Ferrofluids of magnetic multicore nanoparticles for biomedical applications

Article abstract of DOI:10.1016/j.jmmm.2009.02.073

For a variety of magnetically based biomedical applications, it is advantageous to use sedimentation stable suspensions of relatively large (d>20 nm) magnetic core-shell nanoparticles. Water-based suspensions of multicore nanoparticles were prepared by co

Full text of DOI:10.1016/j.jmmm.2009.02.073

ARTICLE IN PRESS
Journal of Magnetism and Magnetic Materials 321 (2009) 1501–1504
Contents lists available at ScienceDirect
Journal of Magnetism and Magnetic Materials
journal homepage: www.elsevier.com/locate/jmmm
Ferrofluids of magnetic multicore nanoparticles for biomedical applications
Ã
Silvio Dutz a, , Joachim H. Clement b, Dietmar Eberbeck c, Thorsten Gelbrich d, Rudolf Hergt a,
Robert Mu¨ller a, Jana Wotschadlo b, Matthias Zeisberger a
a Institute of Photonic Technology, Albert-Einstein-Strasse 9, 07745 Jena, Germany
b Department Internal Medicine II, University Clinic Jena, Erlanger Allee 101, 07740 Jena, Germany
c Physikalisch-Technische Bundesanstalt, Abbestrasse 2-12, 10587 Berlin, Germany
d
¨
¨
Institute of Organic Chemistry and Macromolecular Chemistry, Heinrich Heine University, Universitatsstrasse 1, 40225 Dusseldorf, Germany
a r t i c l e i n f o
a b s t r a c t
Available online 21 February 2009
For a variety of magnetically based biomedical applications, it is advantageous to use sedimentation
stable suspensions of relatively large (d420 nm) magnetic core–shell nanoparticles. Water-based
suspensions of multicore nanoparticles were prepared by coating of the particles (synthesized by means
Keywords:
Cellular uptake
Fractionation
of
a modified alkaline precipitation method) with a carboxymethyldextran shell. The resulting
Hysteresis
ferrofluids were structurally and magnetically characterized. It was found that these fluids show a
specific heating power of about 60 W/g (f ¼ 400 kHz, H ¼ 10 kA/m). This value was increased up to
330 W/g by a simple fractionation method based on centrifugation. Finally, the cellular uptake of the
multicore nanoparticles was demonstrated.
Magnetic loss
Magnetic nanoparticle
Magnetite
Multicore particle
Specific heating power
& 2009 Elsevier B.V. All rights reserved.
0. Introduction
on centrifugation. By this way, for some fractions the SHP
could be increased to a higher value than the SHP of the original
fluid.
Magnetic core–shell nanoparticles are of growing importance
Magnetic nanoparticles (MNP) are a promising tool for a wide
spectrum of magnetically based biomedical applications. For the
majority of applications (e.g. hyperthermia, cell separation, and
drug targeting), it is advantageous to use aqueous suspensions of
relatively large (d420 nm) magnetic nanoparticles. Unfortunately,
MNP of this size show a tendency to form aggregates which
results in the sedimentation of the ferrofluid. To solve this
problem, so-called multicore nanoparticles (MCNP) were synthe-
sized by a modified alkaline precipitation method. Water-based
suspensions of these particles were prepared by coating of the
MCNP with a carboxymethyldextran (CMD) shell. In a previous
work [1] only a weak influence of the CMD shell on the magnetic
properties of the coated particles was found. For reasons of
biocompatibility only iron-oxide-based MCNP were prepared.
For the medical heating applications of MNP (hyperthermia),
the applicable amplitude of the alternating magnetic field is
limited to a value of about 15 kA/m due to medical and technical
restrictions [2,3]. Therefore, MNP with a comparatively high
specific heating power (SHP) at a relatively low field amplitude
are needed. It is well known that the SHP is a function of the mean
particle diameter of the MNP as well as of the size-distribution
width [4,5]. Fractions of different particle diameters were
extracted from the prepared fluid by a separation method based
in labelling and separation of cells. Nanoparticles attached to
antibodies which are directed against cell surface markers are
widely used for magnetic cell separation and allow a specific and
high enrichment of the target cells [6,7]. On the other hand, the
nanoparticle shell itself could be used for cell-specific interaction
[8–10]. Depending on the size of the particles as well as the nature
of the shell MNP are attached to the cells or incorporated [11,12].
