Annealing dependence of magnetic properties in nanostructured particles of yttrium iron garnet prepared by citrate gel process

N

Journal of Magnetism and Magnetic Materials 169 (1997) 56 68

ELSEVIER

~ l ~ Journalof magnellsm and magnetic materials

Annealing dependence of magnetic properties in nanostructured particles of yttrium iron garnet prepared by citrate gel process P. Vaqueir&'*, M.A. Ldpez-Quintela a, J. Rivas a, J.M. G r e n e c h e b "Departamentos de Quimica-Fisica y Fisica Aplicada de Santiago de Compostela, E-15706 Santiago de Compostela, Spain bLaboratoire de Physique de l 'Etat Condense, URA CNRS n 807, Universit~ du Maine, F-72017 Le Mans Cedex. France

Received 20 May 1996; revised 7 November 1996

Abstract Yttrium iron garnet (YIG) particles were first synthesized using a low-temperature method based on the citrate gel process. The annealing temperature dependence of the magnetic behaviour was investigated by means of transmission 5VFe M6ssbauer spectrometry and DC magnetic measurements. The as-prepared particles behaves as an amorphous-like structure. Annealing treatments cause aggregation and crystallization of particles. For low annealing temperatures, one can suggest a nanostructured behaviour of the YIG particles, which progressively disappears when the annealing temperature increases. Depending on the annealing temperature, YIG nanostructured particles of different size, in the range 50-700 nm were obtained. The values of the saturation magnetization as well as the Curie temperature for the YIG particles are close to those observed for well-crystallized YIG. PACS: 76.80. + y; 75.50.Kj; 75.50.G Keywords. Yttrium iron garnet; Nanostructured particles; M6ssbauer effect

1. Introduction Great scientific and technological interests have been devoted to ferrimagnetic garnets since their discovery in 1956 [1] because of their high resistivity as well as their applications in microwave and

* Corresponding author. Tel.: + 34-81-563100ext. 13029;fax: + 34-81-595012;e-mail: [email protected].

magneto-optical devices. Some magnetic performances such as coercive force and initial permeability, are strongly dependent on the microstructural aspects, i.e. the synthesis m e t h o d and the conditions of further annealing and grinding. M u c h attention was paid to the conventional ceramic methods in order to ameliorate the homogeneity and the density of grains. But this m e t h o d requires very high temperatures ( ~ 1 5 7 0 K); consequently, new methods of synthesis have been recently developed to obtain h o m o g e n e o u s fine particles. One

0304-8853/97/$17.00 ~:~ 1997 Elsevier Science B.V. All rights reserved PII S 0 3 0 4 - 8 8 5 3 ( 9 6 ) 0 0 7 2 8 - 7

P. Vaqueiro et al. / Journal of Magnetism and Magnetic Materials 169 (1997) 56 68

of these routes of synthesis is the citrate gel process, mainly used to prepare mixed complex oxides including garnets [-2, 31. We report in the present paper the annealing temperature dependence of the structural and magnetic properties of nanostructured yttrium iron garnet (YIG) particles prepared by a citrate gel process. The investigations were performed using thermogravimetric analysis, vibrating sample magnetometry and M6ssbauer spectrometry techniques. This last one will provide information on the presence of impurity phases, the distribution of the iron ions in the lattice, as well as the hyperfine parameters.

2. Experimental methods and characterization Stoichiometric mixtures of nitrates of iron (III) and yttrium (III) were dissolved in an aqueous

57

solution of citric acid. Gel labelled A results after heating at 353 K for 2 h. Since it is reported by Matsumoto [4] that the presence of ethylene glycol improves the synthesis process, a second gel B was prepared according to this method, with the same previous treatment. For the preparation of the YIG powders, both gels were dried initially at 383 K during 36 h and further annealing treatments were performed in air at different temperatures (between 673 and 1273 K) and for different periods of time (between 2 and 24 h). The conditions of annealing of the different samples are listed in Table 1, where A and B series depend on the initial gel. In addition, a YIG sample was prepared by conventional ceramic process as standard, from iron and yttrium oxides. Stoichiometric quantities of oxides were mixed, ground for 24 h and heated at 1573 K in air for 6 h. The process was repeated until the pure and well-crystallized iron garnet phase was obtained.

