Magnetic properties of iron oxide nanoparticles

30 downloads 0 Views 3MB Size Report
06249; and Madrid Region Council project NANOBIO. MAGNET (S2009/MAT-1726). A. F. acknowledges financial support from the Spanish MICINN through ...
J Nanopart Res (2013) 15:1514 DOI 10.1007/s11051-013-1514-8

RESEARCH PAPER

Magnetic properties of iron oxide nanoparticles prepared by seeded-growth route A. Espinosa • A. Mun˜oz-Noval • M. Garcı´a-Herna´ndez • A. Serrano • J. Jime´nez de la Morena • A. Figuerola • A. Quarta • T. Pellegrino • C. Wilhelm • M. A. Garcı´a

Received: 28 August 2012 / Accepted: 13 February 2013 Ó Springer Science+Business Media Dordrecht 2013

Abstract In this work we investigate the magnetic properties of iron oxide nanoparticles obtained by twostep synthesis (seeded-growth route) with sizes that range from 6 to 18 nm. The initial seeds result monocrystalline and exhibit ferromagnetic behavior with low saturation field. The subsequent growth of a shell enhances the anisotropy inducing magnetic frustration, and, consequently, reducing its magnetization. This increase in anisotropy occurs suddenly at a certain size (*10 nm). Electronic and structural analysis with X-ray absorption spectroscopy indicates a step reduction in the A. Espinosa (&)  M. Garcı´a-Herna´ndez  J. Jime´nez de la Morena Instituto de Ciencia de Materiales de Madrid (ICMM), Consejo Superior de Investigaciones Cientı´ficas, Sor Juana Ine´s de la Cruz 3, Cantoblanco, 28049 Madrid, Spain e-mail: [email protected]; [email protected] A. Mun˜oz-Noval  A. Serrano  J. Jime´nez de la Morena  M. A. Garcı´a Instituto de Cera´mica y Vidrio (ICV), Consejo Superior de Investigaciones Cientı´ficas, Kelsen 5, Cantoblanco, 28049 Madrid, Spain

oxidation state as the particle reaches 10 nm size while keeping its overall structure in spite of the magnetic polydispersity. The formation of antiphase magnetic boundaries due to island percolation in the growing shells is hypothesized to be the mechanism responsible of the magnetic behavior, as a direct consequence of the two-step synthesis route of the nanoparticles. Keywords Magnetic nanoparticles  Superparamagnetism  X-ray absorption spectroscopy  Antiphase boundaries A. Quarta  T. Pellegrino National Nanotechnology Laboratory of CNR-NANO, via per Arnesano km 5, 73100 Lecce, Italy T. Pellegrino Istituto Italiano di Tecnologia, Via Morego 30, 16163 Genoa, Italy C. Wilhelm Laboratoire Matie`re et Syste`mes Complexes (MSC), UMR 7057, CNRS and Universite´ Paris Diderot, 10 rue Alice Domon et Le´onie Duquet, 75205 Paris Cedex 13, France

A. Mun˜oz-Noval SpLine Spanish CRG Beamline at the ESRF, 6 Rue Jules Horowitz, BP 220, 38043 Grenoble Cedex 09, France A. Figuerola Departament de Quı´mica Inorga`nica i Institut de Nanocie`ncia i Nanotecnologia, Universitat de Barcelona, Martı´ i Franque´s 1-11, 08028 Barcelona, Spain

123

Page 2 of 13

J Nanopart Res (2013) 15:1514

Introduction

Experimental

The unique physical properties of magnetic nanoparticles (NPs) have triggered many studies for the thoughtful understanding of issues such as superparamagnetism, monodomain regime, or surface-enhanced anisotropy (Kodama 1999; Lu et al. 2007). Magnetic NPs are useful in many applications in the field of data storage (Sun et al. 2000), catalysis (Lu et al. 2004), or biomedicine (Pankhurst et al. 2003). In the latter field, iron oxide nanoparticles are preferred due to their higher biocompatibility compared to other magnetic materials (Gupta and Gupta 2005). A key issue for the use of iron oxide nanoparticles in biomedical applications is to be monocrystalline and monodisperse (Figuerola et al. 2010; Fortin et al. 2007; Poddar et al. 2002). Indeed, magnetic properties at the nanoscale are strongly dependent on the particle size (Batlle and Labarta 2002), so that polydispersity in size systematically limits the nanomaterial performance. However, grain boundaries in iron oxide NPs are known to induce magnetic frustration which translates into degradation of their saturation magnetization and magnetic behavior (Coey 1971; Shendruk et al. 2007). Among the different methods for the synthesis of iron oxide nanoparticles, the seeded-growth method (Sun and Zheng 2002) is well suited, since it provides monocrystalline nanoparticles with narrow size distribution and easy control of the particle size. However, recent studies have shown that such features (monocrystallinity and narrow size distributions) are necessary conditions but may not be sufficient to provide efficient magnetic properties (Luigjes et al. 2011; Levy et al. 2011). We have recently demonstrated that monodisperse and monocrystalline iron oxide NPs prepared by the seeded-growth method show unexpectedly low heating power for magnetic hyperthermia (Levy et al. 2011), orders of magnitude below theoretical predictions. Preliminary analysis pointed out to a poor magnetic shell structure as a key actor in the magnetic properties decline. In this work we report new analysis of the magnetic properties of these NPs together with a study of their structure by means of X-ray absorption spectroscopy. Taken together, our findings allow correlating the magnetic behavior of the studied NPs with structural features imposed by the two-step synthesis route.