Small-sized MNP interact with human cells in a cell-type and
time-dependant manner, allowing the precise separation of tumor
cells from leukocytes [9].
In the present paper, the preparation route of the MCNP is
described and structural and magnetic properties of the ferrofluid
and the single particles are shown. A fractionation method based
on centrifugation as a tool for the increasing of the SHP is
investigated. First results for the cellular uptake of the CMD-
coated MCNP in tumor cells are given.
1. Methods
1.1. Preparation and fractionation of the ferrofluids
The particles introduced in this paper were prepared similar to
the well-known wet chemical precipitation methods [13,14], but
using another alkaline medium and a slower reaction velocity.
Ã
Corresponding author. Tel.: +49 3641206136; fax: +49 3641206288.
E-mail address: silvio.dutz@ipht-jena.de (S. Dutz).
0304-8853/$ - see front matter & 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.jmmm.2009.02.073

ARTICLE IN PRESS
1502
S. Dutz et al. / Journal of Magnetism and Magnetic Materials 321 (2009) 1501–1504
In detail, a 1 M NaHCO3 solution was slowly added under
sample (iron(III)oxide, gamma, 99+%, AlfaAesar) with high SHL
determined in earlier investigations and magnetosomes from
magnetotactic bacteria—up till now the particles with the highest
SHL at relatively low field amplitudes [15]. The specific heating
power was measured by means of magnetic field calorimetry
(MFC) at a field amplitude of 10 kA/m and a frequency of 400 kHz.
This parameter combination is suitable for hyperthermia treat-
ment [16].
The Brownian relaxation behavior of the MCNP in the fluid was
investigated by magneto relaxometry (MRX) [17]. For these
measurements, the samples were diluted by de-ionized water
(dilution factors from 1:9 to 1:900). A magnetizing field of
permanent stirring to a FeCl2/FeCl3 solution (total Fe-concentra-
tion: 0.625 M; Fe2+/Fe3+ ratio ¼ 1/1.3) with a rate of 0.75 ml/min.
This procedure was stopped when the pH value reached 8. During
this routine, a brownish precipitate was formed. This precipitate
was heated to 100 1C for 5 min and iron oxides with a spinel
structure were formed under the release of CO2. To remove excess
reaction products from the prepared particles, they were washed
with de-ionized water three times.
For the production of sedimentation stable suspensions of the
MCNP, the particles were coated with a CMD shell. After washing
of the particles the pH value of the dispersed particles was
adjusted to pH 2–3 by the addition of diluted HCl. Then the
suspension was homogenized by ultrasonic treatment for a few
seconds (Sonopuls GM200, Bandelin electronic) and then temp-
erated at 45 1C. An aqueous solution of CMD (initial material: CMD
sodium salt, Fluka) with an CMD/MCNP ratio of about 13 was added
to the suspension and stirred for 60 min at 45 1C. Finally, the
coated particles were washed with de-ionized water to remove
the remaining salts.
For the fractionation of the initial MCNP fluid into fractions of
different mean particle sizes, 30 ml of this fluid were filled in a
cylindrically shaped centrifugation vessel made of glass (sample
height E70 mm and diameter E20 mm). The sample was
centrifuged in a laboratory centrifuge (Cryofuge 6000, Heraeus
Sepatech) at 1000g with a temperature of 20 1C. The sediment was
stored and the supernatant was recovered. A portion of the
supernatant was centrifuged again with 1500g. This procedure
was repeated twice with increasing centrifugal accelerations
(2500g, 3000g). In consequence, 4 sediments and 4 supernatants
were obtained representing 8 fractions of the MCNP.
