In situ synthesis and characterization of magnetic nanoparticles in shells of biodegradable polyelectrolyte microcapsules

Materials Science and Engineering C 45 (2014) 225–233

Contents lists available at ScienceDirect

Materials Science and Engineering C journal homepage: www.elsevier.com/locate/msec

In situ synthesis and characterization of magnetic nanoparticles in shells of biodegradable polyelectrolyte microcapsules I.S. Lyubutin a, S.S. Starchikov a,⁎, T.V. Bukreeva a, I.A. Lysenko b, S.N. Sulyanov a, N.Yu. Korotkov a, S.S. Rumyantseva a, I.V. Marchenko a, K.O. Funtov a, A.L. Vasiliev a,b a b

Shubnikov Institute of Crystallography, Russian Academy of Sciences, Leninsky av. 59, Moscow 119333, Russia National Research Center “Kurchatov Institute”, pl. Akademika Kurchatova 1, Moscow 123182, Russia

a r t i c l e

i n f o

Article history: Received 4 March 2014 Received in revised form 26 August 2014 Accepted 11 September 2014 Available online 16 September 2014 Keywords: Nanostructured materials Magnetic properties Mössbauer spectroscopy

a b s t r a c t Hollow microcapsules with the shell composed of biodegradable polyelectrolytes modified with the maghemite nanoparticles were fabricated by in situ synthesis. The nanoparticles were synthesized from the iron salt and the base directly on the capsule shells prepared by “layer by layer” technique. An average diameter of the capsule was about 6.7 μm while the average thickness of the capsule shell was 0.9 μm. XRD, HRTEM, Raman and Mössbauer spectroscopy data revealed that the iron oxide nanoparticles have the crystal structure of maghemite γ-Fe2O3. The nanoparticles were highly monodisperse with medium size of 7.5 nm. The Mössbauer spectroscopy data revealed that the nanoparticles have marked superparamagnetic behavior which was retained up to room temperature due to slow spin relaxation. Because of that, the microcapsules can be handled by an external magnetic field. Both these properties are important for target drug delivery. Based on the Mössbauer spectroscopy data, the spin blocking temperatures TB of about 90 K was found for the particles with size D ≤ 5 nm and TB ≈ 250 K for particles with D ≥ 6 nm. The anisotropy constants K were determined using the superparamagnetic approximation and in the low temperature approximation of collective magnetic excitation. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Magnetic nanostructures are considered as promising materials in high-density magnetic storage devices, magnetic heads, and sensors [1,2]. Recently, more attention was paid to the application of magnetic nanostructures in biology and medicine and focused mainly on the targeted drug delivery [3], hyperthermia treatment [4], magnetic resonance imaging (as contrast media) [5], separation of biochemical products [6] and gene manipulation and immunoassays [7]. For the application of magnetic nanoparticles in these areas, they should be non-toxic, chemically stable and biocompatible with cells, tissues and the whole body. Iron oxides (especially magnetite-Fe3O4, maghemite γ-Fe2O3 and hematite α-Fe2O3) were widely studied [8,9] and are gradually implemented in various fields of biomedicine. The polyelectrolyte capsules prepared by a layer-by-layer technique are promising drug delivery systems [10,11]. But a simpler way for targeting of polyelectrolyte capsules is the magnetic modification, which is possible by the incorporation of magnetic nanoparticles into capsules. One of the main methods for the adaptation of polyelectrolyte capsules for targeted drug delivery by magnetic field is the electrostatic adsorption of previously synthesized magnetic nanoparticles on the surface of oppositely charged layer of the capsule shell [12–14]. In addition ⁎ Corresponding author. E-mail address: [email protected] (S.S. Starchikov).

http://dx.doi.org/10.1016/j.msec.2014.09.017 0928-4931/© 2014 Elsevier B.V. All rights reserved.

to this physical approach there are some chemical approaches for the magnetic modification of polyelectrolyte capsules. Magnetic nanoparticles were synthesized inside the capsules due to the possibility of keeping the component concentrations in a closed volume [15]. But the presence of nanoparticles inside the capsule reduces the effective volume of the cavity for functional compound encapsulation. Only a few works are devoted to the magnetic modification of polyelectrolyte shells by chemical techniques. In Ref. [16], spherical shells composed of polyallylamine and magnetite nanoparticles were obtained by formation of the capsule internal layers from polyallylamine/ citrate–ion complex. Then the anions were replaced with hydroxide ions and capsules were placed in a solution of two- and trivalent iron salts for magnetite preparation followed by removing the external polyelectrolyte layers in a highly alkaline medium. In Ref. [17], the in situ synthesis of magnetite Fe3O4 nanoparticles on polyelectrolyte shells was performed using a palladium catalyst adsorbed on polycation layers. Magnetite was obtained by nitrate ion reduction with the aid of dimethylboranamine in the presence of trivalent iron ions in an aqueous suspension of capsules at 60 °С. Previously, we proposed alternative technique for the magnetic modification of polyelectrolyte capsules [18]. Synthesis of magnetite nanoparticles by chemical condensation of Fe3+ and Fe2+ ions in a suspension of capsules from polyallylamine and polystyrene sulfonate has led to the preparation of magnetic capsules. The process was performed under heating of the reaction mixture at 80 °C. However, dealing with

226

I.S. Lyubutin et al. / Materials Science and Engineering C 45 (2014) 225–233

biochemical materials, in particular for the encapsulation of proteins, such heating is undesirable. In this work, we have modified polyelectrolyte capsules by this method using biopolymer polylysine and dextran sulfate and heating the reagents to a temperature suitable for biomaterials. After formation of the nanocomposite capsules the comprehensive study of their structural and magnetic properties was performed. The particular efforts in this work were devoted to distinguish between magnetite Fe3O4 and maghemite γ-Fe2O3 nanoparticles and to the study the size effect on the magnetic behavior of these materials. Usually, the X-ray diffraction (XRD) and high resolution transmission electron microscopy (HRTEM) are mainly used for the characterization of magnetic nanocapsules. Meanwhile magnetite Fe3O4 and maghemite γ-Fe2O3 have the similar spinel-type crystal structure and their identification by structural methods is hardly possible especially if the system contains a mixture of various iron oxide compounds. In our study different and complementary techniques were used for the evaluation of particle structural, physicochemical, and magnetic properties. Along with XRD and TEM, we applied the Mössbauer spectroscopy and Raman spectroscopy for the particle characterization.