Iron oxide nanoparticles of different sizes were prepared following a previously published procedure (Levy et al. 2011). All syntheses were carried out under air-free conditions, using a standard Schlenk line set-up. For the preparation of 6 nm iron oxide nanoparticles, 1.5 mmol of oleic acid (C17H33CO2H or OLAC, 90 %), 0.75 mmol of oleylamine (C17H33NH2 or OLAM, 70 %), and 1.25 mmol of hexadecane-1,2-diol (C16H32O2 or HDD, technical grade) were dissolved in 10 ml of 1-octadecene (C18H36 or ODE, 90 %) in a three-neck flask and were degassed while stirring at 120 °C for 30 min under vacuum conditions. Under a blanket of nitrogen, the temperature of the reaction mixture is decreased until 60 °C is reached. At this temperature, 1 mmol of iron pentacarbonyl (Fe(CO)5, 98 %) dissolved in 1 ml of degassed ODE is injected into the flask. All the chemicals are purchased from Aldrich. The temperature is then increased to 280 °C at a rate of 10 °C/min and the reaction mixture is left at this temperature for 1 h. The temperature is then decreased to 130 °C and the reaction mixture is exposed to air at this temperature for 1 h to enhance oxidation. The reaction mixture is later cooled to room temperature and the nanoparticles are washed by repeated additions of acetone/isopropanol mixtures (50 % volume) and finally redispersed in toluene. For the synthesis of iron oxide nanoparticles with diameters above 6 nm, the same procedure must be followed until the reaction has proceeded at 280 °C for 1 h. After that, the temperature is decreased to 250 °C and a solution of 2 mmol of Fe(CO)5 and 2.8 mmol of OLAC in 5 ml of ODE is injected dropwise with an automatic syringe pump at a controlled rate of 0.1 ml/min. Once the injection is finished, the reaction mixture is left at 250 °C for 1 h. When the reaction is completed, the size of the iron oxide nanoparticles is close to 8 nm. However, further identical injections allow the nanoparticles to grow until a maximum diameter of 18 nm. The X-ray absorption measurements (X-ray absorption near-edge structure (XANES) and extended x-ray absorption fine structure (EXAFS)) were performed at the Fe K-edge energy at room temperature in conventional transmission mode using ionization chambers as detectors. The experiments were carried out at the BM25 Spanish CRG Beamline (SpLine) of the ESRF

123

J Nanopart Res (2013) 15:1514

(European Synchrotron Radiation Facility, Grenoble, France). Each powder sample was sandwiched between two pieces of kapton tape located on the beam path. The amount of material was calculated to optimize the expected signal-to-noise ratio. Several scans were taken, in order to obtain a good signal-tonoise ratio. Fe metal foil was simultaneously measured for energy calibration. FeO, a-Fe2O3, c-Fe2O3, and Fe3O4 powders were chosen as bulk references. Data analysis was carried out using Athena (Newville 2001) program identifying the beginning of the absorption edge, E0, and the pre-edge and post-edge backgrounds. The spectra were subjected to a background subtraction and normalized by the edge jump. Viper (Klementev 2001) program was used to process the normalized EXAFS signal as a function of the modulus of the photoelectron wavevector, k, v(k) in the range of ˚ -1. A k2 weighting was used to properly 2.5–12 A account for the signal at high-energy region of the spectrum. Previously, the signal was filtered by a Hanning window type and the phase and amplitude have been recalculated using FEFF code (Rehr 1993) version 6.01. The magnetic characterization of the samples was carried out using superconducting quantum interference device (SQUID) (Quantum Design, San Diego, USA). Isothermal magnetization curves at low temperature (5 K) and the thermal dependence of the magnetization of liquid samples were recorded upon zero-field cooling (ZFC) and field cooling (FC) protocols. The concentration of iron oxide NPs in the liquid suspensions was quantified by the induction-coupled plasma atomic emission spectrometer (ICP-AES) with

Page 3 of 13

the instrument iCap 6000 Series (Thermo Ficher, Waltham, USA). This technique was used to normalize the magnetization values to the real content of iron oxide in the samples, where 25 lL of each sample was digested in 2.5 ml of aqua regia and 48 h later, diluted to a known volume with milli-Q ultrapure water. Transmission electron microscopy (TEM) images were acquired on a JEOL JEM-1011 and with a JEOL JEM 2200FS transmission electron microscopes (JEOL, Tokyo, Japan) operating, respectively, at an accelerating voltage of 100 and 200 kV. Colloidal organic solution of iron oxide samples were drop cast onto a thin carbon film supported by a copper TEM grid and then allowing the solvent to evaporate.

Results Structural characterization Transmission electron microscopy (TEM) (Fig. 1) and X-ray diffraction (XRD) characterization of the NPs showed they were monocrystalline with a very narrow size distribution with uniform size that ranges from 6 to 18 nm as previously reported (Levy et al. 2011). The diffraction peaks detected correspond to the standard pattern of a spinel structure. Within the magnetic iron oxides, magnetite (Fe3O4) and maghemite (c-Fe2O3) present a similar inverse spinel structure with almost identical lattice parameters (8.385 ˚ for Fe3O4 and c-Fe2O3 (Wang et al. and 8.346 A 2009), respectively, although with different space groups, Fd3m and P4332). Therefore, distinguishing

Fig. 1 Representative TEM images of iron oxide nanoparticles of different diameters in toluene. From left to right 6.5 nm (SD = 0.1); 8.6 nm (SD = 0.1); 13.4 nm (SD = 0.1). The scale bar is marked at the bottom of each micrograph and represents 50 nm in all cases

123

Page 4 of 13 Fig. 2 XANES spectra at the Fe K-edge of a the references (Fe metal, FeO, Fe3O4, c-Fe2O3, and aFe2O3) and b of the NPs. c Estimated oxidation state of NPs. d Derivatives of NPs spectra compared to those of spinel references Fe3O4 and c-Fe2O3