H ¼ 1300 A/m is applied for t ¼ 1 s. A total of 450
ms after
switching off the field, a highly sensitive low-TC-SQUID sensor
measures the magnetic induction B(t) at a distance of 10 mm
above the sample. The measurement time window is
450
msptp0.45 s. From these relaxation curves, the size distribu-
tion of hydrodynamic diameters of the CMD-coated particles was
calculated by fitting the so-called cluster moment superposition
model (CMSM) to the relaxation data [18]. The CMSM describes
the relaxation of the magnetic moment of an ensemble of
particulate magnetic entities, the magnetic moment of which
´
can relax via the Brownian and Neel mechanism. The distribution
f(dh, ) of the hydrodynamic diameters dh with the median
diameter and the distribution parameter is assumed to be a
m, s
m
s
lognormal one. In previous investigations, it was empirically
found that the hydrodynamic diameter of the entities with the
mean volume is in good agreement with hydrodynamic diameter
obtained by PCS [17].
Furthermore, the MCNP were investigated by AC susceptome-
try (DynoMag, IMEGO) in the frequency range from 1 Hz to
200 kHz at a field amplitude of 0.4 kA/m. The hydrodynamic
diameters distribution was determined by fitting the complex
susceptibility data to a model that assumes Brownian relaxation
of the MCNP.
1.2. Investigation of cellular uptake
The interaction of the MCNP with human cells was studied
with the breast cancer cell line MCF-7. The cells were cultivated in
Dulbecco’s Modified Eagle Medium (DMEM) +10% fetal calf serum.
For incubation experiments, adherent cell culture cells were
detached with a trypsin/EDTA solution and inoculated with MCNP
in PBS/EDTA (2 mmol) for the times indicated. After the treatment,
the magnetically labelled cells were separated by MACS
(SuperMACS and MS-columns, Miltenyi-Biotec). The efflux was
designed as ‘negative fraction’ and the cells retained in the
column as ‘positive fraction’. Total cell numbers were estimated
by cell counting (Coulter Z2, Beckman-Coulter).
2. Results
It was found that during the precipitation primary particles
with a mean diameter of 14 nm (XRD) were formed. Phase
identification by XRD demonstrates that the particles consist of
solid solutions (or of a mix) of maghemite and magnetite. These
particles form clusters (Fig. 1a) of 40–80 nm (TEM) surrounded by
a CMD shell. Clearly, a size distribution of the cluster diameters
could be observed (Fig. 1b). The dynamics of the cluster formation
as well as their magnetization configuration are yet unknown. For
the hydrodynamic diameters of these clusters values of 158 nm
(PCS) and 160 nm (AC susceptometry) were determined. These
values are affected by a hydrate shell on the CMD layer.
1.3. Structural and magnetic characterization
Hydrodynamic diameters (dh) of the CMD-coated particles in
the ferrofluid were determined by using photon correlation
spectroscopy (PCS; HPPS-ET, Malvern Instruments). Dry samples
of the fluids were structurally characterized by field emission
scanning electron microscopy (FE-SEM; JSM 6300-F, JEOL),
transmission electron microscopy (TEM; JEM 2010FEF, JEOL) and
X-ray diffraction (XRD; X’pert-Twin diffractometer, Philips). The
mean sizes (d) of the magnetic cores were calculated from
measurements of the XRD line width by using the Scherrer
method.
For the investigation of the quasistatic parameters, the MCNP
in the fluid were immobilized by drying. Minor loops of the dry
particles were measured by vibrating sample magnetometry
(VSM; Micromag 3900, Princeton Measurement Corporation). By
integrating of the area of the measured minor loops, the specific
hysteresis losses (SHL) per cycle depending on the field amplitude
were calculated. The results were compared with a commercial
Fig. 1. Typical TEM images of a single cluster (a) and an ensemble of clusters (b).
The aggregation of the clusters in Fig. 1b occurs during the sample preparation
(drying of the fluid) for TEM imaging.

