You are on page 1of 8

Applied Thermal Engineering 117 (2017) 409–416

Contents lists available at ScienceDirect

Applied Thermal Engineering


journal homepage: www.elsevier.com/locate/apthermeng

Research Paper

Omitting the need of external heat capacity data in an adiabatic


magnetothermal setup devoted to the characterization of
nanomaterials for magnetic hyperthermia
Eva Natividad ⇑, Irene Andreu 1
Instituto de Ciencia de Materiales de Aragón (ICMA), CSIC – Universidad de Zaragoza, Campus Río Ebro, María de Luna, 3, Zaragoza 50018, Spain

h i g h l i g h t s

 Presence and arrangement of magnetic nanoparticles affect heat capacity of a medium.


 We prove that heat capacity can be determined using an adiabatic magnetothermal setup.
 This allows obtaining released heat power without additional heat capacity data.
 Omitting external heat capacity determination saves measuring time and resources.
 Accuracy increases using the same setup, specimen, conditions and temperature sweep.

a r t i c l e i n f o a b s t r a c t

Article history: The heat power dissipated by magnetic nanoparticles under alternating magnetic fields is a key param-
Received 24 November 2016 eter in biomedical applications like magnetic hyperthermia. The pulse-heating method determines the
Accepted 6 February 2017 temperature increments undergone in adiabatic conditions by a specimen due to magnetic hyperthermia.
Available online 9 February 2017
The heat power can be accurately calculated from these increments provided that the heat capacity of the
specimen is previously known. In this work, we highlight how this heat capacity, assumed to be the sum
Keywords: of the contributions of each specimen component, can present deviations due to several effects. Such
Pulse-heating method
deviations make necessary a systematic heat capacity determination for every specimen, which in turn
Magnetic hyperthermia
Heat capacity
requires adequate equipment and additional time. As an alternative, we present a new method that
Adiabatic not only provides heat capacity data but also can be implemented in the same setup used for quantifying
Radiative heat exchange the temperature increments by the pulse-heating method in adiabatic conditions. Since this method
determines the heat capacity during the time intervals between alternating magnetic field pulses, it
allows to save characterization time and resources. Eventually, it provides higher accuracy: the heating
power is obtained using just one specimen, one setup and one temperature-sweep measurement, avoid-
ing error arising from the specimen composition or measuring conditions.
Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction mass of MNP is mostly referred to as specific absorption rate (SAR),


but also as specific loss power (SLP). While the former focuses on
Like bulk magnetic materials, magnetic nanoparticles (MNPs) the fact that the transduced power is absorbed from the AMF,
are able to convert into heat the electromagnetic energy supplied the second highlights that this power is released as heat.
by an alternating magnetic field (AMF). But, unlike bulk materials, MNPs invest the electromagnetic energy in reversing their mag-
their small size turns them into highly localized heat sources, suit- netic moments. The origin of the released heat it is the non-linear
able for biological applications such as magnetic hyperthermia [1] and delayed response of the MNPs magnetisation with respect to
or heat-induced drug release [2]. The transduced power, P, per unit the AMF. Accordingly, SAR can be determined by either magnetic
or calorimetric methods. In the most used magnetic method, the
area of the magnetisation vs. field hysteresis loop of MNPs under
⇑ Corresponding author. AMF is quantified (dynamic hysteresis loop). The derived energy
E-mail address: evanat@unizar.es (E. Natividad). per cycle is adequately transformed to SAR using the AMF fre-
1
Present address: Simon Fraser University, Department of Chemistry, 8888 quency. The main limitation of this method is the unavailability
University Drive, Burnaby, BC V5A 1S6, Canada.

http://dx.doi.org/10.1016/j.applthermaleng.2017.02.028
1359-4311/Ó 2017 Elsevier Ltd. All rights reserved.
410 E. Natividad, I. Andreu / Applied Thermal Engineering 117 (2017) 409–416