J Nanopart Res (2013) 15:1514

a

b

between both phases is considerably difficult with the above-mentioned techniques. However, magnetite contains a mixed valence of Fe2? and Fe3? with a ratio of Fe2?/Fe3? = ‘ (with an average oxidation state of ?2.67) while in maghemite, Fe ions are fully oxidized to Fe3? with the presence of cation vacancies in octahedral positions to compensate the increased positive charge (Wang et al. 2009; Corrias et al. 2009). In both cases the Fe ions occupy both tetrahedral and octahedral sites. Hence, X-ray absorption spectroscopy techniques such as XANES and EXAFS at the Fe K-edge become powerful techniques to study the structure of the iron oxides (Corrias et al. 2000; 2009; Espinosa et al. 2012; Jiao et al. 2006). XANES provides information of the oxidation state, and EXAFS gives information about the local environment around the absorbing Fe ions, including distances between atoms and coordination numbers of surrounding shells. Consequently, an analysis combining both techniques can reveal the structure of very small systems to account for the Fe short-range geometry and lead to the identification of crystal phases (Jime´nez-Villacorta et al. 2010). The XANES Fe K-edge measurements of the nanoparticles are presented in Fig. 2. The XANES data for the bulk references (Fe foil, FeO, Fe3O4, a-Fe2O3, and c-Fe2O3) are also shown for comparison. As a first observation, all samples have an average oxidation state similar to spinel references (Fig. 2a),

123

c

d

confirming the results from the powder XRD measurement (Levy et al. 2011). However, both the white line (the first peak above the edge) and the second peak (*7148 eV) are very similar to those in the magnetite reference. More precisely, considering the linear interdependence between energy edge shifts and the oxidation state of compounds of the same species, we can estimate the valence state of the samples (Kunzl 1932), interpolating those from the references. The energy edge value was taken from the first maximum of derivative spectra. Also, the comparison of XAS derivatives allows spotting subtle differences (Cabot et al. 2007). This comparison points out that the shape of the derivative spectra reproduces quite well the transition features detected in magnetite. Following these procedures, the calculated oxidation state of the samples with sizes up to 10 nm is around ?2.8 and, as the NPs size increases above 10 nm, the absorption edge shifts to lower energies reaching ?2.67, the value corresponding to reference Fe3O4 (Fig. 2a). A reduction of the oxidation state with increasing the particle size can be explained in terms of finite size effects since magnetite NPs commonly exhibit an oxidized outer shell containing Fe3? ions. Smaller NPs present a larger fraction of surface atoms, and hence, a higher average oxidation. In this case, a progressive reduction of the oxidation state with the particle size would be expected. However, the behavior observed in Fig. 1c exhibits a qualitative change when the particles reach

J Nanopart Res (2013) 15:1514

Page 5 of 13

the 10 nm size suggesting that this behavior is not related to a spontaneous surface oxidation. The pseudo-radial distribution function around the Fe atom (Fig. 4) is obtained by performing Fourier Transform (FT) of the EXAFS signal weighted by k2 in ˚ -1). The FT module is the range of k (2.70–12.16 A related to the coordination spheres of atoms that absorb the radiation and its intensity is proportional to number of surrounded neighbors. It is relevant to note that the analysis of all samples, as well as references, was carried out under the same conditions, so that they are comparable. The EXAFS signals v(k)*k2 of the samples are almost identical to each other and they are pretty similar to that of magnetite (Fig. 3a). Nevertheless, the ˚ -1 present in maghevery subtle decrease around 5 A mite (Corrias et al. 2009) can also be observed in our samples. Therefore, we cannot rule out a fraction of maghemite in all the NPs explored although XANES analysis points out to an oxidation state close to magnetite, particularly for larger NPs. On the other hand, FT module as a function of distance of EXAFS signals (Fig. 3b) presents two main peaks or average distances for both references and samples. The first peak corresponds to the radial distribution of oxygen with respect to iron (Fe–O distance) in either tetrahedral or octahedral sites (Signorini et al. 2003). The second peak is an average overlapping of the positions

Fig. 3 a EXAFS filtered experimental and fitting signals, v(k)*k2 and b Fourier transform function (experimental and fitting (red) for the NPs and Fe3O4 and c-Fe2O3 references. The simulated fitting signals are displayed in red color. (Color figure online)

a

of iron atoms (Fe–Fe distances located in octahedral and tetrahedral sites (Carta et al. 2008)). These average distances are characteristic of Fe3O4 and c-Fe2O3 due to their similar spinel structure, although the differences in octahedral and tetrahedral of Fe positions may modify specially the shape of the second peak. For example, the presence of cation vacancies in maghemite is reflected in a reduction in the FT intensity corresponding to higher Fe–Fe distances. We performed a fit of the experimental data with ˚ . One theoretical signals in the range of R of 0.8–3.8 A ˚ for oxygen shell was used to fit the first peak 1.96 A Fe–O bonds and three distances for the second peak, ˚, for the different Fe–Fe bonds 2.50, 3.20, and 3.70 A respectively (Carta et al. 2008). These values were initially fitted considering Fe oxide references while keeping fixed the number of neighbors. This last analysis rendered the amplitude reduction factor S20 to be used in the fit of NPs (S20 = 0.6). Table 1 gives the obtained EXAFS parameters fitting results. In general, a reduction of coordination of both shells has been detected in the NPs due to the decrease of the total number of neighbors, which may be a consequence of the nanosize of the structure. This is especially noticeable in the decrease of the Fe–Fe relative intensity with respect to Fe–O peak in comparison with spinel references. The radial

b

123

Page 6 of 13

J Nanopart Res (2013) 15:1514

Table 1 Fitting structural parameters of the coordination shells of reference Fe3O4 and nanoparticles at the Fe K-edge Sample