ARTICLE IN PRESS
S. Dutz et al. / Journal of Magnetism and Magnetic Materials 321 (2009) 1501–1504
1503
Immediately after the preparation of the fluid, a sediment
consisting of large aggregates is formed. After removal of this
sediment, the supernatant shows a high stability against sedi-
mentation over a period of several months.
Due to the size distribution of the clusters, the fractionation of
the original fluid based on centrifugation appeared to be useful.
Starting from a mean hydrodynamic diameter of 158 nm (PCS) for
the original fluid eight fractions with a size range from 50 to
164 nm (PCS) were obtained (Table 1). Because of the very low
particle concentration in the fraction with the smallest particles, it
was not possible to determine its magnetic parameters. This
fraction was excluded from further investigations.
In Fig. 2, the field amplitude dependence of SHL per cycle
measured by VSM is shown for immobilized particles of the
original fluid and the fraction with the highest calorimetrically
measured SHP in the interesting field amplitude range from 5 to
20 kA/m. Additionally the curves of magnetosomes [15] and a
commercial sample (Alfa Aesar) with high SHL are presented.
The SHP of the fluid particles was determined by the magnetic
field calorimetry measurements. The original fluid shows
a
moderate SHP of about 60 W/g (at 400 kHz and 11 kA/m). This
value was increased up to 330 W/g for the fraction with dh ¼ 82
nm (PCS, Table 1). For this fraction, hydrodynamic diameters of 83
and 84 nm were determined by AC susceptometry and MRX,
respectively. While the moderate SHP of the original ferrofluid
correlates relatively well with the hysteresis losses measured by
VSM, the fraction with dh ¼ 82 nm shows much larger losses
determined by calorimetry than extrapolated from hysteresis. This
may be explained by the high frequency of calorimetrical
measurements in comparison to the nearly six orders of
magnitude slower VSM measurements. The sample with the
relatively small mean size probably contains a considerable
VSM measurements with immobilized particles reveal
a
thermally blocked behavior of the particles. The clusters of
different mean diameters (consisting of primary cores with a
nearly constant mean diameter of 14 nm, measured by XRD) show
a coercivity depending on the size of the clusters (Table 1). This
implies that in this case the magnetic behavior is mainly
determined by the cluster size and not by the constant size of
the primary cores. As demonstrated earlier [19], the coercivity
(HC) correlates linear with the relative remanence (MR/MS). Hence,
the applied fractionation method allows the adaption of the
relative remanence of the ferrofluid to optimal values for
applications based on magnetic attraction, e.g. cell separation or
drug targeting.
10
Table 1
Hydrodynamic diameter (dh, measured by PCS), fractionation parameters,
coercivity (HC, immobilized particles) and SHP (measured at liquid samples) of
the original fluid and of the prepared fractions.
dh (nm)
Fractionation parameter
–
Hc (kA/m)
SHP (W/g)
60
1
Original fluid
Fractions
158
3.10
50
61
2500 g, supernatant
3000 g, sediment
1500 g, supernatant
2500 g, sediment
1000 g, supernatant
1500 g, sediment
1000 g, sediment
0.86
1.99
2.21
2.55
2.60
2.68
3.33
–
1:30 immobilized
1:90 immobilized
1:300 immobilized
1:900 immobilized
1:30 fluid
1:90 fluid
1:300 fluid
1:900 fluid
274
332
145
152
81
82
95
120
142
164
0.1
10
53
0.001
0.01
0.1
relaxation time [s]
100
10-1
10-2
10-3
10-4
10-5
1:9 fluid
1:9 immobilized
1:30 immobilized
1:90 immobilized
1:300 immobilized
1:30 fluid
1:90 fluid
1:300 fluid
1
magnetosomes
original fluid (158 nm)
fraction 3 (82 nm)
AlfaAesar
1
10
100
1000
0.1
magnetic field amplitude [kA/m]
0.001
0.01
0.1
Fig. 2. Dependence of specific hysteresis losses (SHL) of immobilized particles on
the field amplitude of minor loops for the original fluid (dh ¼ 158 nm) and the
fraction with the highest SHP (dh ¼ 82 nm). For comparison, the curves of
magnetosomes and of a commercial sample with high SHL (Iron(III)oxide, gamma,
99+%, AlfaAesar) are shown. The marked frame indicates the interesting field range
for biomedical heating applications.