of commercial setups working with AMF frequencies around Also, further considerations suggest that the Rcp,i  mi approach
100 kHz and amplitudes up to 10 kA/m (tentative range for biolog- may generate some errors due to a number of reasons. For exam-
ical applications). This is due to technical problems related to the ple, MNPs can induce significant changes in the heat capacity of
required fast acquisition of magnetisation data. Only a few groups the medium they are dispersed in [22,25,26]. Also, the experimen-
worldwide have developed special-purpose devices to overcome tal determination of C of the SAR specimen itself is in many
these limitations [3–8]. instances not possible, for example using DSC, due to differences
However, calorimetric methods are the most used by far in the specimen size and geometry required by both methods.
([9–13] and references therein). Two main methods can be found Moreover, the DSC characterization of volatile liquids, often used
in the literature: the initial-slope method in isoperibol conditions as dispersive medium, is difficult due to sample evaporation. Also,
and the pulse-heating method in adiabatic conditions, based on the scan rate used to determine cp,i(T) can modify the temperature
traditional adiabatic calorimetry [13,14]. In both methods, SAR is at which certain processes occurs, or even hinder them [27,28].
quantified from the temporal evolution of the temperature, T(t), While a standard heating rate for DSC is 10 °C/min, in adiabatic
of the MNPs under the AMF. Also, both require determining sepa- magnetothermia this rate is in general about 100 times lower,
rately the heat capacity (C) of the measured specimen. The former and depends strongly on the heating ability of the specimen. To
has been found to generate unquantified errors in SAR determina- minimize all these sources of experimental errors, it is highly
tion due to the mismatch between the experimental conditions desirable to measure C and P on the same conditions and
and the thermal model used for relating T, P, and C [13,15,16]. specimens.
The main drawback of the latter is the unavailability of a setup cap- For this purpose, in this work we take a further step in the
able of measuring T(t) in adiabatic conditions and under an AMF. determination of SAR(T) by the pulse-heating method in adiabatic
To the authors’ knowledge, there is just one apparatus (adiabatic conditions. Analysing and modifying the heat transfer processes
magnetothermal setup) operating under this method, which has occurring in the magnetothermal setup described in [17,18], we
been shown to provide the most accurate P data [13,17]. have developed the first method to calculate, not only (DT/Dt),
Most scientific research merely evaluates SAR at room temper- but the whole product C  (DT/Dt) in Eq. (2) in adiabatic conditions
ature [9–11], neglecting any SAR variation with temperature. How- and using just one specimen, one temperature-sweep measure-
ever, biological applications involve temperatures from 36 up to ment and one setup. Our method thus suppresses the need of
about 50 °C. Among the abovementioned methods, only the external heat capacity data, allowing obtaining more accurate
pulse-heating method in adiabatic conditions [18] and some meth- SAR data, while saving time and resources. In this paper, we first
ods measuring dynamic hysteresis loops [4,19] have proved to be evaluate the changes that the measuring parameters and/or the
capable of providing SAR data as a function of temperature, SAR presence of magnetic nanoparticles can induce in the heat capacity
(T). The former has the advantage of covering a much wider tem- of different samples. We then describe our method and its imple-
perature range. Indeed, it has allowed to show that the determina- mentation in an adiabatic magnetothermal setup. Finally, the
tion of SAR(T) at wide (even sub-ambient [18]) temperature ranges developed method is assessed on six specimens with different C
provides valuable information about the magnetic state and prop- (T) characteristics.
erties of the MNPs [20–24]. Most importantly, this information,
similar to those provided by the out-of-phase component of the
magnetic susceptibility, is obtained with AMFs suitable for mag- 2. Materials and methods
netic hyperthermia applications.
In ideal adiabatic conditions, it can be assumed that all the heat 2.1. Sample and specimen preparation
power generated by the MNPs is invested in the specimen temper-
ature rise, and therefore the thermal power balance between the Three dispersive media (water, n-dodecane, n-tetracosane)
specimen and its environment reads were used to produce six different samples (see Table 1): three
without magnetic nanoparticles (W, C12 and C24) and three with
dTðtÞ magnetic nanoparticles (W-FF, C12-FF and C24-FF). Water was
P¼C ð1Þ
dt used as representative of biologically compatible medium, n-
As specimen, we consider a sample with a certain mass of mag- dodecane as an example of the organic media used in many MNP
netic nanoparticles, mMNP, dispersed in a medium, together with its synthesis processes, and n-tetracosane was used as a medium
container. Integration of Eq. (1) leads to the expression for SAR cal- which is solid in the whole temperature range, like biological tis-
culation in adiabatic conditions, sue. These media are pure solvents whose heat capacity and melt-
ing temperature (see Table 1) are well-known, [29–33] while
P C DT different enough as to prove that the developed method can be
SAR ¼ ¼  ð2Þ used with other media of diverse cp(T), such as physiological saline,
mMNP mMNP Dt
blood and bone.
where DT is the specimen temperature increment during an AMF MNPs in samples W-FF, C12-FF and C24-FF come from different
application interval Dt, and C is assumed constant in the considered sources. The sample in W-FF is the fluidMAG-UC/C aqueous disper-
DT. The pulse-heating method alternates P – 0 and P = 0 periods sion of Chemicell GmbH. MNPs in C12-FF and C24-FF have been
(i.e., field-on and field-off periods) to obtain DT values at several synthesized in our laboratory adapting the seeded growth method
temperatures and assigns them to the midpoint of the temperature of Sun et al. [34,35] and Luigjes et al. [36]. For this purpose, initial
interval, which is typically of a few degrees. Using Eq. (2) and the 6-nm spherical seeds were obtained by thermal decomposition of
obtained DT values, the SAR in function of temperature, SAR(T), iron acetylacetonate in benzyl ether in the presence of oleic acid,
can be calculated, provided that C(T) is also known [18]. C is oleylamine, and 1,2-hexadecanediol. Starting from these seeds,
obtained as Rcp,i  mi, where cp,i and mi are, respectively, the specific MNPs in samples C12-FF and C24-FF were obtained using identical
heat capacity and the mass of each component of the specimen. For seeded growth steps (2 and 26, respectively) under similar condi-
this purpose, all cp,i(T) must be known, either from literature data or tions, but with reduced amounts of oleic acid and oleylamine,
from specific measurements, using e.g. Differential Scanning which act as surfactants. The detailed procedures are described
Calorimetry (DSC). In the latter case, this involves additional in [23,24], respectively. Information about the size and shape of
resources and characterization time. all MNPs is provided in the supplementary information.
E. Natividad, I. Andreu / Applied Thermal Engineering 117 (2017) 409–416 411

Table 1
Description of the samples (dispersive medium and MNP concentration) and mass composition of the measured specimens (see Section 2).