Bond

˚) Rj (A

Nj

˚ -1) rj (A

Fe3O4

Fe–O

1.97

3.7

0.006

Fe–Fe

3.01

3.2

0.008

Fe–Fe

3.50

6.4

0.008

Fe–Fe

3.72

3.2

0.008

Fe–O

1.94

3.2

0.008

Fe–Fe

3.05

2.5

0.008

Fe–Fe

3.53

1.7

0.010

Fe–Fe

3.68

0.9

0.009

Fe–O

1.94

3.2

0.008

Fe–Fe

3.05

2.4

0.008

Fe–Fe

3.53

1.8

0.010

Fe–Fe

3.66

0.9

0.010

Fe–O

1.94

3.2

0.007

Fe–Fe

3.05

2.4

0.008

Fe–Fe Fe–Fe

3.52 3.67

2.0 0.9

0.010 0.012

Fe–O

1.94

3.2

0.009

Fe–Fe

3.05

3.1

0.009

Fe–Fe

3.51

2.4

0.010

Fe–Fe

3.67

1.1

0.011

Fe–O

1.95

3.2

0.009

Fe–Fe

3.05

3.1

0.010

Fe–Fe

3.52

2.4

0.012

Fe–Fe

3.68

1.1

0.010

Fe–O

1.94

3.2

0.008

Fe–Fe

3.04

3.2

0.010

Fe–Fe

3.51

2.4

0.011

Fe–Fe

3.68

1.2

0.011

6 nm

7 nm

8 nm

10 nm

13 nm

18 nm

N is the coordination no. (±0.4, and fixed for reference), Fe–O and Fe–Fe are the average interatomic distances (±0.02–0.04), and r are the Debye–Waller factors (±2 9 10-3). The overall reduction factor S20 was fixed to 0.6

distribution of distances is mostly centered around Fe3O4 values. The Fe–O bond fitted distance ˚ ) for the samples hardly varies with the (1.94–1.95 A diameter of nanoparticles and it is slightly lower than the corresponding Fe3O4 reference. However, the Fe– O peak is slightly less intense for the largest NPs, which may be indicative of a presence of oxygen vacancies. The fitted Fe–Fe distances are quite similar between samples with slight variations. Since the EXAFS signals v(k)*k2 do not seem to vary with the diameter of the particle, any subtle differences in the

123

Fe–Fe distribution peak between samples can be ascribed to different tetrahedral or octahedral sites occupation, randomly distributed and generating cation vacancies. Therefore, observing the EXAFS analysis, we cannot detect a clear visible trend as the diameter of the NPs increases, concerning the occupation of Fe or O sites. This fact can be explained in terms of a complex migration of different vacancies that contributes to a broad distribution, specially in the Fe–Fe peak, but responsible for the change in the overall oxidation state stemming from the XANES analysis. Thus, the structural characterization of the samples points out a very similar structure without clear differences as a function of the particle size. Magnetic properties Figure 4 presents the magnetization curves of the samples at 300 K and at 5 K upon FC and ZFC conditions. The magnetization was normalized to the iron oxide mass in the sample determined by ICP (Levy et al. 2011). The curves measured at 5 K exhibit larger MS values than those obtained at 300 K except for the largest NPs (*18 nm), for which MS is almost identical at both temperatures. Moreover, for NPs smaller than 10 nm, the curves at 5 K are fully saturated for applied fields around 10 KOe while larger NPs exhibit a non-saturated component even at 50 KOe. The slope of this non-saturated component increases with the particle size (Levy et al. 2011). A numerical analysis of the curves provided the results summarized in Fig. 5. The saturation magnetization (MS) values at 5 K are significantly smaller than those of bulk magnetite and maghemite (87 and 93 emu/g, respectively) (Cullity and Graham 2009). For both, 5 and 300 K, the observed tendency is MS to decrease with the particle size. The remanence magnetization (MR) at 5 K is lower than one half of MS, as it could be expected for a system composed of non-interacting NPs with a random distribution of anisotropy axis. The coercive field (HC) at low temperature (Fig. 5c) increases with the particle size (as expected for this size range) from 35 Oe for the 6 nm seeds, to 400 Oe for the 18 nm NPs. There is clear jump in the trend for the 10 nm size for which HC exhibits a sharp increase with a value of 283 Oe. This non-monotonic trend points out that the system becomes more complex magnetically,

J Nanopart Res (2013) 15:1514 Fig. 4 Magnetization curves at 5 K upon ZFC (black) and FC (Hcooling = 50 KOe) (red) and at 300 K (blue). Insets show a detail of the low-field region. (Color figure online)

Page 7 of 13

a

b

c

d

e

f

behavior probably ascribed to an enhancement of the anisotropy. On the other hand, the initial 6 nm seeds and the 7 nm NPs do not exhibit any exchange bias (HEB) shift. For 8 nm NPs, there is a weak HEB (*8 Oe), while above 10 nm size, the exchange bias field is noticeable (HEB = 50 Oe for 10 nm NPs), and increases with the particle size as summarized in Fig. 5d. The thermal dependence of the magnetization measured upon FC and ZFC conditions is presented in Fig. 6. The curves present the typical profile of superparamagnetic nanoparticles with a well-defined blocking temperature (TB). The curve obtained upon ZFC appears always below that obtained upon FC except for the case of 7 nm NPs for which they appear inverted above TB. This behavior has been associated

with the presence of antiferromagnetic and ferromagnetic correlations, spin frustration, or irreversibility mechanisms (Azad et al. 2005). The inverted difference between ZFC/FC curves in this sample is not very pronounced, but small ferrimagnetic areas of uncompensated spins and disorder may lead to this behavior (Villafuerte-Castrejo´n et al. 2011). A numerical analysis of these curves is presented in Fig. 7. The blocking temperature (TB) increases with the particle size exhibiting a deviation from the monotonic behavior for NPs having sizes larger than 10 nm. The irreversibility temperature (TIR) matches TB for the initial seed, as expected for their narrow size distribution, but not for larger NPs. As already mentioned, for the 7 nm NPs, FC and ZFC curves cross each other so it is not obvious to define the TIR but the crossover