relaxation time [s]
Fig. 3. MRX curves of the original fluid (a) and the sample with the highest SHP
(b). Curves for measurements with fluids (open symbols) as well as immobilized
particles (full symbols) are shown. The dilution factor ferrofluid to de-ionized
water was changed from 1:9 to 1:900.

ARTICLE IN PRESS
1504
S. Dutz et al. / Journal of Magnetism and Magnetic Materials 321 (2009) 1501–1504
100
80
60
40
20
0
The SHL of the immobilized particles measured by VSM are
relatively high in comparison to commercial samples. However,
the curves of the original fluid as well as all fractions show a
similar course of SHL over the field amplitude which does not
reflect the strong differences between the fractions and the
original fluid observed for the SHP of the fluid samples. Hence, the
SHP values are affected by another relaxation mechanism beside
the hysteresis losses—the Brownian relaxation. For heating
applications of the MCNP in medicine (hyperthermia), the
Brownian losses are mainly suppressed due to a relatively high
degree of binding of the particles to the tumor tissue which means
a diminishing of the SHP [21] that may result in an insufficient
temperature increase in the tumor region [22]. The effect of
particle immobilization on the SHP of MCNP has to be investi-
gated in more detail in further experiments.
Investigations on the cellular uptake of the MCNP show that
these particles interact with tumor cells in a similar way as single
core nanoparticles with similar core diameters do. Initial analysis
was carried out with breast cancer cell line cells only, thus further
studies with other cell types are needed in order to assess the
potential of MNCP for differentiation of tumor cells from other
cells, e.g. leukocytes. The amount of adsorbed or incorporated
particles and thus the magnetic moment of the magnetic material
physically connected to the cell is sufficient for a magnetic
separation of the cells with commercial separation units.
multicore MNP
single core MNP
4
8
12
incubation time [min]
Fig. 4. Uptake of MCNP in MCF-7 tumor cells as a function of the incubation time.
For comparison data of single core MNP obtained in an earlier investigation [20]
are shown (hatched bars).
amount of smaller particles which relax during the quasistatic
VSM measurement contributing negligibly to hysteresis.
After dilution with de-ionized water, the particles of the
original fluid show a tendency to aggregation due to the relative
high amount of larger particles in the fluid. Accordingly, the
relaxation curves in MRX measurements (Fig. 3a) show a shift of
the amplitude for different dilutions, but no changes for the
investigation of immobilized particles. For the fraction with the
highest SHP, no significant changes of their properties after strong
dilution by de-ionized water and BSA buffer were found (Fig. 3b).
This means a good stability against sedimentation. Due to the
wider cluster size distribution and a higher amount of larger
clusters in the original fluid, this sample shows a slower Brownian
relaxation than the sample with the highest SHP. For both
samples, a relatively high relaxation amplitude was determined
by MRX compared to the values of commercial ferrofluids.
The interaction of the MCNP with human cells was analyzed
with the breast cancer cell line MCF-7. The tumor cells are labelled
rapidly with the nanoparticles. Within 4 min more than 50% of the
cells are detected in the positive fraction (Fig. 4). Prolonged
incubation lead to an increase of cell content in the positive
fraction up to 85%. These data are in good correlation to previous
results with CMD-coated MNP with a mean diameter of 10 nm
[20]. The data show that MCNP are a suitable tool for cell
separation.