2. Sample preparation and characterization techniques

Fig. 2. Scheme of capsule fabrication.

After the adsorption of each polymer layer the capsules were washed three times with deionized water to remove the excess of a polyelectrolyte. The capsule shell with the composition of (PLL/DS)4 was formed by eight cycles of polyelectrolyte adsorption. To obtain hollow capsules, the CaCO3 core was decomposed by EDTA. Scheme of capsule fabrication is shown in Fig. 2. 2.3. Synthesis of magnetic nanoparticles The in situ synthesis of magnetic nanoparticles in a suspension of polyelectrolyte capsules was performed using the Elmore method of magnetite fabrication [21] (chemical condensation of Fe3 + and Fe2 + ions with molar ratio 2:1 on capsule shells by adding a base):

2.1. Materials 2FeCl3 + FeCl2 + 8NH4OH → Fe3O4↓ + 8NH4Cl + 4H2O. Poly-L-lysine hydrobromide (PLL, Mw 40–60 kDa), sodium chloride, calcium chloride, sodium carbonate, trisodium salt of ethylenediamine tetraacetic acid (EDTA) and iron(II) chloride tetrahydrate were purchased from Sigma-Aldrich (Germany), and iron(III) chloride hexahydrate was purchased from Acros Organics (US). Dextran sulfate sodium salt from Leuconostoc spp. (DS, Mw ~500 kDa) and ammonium hydroxide were purchased from Fluka, Sweden and US, accordingly.

2.2. Preparation of hollow capsules We used the particles of calcium carbonate CaCO3 as cores for microcapsules. The spherical CaCO3 microparticles were prepared by colloidal crystallization from supersaturated solution as previously described in Ref [19]. The rapid mixing of equal volumes of 0.33 M CaCl2 and Na2CO3 solutions with further intensive agitation of reaction mixture on a magnetic stirrer during 30 s and suspension exposure for 10–15 min at room temperature without stirring led to the formation of the CaCO 3 microspherulites with a typical size of 5–6 μm (Fig. 1(a) and (b)). Polyelectrolyte capsules were fabricated by layer-by-layer deposition technique [20]. PLL and DS were alternately adsorbed on the surface of core particles from solutions with polymer concentration of 2 mg/mL and NaCl concentration of 11.7 mg/mL (see Fig. 2).

Here (↓) means precipitation. In our work 0.15 М iron(III) and iron(II) chloride solutions and suspension of hollow capsules were mixed and heated up to 37 °C in the reaction vessel. Then the vessel was put into a water bath at 37 °C which was placed on a magnetic stirrer. Total mass of iron ions was 0.25%, the number of capsules was approximately 108/ml. At intensive stirring 0.75 mL of 25% ammonium hydroxide solution was added to the suspension and the obtained mixture was stirred for 6 min. After that the capsules with iron oxide nanoparticles formed in the shells were washed three times with deionized water under separating the capsules from the supernatant by centrifugation (2000 rpm for 1 min). For further investigation, the nanocomposite capsules were centrifuged and the residue was lyophilized. 2.4. Characterization techniques The microstructure, phase composition, magnetic and electronic properties of the samples were studied by scanning (SEM) and transmission (TEM) electron microscopy including high resolution TEM (HRTEM), X-ray powder diffraction, Raman spectroscopy, and Mössbauer spectroscopy. X-ray diffraction studies were performed on a “Belok” station installed on the synchrotron source from a bending magnet of the storage ring in

Fig. 1. SEM images of spherical microparticles of the CaCO3 spherulites used as the cores for microcapsules: (a) low magnification image demonstrating the uniform size of the particles and (b) higher magnification image of the particle surface.

I.S. Lyubutin et al. / Materials Science and Engineering C 45 (2014) 225–233

the National Research Center “Kurchatov Institute” (NRC-KI), Moscow, Russia [22]. Two-dimensional recording of powder XRD was then transformed into a one-dimensional pattern, which has ensured high statistical accuracy in determining of the I(2θ) dependence [23,24]. The Rietveld method was used for the crystal structure refinement. For scanning electron microscope the JSM 8440 (JEOL, Japan) was used. TEM analyses were conducted on the powders which were dispersed in alcohol by sonicating for half an hour. The resulting dilute suspensions were deposited onto copper TEM grids coated with a thin Formvar-carbon support film, and allowed to air-dry. Preliminary microstructural analyses were performed in a Tecnai G230ST (FEI, US) transmission/scanning electron microscope (TEM/STEM) operating at an accelerating voltage of 300 kV. Several samples were studied using a transmission/scanning electron microscope Titan 80-300 TEM/STEM (FEI, Oregon, US) equipped with a spherical aberration (Cs) corrector (electron probe corrector), a high angle annular dark field (HAADF) detector, an atmospheric thin-window energy dispersive X-ray (EDX) spectrometer (Phoenix System, EDAX, Mahwah, NJ, US), and post-column Gatan energy filter (GIF; Gatan, Pleasanton, CA, US). The TEM analyses were performed at 300 kV. Digital Micrograph (DM) (Gatan, Pleasanton, CA, US) software was used for the micrograph and diffraction pattern analysis. JEMS software [25] was used for image simulations. Raman spectroscopy measurements were carried out at room temperature with a 473 nm Optronic DPSS blue laser as an excitation source and a 647.1 nm Spectra-Physics Beamlock 2080 Krypton red laser. Princeton Instruments Acton SP2500 Monochromator/Spectrograph and Spec-10 system with a nitrogen cooled CCD detector were used to collect the Raman signal. The Mössbauer absorption spectra from 57Fe nuclei were recorded at temperatures between 10 and 295 K using a standard MS1100Em spectrometer operating in the constant acceleration mode. The gamma-ray source 57Co(Rh) was at room temperature. Isomer shifts were measured relative to the reference α-Fe sample (18-μm-thick iron foil annealed in hydrogen) at room temperature. Computer processing of spectra was carried out using Univem MS software. For the low temperature Mossbauer measurements the closed-cycled helium cryostat was used [26].