123

Page 8 of 13

J Nanopart Res (2013) 15:1514

a

b

c

d

Fig. 5 a Saturation magnetization of the NPs at 5 K and 300 K, b remanence magnetization, c coercive field at 5 K, and d exchange bias field at 5 K

seems to take place very close to TB. For larger NPs, the TIR results are significantly larger than TB, as a consequence of a broad distribution of sizes and interparticle correlations (Guardia et al. 2011; Batlle et al. 1993). The full width at half maximum (FWHM) of the ZFC curve as a function of the particle size is shown in Fig. 7b. The tendency is an increase of the FWHM with the particle size, as expected. Summing up, for NPs larger than 10 nm size, a clear deviation of the monotonic behavior with a sudden increase of the widths is observed. Also an increase of the corresponding coercive fields, exchange bias fields, and blocking temperatures is detected. This FWHM is associated with the distribution of energy barriers to overcome spin inversion. Therefore, the results summarized in Fig. 7 point out a significant variation of the anisotropy when the particles reach 10 nm size.

Discussion The magnetic characterization of the NPs points out two remarkable features: the progressive reduction of MS as the particle size increases, and the qualitative

123

change at 10 nm size, where most of the magnetic parameters exhibit a singularity. The latter was also observed in the oxidation state measured from XANES spectra (Fig. 2). However, no clear differences were observed in the structural characterization carried out with EXAFS, XRD, or TEM (Levy et al. 2011). Therefore, the samples exhibit some magnetic disorder without structural disorder but related with their electronic configuration (ultimately responsible for the magnetic behavior). The initial seeds show very low MS values, about 40 emu/g, which is less than half of the bulk value. NPs usually exhibit reduced MS values in comparison with bulk values due to size and surface effects. These effects have been commonly accepted as unavoidable, but recent works (Batlle et al. 2011) showed that it is possible to reach MS values very close to those of the bulk material (*90 %) by the appropriated preparation methods. Thus, the initial seeds already exhibit some kind of magnetic disorder despite their excellent structural properties. This magnetic disorder is likely to be at the surface where the lack of symmetry favors the increase of oxygen and cation vacancies inducing magnetic frustration and, consequently, the formation

J Nanopart Res (2013) 15:1514 Fig. 6 Thermal dependence of the magnetization for the nanoparticles applying a field of Hmeas = 100 Oe upon ZFC (black) and FC (Hcooling = 50 KOe) (red). (Color figure online)

of a shell with reduced magnetization (Batlle and Labarta 2002; Shendruk et al. 2007; Pellegrino et al. 2004; Guardia et al. 2007; Pankhurst and Pollard 1991; Morales et al. 1999; Linderoth et al. 1994). Actually, the oxidation state of the Fe cations obtained from XAS measurements is close to that of magnetite, with a possible fraction of maghemite. This suggests the existence of some oxygen and cation vacancies (electrically compensated by changes in the Fe oxidation state) that accounts for the reduced magnetization layer (Shendruk et al. 2007). However, the MS values for the larger NPs grown on these seeds cannot be ascribed solely to surface effects. In that case, MS should increase with the particle size since this shell amounts for a smaller fraction of the whole NPs, while the opposite trend is

Page 9 of 13

a

b

c

d

e

f

actually observed. Because of the reduced MS value of the initial seed, the shell grows on an already magnetically poor surface, yielding a magnetic layer with magnetic frustration. Luigjes et al. prepared monocrystalline iron oxide particles following the same synthesis method and for a size larger than 10 nm, their ferromagnetic diameters were also smaller than expected from TEM, therefore, exhibiting magnetic polydispersity. They correlated the presence of defects in the initial seeds with a degradation of the magnetic properties. The other key feature of the NPs is the observed qualitative change for magnetic parameters in NPs larger than 10 nm size. The sudden increase in the coercive field and blocking temperatures for NPs of 10 nm size suggest that this is not due to size effects,

123

Page 10 of 13

J Nanopart Res (2013) 15:1514

a

b

Fig. 7 a Blocking temperature (TB) and irreversibility temperature (TIR) and b full width at half maximum (FWMH) of the ZFC curve as a function of the particle size

but to the development of an additional source of anisotropy for larger sizes. The anisotropy constant can be determined from the thermal dependence of the magnetization upon ZFC according to: Ku ¼ 25

kB T V

ð1Þ

V being the nanoparticle volume and kB, the Boltzmann constant. The anisotropy can also be calculated from the coercive field at low T according to:  1=2 ! 2Ku T Hc ¼ a 1 ð2Þ TB MS with a = 0.48 for NPs with a random distribution of anisotropy. We calculated the anisotropy using both equations for the different particle sizes. Results are presented in Fig. 8. The anisotropy constant was determined according to Eq. (1) using both the real volume of the particles obtained from the TEM analysis and also the effective magnetic volume: since the magnetization of the particles is smaller than the bulk one, we assume that just a fraction of the particle volume is ferromagnetically ordered. This magnetic volume is obtained multiplying the TEM volume by the experimental MS of the particles and dividing by the saturation magnetization of bulk magnetite (87 emu/g). Similarly, the calculation of the anisotropy with Eq. (2) was performed using the experimental value of MS and that of the bulk MS-bulk. This makes sense since the NPs have a magnetically frustrated layer and a ferromagnetic core. This core is the main contributor to the saturation magnetization and its contribution is close to the bulk value.