Table 1 Samples prepared with a treatment of 2 and 12 h Sample

Annealing temperature(K)

Time(h)

Phases

A1 A2 A3 A4

673 673 873 873

2 12 2 12

Amorphous-like +magnetite

A5 A6 A7 A8 A9 A10 All A12 BI B2 B3 B4 B5 B6 B7 B8 B9 BI0 Bll B12 BI3

973 973 1073 1073 1173 1173 1273 1273 573 673 673 873 873 973 973 1073 1073 1173 1173 1273 1273

2 12 2 12 2 12 2 12 2 2 12 2 12 2 12 2 12 2 12 2 12

Amorphous-like + YIG YIG

Amorphous Amorphous-like + magnetite

Amorphous-like + YIG YIG+ ~ ortho~rrite

P. Vaqueiro et al. /Journal of Magnetism and Magnetic Materials" 169 (1997) 56-68

58

Thermogravimetric analysis (TGA) measurements were recorded with an equipment PerkinElmer TGA7, with a scanning rate of 10 K/rain and N2 atmosphere; sample weights were 2 mg. Combined T G A - D T A measurements were carried out with an equipment T G A - D T A (differential thermal analysis) V1.0B TA Inst. 2000, over the range 300 1573 K, in Ar atmosphere, at 10 K/rain and with a sample weight of 15 rag. The M6ssbauer absorption studies were carried out over the range 4.2-573 K in transmission geometry using a constant acceleration signal and with a source of 57Co diffused into a rhodium matrix. The isomer shift values are given relative to metallic iron at 300 K. The spectra were fitted using the Mosfit program [5]. Magnetic measurements were performed using a vibrating sample magnetometer (VSM) model DMS 1660, and hysteresis loops were recorded at room temperature until 2 kOe. Measurements of saturation magnetization versus temperature were carried out over the range 300 570K with an applied field of 2 kOe. The structure and microstructure of the samples was characterized with X-ray diffraction (XRD) and transmission electron microscopy (TEM); the main results are presented elsewhere [-6]. On XRD patterns, one can observe the presence of peaks characteristic of crystalline YIG for the samples heated above 973 K, while at lower temperatures the samples contain an amorphous-like iron garnet phase mixed with iron oxides as magnetite or maghemite, Fe304 and 7-Fe203 respectively. Samples B contain larger quantities of these oxides; a small fraction of orthoferrite (YFeO3) is also observed in samples B after treatment above 973 K. The mean size of the particles, estimated from XRDTable 2 Mean XRD and TEM sizes Sample A4 A6 A7 A8 A10 A12

XRD size (nm)

73 96 150 690

TEM size (nm) 20 50 100

500

line broadening and from TEM measurements, are listed in Table 2 according to the annealing treatment. In all the studied samples, particles exhibit rounded irregular shapes, generally lengthened.

3. Results

3.1. Thermal analysis Measurements of combined T G A - D T A were carried out on the sample A1, in Ar atmosphere and at 10 K/rain. The TGA curve gives evidence for a weight loss of 10%, mainly due to the loss of organical residuals from the sample. A previous TGA study of the gel shows that decomposition mainly occurs over the range 473-673 K. IR spectra of samples annealed at 673 K exhibit bands typical of carbonates or ionized carboxylates [7], in good agreement with TGA results of sample A1, where a weight loss appears due to the decomposition of these carbonates or carboxylates. No exothermic peak occurs on the DTA curve of sample A1; indeed, one would expect the presence of a crystallization peak, because the XRD pattern of the sample before the DTA measurements looks like that of a typical amorphous while that of the sample resulting after DTA measurement exhibits well-defined Bragg peaks, consistent with a crystalline YIG phase. The absence of exothermic peak would indicate that particles are crystalline, but XRD patterns before DTA measurements are like those of amorphous materials. One can suggest that the material results from the presence of ultrafine crystallites (about 10 nm diameter) and consequently from a large contribution of boundaries with disordered structure; we called it amorphouslike behaviour. Thermogravimetric analysis was carried out in the presence of a magnetic field on the samples heated at T A = 673 K and TA = 873 K, in order to determine the magnetic ordering temperature. Measurements were recorded on each sample from room temperature until T = 1023 K, at 10 K/min and in air atmosphere. On each sample at least four consecutive measurements were recorded, because a weight loss due to gas emission occurred in the first measurements. From the first thermal

P. Vaqueiro et al. /Journal of Magnetism and Magnetic Materials 169 (1997) 56 68

measurement, a temperature of magnetic ordering was detected at 843 K. However, from the following thermal measurements recorded on the same sample, two temperatures of magnetic ordering were determined, located at T - 5 3 0 K and T = 843 K, respectively. They can be attributed to the Curie temperatures of Y | G and magnetite, respectively. Thus, the presence of magnetite can be established, whose Curie temperature is T = 850 K, and the presence of maghemite can be discarded, as it becomes hematite above T - 673 K, with a Curie temperature of T = 953 K. The transition temperature of all studied samples are found in the range 525 530 K, smaller than the Curie temperature of bulk crystalline YIG (Tc = 555 K), and much higher than the freezing magnetic temperature of amorphous YIG (To < 80 K according to the amorphization process [8]). The small lowering of Tc can be first attributed to the presence of both disordered boundaries and ultrafine crystallites. After these T G A measurements, XRD analyses were carried out, and the patterns exhibited Bragg peaks attributed to the crystalline YIG phase. 3.2. Miissbauer spectrometry