227

3. Results and discussion 3.1. Synthesis Hollow capsules with the shell composed of biodegradable polyelectrolytes (PLL/DS)4 modified with iron oxide nanoparticles in the shell were fabricated. Fig. 1 shows SEM images of spherical microparticles of the CaCO3 spherulites used as the cores for microcapsules. The particles are uniform with a typical size of 5–6 μm. The subsequent fabrication of capsules and the in situ synthesis of magnetic nanoparticles in the capsule shells are shown schematically in Fig. 2. It can be assumed that during the in situ synthesis of maghemite nanoparticles on the shells of polyelectrolyte capsules by proposed method the following processes take place. After adding iron chloride solutions to capsule suspension iron cations adsorb partly on the surface of the outer negatively charged layer of DS and partly inside the shell on noncompensated negatively charged groups of DS. Some iron ions remain in the solution. After adding a base, nanoparticles are formed on the surface and inside the shell and also in the solution. Nanoparticles formed in the solution are adsorbed onto the surface of capsules and partly stabilized by polyelectrolytes of the shell. The sediment of the capsules was observed to move in the magnetic field of a constant magnet. 3.2. TEM observations The bright field TEM image of the capsule with iron oxide particles inside is presented in Fig. 3(a). We found that an average diameter of the capsule was around 7 μm. The thickness of the capsule shell varied between 0.2 and 1 μm. The examination of the areas inside and outside the shell demonstrated that mostly the particles were inside the shell, however some particles were clustered outside the shells (Fig. 3(b)). TEM and especially HRTEM studies of these groups of the particles were more convenient because their images were not overlapped with the capsule shells (Fig. 3(c) and (d)). These groups of the particles

Fig. 3. Bright field TEM images of: (a) — the microcapsule with iron oxide nanoparticles inside the shell of microcapsule; (b) — the shell of the microcapsule with the particles inside and outside; (c) — the group of iron oxide particles inside the capsule; (d) — the group of iron oxide particles exhibited faceted morphology. In few areas the Moiré patterns appeared due to the particle overlapping, and (e) — the particle size distribution.

228

I.S. Lyubutin et al. / Materials Science and Engineering C 45 (2014) 225–233

Fig. 4. Selected area diffraction pattern (SADP) from the group of the particles. The results of SADP analysis are in Table 1.

were used to study particle morphology and size distribution (Fig. 3e). The conventional TEM and HRTEM investigations demonstrated that particles adopted polyhedral morphology, and typical size histogram demonstrated a sufficiently homogeneous distribution. The nanoparticles in shells are rather monodisperse. About 94% of the particles have the size in the range of 6–9 nm with a medium size of 7.5 nm (Fig. 3e). To determine the structure of nanoparticles the electron diffraction was applied as a first step. Typical selected area diffraction pattern (SADP) from the group of the particles is shown in Fig. 4. The shape of the SADP rings indicated that the particles were positioned on a substrate randomly. For the structure identification, we examined more than 20 compounds of Fe oxides and few hydroxides adopted cubic structure. It was found that all of the SADP rings could be as arising from four compounds: Fe3O4-magnetite, FeOOH-iron(III) hydroxide, γ-Fe2O3-maghemite and α-Fe2O3-hematite (see Table 1). The last compound had the worse fit to the experimental data due to relatively large difference in interplanar distances correspondent to the 1st and 2nd

diffraction rings. However, we note that it was difficult to conclude, because the major distinctions between the diffraction patterns of these structures are the difference in intensities of the rings and subtle difference in lattice parameters. Further set of HRTEM experiments was performed to select the crystal structure among the four candidates. Several HRTEM images were analyzed and one example is presented in Fig. 5a. Fast Fourier Transform (FFT) obtained from the HREM image of individual particle (Fig. 5b) was compared to simulated SADP (Fig. 5c). Such an analysis of several particles pointed to FeOOH-iron(III) hydroxide (space group Fd3m, a = 0.836 nm) and γ-Fe2O3-maghemite (space group Fd3m , a = 0.833 nm) as possible candidates with the same spacings and systematic absences of diffraction maxima as observed experimentally. Two other compounds were withdrawn because of the discrepancy of FFT and simulated SADP. Further, we tried HREM image simulation for the final trial of two compounds. A series of simulated HRTEM images of the maghemite γ-Fe2O3 particle in [310] projection is presented in Fig. 5d. These data matches extremely well at crystal thickness t = 4–8 nm and defocus values Δf = 60–62 nm. That range of thickness corresponded to the particle sizes and experimental defocus values were close to the Scherzer one. The image simulations presented in Fig. 5d were obtained for slightly tilted (~ 2.75°) zone axis away B = [310] toward the [133]. To further test the model, the image simulation results for defocus Δf = 61 nm and thickness t = 4.5 nm were inserted in the experimental image (Fig. 5e) and excellent match was found between the two images. The simulated images of FeOOH-iron(III) hydroxide fitted worse, but finally we could not rule out that compound. Thus, after the ED and HREM analysis the most probable candidate was γ-Fe2O3-maghemite and the less probable was FeOOH-iron(III) hydroxide. 3.3. XRD data The crystal structure and phase purity of the sample were examined by X-ray diffraction, and the XRD pattern of microcapsules is shown in Fig. 6. We used the Rietveld method for the crystal structure refinement, and for the peak interpolation, the symmetric pseudo-Voigt function was applied. The Bragg R-factor RBr = 2.1% was obtained. All reflections of the sample can be readily indexed to the cubic phase with a spinel type structure (the space group Fd3m). The calculated unit cell parameter is a = 8.3545(3) Å. An estimation of the crystallite size by Scherrer's

Table 1 The comparative analysis of the obtained SADP (shown in Fig. 4) with the acceptable data for iron oxides and hydroxide. Sample

α-Fe2O3 Hematite

Fe3O4 Magnetite  Fd3m

R3cH [28]

[27]

γ-Fe2O3 Maghemite  Fd3m

FeO(OH) Iron(III) hydroxide  Fd3m

[29]

[30]

Ring, №.

