Zn,Ni ferrite/NiO nanocomposite powder obtained from acetylacetonato complexes

INSTITUTE OF PHYSICS PUBLISHING

NANOTECHNOLOGY

Nanotechnology 17 (2006) 4877–4884

doi:10.1088/0957-4484/17/19/017

Zn,Ni ferrite/NiO nanocomposite powder obtained from acetylacetonato complexes M Vuˇcini´c-Vasi´c1 , B Antic2,8 , A Kremenovi´c2,3 , A S Nikolic4 , M Stoiljkovic5, N Bibic6 , V Spasojevic2 and Ph Colomban7 1

Faculty of Technical Sciences, University of Novi Sad, Trg D. Obradovica 6, 21000 Novi Sad, Serbia and Montenegro 2 Solid State Physics Laboratory, Institute of Nuclear Sciences ‘Vinca’, PO Box 522, 11001 Belgrade, Serbia and Montenegro 3 Faculty of Mining and Geology, Laboratory for Crystallography, University of Belgrade, Djusina 7, 11000 Belgrade, Serbia and Montenegro 4 Faculty of Chemistry, Department of Inorganic Chemistry, University of Belgrade, PO Box 158, 11001 Belgrade, Serbia and Montenegro 5 Physical Chemistry Laboratory, Institute of Nuclear Sciences ‘Vinca’, PO Box 522, 11001 Belgrade, Serbia and Montenegro 6 Atomic Physics Laboratory, Institute of Nuclear Sciences ‘Vinca’, PO Box 522, 11001 Belgrade, Serbia and Montenegro 7 LADIR, UMR 7075 CNRS, and Universit´e Pierre and Marie Curie, 94230 Thiais, France E-mail: [email protected]

Received 23 May 2006, in final form 13 August 2006 Published 11 September 2006 Online at stacks.iop.org/Nano/17/4877 Abstract The results on the synthesis, microstructure, structure and DC magnetization studies of nanocomposite Zn,Ni ferrite/NiO powder obtained by thermal decomposition of acetylacetonato complexes are reported in this paper. According to the results obtained by inductively coupled plasma optical emission spectroscopy (ICP-OES) element analysis and multiphase Rietveld refinement, the three samples made are composed of spinel-ferrite (86.7%–96.7%) and NiO (3.3%–13.3%) phases. The compositions of the spinel-ferrite (SP) phase in the investigated samples, S1–S3, are Zn0.72 Ni0.24 Fe1.98 O4 , Zn0.56 Ni0.29 Fe2.07 O4 and Zn0.40 Ni0.40 Fe2.10 O4 , respectively. Due to the cation deficiency in spinels, created vacancies induce a partial change in the cation valence, Ni2+ → Ni3+ . The vacancy distribution is found to be random at 8a and 16d cation sites, except in sample S3, where all vacancies are over octahedral sites. The x-ray line broadening due to crystallite size effect is found to be isotropic for all spinels, while the x-ray line broadening due to the strain effect is anisotropic. A correlation between the Zn2+ occupancy of the tetrahedral site and the 650 cm−1 Raman peak intensities is shown. The observed coercivity decrease and shift in hysteresis loop in the samples are caused by the interaction between spinel and NiO phase. The results of M(H) measurements point to the properties of an ensemble of interacting nanoparticles. High saturation magnetization values and superparamagnetic behaviour at room temperature point to the technological significance of the title compounds.

1. Introduction Magnetic materials are one of the most vital and fastest growing areas of research in the field of nanotechnology. 8 Author to whom any correspondence should be addressed.

0957-4484/06/194877+08$30.00

Their properties are significantly modified in comparison with those of the bulk counterpart. Some new magnetic properties and phenomena, such as superparamagnetism, spin canting and core/shell structure, are characteristic only of nanoscale magnetic materials. These properties depend on a number of