Dispersive medium Tmelting (K) Sample/specimen mglass (g) mad (g) mmed (g) mMNP (g) c (mg/g)
Water 273 W 0.34352 0.01511 0.09268 ... ...
W-FF 0.30376 0.02549 0.09432 2.36  103 24.4
n-Dodecane 263 C12 0.37078 0.01080 0.08447 ... ...
C12-FF 0.31907 0.02532 0.07546 8.3  104 10.9
n-Tetracosane 324 C24 0.38361 0.02251 0.09495 ... ...
C24-FF 0.35431 0.03868 0.09587 3.9  104 4.1

Six different specimens for SAR measurement were obtained by DSC. It is observed that the cp of the silica glass bought from two
from those six samples (and named after them). To prepare these different suppliers (silica glass (1) and silica glass (2)) deviates by
specimens, a portion of every sample (0.1 mL of dispersive med- about 25%. Also, that the cp of the N27 adhesive slightly varies upon
ium + MNPs if any) was enclosed inside similar silica glass contain- aging (fresh vs. after 14 days), and that two fresh preparations (N27
ers and sealed with Nural 27 (N27, Pattex), a two-component (1) and N27(2)) of this two-component adhesive yield a cp differ-
epoxy adhesive (see Table 1 and Fig. S1 in the supplementary infor- ence of about 13%. This is most probably due to variations in the
mation). Table 1 collects, for each specimen, the mass of the silica adhesive component proportions leading to different curing speeds
glass container (mglass), mass of the adhesive (mad), mass of the dis- and making the adhesives be in different polymerization stages at
persive medium (mmed) and mass of magnetite MNPs in the sample the moment of DSC measurement. These results highlight the con-
(mMNP). venience of a systematic characterization of both components in
order to reduce errors in SAR(T) coming from C(T) deviations.
2.2. Sample and specimen characterization Fig. 1(b) is related with samples W, W-FF, C24 and C24-FF. On
one hand, it compares the cp of pure n-tetracosane determined in
The iron content of samples C12-FF and C24-FF was determined an aneroid calorimeter [29] and the cp of sample C24-FF (n-
by Inductively Coupled Plasma – Optical Emission Spectrometry tetracosane + MNPs), measured in this work with DSC. The vertical
(ICP-OES), using a Jobin Yvon 2000 setup. Samples were previously golden line indicates the melting temperature of n-tetracosane
digested in aqua regia overnight. The concentration of MNPs in according to [29]. The cp data of magnetite is also provided [30],
each sample was then calculated assuming stoichiometric compo- although the MNP concentration of sample C24-FF is so low
sition of the Fe3O4 nanoparticles. This concentration (c) is collected (4.1 mg/g) that the additive contribution of MNPs to C(T) is
in Table 1 in mg of MNPs per g of sample (MNPs + dispersive med- negligible. Although at low temperature the data series of
ium). The detailed characterization used to elucidate the MNP crys- n-tetracosane and C24-FF deviate slightly, this deviation increases
talline phase of samples C12-FF and C24-FF is the same followed in with temperature, reaching 27% at 300 K. The origin of this
[23,24], respectively (see supplementary information). MNP crys- deviation (setup, measuring conditions, presence of MNPs,. . .)
talline phase and concentration of sample W-FF is provided by [22,25–28] is beyond the scope of this work, but it is evident that
the supplier. Even if magnetite was found to be the main crys- the cp of C24-FF cannot be obtained as a weighted sum of those of
talline phase in the three samples, the presence of traces of maghe- pure n-tetracosane and Fe3O4. On the other hand, this figure com-
mite cannot be discarded. However, this presence does not alter pares the cp of the frozen sample W-FF (water + MNPs) measured
the work results, given the negligible contribution of the MNPs in this work with that of water [31,32]. It is observed that sample
to the additive C of each specimen, and also to the similarity of W-FF undergoes premelting at about 250 K, likely due to the pres-
the specific heat capacities of magnetite and maghemite. ence of the MNPs, and thus its cp deviates from that of ice from this
The specific heat capacity data of samples and some compo- temperature on.
nents of the specimens were either obtained from the literature Fig. 1(c) shows three measurements of the cp of pure n-
or measured by DSC, using a Q1000 device from TA Instruments dodecane (sample C12). The measurement in reference [33] was
and indium and sapphire as calibrants. performed in adiabatic conditions, while the other two were
The SAR data of the specimens were measured using a special- obtained for the current work using DSC, either in a standard alu-
purpose magnetothermal setup based on the pulse-heating minium hermetic pan (without o-ring) or in a hermetic stainless
method in adiabatic conditions [17,18]. steel high volume pan (with o-ring). Comparing the results of the
first (blue2 continuous line) and second (black dashed line) series,
3. Results and discussion we observe that the DSC provides lower cp values when the n-
dodecane is solid (up to 12%) and just after the melting (17%). Above
3.1. Deviations among heat capacity data the melting, the cp obtained with DSC and the standard aluminium
hermetic pan displays an upward trend, presumably due to the
As explained in the introduction, the calorimetric determina- vaporization heat of the n-dodecane that was able to escape the
tion of SAR(T) values requires external C(T) data. This data is usu- pan, given that leakage of n-dodecane could not be avoided with this
ally obtained as Rcp,imi, where cp,i and mi are, respectively, the kind of pan. The hermetic pan with o-ring was able to prevent the
specific heat capacity and the mass of each component of the spec- leakage of n-dodecane, although the obtained data displays a wider
imen. In this section, we compare cp (T) data of some samples and temperature range for the melting of C12. Even after scaling the cp
some components of the specimens in Table 1 with the aim of eval- data obtained with this pan (red continuous line) to the literature
uating the convenience of using this additive approach. Note that values in solid state, the cp values of the liquid n-dodecane appear
all ferrofluid samples have a MNP mass concentration lower than lower than those reported in [33].
25 mg per gram of sample and, therefore, the additive contribution
of the MNPs to the heat capacity is negligible.
Fig. 1(a) displays the cp values corresponding to the sample con- 2
For interpretation of color in Fig. 1, the reader is referred to the web version of
tainer and the sealant used in the specimens measured in this work this article.
412 E. Natividad, I. Andreu / Applied Thermal Engineering 117 (2017) 409–416

Fig. 1. Specific heat capacity data of some samples and some components of the specimens in Table 1 (see Section 3.1 for further details).