Intensity

d, A

d, A

Int

hkl

d, A

Int

hkl

d, A

Int

hkl

d, A

Int

hkl

1

Low

4.99

4.84

8.1

111

3.68

28.9

102

111

3.08

2.96

28.1

220

2.7

100

104

2.96

27.4

022

3

High

2.63

110

2.52

60.3

113

2.18

113 222 400

73.7

Medium

100 8.2 20.6

2.52

4

2.53 2.42 2.1

2.29 2.21 2.07

2.1 18 1.9

006 113 202

2.09

60.5

004

5

Low

1.76

1.71

9.6

224

1.84

37.8

204

1.71

11.3

224

6

Medium

1.67

1.61

32.3

115

7

Medium

1.53

1.48 1.41

44.8 1.1

440 531

1.69 1.6 1.49 1.45

46.1 8.8 32.6 31.8

116 108 214 300

110 111 210 112 220 310 113 222 320 321 400 421 224 115

0.8

Medium

3.1 3.6 8 3.9 35.4 2.9 100 4.1 1.1 1.3 17.3 2.3 11.9 33.1

4.83

2

5.89 4.81 3.73 3.41 2.95 2.63 2.52 2.41 2.31 2.22 2.09 1.82 1.7 1.61 1.55 1.52 1.47

1.4 1.1 45

520 521 440

1.61 1.61 1.48

30.3 2.5 100

115 333 044

I.S. Lyubutin et al. / Materials Science and Engineering C 45 (2014) 225–233

229

Fig. 5. HRTEM study of maghemite γ-Fe2O3: (a) HRTEM image of maghemite γ-Fe2O3 particle in [310] projection, red line denotes the area of FFT analysis; (b) — corresponding FFT power spectra; (c) — simulated [310] zone axis diffraction pattern; (d) simulations showing the effect of thickness and objective lens defocus on contrast in HRTEM images obtained in [310] zone axis and (e) enlarged HRTEM image with simulations obtained using structural model of maghemite γ-Fe2O3 in the inset outlined with white. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

formula of the peak broadening [31] assuming the spherical shape of the particles gives an average size D = 12(1) nm. This correlates with the TEM data if some particle aggregation is taken into account. It is known that both magnetite Fe3O4 and maghemite γ-Fe2O3 have a spinel-type crystal structure, and it is difficult to separate these phases in XRD pattern. Stoichiometric magnetite has a cubic close-packed structure with the unit-cell parameter of 8.396–8.400 Å [32,33]. In maghemite, the unit cell is somewhat smaller (8.33–8.34 Å) because of the formation of cation vacancies and the smaller size of Fe3+ ions as compared to Fe2+ [32,34]. The unit cell value obtained for our nanoparticles is very close to the maghemite value which implies that the iron oxide nanoparticles in the shell of microcapsules have the γ-Fe2O3 structure. Depending on the method of sample preparation, cation vacancies in maghemite can distribute randomly or orderly over the octahedral sites of the spinel structure [32,33,35,36]. The vacancy ordering may result in

additional peaks in XRD patterns of γ-Fe2O3. In particular, the peaks (110), (210) and (211) can appear in a small angle region. In our study, these peaks were not found on the XRD pattern, which may indicate random distribution of cation vacancies in γ-Fe2O3. On the other hand, the background in the small angle region is high (see inset in Fig. 6), and these reflexes can be hardly visible because of their small intensity. To make the confident conclusion about the formation of the γ-Fe2O3 nanoparticles we applied the Raman and Mössbauer spectroscopy techniques for investigation of the nanoparticles incorporated into the microcapsules. 3.4. Raman spectroscopy data Raman spectroscopy was applied to study the phase composition of nanoparticles. The Raman spectra were recorded with blue and red

230

I.S. Lyubutin et al. / Materials Science and Engineering C 45 (2014) 225–233

broad maximum developing at about 1070 cm−1 does not overlap the peaks of magnetite.

3.5. Mössbauer spectroscopy data

Fig. 6. XRD pattern of iron oxide nanoparticles in shells of microcapsules and the Rietveld refinement of the crystal structure (the red solid line) corresponding to the spinel-type maghemite γ-Fe2O3. In insert, the small intensive (111) reflex of maghemite is shown in the small angle region at high background. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

lasers and a representative pattern is shown in Fig. 7. Three intensive peaks corresponding to the Raman active phonon modes are present in the spectra. The frequencies of the main bands are in the range 360–370, 505–515 and 690–710 cm−1. These values match well the Raman spectra for maghemite previously reported for bulk and nano materials [37–39]. Three Raman active phonon modes at 365, 511 and 700 cm−1 have been observed in γ-Fe2O3 maghemite which relate to the T2g, Eg and A1g optical transitions in iron ions, respectively [37]. Thus, the main bands observed in Raman spectra of iron nanoparticles in the capsule shells correspond to maghemite. Frequencies of the modes can vary with the method of material preparation, with the size of particles [40], and also with the distribution of vacancies in the maghemite crystal unit cell [37,38]. In the Raman spectra of magnetite Fe3O4 only mode at 670 cm−1 (A1g) is pronounced [37]. This line cannot be excluded from the spectrum of our nanoparticles since the observed vibration bands are rather broadening. However, from the Raman vibrational spectra we may conclude that the dominate phase in the iron oxide nanoparticles is maghemite γ-Fe2O3. In the inset of Fig. 7, the Raman spectrum of hollow polyelectrolyte capsules without magnetic nanoparticles in the shells is also shown in the same frequency range. The pattern is smooth without visible peaks. A

Fig. 7. The room temperature Raman spectrum of iron oxide nanoparticles in the shells of microcapsules. Arrows indicate the positions of phonon modes in γ-Fe2O3 maghemite related to the T2g, Eg and A1g optical transitions in iron ions. In the inset, the Raman spectrum of hollow polyelectrolyte capsules without magnetic nanoparticles in the shells is shown. The spectra were obtained with the 473 nm blue laser.

The Fe57-Mössbauer spectra of the samples were recorded at temperature ranging from 10 to 297 K, and the spectra evolution is shown in Fig. 8. At low temperatures a well-defined magnetic sextet pattern is observed with slightly broaden and asymmetric lines. At the lowest temperature of 10 K, the Mössbauer spectrum can be well fit to two dominating six-line magnetic components A and B. A small-intensive component (C) with broaden lines and lower magnetic splitting can be added for the best spectra fitting (Fig. 9a). The hyperfine parameters of the A and B components are as follows: isomer shifts δ(A) = 0.41 mm/s, δ(B) = 0.46 mm/s, quadrupole shifts ε(A) = − 0.03 mm/ s, ε(B) = 0.01 mm/s, and magnetic hyperfine fields Hhf(A) = 50.5 T and Hhf(B) = 52.6 T. These parameters are the typical characteristics of Fe3+ ions in maghemite γ-Fe2O3 nanoparticles [36]. For the small intensive magnetic component δ(C) = 0.44 mm/s, ε(C) = −0.06 mm/s, and bHhf(C)N = 47.0 T. No traces of ferrous ions (related to magnetite) were observed. Obviously, the A and B spectral components should be attributed to the tetrahedral and octahedral sites of maghemite, respectively. Usually, the δ and Hhf values for iron ions in the tetrahedral sites are less than in the octahedral sites because of the shorter Fe\O bond-length and the higher degree of covalence in the Fe\O bonds at A sites compared with B sites of the spinel structure [41]. It seems that the third component C with the lowest Hhf(C) value corresponds to the iron ions in the surface layer. The area of this component is about 15% of the total iron content in the sample. For the ball-shape particles with a diameter of 7.5 nm, the corresponding thickness of the surface iron layer is about

Fig. 8. The Fe57-Mössbauer spectra of the iron oxide nanoparticles in shells of microcapsules recorded at temperatures between 10 and 297 K.