© 2006 IOP Publishing Ltd Printed in the UK

4877

M Vuˇcini´c-Vasi´c et al

factors such as composition, shape, size, surface morphology, anisotropy, and interparticle interactions [1, 2]. Nanometre scale magnetic materials are important in the following applications: high density information storage media, drug delivery, medical diagnostics, ferrofluids, electronic devices, catalysts [3]. They are also of current interest in basic research, since they form suitable systems for studying nanomagnetism. There is a reason to expect that the title compounds would in the future be applicable in microwave devices because Zn,Ni ferrite has high resistivity, high Curie temperature, chemical stability and good magnetic properties at high frequencies [4]. Ferrites with the spinel structure have a general formula MFe2 O4 . In a spinel unit cell, oxygen ions form close packing units with M2+ and Fe3+ ions distributed between tetrahedral A (8a) and octahedral B (16d) interstitial sites in the space group (SG) Fd 3m . In most cases ferrites with spinel structure have (M, Fe)3 O4 stoichiometry, where the cation/anion ratio is 3:4, although deviation from stoichiometry is possible [5–7]. In cation deficient spinels, the presence of vacancies leads to the modification of the cation valence, thus influencing the change of physical properties. For example, an increase in the initial permeability of Zn,Mn ferrites with increasing nonstoichiometry [5], or a significant variation of permeability in Mg,Mn ferrites, was found even for a small deviation of stoichiometry [6]. In ultrafine Zn,Mn ferrites, cations deficiency leads to a certain change in blocking (freezing) temperature and magnetization value [8]. Magnetic properties are sensitive to microstructure and structure parameters, which, themselves, strongly depend on the preparation method used. Consequently, different methods of preparation of nanosize powders are described in the literature [9, 10]. One of the most interesting methods of preparation of ultrafine powders is from complex compounds as precursors. The prepared complex compounds have relatively low temperatures of thermal decomposition giving ultrafine powder as a product [7, 11]. One of the aims of this work is to produce nanoparticles using the method based on the thermal decomposition of acetylacetonato complexes. Composites made of magnetic particles in a non-magnetic matrix are the subject of current investigations because of their great technological importance. In particular, composites made of two magnetic phases are significant in applications (particularly thin films) because of their unusual properties, generally governed by magnetic interactions. For example, Zeng et al [12] studied nanocomposites made of magnetically hard and soft powder phases (exchange-spring magnets), that interact by magnetic exchange coupling. The composites had large energy product compared to traditional single phase materials. Frandsen et al [13] reported the magnetic behaviour of nanopowder composite γ -Fe2 O3 /NiO (CoO), with large influences of antiferromagnetic (AF) NiO (CoO) on the ferrimagnetic (FM) phase of γ -Fe2 O3 via exchange coupling. They obtained larger (smaller) coercivity in composite γ -Fe2 O3 /NiO (γ -Fe2 O3 /CoO) in comparison with γ -Fe2 O3 . Thus, we were motivated to study composites that are composed of cation deficient spinel (FM) and nickel oxide (AFM) and to compare their properties with bulk and nanoscale materials obtained by different synthesis routes. In the samples crystallizing in spinel structure type, with 56 atoms in the unit cell and only 14 atoms in the asymmetric 4878

unit, 42 vibrational modes are expected. Group theory predicts the following modes [14]: A1g + Eg + T1g + 3T2g + 2A2u + 2Eu + 5T1u + 2T2u . The T1g , A2u , Eu and T2u modes are silent. There are five Raman-active modes (A1g + Eg + 3T2g ) and five infrared-active modes (5T1u ). The presence of an inversion centre in the space group Fd 3¯ m implies mutual exclusion of Raman and infrared activities for the same vibration modes. In the spinel only the sites ¯ m and 3m , occupied by a cation in with symmetries 43 tetrahedral coordination and O2− respectively, contribute to Raman activity [15]. The goals of this work were: (i) to synthesize nanoscale ferrites using complex compounds with acetylacetonato ligands as precursors, (ii) to determine the structure, particularly cation and vacancy distribution, (iii) to analyse the x-ray powder diffraction (XRPD) line broadening due to size and strain effects, (iv) to investigate samples by Raman spectroscopy, (v) to study magnetic properties using DC magnetization measurements, (vi) to examine the relation between magnetization and structure and/or microstructure and (vii) to point to the technological importance of the nanocomposites obtained.

2. Experimental details 2.1. Sample preparation The appropriate complex compounds, [Zn(AA)2 ], [Ni(AA)2 ] and [Fe(AA)3 ], with acetylacetonato (AA) ligands, were prepared and mixed in stoichiometric ratios: 0.75:0.25:2, 0.5:0.5:2 and 0.25:0.75:2, respectively. The three mixtures were then thermally decomposed under air atmosphere. The heating rate to the final temperature of 500 ◦ C was 10 ◦ C min−1 . The mixtures were kept for one minute at final temperature before cooling at ambient temperature (cooling rate 20 ◦ C min−1 ). 2.2. Chemical quantitative analysis Chemical quantitative analysis was performed by inductively coupled plasma optical emission spectroscopy (Spectroflame ICP, 2.5 kW, 27 MHz). ICP-OES analysis was performed by measuring the intensity of radiation of the specific wavelengths emitted by each element. The samples dispersed in liquid were introduced into the plasma as aerosol, where they were vaporized, atomized and excited. 2.3. X-ray data collection, TEM measurements and DC magnetization measurements X-ray diffraction data were collected on a Philips PW 1710 automated x-ray powder diffractometer using Cu Kα radiation, graphite monochromator and Xe-filled proportional counter. Data were collected every 0.03◦ in the angle range 15◦ –135◦ in 2θ . The counting time was fixed at 15 s per step for all samples. Microstructural characterization was performed by transmission electron microscopy (TEM), using a Philips EM 400 microscope (operated at 120 kV). Hysteresis loops were measured at 2 K after field-cooling using an MPMS XL-5 SQUID magnetometer. The samples

