE CITE THIS ARTICLE AS DOI: 10.1063/1.5120609
5 / 5
Johansson, IEEE Trans. Magn. 53, 6100904 (2017).
26S. Draack, N. Lucht, H. Remmer, M. Martens, B.
Fischer, M. Schilling, F. Ludwig, and T. Viereck, J.
Phys. Chem. C 123, 6787 (2019).
27N.-S. Cheng, Ind. Eng. Chem. Res. 47, 3285 (2008).
28J. Dieckhoff, M. Schilling, and F. Ludwig, Appl.
Phys. Lett. 99, 112501 (2011).
29S. Havriliak and S. Negami, Polymer 8, 161 (1967).
30J. Landers, S. Salamon, H. Remmer, F. Ludwig, and
H. Wende, ACS Appl. Mater. Interfaces 11, 3160
(2019).
31T. Wawrzik, T. Yoshida, M. Schilling, and F.
Ludwig, IEEE Trans. Magn. 51, 5300404 (2015).
32R. M. Ferguson, K. R. Minard, A. P. Khandhar, K. M.
Krishnan, Med. Phys. 38, 1619 (2011).
This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.
PLEASE CITE THIS ARTICLE AS DOI: 10.1063/1.5120609
This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.
PLEASE CITE THIS ARTICLE AS DOI: 10.1063/1.5120609
This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.
PLEASE CITE THIS ARTICLE AS DOI: 10.1063/1.5120609
This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.
PLEASE CITE THIS ARTICLE AS DOI: 10.1063/1.5120609
This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.
PLEASE CITE THIS ARTICLE AS DOI: 10.1063/1.5120609
This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.
PLEASE CITE THIS ARTICLE AS DOI: 10.1063/1.5120609