I.S. Lyubutin et al. / Materials Science and Engineering C 45 (2014) 225–233

231

Fig. 9. The Fe57-Mössbauer spectra of the iron oxide nanoparticles in the shells of microcapsules recorded at temperatures of 10 K (a) and 297 K (b). Two solid lines (green and red) are the calculated spectrum components corresponding to the tetrahedral (A) and octahedral (B) sites of Fe3+ ions in maghemite. The third small-intensive line corresponds to iron ions at the surface of the nanoparticles. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

0.25 nm. This means that only thin atomic layer is subjected to the surface effect. The chemical formula of maghemite γ-Fe2O3 can be represented as nonstoichiometric magnetite containing vacancies □ in the octahedral sites (Fe)[Fe5/6□1/6]2O4. The intensity ratio of the (A) and [B] Mössbauer components in the vacant γ-Fe2O3 is expected to be near 1:1.67. In our case, the observed A/B ratio in nanoparticles is about 1.65 that is very close to the expected value for maghemite. This is an additional support for the maghemite phase present in the shells of the microcapsules obtained. As temperature increases up to 120 K, the shape of the spectra remains the same (Fig. 8). No trace of the Verwey transition, characteristic feature of magnetite, was observed. However, a paramagnetic doublet component appears above 70 K signaling the transition of some part of nanoparticles from magnetic ordering to a paramagnetic state (Fig. 8). The hyperfine parameters of the doublet are as follows: the isomer shift δ = 0.43(1) mm/s, the quadrupole splitting Δ = 0.79 (1) mm/s, and the half-line width Г = 0.81(2) mm/s. These are the typical characteristics of the high spin Fe3+ ions in maghemite [36,41,42]. Obviously, this paramagnetic component appears from small particles which spin-blocking temperature TB is below 70 K. From the area of the spectrum lines at 90 K, the fraction of paramagnetic particles can be evaluated as ~ 7% (of the total iron content in the sample). This value corresponds to the relative content of particles with the size below 5 nm obtained from the particle-size distribution function found from the TEM data (see Fig. 3e). Thus, we can decide that the value TB ≈ 90 K is a characteristic of maghemite nanoparticles with the size D ≤ 5 nm. Fig. 10 shows the temperature dependence of the area of the paramagnetic doublet component. This area remains almost stable (only slightly increasing from 9 to 11%) at temperatures between 90 and 250 K, and then it drastically increases to about 18%. Taking the TEM data, this increment in the area corresponds to the fraction of particles with the size D ≈ 6 nm. This indicates that the spin blocking temperature for the 6 nm particles of maghemite in microcapsules is TB ≈ 250 K (Fig. 10). At room temperature the hyperfine parameters of the paramagnetic doublet are δ = 0.34(1) mm/s, Δ = 0.70(1) mm/s and Г = 0.61(1) mm/s, which are close to values reported before for nano-maghemite [36,42]. At temperatures above 120 K, the lines of magnetic sextet became broadened extending to the inner part of the spectrum (Fig. 8), and an additional broad magnetic component should be added for the adequate spectra fitting. The line broadening increases as temperature rises, which is the signature of superparamagnetic relaxation of the iron magnetic moments. At room temperature, the six-line shape of the superparamagnetic component is barely discerned (Fig. 9b). This indicates that the main fraction of nanoparticles (about 80%) is in the superparamagnetic state at 297 K. According to the TEM data, this

fraction is mainly composed from the particles with the size of 7– 9 nm. Thus it means that (due to slow spin relaxation) the blocking temperature for such particles is above the room temperature. Here we remind that the Neel temperature of bulk maghemite is very high TN ≈ 863 K [43]. The blocking temperature is defined as the temperature below which the magnetization is stable and the particles behave as a magnetically ordered crystal [44,45]. The conception of stability is defined by the time of thermal fluctuation τ of magnetic moments among the various easy magnetization directions [46] τ ¼ τ0 exp ðV K=k T Þ

ð1Þ

where τ0 is the material constant (~ 10− 9 s) [44], V is the volume of nanoparticles, K is the anisotropy constant, k is the Boltzmann constant and T is the temperature. This time should be confronted with the time scale of the experimental method used for study of the magnetic properties. In Mössbauer measurements this time is of the order of Larmor precession time of the Fe nuclei magnetic moment (τ ~ 10−8 s), and it is much shorter than in magnetic measurements (τ ~ 102 s). Then, the expression for thermal equilibrium state of the assembly of uniaxial singledomain nanoparticles with the stable magnetization can be given in the form [44]: VK ¼ 2:3 k T B ;

ð2Þ

where TB is the blocking temperature. Using the values of V and TB obtained in our experiments we estimated the anisotropy constants K = 4 × 105 erg/сm3 and 7 × 105 erg/сm3 for the particles with the sizes of 5 and 6 nm, respectively. These data show that the surface anisotropy (which relative contribution to the total anisotropy in 5 nm particles is higher than in 6 nm particles) is less than the core particle anisotropy. This can be attributed to the coating of the nanoparticles by nonmagnetic polyelectrolyte media. The K values obtained for our nanoparticles are close to those found previously for the nanomaghemite with size of 5 nm [42] and 7 nm [47]. On the other hand, the K values in our nano-maghemite are about one order of magnitude higher than in the bulk material (which is about 4.7 × 104 erg/cm3) [48,49]. The temperature dependence of magnetic fields Hhf for octahedral and tetrahedral sites in maghemite nanoparticles is shown in Fig. 11. We found that at low temperatures the Hhf (T) dependence for A and B sites can be well approximated by a linear law (Fig. 11). As suggested by Mørup et al. [45,50], such behavior is a characteristic of collective magnetic excitation (i.e., small spin fluctuation around an easy direction) which should be differentiated from superparamagnetic relaxation when the magnetization direction fluctuates among the various