Zn,Ni ferrite/NiO nanocomposite powder obtained from acetylacetonato complexes

were cooled in a 50 kOe field down to low temperature, and measurements were made from 50 up to −50 kOe, and vice versa. For sample S3 the hysteresis loops were measured at different temperatures, 2–18 K in the zero-field cooled (ZFC) regime. 2.4. Raman measurements Two different instruments were used: a high-resolution XY spectrograph (Dilor, Lille, France) equipped with a double monochromator filter and a back-illuminated, liquid-nitrogen cooled, 2000 × 256 pixels CCD detector (Spex, Jobin-Yvon– Horiba Company) and an high sensitivity multichannel notchfiltered INFINITY spectrograph (Jobin-Yvon–Horiba SAS, Longjumeau, France) equipped with a Peltier cooled CCD matrix. These instruments were used to record Raman spectra between 10 (XY instrument)/150 (INFINITY instrument) and 2000 cm−1 , using 514.5, 532, 632 and 647.1 nm exciting lines (Ar+ –Kr+ , YAG, and He–Ne lasers). Backscattering illumination and collection of the scattered light were made through an Olympus confocal microscope (long focus Olympus 10× or 50× objective, total magnification 500×). Because of the strong coupling between dark materials and the laser light, careful attention was paid to the power of illumination used. Examination was made on small particle aggregates, a procedure preferentially used for dark material. With the XY instrument (2.5 mW) a special scanning mirror was added to move the laser spot on a ∼100 μm line to decrease the light induced heating. The used power of illumination ranged between 0.1 and 2 mW (INFINITY) and 1–10 mW (XY); the laser spot is ∼20 μm2 .

3. Results and discussion 3.1. Structure refinement. Cation valence determination The element analysis of synthesized samples was performed by inductively coupled plasma optical emission spectroscopy. The obtained results were different from those calculated for stoichiometric Zn,Ni ferrites (Zn1−x Nix Fe2 O4 ). The percentage of nickel is higher than one expected in ferrites. On the basis of this fact it is concluded that the samples are nonstoichiometric or/and the samples are multiphase. It is not rare that some synthesis procedures influence a deviation from stoichiometry [7, 16]. A deviation from stoichiometry can be due to cation deficiency, e.g. (Zn, Mn, Fe)3−δ O4 (δ = 0.18–0.30) [7] or cation surfeit, e.g. Fe1.7 Ni1.43 O4 [17] and Ni1.25 Fe1.85 O4 [18]. In the latter case, the presence of cations in 16c sites (empty in stoichiometric spinel structure, SG Fd 3m ) results in an increase in relative intensities of reflections (222), (400) and (440) [17, 18]. The crystal structures of the as-prepared samples were checked by the x-ray powder diffraction (XRPD) method. In the obtained diffraction patterns all reflections could be indexed in the SG Fd 3m . However, the results of ICP-OES analysis suggested that the phase which in the diffractogram looks like a pure spinel should be analysed carefully. Several structure models were tested in order to determine possible crystal phases in the samples and to refine their structures. Single phase Rietveld refinements for nonstoichiometric spinel phase, as well as a structural

Figure 1. Comparison between observed (•) and calculated (——) intensities for sample S1. The vertical bars indicate the reflection positions for each phase separately. The difference pattern appears below. The first row corresponds to the ferrite while the second corresponds to NiO.

model with ‘additional’ Ni in the 16c position, failed. While searching the inorganic crystal structure database [19] we found that the strongest XRPD lines for NiO (SG Fm 3m a ≈ ˚ appeared at approximately the same 2θ positions 4.18 A) as spinel reflections (222), (400) and (440). The Rietveld refinement based on a two-phase model (nonstoichiometric spinel + NiO) gave the best results for all samples. The Rietveld refinement analyses of XRPD data were performed by the Fullprof computer program [20]. A TCH pseudo-Voigt profile function was used. Figure 1 shows a comparison between observed and calculated intensities of sample S1. The results of the Rietveld refinement procedure, given in table 1 and figure 1, confirm good agreement between the structural model and the observed data. Quantitative phase analyses of samples were done by the Rietveld refinement procedure. In table 2, mass percentages of each phase are presented. On the basis of these results the mass of each cation per gram of sample was calculated. The obtained values are in good agreement with the ICPOES element analysis results; see column ‘Fullprof chemical analysis’ in table 2. This fact, together with considerably low agreement factors of refinements (table 1), confirms the validity of the models used. The cation distribution over 8a and 16d crystallographic sites, in the spinel phase, was investigated by the refinement of the occupation numbers (N ). The results of occupation number refinement showed that in Zn,Ni ferrites, all ions occupy both cation sites. Zinc ions preferentially occupy tetrahedral 8a sites, while nickel ions preferentially occupy octahedral 16d sites. The presence of vacancies was detected in the structures of all spinels. In other words the obtained spinel phases are cation deficient, (Zn, Ni, Fe)3−δ O4 with parameter δ = 0.06–0.10. The sum of occupation numbers of cations in 8a and 16d positions (table 1) indicates that the vacancy concentrations are approximately the same at both mentioned positions in spinels S1 and S2. On the other hand, spinel S3 shows that 8a positions are fully occupied, while all vacancies are distributed over octahedral sites. The sum of net formal charges for ferrite phases in samples S1–S3 is not valid in the case when all iron has 4879