Fig. 1(d) compares the cp series of pure n-dodecane according to the adiabatic magnetothermal setup described in [17,18] and
[33], with two successive series of that of sample C12-FF (n- schematized in Fig. 2. In this setup, the specimen is first set under
dodecane + MNPs), measured in this work with DSC. The first vacuum and cooled down using liquid nitrogen. Afterwards, the
C12-FF series (black dashed line) was obtained after the sample adiabatic control is enabled: the specimen environment (adiabatic
was quenched in liquid nitrogen in presence of a static magnetic shield) is controlled to keep a certain small temperature difference
field, so that a texture is presumably acquired and preserved by with the specimen, dT, to minimize radiation heat transfer. Finally,
the MNP in the frozen medium. Such texture must be lost after- the alternating magnetic field pulse sequence is launched. The adi-
wards, upon melting of n-dodecane, given that these MNPs are abatic shield consists in a ceramic piece provided with a thin film
already superparamagnetic at that temperature (supplementary resistor to minimize uncontrolled self-heating in presence of the
Fig. S3). The second C12-FF series (red continuous line) was AMF due to Foucault currents. As observed in Fig. 3, the net upward
recorded subsequently, after the sample was cooled inside the temperature sweep is achieved by the specimen self-heating under
DSC, i.e., the sample is non-quenched and has no texture. It is first the successive AMF pulses, since the sample cannot be heated by
observed that the slope of the DSC series is different to that in [33], traditional means (e.g. a resistive heater) [17].
due to the presence of MNPs. In addition, before the melting of n- First, we have analysed the heat transfer processes occurring in
dodecane, the cp values of the textured C12-FF sample are up to the setup. If we focus on Fig. 3(a) and (c), for dT = 0 (specimen
38% lower than those of the non-textured C12-FF, due to the differ- and adiabatic shield at the same temperature), we observe
ent arrangements of the MNPs. After the melting of n-dodecane, that the specimen temperature has a downward linear trend
both C12-FF series are very similar, as expected due to the lack (0.1 °C/min) when the AMF is switched off. The origin of this
of texture, and display lower cp values (up to 8%) that those in trend is that, in non-ideal but real adiabatic systems, very weak
[33]. This figure then shows how, not only the presence, but also thermal losses occur to an outer environment.
the arrangement of the MNPs can affect the cp values in such a In these conditions and in absence of AMF the observed
way that the expression Rcp,imi does not apply. downward slope, dT(t)/dt, is described as,
dTðtÞ
3.2. Method for accounting for heat capacity variations in SAR data C  ¼ L½TðtÞ  T out  ð3Þ
dt
Based on the above, it seems clear that the most adequate where L(W/K) is a coefficient accounting for losses and Tout stands
determination of SAR(T) involves measuring C(T) and DT/Dt in for the temperature of the outer environment.
Eq. (2) on the same conditions, i.e., on the same specimen (same Such weak downward trends have been considered and allowed
sample portion, container and sealant), at the same temperature in traditional experimental adiabatic determination of heat [14],
sweep rate and subjected to the same magnetic field. provided that losses are small enough as to provide a linear T(t)
In order to make this possible, we have developed a method to trend in absence of AMF and keep acceptable adiabatic conditions.
calculate P(T) using just one specimen, one temperature-sweep In these non-ideal conditions, it is still possible to use Eq. (2) to cal-
measurement and one setup. This method was implemented in culate SAR through the pulse-heating method. Each DT value must
E. Natividad, I. Andreu / Applied Thermal Engineering 117 (2017) 409–416 413

Fig. 2. Partial scheme of: (a) the adiabatic magnetothermal setup in [18]; (b) the modification made to the same setup (see end of Section 3.2).