Acknowledgements
The authors gratefully acknowledge financial support by the
Deutsche Forschungsgemeinschaft (FKZ: ZE825/1-1, CL202/1-2,
SCHM1747/2-1, TR408/4-1) and by the BMBF (FKZ-13N9150). They
thank Ch. Oestreich from the Technical University Bergakademie
Freiberg for TEM imaging and Ch. Schmidt from the IPHT Jena for
XRD investigations. We highly appreciate the AC susceptometry
analysis carried out at the Imego Institute in Goteborg by Ch.
Johansson, K. Petterson and A. Prieto Astalan.
References
[1] S. Dutz, W. Andra¨, R. Hergt, et al., J. Magn. Magn. Mater. 311 (2007) 51.
[2] I.A. Brezovich, Med. Phys. Mon. 16 (1988) 82.
[3] R. Hergt, W. Andra¨, Magnetism, in: W. Andra¨, H. Nowak (Eds.), Medicine—
Second Edition, Wiley-VCH, Berlin, 2007.
[4] R. Hergt, R. Hiergeist, M. Zeisberger, et al., J. Magn. Magn. Mater. 280 (2004)
358.
[5] G. Glo¨ckl, R. Hergt, M. Zeisberger, et al., J. Phys.: Condens. Matter 18 (2006)
2935.
[6] S. Miltenyi, W. Muller, W. Weichel, et al., Cytometry 11 (2) (1990) 231.
[7] I. Safarik, M. Safarikova, J. Chromatogr. B: Biomed. Sci. Appl. 722 (1–2)
(1999) 33.
[8] A.K. Gupta, M. Gupta, Biomaterials 26 (18) (2005) 3995.
[9] J.H. Clement, M. Schwalbe, N. Buske, et al., J. Cancer Res. Clin. Oncol. 132
(2006) 287.
3. Discussion
[10] A.H. Lu, E.L. Salabas, F. Schuth, Angew. Chem. Int. Ed. Engl. 46 (8) (2007) 1222.
[11] K. Wagner, A. Kautz, M. Ro¨der, et al., Appl. Organomet. Chem. 18 (2004) 514.
[12] K. Aurich, M. Schwalbe, J.H. Clement, et al., J. Magn. Magn. Mater. 311
(2007) 1.
[13] S.E. Khallafalla, G.W. Reimers, IEEE Trans. Magn. 16/2 (1980) 178.
[14] R. Massart, IEEE Trans. Magn. 17 (1981) 1247.
[15] R. Hergt, R. Hiergeist, M. Zeisberger, et al., J. Magn. Magn. Mater. 293
(2005) 80.
[16] I. Hilger, W. Andra¨, R. Hergt, et al., Radiology 218 (2001) 570.
[17] D. Eberbeck, F. Wiekhorst, U. Steinhoff, et al., J. Phys.: Condens. Matter 18
(2006) 2829.
In the XRD measurements, the peak position of the (4 4 0)
reflex was used to distinguish between the magnetic phases
maghemite and magnetite. The obtained diffractogram show a
relatively broad single peak between the theoretical angles
for maghemite and magnetite. This means that the resulting
particles presumably consist of solid solutions of both phases
(one resulting peak). However, it is also possible that particles
consist of
a mix of maghemite and magnetite particles
(two separated peaks). In principle, a differentiation between
these both cases is feasible, but due to the peak broadening of the
small particles, a superposition of both peaks is possible. For a
better determination of the phase composition, another analytic
method is needed, e.g. Moessbauer spectroscopy.
[18] D. Eberbeck, C. Bergemann, F. Wiekhorst, et al., Magnetohydrodynamics 41
(2005) 305.
[19] S. Dutz, R. Hergt, J. Mu¨rbe, et al., J. Magn. Magn. Mater. 308 (2007) 305.
[20] M. Schwalbe, N. Buske, M. Vetterlein, et al., Z. Phys. Chem. 220 (2006) 125.
[21] J.-P. Fortin, F. Gazeau, C. Wilhelm, Eur. Biophys. J. 37 (2008) 223.
[22] R. Hergt, S. Dutz, J. Magn. Magn. Mater. 311 (2007) 187.