M Vuˇcini´c-Vasi´c et al

Table 1. Crystal structure and microstructure parameters for spinels. (Calculated cation–anion distances: ∗ from Fullprof and ∗∗ on the basis of cation distribution and ionic radii (see text).) Sample

S1

S2

S3

Spinel composition ˚ Lattice parameter, a (A)

Zn0.72 Ni0.24 Fe1.98 O4 8.4222(1)

Zn0.56 Ni0.29 Fe2.07 O4 8.4123(1)

Zn0.40 Ni0.40 Fe2.10 O4 8.39137(9)

1.942(2) 2.039(2) 1.936(7) 2.031(7)

1.933(2) 2.040(2) 1.925(7) 2.031(7)

1.910(2) 2.045(2) 1.902(7) 2.036(7)

0.1413(6) 0.0090(6) 0.0943(6) 0.0392(6) 0.0501(6) 0.3997(6)

0.1101(7) 0.0219(7) 0.1115(7) 0.0295(7) 0.0501(7) 0.4059(7)

0.0621(5) 0.0184(5) 0.1698(5) 0.0379(5) 0.0816(5) 0.3562(5)

147.36(7) 17(2)

131.9(1) 32(2)

161.65(7) 31(3)

˚ Cation–anion distances, d (A) d(M8a –O)∗ d(M16d –O)∗ d(M8a –O)∗∗ d(M16d –O)∗∗ Occupation parameters, N N(Zn)8a N(Ni)8a N(Fe)8a N(Zn)16d N(Ni)16d N(Fe)16d ˚ Average apparent size (A) Average mixing strain × 104

Table 2. Results of ICP-OES and Fullprof quantitative analysis. ICP-OES chemical analysis (%)

Fullprof chemical analysis (%)

Percentage of the phase (%)

Sample

Fe

Zn

Ni

Fe

Zn

Ni

Spinel

NiO

S1 S2 S3

43(2) 42(2) 42(2)

17.4(5) 12.2(3) 8.0(4)

7.6(8) 11.8(6) 17.6(6)

43(2) 44(2) 44(2)

18(2) 13(2) 9.5(2)

9(2) 13(2) 19(2)

96.7(3) 91.5(5) 86.7(3)

3.3(1) 8.5(2) 13.3(1)

oxidation state +3 and all nickel and zinc have oxidation state +2, suggesting a change in cation valence. Zinc has +2 stable valence, which suggests that nickel and/or iron ions change their valence state through the oxidation/reduction process. During the refinement it was found that all cations populated both tetrahedral 8a and octahedral 16d position (table 1). The valence state of nickel in tetrahedral coordination is +2, while for octahedral coordination nickel oxidation states could be +2, +3 and +4 [24]. Nickel in the +4 oxidation state could rarely be found in a spinel structure and therefore we excluded the possibility of Ni oxidation from +2 into +4. The valence states of iron could be +2 and +3 in both tetrahedral and octahedral coordination. Assuming that only the oxidation process, Ni2+ → Ni3+ , takes place, on the basis of the sum of net formal charges and cation distribution, we found in S1, S2 and S3 that the percentage of nickel in octahedral sites in the state +3 was 68%, 46% and 31%, respectively. In order to determine the oxidation state of nickel and iron, we calculated cation–anion bond lengths (d ∗∗ ) respecting the cation distribution found and the literature values of ionic radii [21], and assuming that all iron ions are in the +3 oxidation state. The calculated cation–anion distances (d ∗∗ ) are given in table 1 together with the values obtained from the Rietveld refinement (d ∗ ). Good agreement between them (equal within around 3 esd) confirmed the assumption that nickel partly oxidized from +2 to +3. Taking the cation–anion distances obtained during the Rietveld refinement as reference values, all calculated distances are shifted towards smaller values. This could be a sign of iron oxidation state reduction that was neglected in the calculations. Consequently, a possibility of partial reduction of iron was considered. Assuming that all nickel in octahedral 4880

sites has oxidation state +3, a small part of iron, less than 5%, will have state +2 for spinel-ferrite in S1 and S2, while for spinel-ferrite in the sample S3, 10% of the iron is reduced. With this change in cation valences, the calculated distances are again equal within the error (3 esd) to d ∗ . It can be concluded that oxidation of nickel took place (the percentage of Ni3+ decreases with the increase in nickel concentration in spinels) and the reduction of iron in low percentage also cannot be excluded. 3.2. The size–strain microstructure analysis X-ray line broadening analysis was used to obtain information about XRPD line broadening due to crystallite size and strain effects. In the Fullprof computer program x-ray line broadening was analysed through the refinement of regular TCH-pV function parameters (isotropic effects) and the refinement of multipolar functions, i.e., symmetrized cubic harmonics [22–24]. The starting point for size–strain microstructure analysis is the fact that x-ray line broadening is influenced by both size and strain effects. Taking into consideration that we have nanocomposite materials with several different cations, which occupy both special positions in crystal lattice, we a priori expected an anisotropic x-ray line broadening due to a strain effect. The standard deviations of calculated average apparent sizes for all samples are small, indicating negligible anisotropy of x-ray line broadening provoked by the size effect. The size/strain standard deviation obtained by using Fullprof represents a measure of the degree of anisotropy [24]. The results of the size–strain microstructure analysis are given in table 1. The average apparent size of mixed Zn,Ni