Fig. 3(c)). But the advantages of these sweep conditions go much


further, since they are the starting point to the direct quantification
of P and thus of SAR, as explained in the following.
When dT > 0 the power balance can be expressed as,
dTðtÞ
C ¼ P þ R  dT  L½TðtÞ  T out  ð4Þ
dt
where the term RdT represents the heat power provided by the
radiation shield, valid for dT  T, and R(W/K) is the coefficient
accounting for the radiative power from the adiabatic shield to
the specimen (conductive and convective heat transfer between
adiabatic shield and sample is negligible in the used setup). In Eq.
(4) there are several unknown quantities: P, C and the whole last
term, which is a function that may vary with temperature in an irre-
producible manner. When the AMF is switched off, P = 0, and only C
and the last term remain unknown. In these conditions, if this last
term were negligible with respect to RdT, the heat capacity of the
specimen could be expressed as,
R  dT
C¼ ð5Þ
adT
where adT is the slope of the T(t) trend (dT/dt) when the AMF is
switched off and the temperature of the radiation shield is dT
degrees higher than that of the sample. According to this, P(T) could
be calculated from,
RðTÞ  dT DTðTÞ
PðTÞ ¼  ð6Þ
Fig. 3. T(t) curves of sample C12-FF obtained using the pulse-heating method in adT ðTÞ Dt
adiabatic conditions for (a) dT = 0 K and (b) dT = 1 K. (c) Whole measurement of the
same sample under both conditions. Inset in (c): calculated DT/Dt values under avoiding a previous determination of C(T). For this, the weak heat
both conditions.
power provided by the radiation shield (RdT) must be calculated.
dT is controlled and measured (and therefore known) during the
whole temperature-sweep measurement. Besides, if the parameter
be determined extrapolating the backward and forward tempera- R is a radiative coefficient, it must follow the law,
ture trends of every pulse, and calculating the difference of these
R ¼ r  G  T3 ð7Þ
extrapolations at the midpoint of the heating interval Dt, as shown
8
in Fig. 3(a). where r is the Stefan-Boltzmann constant (= 5.670  10 W/
The downward linear trend in absence of AMF is undoubtedly a m2 K4), and G is a factor including emissivity and geometrical prop-
drawback to achieve an efficient net upward trend in SAR(T) mea- erties of both the specimen and the adiabatic shield. Calculation of
surements and therefore be able to sweep the whole temperature G then allows quantifying the heat power provided by the radiation
range for SAR(T). Nevertheless, this situation can be reversed. Tak- shield.
ing advantage of radiative heat transfer, the adiabatic shield can be Fig. 3(a) and (b) highlight how the last term in Eq. (4) is, unfor-
used as specimen heater. If it keeps a small positive temperature tunately, of the same order of magnitude than (and thus not neg-
difference with the specimen, dT > 0, an upward trend can be ligible with respect to) RdT in the measurements performed with
established even in absence of AMF. In the setup described in the setup schematized in Fig. 2. For example, at about 20 °C, the
[17,18] and with typical specimens (supplementary Fig. S1) it slope in Fig. 3(a) (dT = 0 K), is 0.101 K/min, while that of Fig. 3
was found that acceptable adiabaticity was still maintained with (b) (dT = 1 K) is 0.084 K/min, that is, the term accounting for uncon-
dT  1.2 K. Accordingly, Eq. (2) can be used to calculate SAR using trolled losses represents 54.6% of the heat gain provided by the adi-
the pulse-heating method, as shown in Fig. 3(b) for dT = 1 K (tem- abatic shield when dT = 1 K.
perature of the adiabatic shield 1 K higher than that of the speci- In order to reduce the contribution of these losses, a new cera-
men). Fig. 3(c) evidences that the measurement at dT = 1 K is mic shield (outer shield) was placed surrounding the adiabatic
much less time-consuming, taking just half the time than that at shield, providing a controlled-temperature outer environment for
dT = 0 K, while giving exactly the same DT/Dt values (inset of the sample (see Fig. 2). The effect of this outer environment, whose
414 E. Natividad, I. Andreu / Applied Thermal Engineering 117 (2017) 409–416

were either obtained from the literature (n-tetracosane [29], water


[31,32], n-dodecane [33]) or measured by DSC (silica glass, adhe-
sive). Good fits to Eq. (8) were achieved in the three cases indicat-
ing, first, that our model for radiative heat exchange is valid, and
second, that for the samples without MNPs there is a good accor-
dance between the heat capacity calculated as Rcp,i  mi and that
obtained directly from adT values using Eq. (5). However, while
the obtained G values for specimens W and C12 are very similar,
that of C24 differs in 3%. This could be assigned to three main rea-
sons. First, the glass of the container is transparent to certain infra-
red wavelengths. This implies that the radiation coming from the
sample inside the container can be transmitted across it, this
resulting in slightly different emissivity values. Second, subtle vari-
Fig. 4. Slope of the field-off temperature trend of sample C12-FF without outer
ations in the specimen-shield relative position can influence the
shield (nOS) and dT = 1 K (squares) or dT = 0 K (triangles), and with outer shield
control (OS) and dT = 1 K (circles) or dT = 0 K (diamonds). value of G, since it depends on geometrical factors. And third, the
small thermal losses still present in the system (last term in Eq.
(4)) can also affect the G value. Note, however, that the product
temperature follows that of the specimen, is evident (see Fig. 4): C  adT against T for specimen C12 without the outer shield control
now, the slope at about 20 °C with dT = 0 K is 0.012 K/min, while (inset in Fig. 5(a)) does not follow a T3 law, since it includes larger
that at dT = 1 K is 0.186 K/min, that is, the term accounting for uncontrolled losses. This highlights the fact that, without the use of
uncontrolled losses represents only 6.1% of the heat gain provided the outer shield control, C cannot be obtained by the simple Eq. (5)
by the adiabatic shield when dT = 1 K, so it can be neglected. P(T) and a unique constant G. According to the above, we considered
can now be calculated using Eqs. (6) and (7). that assuming a mean G value ((9.2 ± 0.2)  104 m2 in our setup)
for calculations is the best option to minimize errors. Per cent
3.3. Assessment of the method errors in Table 2 are calculated with respect to this mean value.
Fig. 5(b) displays C  adT vs. T plots of the studied specimens with
In order to assess the method outlined above, the value and MNPs, obtained as Rcp,imi using the same contributions than those
reproducibility of G was evaluated on the six specimens described in Fig. 5(a), and in addition that of magnetite [29–33]. Note, how-
in Table 1. It was calculated for each sample by fitting the obtained ever, that the term accounting for the heat capacity of MNPs is neg-
experimental data to the expression, ligible, due to the low mass concentrations. Similarly to specimens
without MNPs, W-FF shows good fits to Eq. (8), although with a
C  adT ¼ r  G  dT  T 3 ð8Þ lower G value (Table 2). C12-FF and C24-FF display, however, devi-
where C in this equation was obtained as Rcp,i  mi. ations from the T3 law at high T, pointing to a non-negligible influ-
Fig. 5(a) shows C  adT vs. T plots of the studied specimens with- ence of the MNPs in the heat capacity of the samples, not
out MNPs. Specific heat capacity values for Rcp,imi calculations accounted by simple addition of the contributions of the compo-
nents of the sample.
Fig. 5(c) displays C  adT vs. T plots of the studied specimens with
MNPs, obtained as Rcp,i  mi using the same contributions than
those in Fig. 5(a), except for the contributions of samples C12-FF
(n-dodecane + MNPs) and C24-FF (n-tetracosane + MNPs), for
which the DSC cp values displayed in Fig. 1 (b: C24-FF) and (d:
C12-FF, non-quenched) were used. It is observed that now the
three fits are very similar, and that deviations from the T3 law at
high T have been partially and totally corrected for samples C12-
FF and C24-FF, respectively. These results prove that the developed
method accounts for the deviations between heat capacity data
observed in DSC measurements.
In particular, DSC revealed that the cp of sample C24-FF is up to
27% lower than that of pure n-tetracosane when approaching its
melting point. Therefore, the C  adT product is overestimated when
the cp of pure n-tetracosane [29] is used to calculate C of sample
C24-FF, as seen in Fig. 5(b). The same happens to sample C12-FF
just after the melting of n-dodecane.