123

Fig. 8 Uniaxial anisotropy constant for the nanoparticles obtained from the values of HC (Eq. 1) and from the blocking temperature (Eq. 2). In the first case the calculation was carried out using the NPs volume obtained from TEM (HC, VTEM) and the magnetic volume determined from the experimental MS (HC, VMAG) as described in the text. For the second, calculation was performed using the experimental MS value (TB, MS-EXP) and the MS bulk one for magnetite (TB, MS-BULK)

While it is tough to determine which is the best method to calculate the anisotropy, there are some common trends in the size dependence of the anisotropy constant K. Initially, the anisotropy decreases with the particle size but there is a sudden increase when reaching the 10 nm size. Actually this increase stars already for the 8 nm NPs. The K values for the initial seeds are in the range 35–55 kJ/m3 depending on the calculation method. These values are higher than the anisotropy constant for bulk magnetite (11–13 kJ/m3) (Goya et al. 2003) or maghemite (5–15 kJ/m3) (Cornell and Schwertmann 2003; Caizer et al. 2003; Shendruk et al. 2007). It has been reported that for iron oxide nanoparticles the lack of symmetry

J Nanopart Res (2013) 15:1514

at the surface increases the anisotropy leading to a surface shell with enhanced anisotropy (surface anisotropy) (Shendruk et al. 2007). The surface is a preferential site for vacants migration inducing magnetic frustration, spin canting, and spin glasses. As the particle size increases, this shell represents a smaller fraction of the whole nanoparticle volume, so the effective anisotropy decreases toward the volume value. However, the sudden increase of anisotropy observed when reaching the 10 nm size cannot be explained solely by this surface effect. This trend is confirmed for both calculations methods, pointing out to some kind of qualitative change for this size, and suggesting that this phenomenon is not intrinsic of the layers, but related to their growing process. The increase of anisotropy coincides with the onset of the exchange bias (Fig. 5d), a phenomenon that takes place when two magnetic phases with different features comes in contact (Meiklejohn and Bean 1957; Nogue´s and Schuller 1999; Nogue´s et al. 2005). Actually, exchange bias has also been observed in phases with the same composition if the interface shows some kind of magnetic disorder (Martı´nez et al. 1998). These interfaces cannot correspond to the seedgrowing layer one. In this case, the exchange bias should be observed already for all the NPs with a grown layer, in particular for the 8 nm NPs (2 nm thickness is enough to observe exchange bias). Antiphase boundaries (APB) are a well-known magnetic defect that does not require structural defects. APB are been extensively observed and studied in magnetite films epitaxially grown on MgO (Arora et al. 2005; Novakova et al. 2003; Margulies et al. 1997; Eerenstein et al. 2003). Within the seeded-growth method, islands of magnetite epitaxially grow over the initial seed facets until they percolate forming a continuously film. Our ‘‘substrate’’ is not flat but rounded, inducing slight mismatches at the edges where the island percolate favoring the formation of APB (in the case of MgO substrates, the difference in lattice parameter induces the mismatch). The presence of these antiphase boundaries in the magnetite film grown over MgO promotes exchange bias effects as it is indeed observed. Moreover, the magnetization curves at low temperature upon FC and ZFC exhibit a shape very similar to that we found here (see Fig. 4f in reference Arora et al. 2005). Hence, we conclude that the origin anisotropy enhancement is the appearance of APB when the layers grow over the initial seeds.

Page 11 of 13

The appearance of the APB could be partially at the root of the observed decrease of MS with the particle size. In fact, the development of APB induces some magnetic frustration that takes place abruptly at a certain size but is not the main one responsible for the observed decrease MS. The nanoparticles seem to exhibit magnetic memory. Such a memory could be erased by heating over their order temperature (850 °C), but this process is not possible since the nanoparticles would aggregate while growing. Consequently, there is no chance to delete the magnetic complexity of the nanoparticles, suggesting that two pot methods are not convenient to obtain iron oxide nanoparticles suitable for biomedical applications.

Conclusions In summary, although the controlled design of iron oxide nanoparticles using a two-step seeded-growth route is attractive, the drop in the magnetic properties limits their subsequent applications for the biomedical field. On the one hand, the presence of defects in the initial seeds induces a shell with a reduced magnetization. The layers growing on this depleted magnetic shell exhibit low magnetic properties yielding to a progressive decrease of the saturation magnetization when increasing the particle size. This problem can be overcome by using seeds with excellent crystal quality as recently demonstrated (Batlle et al. 2011). On the other hand, when the islands nucleated on the initial seeds surfaces percolate they form antiphase boundaries that increase the anisotropy and induce magnetic frustration. Consequently, the nanoparticles undergo qualitative changes in the magnetic properties for a critical size (*10 nm). Finally, the NPs growth as a two-step process results in a magnetic signature reflecting their dual structure; the initial seed and the layer subsequently grown. Overcoming this issue is not obvious, so that, despite excellent structural properties, NPs produced by two pot methods are for now not suitable for biomedical applications. Acknowledgments This work was supported by the European project MAGNIFYCO (Contract NMP4-SL-2009-228622); Spanish Ministerio de Economı´a y Competitividad grants CSD2009-00013, MAT2011-27470-C02-02, and FIS-200806249; and Madrid Region Council project NANOBIO MAGNET (S2009/MAT-1726). A. F. acknowledges financial support from the Spanish MICINN through CTQ2009-06959

123

Page 12 of 13 and for a Ramo´n y Cajal Fellowship (RYC-2010-05821). We thank F. Gazeau and M. Levy for their essential contributions, and C. Prieto for helpful discussions on the XAS measurements. We acknowledge ESRF for beamtime and BM25 personnel for technical support.