Zn,Ni ferrite/NiO nanocomposite powder obtained from acetylacetonato complexes

Figure 2. Projections of the three-dimensional body that represents ‘average max-strain’ in the (001) crystallographic plane; (a) Zn0.72 Ni0.24 Fe1.98 O4 , (b) Zn0.56 Ni0.29 Fe2.07 O4 and (c) Zn0.40 Ni0.40 Fe2.10 O4 .

Figure 3. TEM bright field image of Zn0.56 Ni0.29 Fe2.07 O4 /NiO composite.

ferrites is found to be ∼15 nm, that is smaller in comparison ˚ obtained by using the with the value of ZnFeO4 (277(10) A) same method of synthesis [7]. The obtained max-strain has the smallest value for the spinel phase Zn0.72 Ni0.24 Fe1.98 O4 in S1. The x-ray line broadening anisotropy due to the strain effect in S2 is found to be small, while the anisotropy of xray line broadening due to the strain effect in S1 and S3 is approximately 10%. The apparent sizes/strains along different directions can be reconstructed from the refined spherical harmonic (quadratic form) coefficients. An average ‘apparent shape or strain’ of the coherent domains could be obtained. We used program GFOURIER [25], incorporated in WINPLOT [26], to visualize the anisotropy of x-ray line broadening due to crystallite strain. The projections of the three-dimensional body representing ‘average max-strain’ in the (001) crystallographic plane are given in figure 2 in ferrite phases of samples S1–S3. In the ferrite phase in samples S1 and S3 the strain is the highest in crystal axes directions, as seen in figure 2. 3.3. Particle shape. Particle versus crystallite size The particle size and morphology of one selected sample were characterized by transmission electron microscopy. A typical

Figure 4. Raman spectra for the Fe2.85 Y0.15 O4 , S1, S2 and S3 samples.

bright field image of Zn0.56 Ni0.29 Fe2.07 O4 /NiO composite is shown in figure 3. TEM analysis revealed that the particles are isotropic, with a relatively uniform size distribution. It can be seen that the mean particle size is in the range 10– 15 nm, indicating that on average one grain is composed of one crystallite. This implies that the nanomaterial produced is well crystallized and free of defects. A more detailed TEM analysis enabled us to clarify the morphology of the synthesized particles. The particles are nearly spherical in shape with faceted morphology. 3.4. Raman spectra Figure 4 shows representative Raman spectra recorded with 647.1 nm excitation (1.25 mW) using a 50× objective (total magnification 500×, i.e., the analysed surface is ∼30–50 μm2 ). The analysis was made in different locations and with different power of illuminations between 0.5 and 10 mW without observing any changes. This indicates 4881

M Vuˇcini´c-Vasi´c et al

Table 3. Raman wavenumbers and expected symmetry for spinel phases in S1–S3. (Note: w—weak, m—medium, S—strong, sh—shoulder.)