232

I.S. Lyubutin et al. / Materials Science and Engineering C 45 (2014) 225–233

correlates with the observed decreasing of K value with increasing the particle volume, which leads to the K of 4.7 · 104 erg/cm3 in the bulk material [48,49]. In these samples the maghemite nanoparticles are in the shells of microcapsules and are separated from each other by nonmagnetic polyelectrolyte media. Such a coating results in the decreasing of the interparticle magnetic interaction or even can exclude it at all [36,51]. The particles are small enough (D b 10 nm) to be considered as the single domain magnetic nanoparticles [45]. In addition, we have found that the resonance absorption in the Mössbauer spectra decreases with temperature rise (Fig. 8), which can be explained by the particle thermal vibrations that cause a lowering of the recoil free fraction [52]. This can be associated with the presence of polyelectrolyte media coating that reduces the magnetic and mechanical coupling of the particles. 4. Conclusions Fig. 10. Effect of temperature on the area of the paramagnetic component in the Mössbauer spectra of maghemite nanoparticles in the shells of the microcapsules. TB (1) and TB (2) are the spin blocking temperatures for the maghemite nanoparticles with the size D ≤ 5 and 6 nm, respectively. Solid line is guide for the eye.

easy directions. The correlation time of collective magnetic excitation is much shorter than the observation time in Mössbauer measurements, and the average value of the magnetic hyperfine field in the low temperature limit (KV / kT ≫ 1) can be given as [45] H hf ðV; T Þ ≈ Hhf ðV bulk ; T Þ ½1−kT=2KV ;

ð3Þ

where Hhf (Vbulk, T) is the hyperfine field in a large crystal at the same temperature (i.e., in the absence of collective magnetic excitations). Using Eq. (3) we can estimate the values of the anisotropy constant of the maghemite nanoparticles with the average size of 7.5 nm: 5

3

5

3

K octa ¼ 6:1  10 erg=сm and K tetra ¼ 4:8  10 erg=сm : It is interesting that the K values obtained in approximation of collective magnetic excitations are a little bit different for the octahedral and tetrahedral sites. It may imply that the spin systems of A and B sublattices in ferrimagnetic maghemite fluctuate somewhat incoherently due to intra sublattice exchange coupling. The K values obtained for small particles (5 and 6 nm) from the superparamagnetic approximation are close to the value for larger particles (7.5 nm) found in the low temperature approximation of collective magnetic excitation. This

Hollow microcapsules from biodegradable polyelectrolytes poly-Llysine and dextran sulfate were fabricated by layer-by-layer adsorption technique. The capsule shells (PLL/DS)4 were modified with the maghemite nanoparticles by in situ synthesis. XRD, HRTEM, Raman and Mössbauer spectroscopy data revealed that the iron oxide nanoparticles have the crystal structure of maghemite γ-Fe2O3. TEM images show that an average diameter of the capsule is about 6.7 μm while the average thickness of the capsule shell is 0.9 μm. The maghemite nanoparticles formed in the capsule shell are rather monodisperse with the medium size of 7.5 nm, indicating a unique and efficient mechanism of formation of the microencapsulation. Furthermore, the most important data provided by Mössbauer spectroscopy, reveals that approximately 80% of all maghemite nanoparticles with the size of 7–9 nm have marked superparamagnetic behavior which is retained up to room temperature due to slow spin relaxation, this allows directing the microcapsules to a place determined with the presence of a magnetic field of a constant magnet. Additionally, the porous nature of the core and dissolution of CaCO3 after coating of the layerby-layer, makes it possible to absorb molecules that may remain trapped within the microcapsule. Therefore, mild conditions in the synthesis of magnetic nanoparticles incorporated in the microcapsules may enable encapsulating bioactive substances without loss of its biological activity. Finally, because of the characteristics of the microcapsules obtained, it may suggest its application in the administration of drugs, emphasizing therapies for hyperthermia. Acknowledgments Support by the Russian Scientific Foundation (Project #14-1200848) is acknowledged. T.V.B., I.A.L., S.S.R., and I.V.M. appreciate the support of the Russian Foundation for Basic Research (grant 14-0201217a). References

Fig. 11. The temperature dependence of magnetic hyperfine fields Hhf for octahedral and tetrahedral sites in maghemite nanoparticles obtained from Mössbauer spectra. Dashed lines are the approximation to a linear law of collective magnetic excitation. Solid lines are guide for the eye.

[1] H. Zeng, J. Li, J.P. Liu, Z.L. Wang, S. Sun, Exchange-coupled nanocomposite magnets by nanoparticle self-assembly, Nature 420 (2002) 395–398, http://dx.doi.org/10. 1038/nature01208. [2] R.C. O'Handley, Modern Magnetic Materials: Principles and Applications, John Wiley, New York, 2000. [3] H.-S. Cho, Z. Dong, G.M. Pauletti, J. Zhang, H. Xu, H. Gu, et al., Fluorescent, superparamagnetic nanospheres for drug storage, targeting, and imaging: a multifunctional nanocarrier system for cancer diagnosis and treatment, ACS Nano 4 (2010) 5398–5404, http://dx.doi.org/10.1021/nn101000e. [4] S. Mornet, S. Vasseur, F. Grasset, E. Duguet, Magnetic nanoparticle design for medical diagnosis and therapy, J. Mater. Chem. 14 (2004) 2161–2175, http://dx.doi.org/10. 1039/B402025A. [5] F. Hu, K.W. MacRenaris, E.A. Waters, E.A. Schultz-Sikma, A.L. Eckermann, T.J. Meade, Highly dispersible, superparamagnetic magnetite nanoflowers for magnetic resonance imaging, Chem. Commun. 46 (2010) 73–75.