References Arora SK, Sofin RGS, Nolan A, Shvets IV (2005) Antiphase boundaries induced exchange coupling in epitaxial Fe3O4 thin films. J Magn Magn Mater 286:463–467 Azad AK, Mellerga˚rd A, Eriksson SG, Ivanov SA, Yunus SM, Lindberg F, Svensson G, Mathieu R (2005) Structural and magnetic properties of La2FeCrO6 studied by neutron diffraction, electron diffraction and magnetometry. Mater Res Bull 40:1633–1644 Batlle X, Labarta A (2002) Finite-size effects in fine particles: magnetic and transport properties 15:R15–R42 Batlle X, Garcı´a del Muro M, Tejada J, Pfeiffer H, Go¨rnert P, Sinn E (1993) Magnetic study of M-type doped barium ferrite nanocrystalline powders. J Appl Phys 74(5):3333 Batlle X, Pe´rez N, Guardia P et al (2011) Magnetic nanoparticles with bulklike properties (invited). J Appl Phys 109(7):07B 524 Cabot A, Puntes VF, Shevchenko E, Yin Y, Balcells L, Marcus M a, Hughes SM, Alivisatos P (2007) Vacancy coalescence during oxidation of iron nanoparticles. J Am Chem Soc 129(34):10358–10360 Caizer C, Savii C, Popovici M (2003) Magnetic behaviour of iron oxide nanoparticles dispersed in a silica matrix. Mater Sci Eng, B 97:129–134 Carta D, Casula MF, Mountjoy G, Corrias A (2008) Formation and cation distribution in supported manganese ferrite nanoparticles: an X-ray absorption study. Phys Chem Chem Phys 10(21):3108–3117 Coey JMD (1971) Noncollinear Spin Arrangement in Ultrafine Ferrimagnetic Crystallites. Phys Rev Lett 17:1140–1142 Cornell RM, Schwertmann U (2003) The Iron Oxides, 2nd edn. Wiley, Hoboken Corrias A, Ennas G, Mountjoy G, Paschina G (2000) An X-ray absorption spectroscopy study of the Fe K edge in nanosized maghemite and in Fe2O3–SiO2 nanocomposites. Physical Chemistry Chemical Physics 2:1045–1050 Corrias A, Mountjoy G, Loche D, Puntes V, Falqui A, Zanella M, Parak WJ, Casula MF (2009) Identifying Spinel Phases in Nearly Monodisperse Iron Oxide Colloidal Nanocrystal. J Phys Chem C 113(43):18667–18675 Cullity BD, Graham CD (2009) Introduction to magnetic materials, 2nd edn. Wiley-IEEE Press, Hoboken Eerenstein W, Palstra T, Hibma T, Celotto S (2003) Diffusive motion of antiphase domain boundaries in Fe3O4 films. Phys Rev B 68(1):1–7 Espinosa A, Serrano A, Llavona A, Jime´nez de la Morena J, Abuin M, Figuerola A, Pellegrino T, Ferna´ndez JF, Garcı´aHerna´ndez M, Castro GR, Garcı´a MA (2012) On the discrimination between magnetite and maghemite by XANES measurements in fluorescence mode. Meas Sci Technol 23(1):015602 Figuerola A, Di Corato R, Manna L, Pellegrino T (2010) From iron oxide nanoparticles towards advanced iron-based

123

J Nanopart Res (2013) 15:1514 inorganic materials designed for biomedical applications. Pharmacological research: the official journal of the Italian Pharmacological Society 62(2):126–143 Fortin J-P, Wilhelm C, Servais J, Me´nager C, Bacri J-C, Gazeau F (2007) Size-sorted anionic iron oxide nanomagnets as colloidal mediators for magnetic hyperthermia. J Am Chem Soc 129(9):2628–2635 Goya GF, Berquo´ TS, Fonseca FC, Morales MP (2003) Static and dynamic magnetic properties of spherical magnetite nanoparticles. J Appl Phys 94(5):3520–3528 Guardia P, Batlle-Brugal B, Roca AG, Iglesias O, Morales MP, Serna CJ, Labarta A, Batlle X (2007) Surfactant effects in magnetite nanoparticles of controlled size. J Magn Magn Mater 316(2):e756–e759 Guardia P, Labarta A, Batlle X (2011) Tuning the Size, the Shape, and the Magnetic Properties of Iron Oxide Nanoparticles. J Phys Chem C 115:390–396 Gupta AK, Gupta M (2005) Synthesis and surface engineering of iron oxide nanoparticles for biomedical applications. Biomaterials 26(18):3995–4021 Jiao F, Jumas J-C, Womes M, Chadwick AV, Harrison A, Bruce PG (2006) Synthesis of ordered mesoporous Fe3O4 and gamma-Fe2O3 with crystalline walls using post-template reduction/oxidation. J Am Chem Soc 128(39):12905– 12909 Jime´nez-Villacorta F, Sa´nchez-Marcos J, Ce´spedes E, Garcı´aHerna´ndez M, Prieto C (2010) Effects of interparticle interactions in magnetic Fe/Si3N4 granular systems. Phys Rev B 82(13):1–12 Klementev KV (2001) Extraction of the fine structure from X-ray absorption spectra. J Phys D Appl Phys 34:209–217 Kodama RH (1999) Magnetic nanoparticles. J Magn Magn Mater 200:359–372 Kunzl V (1932) A linear dependence of energy levels on the valency of elements. Collection des Travaux Chimiques de Tche´coslovaquie 4:213–224 Levy M, Quarta A, Espinosa A, Figuerola A, Wilhem C, Garcı´aHerna´ndez M, Genovese A, Falqui A, Alloyeau D, Buonsanti R, Cozzoli PD, Garcı´a MA, Gazeau F, Pellegrino T (2011) Correlating Magneto-Structural Properties to Hyperthermia Performance of Highly Monodisperse Iron Oxide Nanoparticles Prepared by a Seeded-Growth Route. Chem Mater 23:4170–4180 Linderoth S, Hendriksen PV, Bodker F, Wells S, Davies K, Charles SW, Morup S (1994) On spin-canting in maghemite particles. J Appl Phys 75(10):6583–6585 Lu A-H, Schmidt W, Matoussevitch N, Bo¨nnemann H, Spliethoff B, Tesche B, Bill E, Kiefer W, Schu¨th F (2004) Nanoengineering of a magnetically separable hydrogenation catalyst. Angew Chem Int Ed Engl 43(33):4303–4306 Lu A-H, Salabas EL, Schu¨th F (2007) Magnetic nanoparticles: synthesis, protection, functionalization, and application. Angew Chem Int Ed Engl 46(8):1222–1244 Luigjes B, Woudenberg SMC, Groot RD, Meeldijk JD, Galvis HMT, Jong KPD, Philipse AP, Ern BH (2011) Diverging Geometric and Magnetic Size Distributions of Iron Oxide Nanocrystals. J Phys Chem C 115:14598–14605 Margulies DT, Parker FT, Rudee ML, Spada FE, Chapman JN, Aitchison PR, Berkowitz AE (1997) Origin of the Anomalous Magnetic Behavior in Single Crystal Fe3O4 Films. Phys. Rev. Lett. 79(25):5162–5165