T12g T32g Eg A1g T22g A1g A1g

S1

S2

S3

180 w 330 m 420 w 480 S 560 w 650 S 685 S

190 w 305 m 400 w 470 S 555 w 645 sh 680 S

195 m 300 m 390 w 465 S 555 w 640 sh 680 S

homogeneous materials, with a rather good stability versus oxidation, contrary to the findings in previous studies on spinels: Fe3 O4 , Fe2.55 In0.45 O4 , Fe2.85 Y0.15 O4 [27]. In particular there was no evidence of the formation of crystalline Fe2 O3 (haematite) by an increase in power of illumination. Fluorescence was rather important; it requires waiting for ∼30 min of illumination before the spectra could be recorded. The fluorescence background was rather linear and could be easily removed by subtraction. The comparison of Raman spectra of spinels in S1– S3 with reference Fe3 O4 or Fe2.85 Y0.15 O4 nanopowder [27] spectra shows a strong broadening of the main peak and additional bands, indicating a lowering of the symmetry or the presence of the second phase. In sample S1, some weak and broad bumps at ∼190 and 280 cm−1 were observed. Thus a trace of haematite phase is not excluded, but the very low intensity of these bumps certifies that the amount could be very small (less than 1%). These bands are not observed in S2 and S3. Very similar Raman spectra with a strong 450–500 cm−1 and medium 340 and 215 cm−1 peaks have been reported for NiFe2 O4 single crystal [28] and assigned to a strong distortion of the ideal D3d symmetry. The observation of a 215 cm−1 peak in Ni-ferrite single crystal indicates that the peak expected to be the signature of haematite is rather a signature of a ferrite. The NiFe2 O4 single crystal spectrum also shows a medium peak at ∼575 cm−1 and a rather strong splitting of the main ∼700 cm−1 peak. In S1–S3 the presence of NiO phase is not excluded because the main peak [29] of NiO is usually observed at ∼575 cm−1 . The intensity of this peak increases when the Ni content increases, which is consistent with this assignment. Thus it can be concluded that all the observed bands belong to the spinel (table 3) and NiO phases, except the broad bumps at 190 and 280 cm−1 in S1. The band broadening is intrinsic to many nanophased oxides [29–32], because a small particle size hinders the phonon propagation and hence induces a Brillouin zone folding which makes all the phonons Raman active. A very similar feature is provoked by a disordered substitution of atoms either by heavier ones or by vacancies. Thus, the observed broadening in S1–S3 can be due (i) to the nanophase character, (ii) to the Fe/Zn/Ni substitution, (iii) to the presence of vacancies (or to the sum of all three). By comparison with other spinel Raman signatures, the bandwidth (full width at half height ∼80 cm−1 ) is rather equivalent to or even weaker than those measured in other ferrites. This indicates a good crystallinity and that the small sizes of the samples do not contribute to the broadening significantly. 4882

Figure 5. Population of tetrahedral position by Zn versus relative intensity of the vibration mode at ∼650 cm−1 .

In a very simple molecular approximation, the highest wavenumber peak—this mode may have a strong A1g symmetry character—can be assigned to the stretching mode of the vibrational unit composed of the shortest covalent bonds. In this model the observed splitting can be associated with the different atoms. Inter-ionic distances between cations in tetrahedral (8a) position and oxygen are: r (Fe3+ –O2− ) = ˚ r (Zn2+ –O2− ) = 1.95 A; ˚ r (Ni2+ –O2− ) = 1.90 A. ˚ Ions 1.84 A; Zn2+ and Fe3+ predominantly populate the tetrahedral position (8a). Hence, there are two peaks at about 670 cm−1 for Fe–O vibrations and at about 640 cm−1 for Zn–O vibrations. It is clear that the ∼650 cm−1 peak intensity is linearly correlated with Zn2+ content in tetrahedral coordination, figure 5. The assignment is consistent with the longer Zn–O distance. 3.5. DC magnetization versus field From the magnetic point of view, the studied samples (S1, S2 and S3) are composed of two magnetic phases, ferrimagnetic spinel and antiferromagnetic NiO. A bulk antiferromagnet, such as NiO, has zero net magnetic moment. However, an ensemble of nanoparticle NiO exhibits superparamagnetism [33] and spin–glass properties [34]. Namely, due to the surface uncompensated moments, NiO particles have no zero net magnetic moments. To study the magnetic behaviour of SP/NiO composites with emphasis on FM–AFM interactions, we carried out magnetization measurements, M(H ), at 2 K, and at different temperatures for sample S3. The obtained hysteresis loops are smooth, without any sign of the presence of more than one phase, indicating strong magnetic interactions between particles [13], so the composite SP + NiO behaves magnetically as a single phase (figure 6). The obtained curves are characterized by nonsaturated magnetization up to H = 50 kOe, indicating the existence of non-ordered spins in the surface layer of the particles. The high irreversibility was not observed in M(H ) measurements. The lower and the upper M(H ) branch coincided in fields bigger than 10 kOe. Magnetization versus field at 300 K (shown in the inset of figure 6) is typical for SPM

Zn,Ni ferrite/NiO nanocomposite powder obtained from acetylacetonato complexes

Figure 6. Hysteresis loops of the investigated samples measured at 2 K. The inset: hysteresis loops at 300 K.

systems: up to 50 kOe the magnetization does not saturate, and both remanence and coercivity are zero. Saturation magnetization MS was obtained by extrapolating the M(1/H ) dependence on the 1/H = 0 value. The MS values found at 2 K (300 K), 75 emu g−1 for (33 emu g−1 ) S1, 79 emu g−1 (41 emu g−1 ) for S2 and 77 emu g−1 (49 emu g−1 ) for S3, are lower compared to those found for bulk Zn1−x Nix Fe2 O4 with similar concentration of magnetic ions [34]. However, our inspection of literature points to high observed values in comparison to similar spinel composition with similar particle size; see e.g. [35–37]. The following effects on the magnetization value should be considered: (i) the presence of parasite phases, (ii) particle size effect, (iii) cation distribution, (iv) cation valence change and (v) interactions. The effects can have opposite contribution. Parasite phases, such as Fe2 O3 , Fe, Ni, have not been noticed in diffraction patterns of the samples. The Raman spectra of the samples cannot exclude the presence of haematite in S2 sample in very small percentage. Size effect in nanoparticles is often a cause of the magnetization reduction, in comparison with the values obtained for the bulk counterparts. Namely, the decrease in particle size increases the surface to volume ratio, forming the surface uncompensated magnetic moments (shell) which are responsible for significant reduction of net magnetization. Cation distribution deviation in nanosize samples from that expected of the bulk counterpart can lead to magnetization enhancement/reduction. Bulk Ni ferrite is a well known inverse spinel with cation distribution (Fe3+ )A [Ni2+ Fe3+ ]B . Zn2+ has the largest preference to tetrahedral sites in the spinel structure, so in bulk Zn,Ni ferrite tetrahedral sites are occupied by Zn2+ and Fe3+ ions. The results of crystal structure refinements show a cation redistribution, i.e. different to the expected result when taking into account the known cation site preferences (see table 1). The intensity of exchange interactions (8a–16d, 8a–8a and 16d–16d) is determined by cation occupancies, so that the net magnetization depends on the cation distribution. The change of cation distribution in nanoparticles leads not rarely to enhancement of the saturation magnetization compared to that of the bulk counterpart.