I.S. Lyubutin et al. / Materials Science and Engineering C 45 (2014) 225–233 [6] J. Ugelstad, A. Berge, T. Ellingsen, R. Schmid, T.-N. Nilsen, P.C. Mørk, et al., Preparation and application of new monosized polymer particles, Prog. Polym. Sci. 17 (1992) 87–161, http://dx.doi.org/10.1016/0079-6700(92)90017-S. [7] H. Nakayama, A. Arakaki, K. Maruyama, H. Takeyama, T. Matsunaga, Singlenucleotide polymorphism analysis using fluorescence resonance energy transfer between DNA-labeling fluorophore, fluorescein isothiocyanate, and DNA intercalator, POPO-3, on bacterial magnetic particles, Biotechnol. Bioeng. 84 (2003) 96–102, http://dx.doi.org/10.1002/bit.10755. [8] C. Fang, M. Zhang, Multifunctional magnetic nanoparticles for medical imaging applications, J. Mater. Chem. 19 (2009) 6258–6266, http://dx.doi.org/10.1039/ B902182E. [9] S. Yang, X. Feng, S. Ivanovici, K. Müllen, Fabrication of graphene-encapsulated oxide nanoparticles: towards high-performance anode materials for lithium storage, Angew. Chem. Int. Ed. 49 (2010) 8408–8411, http://dx.doi.org/10.1002/anie. 201003485. [10] W. Tong, X. Song, C. Gao, Layer-by-layer assembly of microcapsules and their biomedical applications, Chem. Soc. Rev. 41 (2012) 6103–6124, http://dx.doi.org/10. 1039/C2CS35088B. [11] L.J. De Cock, S. De Koker, B.G. De Geest, J. Grooten, C. Vervaet, J.P. Remon, et al., Polymeric multilayer capsules in drug delivery, Angew. Chem. Int. Ed. 49 (2010) 6954–6973, http://dx.doi.org/10.1002/anie.200906266. [12] F. Caruso, M. Spasova, A. Susha, M. Giersig, R.A. Caruso, Magnetic nanocomposite particles and hollow spheres constructed by a sequential layering approach, Chem. Mater. 13 (2001) 109–116, http://dx.doi.org/10.1021/cm001164h. [13] B. Zebli, A.S. Susha, G.B. Sukhorukov, A.L. Rogach, W.J. Parak, Magnetic targeting and cellular uptake of polymer microcapsules simultaneously functionalized with magnetic and luminescent nanocrystals, Langmuir 21 (2005) 4262–4265, http://dx.doi. org/10.1021/la0502286. [14] D.A. Gorin, S.A. Portnov, O.A. Inozemtseva, Z. Luklinska, A.M. Yashchenok, A.M. Pavlov, et al., Magnetic/gold nanoparticle functionalized biocompatible microcapsules with sensitivity to laser irradiation, Phys. Chem. Chem. Phys. PCCP. 10 (2008) 6899–6905, http://dx.doi.org/10.1039/b809696a. [15] S. Erokhina, T. Berzina, L. Cristofolini, D. Shchukin, G. Sukhorukov, L. Musa, et al., Patterned arrays of magnetic nano-engineered capsules on solid supports, J. Magn. Magn. Mater. 272–276 (2004) 1353–1354. [16] D.G. Shchukin, G.B. Sukhorukov, H. Möhwald, Smart inorganic/organic nanocomposite hollow microcapsules, Angew. Chem. Int. Ed. 42 (2003) 4472–4475, http:// dx.doi.org/10.1002/anie.200352068. [17] M. Nakamura, K. Katagiri, K. Koumoto, Preparation of hybrid hollow capsules formed with Fe3O4 and polyelectrolytes via the layer-by-layer assembly and the aqueous solution process, J. Colloid Interface Sci. 341 (2010) 64–68, http://dx.doi. org/10.1016/j.jcis.2009.09.014. [18] T.V. Bukreeva, O.A. Orlova, S.N. Sulyanov, Y.V. Grigoriev, P.V. Dorovatovskiy, A new approach to modification of polyelectrolyte capsule shells by magnetite nanoparticles, Crystallogr. Rep. 56 (2011) 880–883, http://dx.doi.org/10.1134/S1063774511050051. [19] D.V. Volodkin, A.I. Petrov, M. Prevot, G.B. Sukhorukov, Matrix polyelectrolyte microcapsules: new system for macromolecule encapsulation, Langmuir 20 (2004) 3398–3406, http://dx.doi.org/10.1021/la036177z. [20] G.B. Sukhorukov, E. Donath, H. Lichtenfeld, E. Knippel, M. Knippel, A. Budde, et al., Layer-by-layer self assembly of polyelectrolytes on colloidal particles, Colloids Surf. Physicochem. Eng. Asp. 137 (1998) 253–266, http://dx.doi.org/10.1016/ S0927-7757(98)00213-1. [21] W.C. Elmore, Ferromagnetic colloid for studying magnetic structures, Phys. Rev. 54 (1938) 309–310, http://dx.doi.org/10.1103/PhysRev.54.309. [22] D.M. Kheiker, M.V. Kovalchuk, Y.N. Shilin, V.A. Shishkov, S.N. Sulyanov, P.V. Dorovatovskiĭ, et al., The belok station for protein crystallography on the synchrotron radiation beam from the bending magnet in the Sibir-2 storage ring, Crystallogr. Rep. 52 (2007) 358–364, http://dx.doi.org/10.1134/S1063774507020320. [23] S.N. Sulyanov, A.N. Popov, D.M. Kheiker, Using a two-dimensional detector for X-ray powder diffractometry, J. Appl. Crystallogr. 27 (1994) 934–942, http://dx.doi.org/10. 1107/S002188989400539X. [24] S. Sulyanov, H. Boysen, C. Paulmann, E. Sulyanova, A. Rusakov, Z. Kristallogr. Proc. 1 (2011) 175–180. [25] P.A. Stadelmann, EMS — a software package for electron diffraction analysis and HREM image simulation in materials science, Ultramicroscopy 21 (1987) 131–145, http://dx.doi.org/10.1016/0304-3991(87)90080-5. [26] P.G. Naumov, I.S. Lyubutin, K.V. Frolov, E.I. Demikhov, A closed-cycle cryostat for optical and Mössbauer spectroscopy in the temperature range 4.2–300 K, Instrum. Exp. Tech. 53 (2010) 770–776. [27] M.E. Fleet, The structure of magnetite: symmetry of cubic spinels, J. Solid State Chem. 62 (1986) 75–82, http://dx.doi.org/10.1016/0022-4596(86)90218-5.