J Nanopart Res (2013) 15:1514 Martı´nez B, Obradors X, Balcells L, Rouanet A, Monty C (1998) Low Temperature Surface Spin-Glass Transition in g-Fe2O3 Nanoparticles. Phys Rev Lett 80:181–184 Meiklejohn WH, Bean CP (1957) New Magnetic Anisotropy. Physical Review 105:904–913 Morales MP, Veintemillas-Verdaguer S, Montero MI, Serna CJ, Roig A, Casas L, Martinez B, Sandiumenge F (1999) Surface and Internal Spin Canting in c-Fe2O3 Nanoparticles. Chem Mater 12:3058–3064 Newville M (2001) Ifeffit: interactive exafs analysis and feff fitting. J Synchrotron Radiat 8:322–324 Nogue´s J, Schuller IK (1999) Exchange Bias. J Magn Magn Mater 192:203–232 Nogue´s J, Sort J, Langlais V, Skumryev V, Surin˜ach S, Mun˜oz JS, Baro´ MD (2005) Exchange bias in nanostructures. Phys Rep 422(3):65–117 Novakova AA, Lanchinskaya VY, Volkov AV, Gendler TS, Kiseleva TY, Moskvina MA, Zezin SB (2003) Magnetic properties of polymer nanocomposites containing iron oxide nanoparticles. 259:354–357 Pankhurst QA, Pollard RJ (1991) Origin of the Spin-Canting Anomaly in Small Ferrimagnetic Particles. Phys Rev Lett 67:248–250 Pankhurst QA, Connolly J, Jones SK, Dobson J (2003) Applications of magnetic nanoparticles in biomedicine. J Phys D-Appl Phys 36:R167–R181 Pellegrino T, Manna L, Kudera S, Liedl T, Koktysh D, Rogach AL, Keller S, Ra¨dler J, Natile G, Parak WJ (2004) Hydrophobic Nanocrystals Coated with an Amphiphilic Polymer Shell : A General Route to Water Soluble Nanocrystals. Nano Lett 4:703–707

Page 13 of 13 Poddar P, Fried T, Markovich G (2002) First-order metalinsulator transition and spin-polarized tunneling in Fe3O4 nanocrystals. Phys Rev B 65(17):3–6 Rehr JJ (1993) Recent Developments in Multiple-Scattering Calculations of XAFS and XANES. Jpn J Appl Phys 32:8–12 Shendruk TN, Desautels RD, Southern BW, van Lierop J (2007) The effect of surface spin disorder on the magnetism of cFe2O3 nanoparticle dispersions. Nanotechnology 18(45): 455704 Signorini L, Pasquini L, Savini L, Carboni R, Boscherini F, Bonetti E, Giglia A, Pedio M, Mahne N, Nannarone S (2003) Size-dependent oxidation in iron/iron oxide coreshell nanoparticles. Phys Rev B 68(19):1–8 Sun S, Zheng H (2002) Size-Controlled Synthesis of Magnetite Nanoparticles. J Am Chem Soc 124:8204–8205 Sun SH, Murray SB, Weller D, Folks L, Moser A (2000) Monodisperse FePt Nanoparticles and Ferromagnetic FePt Nanocrystal Superlattices. Science 287(5460):1989–1992 Villafuerte-Castrejo´n ME, Garcı´a-Guaderrama M, Fuentes L, ´ , Garcı´aPrado-Gonjal J, Gonza´lez AM, de la Rubia MA Herna´ndez M, Mora´n E (2011) New Fe3?/Cr3? perovskites with anomalous transport properties: the solid solution La(x)Bi(1-x)Fe(0.5)Cr(0.5)O3 (0.4 B x B 1). Inorg Chem 50(17):8340–8347 Wang C, Baer DR, Amonette JE, Engelhard MH, Antony J, Qiang Y (2009) Morphology and electronic structure of the oxide shell on the surface of iron nanoparticles. J Am Chem Soc 131(25):8824–8832

123