Figure 7. Hysteresis loops for S3 (Zn0.40 Ni0.40 Fe2.10 O4 /NiO) measured after FC and ZFC at 2 K. The inset: hysteresis loops for S1 (Zn0.72 Ni0.24 Fe1.98 O4 /NiO) and S2 (Zn0.56 Ni0.29 Fe2.07 O4 /NiO) measured after FC at 2 K.

The formation of cation deficient spinels with vacancies both at A and B sites is accompanied by cation valence change, Ni2+ → Ni2+ε (0.3  ε  0.6). A partial valence change of nickel ions from +2 to +3 assumes a change of magnetic moments: 2μB (Ni2+ , 3d8 ) → 3μB (Ni3+ , 3d7 ). The reduction of Fe3+ ions in small percentage is also possible. This could be a significant factor that influences the total magnetization value. The existence of vacancies can cause spin canting in the whole volume of the particles, influencing the decrease of magnetization. Magnetization is affected by different types of anisotropy. Magnetic interactions of the two phases can create a preferential direction in the FM phase. Hence, the exchange anisotropy from SP–NiO and SP–SP interactions should be considered. To examine interparticle and intraparticle interactions we measured the hysteresis after field-cooling at 2 K for all samples, and at different temperatures (2–18 K) for S3 after zero-field cooling. Asymmetry of magnetization loops under the FC regime, as well as a shift of the FC branch with respect to the ZFC branch, was observed; see figure 7. The hysteresis loop shift, HC− (FC)– HC+ (FC), is 60 Oe for S3 (30 Oe for S1 and 70 for S2). The obtained HC (2 K) value as well as the ratio (HC− (FC)– HC+ (FC))/ HC (average) is the lowest in sample S3, containing the most NiO, indicating the significant influence of interparticle interactions on coercivity. It is worth mentioning the results reported by Frandsen et al [13] on nanocomposite γ -Fe2 O3 /NiO. They found lower coercivity in the composite than in pure γ -Fe2 O3 , and the opposite ratio in composite γ -Fe2 O3 /CoO due to exchange interactions [13]. Consequently, the values of coercivity point to a significant influence of interparticle interactions. The change of HC (ZFC) with temperature in sample S3 is presented in figure 8. HC decreases with an increase in temperature, as could be expected. In non-interacting nanoparticle ensembles, the HC (T ) dependence should obey the relation [38] HC = HC0 [1 − (T / TB )1/2 ]. The deviation in linearity of HC (T 1/2 ) is an another indication of the interparticle interactions. 4883

M Vuˇcini´c-Vasi´c et al

References

Figure 8. Variation of coercivity with temperature for S3 (Zn0.40 Ni0.40 Fe2.10 O4 /NiO) measured after ZFC.

4. Summary Using a synthesis routine based on thermal decomposition of complexes with acetylacetonato ligands, nanocomposites with the spinel-ferrite with the concentration between 86.7% and 96.7% and NiO with 3.3% and 13.3% were obtained. Precise analysis of the crystal structure was obtained by applying Rietveld methods. From this analysis it was possible to determine the crystal structure parameters and obtain information on the microstructure (size–strain). The additional Raman investigations confirmed the structural data concerning the occupancy of the tetrahedral site by Zn atoms. Magnetic data indicate the complex properties of nanoparticle systems. Analysis of these data confirms that the samples consist of interacting ferrite nanoparticles. It was found that interparticle interaction between ferrite (FM) and NiO (AFM) influence the magnetization, hysteresis loop shift and coercivity values, which are important in nanoscale material applications. Zn,Ni ferrite/NiO nanocomposites have smaller magnetization compared with bulk Zn,Ni ferrite. This is expected, because of surface effects and spin canting in the particle core. However, in comparison with the Zn,Ni ferrite nanoparticles with similar particle size, the title composites generally show higher saturation magnetization values. The synthesized composites are superparamagnetic at room temperature, which points to their technological importance. From the theoretical point of view the title composites are good model systems for studying the influence of interparticle interactions on two main magnetic characteristics, coercivity and magnetization. We believe that composite powders with two different magnetic phases will be increasingly the subject of investigation, in order to improve the magnetic characteristics of basis compounds, and to study magnetism at the nanoscale level.