Table 2
Results of G calculation and per cent error (Er) of each G value (see Section 3.3).

Specimen Figure G (m2) Er (%)


W 5(a) 9.297  104 1.2
W-FF 5(b, c) 8.883  104 3.3
C12 5(a) 9.260104 0.8
C12-FF 5(b) 9.201  104 0.2
5(c) 8.943  104 3.7
C24 5(a) 8.998  104 2.0
Fig. 5. C  adT against T data (with dT = 1 K) fitted according to Eq. (8). (a) Specimens
C24-FF 5(b) 9.166  104 0.2
without MNPs. Inset: C12 with (OS) and without (nOS) outer shield control. (b) and
5(c) 8.847  104 2.6
(c) Specimens with MNPs (see Section 3.3).
E. Natividad, I. Andreu / Applied Thermal Engineering 117 (2017) 409–416 415

since this magnitude can be affected appreciably by the presence


and arrangement of magnetic nanoparticles, or when using differ-
ent measuring conditions.
As an alternative to the systematic determination of heat capac-
ity for each specimen, a method is proposed that allows the direct
calculation of the heat power without the need of additional heat
capacity data. This method not only suppresses the necessity of
knowing previously the mass and specific heat capacity of each
specimen component, but also allows the determination of all
the required data for heat power calculation with the same setup,
on the same specimen, under the same measuring conditions and
during the same temperature-sweep run. This, in turn, avoids dis-
Fig. 6. SAR(T) data for sample C12-FF, measured with AMF amplitude and
frequency of 3.3 kA/m and 108 kHz, respectively (see text). Data at the melting
crepancies arising from the use of different specimens or measur-
temperature interval is omitted. ing conditions to determine temperature increments and heat
capacity, and thus increases accuracy in SAR(T) data.