233

[28] D.E. Cox, W.J. Takei, R.C. Miller, G. Shirane, A magnetic and neutron diffraction study of the Fe2O3–V2O3 system, J. Phys. Chem. Solids 23 (1962) 863–874, http://dx.doi. org/10.1016/0022-3697(62)90143-9. [29] C. Pecharromán, T. González-Carreño, J.E. Iglesias, The infrared dielectric properties of maghemite, γ-Fe2O3, from reflectance measurement on pressed powders, Phys. Chem. Miner. 22 (1995) 21–29, http://dx.doi.org/10.1007/BF00202677. [30] E. Moran, M.C. Blesa, M.-E. Medina, J.D. Tornero, N. Menendez, U. Amador, Nonstoichiometric spinel ferrites obtained from α-NaFeO2 via molten media reactions, Inorg. Chem. 41 (2002) 5961–5967. [31] J.I. Langford, A.J.C. Wilson, Scherrer after sixty years: a survey and some new results in the determination of crystallite size, J. Appl. Crystallogr. 11 (1978) 102–113, http://dx.doi.org/10.1107/S0021889878012844. [32] C.A. Gorski, M.M. Scherer, Determination of nanoparticulate magnetite stoichiometry by Mössbauer spectroscopy, acidic dissolution, and powder X-ray diffraction: a critical review, Am. Mineral. 95 (2010) 1017–1026, http://dx.doi.org/10.2138/am.2010.3435. [33] K. Voleník, M. Seberíni, J. Neid, A Mössbauer and X-ray diffraction study of nonstoichiometry in magnetite, Czechoslov. J. Phys. B. 25 (1975) 1063–1071, http://dx.doi.org/10.1007/BF01597585. [34] W. Kim, C.-Y. Suh, S.-W. Cho, K.-M. Roh, H. Kwon, K. Song, et al., A new method for the identification and quantification of magnetite–maghemite mixture using conventional X-ray diffraction technique, Talanta 94 (2012) 348–352, http://dx.doi. org/10.1016/j.talanta.2012.03.001. [35] J.-E. Jørgensen, L. Mosegaard, L.E. Thomsen, T.R. Jensen, J.C. Hanson, Formation of γFe2O3 nanoparticles and vacancy ordering: an in situ X-ray powder diffraction study, J. Solid State Chem. 180 (2007) 180–185, http://dx.doi.org/10.1016/j.jssc. 2006.09.033. [36] A.G. Roca, J.F. Marco, M. del P. Morales, C.J. Serna, Effect of nature and particle size on properties of uniform magnetite and maghemite nanoparticles, J. Phys. Chem. C 111 (2007) 18577–18584. [37] A.M. Jubb, H.C. Allen, Vibrational spectroscopic characterization of hematite, maghemite, and magnetite thin films produced by vapor deposition, ACS Appl. Mater. Interfaces 2 (2010) 2804–2812, http://dx.doi.org/10.1021/am1004943. [38] D.L.A. de Faria, S. Venâncio Silva, M.T. de Oliveira, Raman microspectroscopy of some iron oxides and oxyhydroxides, J. Raman Spectrosc. 28 (1997) 873–878, http://dx. doi.org/10.1002/(SICI)1097-4555(199711)28:11b873::AID-JRS177N3.0.CO;2-B. [39] I. Chamritski, G. Burns, Infrared- and Raman-active phonons of magnetite, maghemite, and hematite: a computer simulation and spectroscopic study, J. Phys. Chem. B 109 (2005) 4965–4968, http://dx.doi.org/10.1021/jp048748h. [40] W.B. White, The structure of particles and the structure of crystals: information from vibrational spectroscopy, J. Ceram. Process. Res. 6 (2005) 1–9. [41] G.M. Da Costa, E. De Grave, L.H. Bowen, R.E. Vandenberghe, P.M.A. De Bakker, The center shift in Mössbauer spectra of maghemite and aluminum maghemites, Clays Clay Miner. 42 (1994) 628–633, http://dx.doi.org/10.1346/CCMN.1994.0420515. [42] K. Závěta, A. Lančok, M. Maryško, E. Pollert, D. Horák, Superparamagnetic properties of γ-Fe2O3 particles: Mössbauer spectroscopy and d.c. magnetic measurements, Czechoslov. J. Phys. 56 (2006) E83–E91. [43] F. Gazeau, J. Bacri, F. Gendron, R. Perzynski, Y. Raikher, V. Stepanov, et al., Magnetic resonance of ferrite nanoparticles: evidence of surface effects, J. Magn. Magn. Mater. 186 (1998) 175–187, http://dx.doi.org/10.1016/S0304-8853(98)00080-8. [44] B.D. Cullity, Introduction to Magnetic Materials, Addison-Wesley Pub. Co., 1972 [45] S. Mørup, J.A. Dumesic, H.C. Tøpsoe, Magnetic microcrystals, in: R.L. Cohen (Ed.), Appl. Mössbauer Spectrosc, Academic Press, New York, 1980, p. 28. [46] L. Neel, Théorie du trainage magnétique des ferromagnétiques en grains fins avec applications aux terres cuites, Ann. Geophys. 5 (1949) 99–136. [47] T.N. Shendruk, R.D. Desautels, B.W. Southern, J. van Lierop, The effect of surface spin disorder on the magnetism of γ-Fe2O3 nanoparticle dispersions, Nanotechnology 18 (2007) 455704, http://dx.doi.org/10.1088/0957-4484/18/45/455704. [48] S. Krupicka, K. Závěta, Magnetic Oxides, in: D.J. Craik (Ed.), Wiley, New York, 1975, (Ch 5). [49] H. Takei, S. Chiba, Vacancy ordering in epitaxially-grown single crystals of γ-Fe2O3, J. Phys. Soc. Jpn. 21 (1966) 1255–1263, http://dx.doi.org/10.1143/JPSJ.21.1255. [50] F. Bødker, M.F. Hansen, C.B. Koch, K. Lefmann, S. Mørup, Magnetic properties of hematite nanoparticles, Phys. Rev. B 61 (2000) 6826–6838, http://dx.doi.org/10.1103/ PhysRevB.61.6826. [51] A. Cabot, A.P. Alivisatos, V.F. Puntes, L. Balcells, Ò. Iglesias, A. Labarta, Magnetic domains and surface effects in hollow maghemite nanoparticles, Phys. Rev. B 79 (2009) 094419. [52] S. Mørup, C.A. Oxborrow, P.V. Hendriksen, M.S. Pedersen, M. Hanson, C. Johansson, Magnetic and mechanical coupling between ultrafine maghemite particles, J. Magn. Magn. Mater. 140–144 (Part 1) (1995) 409–410, http://dx.doi.org/10.1016/ 0304-8853(94)00963-5.