Acknowledgment The Serbian Ministry of Science, Technology and Development supported this work financially. 4884

[1] De Heer W A 2000 Nanomagnetism Characterization of Nanophase Materials ed Z L Wang (Germany: Wiley–VCH) [2] Nalwa H S (ed) 2002 Magnetic Nanostructures (USA: American Scientific Publishers) chapter 1–4 [3] Vestal C R and Zhang Z J 2002 Chem. Mater. 14 3817 [4] Ponpandian N et al 2005 Appl. Phys. Lett. 86 192510 [5] Tanaka T 1978 Japan. J. Appl. Phys. 17 349 [6] Kang S H and Yoo H L 2000 J. Appl. Phys. 88 4754 [7] Antic B, Kremenovic A, Nikolic A S and Stoiljkovic M 2004 J. Phys. Chem. B 108 12646 [8] Kremenovic A, Antic B, Spasojevic V, Vucinic-Vasic M, Jaglicic Z, Pirnat J and Trontelj Z 2005 J. Phys.: Condens. Matter 17 4285 [9] Caruntu D et al 2002 Inorg. Chem. 41 6137 [10] Han M, Vestal C R and Zhang Z J 2004 J. Phys. Chem. B 108 583 [11] Vucinic-Vasic M, Antic B, Blanusa J, Rakic S, Kremenovic A, Nikolic A S and Kapor A 2006 Appl. Phys. A 82 49 [12] Zeng H, Jing L, Liu J P, Zhong L, Wang L and Shouheng S 2002 Nature 420 395 [13] Frandsen C, Ostenfeld C W, Xu M, Jacobsen C S, Keller L, Lefmann K and Morup S 2004 Phys. Rev. B 70 134416 and references therein [14] Shebanova O N and Lazor P 2003 J. Solid State Chem. 174 424 and references therein [15] Fateley W G, Dollish F R, McDevitt N T and Bentley F F 1972 Infrared and Raman Selection rules for Molecular and Lattice Vibrations: The Correlation Method (New York: Wiley) [16] Sharma D R, Mathur R, Vadera S R, Kumar N and Kutty T R N 2003 J. Alloys Compounds 358 193 [17] Blesa M C, Amador U, Moran E, Menendez N, Tornero J D and Rodriguez-Carvajal J 1993 Solid State Ion. 63–65 429 [18] Blesa M C, Moran E, Amador U and Andersen N H 1997 J. Solid State Chem. 129 123 [19] Inorganic Crystal Structure Database 2002 (Gaithersburg, MD: National Institute of Standard and Technology) [20] Rodriguez-Carvajal J 1998 FullProf computer program ftp:// charybde.saclay.cea.fr/pub/divers/fullprof.98/windows/ winfp98.zip [21] Shanon R D 1976 Acta Crystallogr. A 32 751 [22] Honkim¨aki V and Surotti P 1999 Effects of instrument function, crystallite size and strain on reflection profiles Defect and Microstructure Analysis by Diffraction (IUCR Book Series) ed R L Snyder, J Fiala and H J Bunge (New York: Oxford University Press) [23] Stephens P W 1999 J. Appl. Crystallogr. 32 281 [24] http://www-llb.cea.fr/fullweb/fp2k/fp2k divers.htm file Fullprof-Manual.zip [25] http://www-llb.cea.fr/fullweb/others/newfour.htm [26] http://www-llb.cea.fr/winplotr/winplotr.htm [27] Cvejic Z, Rakic S, Kremenovic A, Antic B, Jovalekic C and Colomban Ph 2006 Solid State Sci. 8 908 [28] Graves P R, Johnston C and Campaniello J J 1988 Mater. Res. Bull. 23 1651 [29] Colomban Ph, Jullian S, Parlier M and Monge-Cadet P 1999 Aerosp. Sci. Technol. 3 447 [30] Colomban Ph 2004 Mater. Sci. Forum 453/454 269 [31] Colomban Ph 2003 Spectrosc. Eur. 15 8 [32] Kosacki I, Suzuki T, Anderson H and Colomban Ph 2002 Solid State Ion. 149 99 [33] Richardson T, Yiagas D I, Turk B, Foster K and Twigg M V 1991 J. Appl. Phys. 70 6977 [34] Tiwari S D and Rajeev K P 2005 Phys. Rev. B 72 104433 [35] Caizer C 2003 Mater. Sci. Eng. B 100 63 [36] Kale A, Nathani H, Srivastava S R and Misra K D R 2004 Mater. Sci. Technol. 20 999 [37] Albuquerque S S, Ardisson D J, Macedo A A W and Alves M C M 2000 J. Appl. Phys. 87 4352 [38] Fonseca F C et al 2002 Phys. Rev. B 66 104406