The deviations from the T3 law at high T still observed in Fig. 5


Acknowledgements
(c) for C12-FF could have an origin in the measuring conditions
during SAR determination. On one hand, DSC measurements
This work has been funded by the Spanish MINECO and
showed that sample C12-FF (n-dodecane + MNPs) presents a
FEDER under project MAT2014-53961-R. I. Andreu thanks the
higher cp than that of n-dodecane in the frozen state, but may
Spanish CSIC for her JAE-Predoc contract. The authors thank
undergo reductions of up to 38% upon MNP texturing under a static
O. Roubeau for the DSC measurements. The use of Servicio General
magnetic field. Even if SAR measurements do not take place under
de Apoyo a la Investigación-SAI, Universidad de Zaragoza is also
static fields like the ones used for texturing, AMFs can also induce
acknowledged.
organization of MNPs [37], whose effect on the real C cannot be
easily quantified with commercial DSC equipment. Contrary to
Appendix A. Supplementary material
what it is observed in Fig. 1(d), the influence of the AMF effect
should only be present at high T. In the magnetothermal setup,
Supplementary data associated with this article can be found, in
the specimen C12-FF was solidified in absence of AMF, so that
the online version, at http://dx.doi.org/10.1016/j.applthermaleng.
organization due to the AMF can occur only during the SAR mea-
2017.02.028.
surement, and only after the melting of n-dodecane. Therefore this
effect can explain the deviations observed at high temperature. On
the other hand, the volume of the glass container of specimen C12- References
FF is occupied with ferrofluid to a high percentage, see supplemen- [1] R. Ivkov, Magnetic nanoparticle hyperthermia: a new frontier in biology and
tary Fig. S1. It is also possible that pressure effects can affect cp medicine?, Int J. Hyperther. 29 (2013) 703–705.
after the melting of n-dodecane. [2] S. Laurent, S. Dutz, U.O. Häfeli, M. Mahmoudi, Magnetic fluid hyperthermia:
focus on superparamagnetic iron oxide nanoparticles, Adv. Colloid Interface
Fig. 6 shows the SAR values of sample C12-FF obtained using
Sci. 166 (2011) 8–23.
‘‘calculated” heat capacity (last term in Eq. (2) and Rcp,i  mi) and [3] E. Garaio, J.M. Collantes, J.A. Garcia, F. Plazaola, S. Mornet, F. Couillaud, O.
‘‘directly measured” adT values (middle term in Eqs. (2) and (6)) Sandre, A wide-frequency range AC magnetometer to measure the specific
absorption rate in nanoparticles for magnetic hyperthermia, J. Magn. Magn.
without any additional cp data. The ‘‘calculated (1)” series was
Mater. 368 (2014) 432–437.
obtained using cp values obtained from the literature (n- [4] M. Bekovic, I. Ban, A. Hamler, Assessment of magnetic fluid losses out of
dodecane [33], magnetite [30]) and measured by DSC (silica glass, magnetic properties measurement, J. Phys.: Conf. Ser. 200 (2010) 072010.
adhesive). The ‘‘calculated (2)” series was obtained using cp values [5] H. Kobayashi, K. Ueda, A. Tomitaka, T. Yamada, Y. Takemura, Self-heating
property of magnetite nanoparticles dispersed in solution, IEEE Trans. Magn.
measured by DSC (sample C12-FF, silica glass, adhesive). Even if 47 (2011) 4151–4154.
data at the melting temperature interval of n-dodecane are omit- [6] B. Mehdaoui, J. Carrey, M. Stadler, A. Cornejo, C. Nayral, F. Delpech, B. Chaudret,
ted, due to the impossibility of obtaining correct adT values at this M. Respaud, Influence of a transverse static magnetic field on the magnetic
hyperthermia properties and high-frequency hysteresis loops of ferromagnetic
interval (T is nearly constant during the melting), data down to FeCo nanoparticles, Appl. Phys. Lett. 100 (2012) 052403.
10 K below the melting temperature differ appreciably between [7] I. Hilger, In vivo applications of magnetic nanoparticle hyperthermia, Int. J.
calculated and directly measured values. This is most probably Hyperther. 29 (2013) 828–834.
[8] B. Kozissnik, A.C. Bohorquez, J. Dobson, C. Rinaldi, Magnetic fluid
due to the particular measuring conditions in the adiabatic magne- hyperthermia: advances, challenges, and opportunity, Int. J. Hyperther. 29
tothermal setup (e.g., very low temperature sweep rate), which (2013) 706–714.
slightly affect the melting process of n-dodecane. Therefore, the [9] C. Sanchez, D. El Hajj Diab, V. Connord, P. Clerc, E. Meunier, B. Pipy, B. Payré, R.
P. Tan, M. Gougeon, J. Carrey, V. Gigoux, D. Fourmy, Targeting a G-protein-
SAR values obtained using the measured adT values not only
coupled receptor overexpressed in endocrine tumors by magnetic
account for the above mentioned variations in C, but also provide, nanoparticles to induce cell death, ACS Nano 8 (2014) 1350–1363.
in this case, a better defined SAR maximum, located just at the [10] R. Di Corato, A. Espinosa, L. Lartigue, M. Tharaud, S. Chat, T. Pellegrino, C.
Ménager, F. Gazeau, C. Wilhelm, Magnetic hyperthermia efficiency in the
melting interval of n-dodecane. In sum, the technique here pre-
cellular environment for different nanoparticle designs, Biomaterials 35
sented outperforms the traditional way of quantifying SAR through (2014) 6400–6411.
the combination of cp values (DSC, literature) and temperature [11] P. de la Presa, Y. Luengo, V. Velasco, M.D.P. Morales, M. Iglesis, S. Veintemillas-
increments during AMF application (adiabatic magnetothermal Verdaguer, P. Crespo, A. Hernando, Particle interactions in liquid magnetic
colloids by zero field cooled measurements: effects on heating efficiency, J.
setup). Phys. Chem. C (2015).
[12] E.A. Périgo, G. Hemery, O. Sandre, D. Ortega, E. Garaio, F. Plazaola, F.J. Teran,
Fundamentals and advances in magnetic hyperthermia, Appl. Phys. Rev. 2
4. Conclusion (2015) 041302.
[13] I. Andreu, E. Natividad, Accuracy of available methods for quantifying the heat
This work has pointed out that the determination of the heat power generation of nanoparticles for magnetic hyperthermia, Int. J.
Hyperther. 29 (2013) 739–751.
capacity can be a source of error in the calculation of the heat [14] E. Gmelin, Modern low-temperature calorimetry, Thermochim. Acta 29 (1979)
power dissipated by nanomaterials for magnetic hyperthermia, 1–39.
416 E. Natividad, I. Andreu / Applied Thermal Engineering 117 (2017) 409–416

[15] E. Natividad, M. Castro, A. Mediano, Adiabatic vs. non-adiabatic determination [26] N.S.S. Mousavi, S. Kumar, Effective heat capacity of ferrofluids – analytical
of specific absorption rate of ferrofluids, J. Magn. Magn. Mater. 321 (2009) approach, Int. J. Therm. Sci. 84 (2014) 267–274.
1497–1500. [27] S. Strella, P.F. Erhardt, Rate effects in the measurement of polymer transition
[16] S. Huang, S.Y. Wang, A. Gupta, D.A. Borca-Tasciuc, S.J. Salon, On the by Differential Scanning Calorimetry, J. Appl. Polym. Sci. 13 (1969) 1373–1380.
measurement technique for specific absorption rate of nanoparticles in an [28] G.V. Poel, V.B.F. Mathot, High-speed/High performance differential scanning
alternating electromagnetic field, Meas. Sci. Technol. 23 (2012) 035701. calorimetry (HPer DSC): temperature calibration in the heating and cooling
[17] E. Natividad, M. Castro, A. Mediano, Accurate measurement of the specific mode and minimization of thermal lag, Thermochim. Acta 446 (2006) 41–54.
absorption rate using a suitable adiabatic magnetothermal setup, Appl. Phys. [29] G.S. Parks, G.E. Moore, M.L. Renquist, B.F. Naylor, L.A. McClaine, P.S. Fujii, J.A.
Lett. 92 (2008) 093116. Hatton, Thermal data on organic compounds. XXV. Some heat capacity,
[18] E. Natividad, M. Castro, A. Mediano, Adiabatic magnetothermia makes possible entropy and free energy data for nine hydrocarbons of high molecular weight,
the study of the temperature dependence of the heat dissipated by magnetic J. Am. Chem. Soc. 71 (1949) 3386–3389.
nanoparticles under alternating magnetic fields, Appl. Phys. Lett. 98 (2011) [30] E.F. Westrum Jr, F. Grønvold, Magnetite (Fe3O4) Heat capacity and
243119. thermodynamic properties from 5 to 350 K, low-temperature transition, J.
[19] E. Garaio, O. Sandre, J.-M. Collantes, J.A. Garcia, S. Mornet, F. Plazaola, Specific Chem. Thermodyn. 1 (1969) 543–557.
absorption rate dependence on temperature in magnetic field hyperthermia [31] W.F. Giauque, J.W. Stout, The entropy of water and the third law of
measured by dynamic hysteresis losses (ac magnetometry), Nanotechnology thermodynamics. The heat capacity of ice from 15 to 273 K, J. Am. Chem.
26 (2015) 015704. Soc. 58 (1936) 1144–1150.
[20] E. Natividad, M. Castro, G. Goglio, I. Andreu, R. Epherre, E. Duguet, A. Mediano, [32] N.S. Osborne, H.F. Stimson, D.C. Ginnings, Heat capacity of water, Handbook of
New insights into the heating mechanisms and self-regulating abilities of Chemistry and Physics, Cleveland, Ohio, 1939, pp. 1972–1973.
manganite perovskite nanoparticles suitable for magnetic fluid hyperthermia, [33] H.L. Finke, M.E. Gross, G. Waddington, H.M. Huffman, Low-temperature
Nanoscale 4 (2012) 3954–3962. thermal data for the nine normal paraffin hydrocarbons from octane to
[21] G.T. Landi, Simple models for the heating curve in magnetic hyperthermia hexadecane, J. Am. Chem. Soc. 76 (1954) 333–341.
experiments, J. Magn. Magn. Mater. 326 (2013) 14–21. [34] S. Sun, H. Zeng, Size-controlled synthesis of magnetite nanoparticles, J. Am.
[22] I. Andreu, E. Natividad, C. Ravagli, M. Castro, G. Baldi, Heating ability of cobalt Chem. Soc. 124 (2002) 8204–8205.
ferrite nanoparticles showing dynamic and interaction effects, RSC Adv. 4 [35] S. Sun, H. Zeng, D.B. Robinson, S. Raoux, P.M. Rice, S.X. Wang, G. Li,
(2014) 28968–28977. Monodisperse MFe2O4 (M = Fe Co, Mn) nanoparticles, J. Am. Chem. Soc. 126
[23] I. Andreu, E. Natividad, L. Solozábal, O. Roubeau, Nano-objects for addressing (2004) 273–279.
the control of nanoparticle arrangement and performance in magnetic [36] B. Luigjes, S.M.C. Woudenberg, R. de Groot, J.D. Meeldijk, H.M. Torres Galvis, K.
hyperthermia, ACS Nano 9 (2015) 1408–1419. P. de Jong, A.P. Philipse, B.H. Erné, Diverging geometric and magnetic size
[24] I. Andreu, E. Natividad, L. Solozábal, O. Roubeau, Same magnetic nanoparticles, distributions of iron oxide nanocrystals, J. Phys. Chem. C 115 (2011) 14598–
different heating behavior: influence of the arrangement and dispersive 14605.
medium, J. Magn. Magn. Mater. 380 (2015) 341–346. [37] B. Mehdaoui, R.P. Tan, A. Meffre, J. Carrey, S. Lachaize, B. Chaudret, M. Respaud,
[25] G. Cordoyiannis, L.K. Kurihara, L.J. Martínez-Miranda, C. Glorieux, J. Thoen, Increase of magnetic hyperthermia efficiency due to dipolar interactions in
Effects of magnetic nanoparticles with different surface coating on the phase low-anisotropy magnetic nanoparticles: theoretical and experimental results,
transitions of octylcyanobiphenyl liquid crystal, Phys. Rev. E 79 (2009) Phys. Rev. B 87 (2013) 174419.
011702.

You might also like