The crystalline structure of the garnet belongs to the space group Oh1° (Ia3d). One can distinguish three types of cationic sites: tetrahedra (d), octahedra (a) and dodecahedra (c). In YIG, Fe 3 + ions occupy octahedral (16a) and tetrahedral (24d) sites and y3+ ions (24c) dodecahedral sites. The 57Fe M6ssbauer spectra of polycrystalline YIG exhibit three high spin state Fe 3+ components, one attributed to tetrahedral sites, and the others two to octahedral sites, the relative absorption of each component in agreement with the population of the different iron sites. Two gels B, designated as GelB1 and GelB2, and a gel A were studied at 77 K as a frozen solution. The gels B result from the same initial solution, the only difference is that this solution was heated for a longer time for the preparation of GelB2. One can expect the synthesis of the same compound for the two gels B. Citric acid form complexes with cations which prevent precipitation from solution, but also polymerizes with ethylene glycol; nitrogen is par-

59

tially lost as nitrogen oxides when the solution is heated to obtain the gel. The gel A does not contain ethylene glycol, so the polymerization reaction does not occur. However, M6ssbauer spectroscopy gives clear evidence for the presence of Fe 2+ in GelB2 and in gel A, as shown in Fig. 1 and in Table 3. But, no Fe 2+ phase is observed on GelB1 as well as on the different samples obtained after further annealing treatment from these gels, except in sample B1 (see Table 3). One can suggest the occurrence of Fe 2÷ seems dependent on heating time and/or temperature of the initial solution. Let us mention some Massbauer studies of photolysis and radiolysis on ferric citrate, whose hyperfine data are in a good agreement with those obtained here for the gels containing iron citrate [9, 10]. The non-irradiated iron citrate spectrum exhibits a paramagnetic doublet of Fe 3 +. After irradiation, with electrons as well as ultraviolet radiation, the Fe 3+ lines have decreased in intensity, whereas a quadrupolar doublet of Fe 2÷ appears, with a quadrupolar splitting value dependent on the Fe2 +/Fe 3+ ratio. The proposed mechanism is a oxidative decomposition of the citrate coupled with the reduction of the Fe 3 + ion; due to irradiation the citrate would form radicals that would disappear very quickly by reaction with the Fe 3"- ions. Recently, it has been reported that photochemical reactions of Fe(llI) citrates complexes occur on time scale of minutes in sunlight [11]. However, the presence of 02 in the solutions leads to Fe(II) yields substantially lower than those in absence of 02. Our initial solutions are exposed to sunlight during heating, but are not purged: one can expect the presence of 0 2 favours the decrease of the Fe(II) concentration. The involved mechanisms in sunlight photolysis are quite complex, but the initial proposed reaction is the formation of a citrate radical like in electrons and ultraviolet irradiation. The spectra, recorded at 77 and 300 K. of both series A and B samples, heated at 673 and 873 K, exhibit a predominant asymmetrical doublet with broad lines and a low-intensity sextet. The quadrupolar component was reproduced with a continuous discrete distribution of quadrupolar splitting P(A). The obtained mean values of the hyperfine parameters of the doublet are similar to those reported in the literature for the amorphous YIG

60

P. Vuqueiro et al. / Journal qf Magnetism and Magnetic Materials 169 (1997) 56-68

\ -5

Vetocity (ram/s) Fig. 1. M6ssbauer spectrum at 77 K of sample GelB2.

Table 3 Hyperfine parameters of samples containing Fe 2÷ Sample

Site

GelB1 GelB2

Fe 3+ Fe 2+ Fe 3+ Fe 2+ Fe 3+ Fe2 + Fe 3+ Fe 2+ Fe 3+ Fe 3+ Fe z + Fe 3+ Fe 2+ Fe 3+

GelA B1 B1 Iron citrate [8] Iron citrate photolytic [8] Iron citrate irradiated [-8]

Temperature (K) 77 77 77 300 77 77 77 77

[12]. F u r t h e r m o r e , the v a l u e s of t h e h y p e r f i n e p a r a m e t e r s of the sextet differ f r o m t y p i c a l v a l u e s o b s e r v e d o n p o l y c r y s t a l l i n e Y I G , b u t t h e y are close to t y p i c a l v a l u e s of m a g n e t i t e . In a d d i t i o n , the 4.2 K s p e c t r a w h i c h e x h i b i t b r o a d sextets (one is i l l u s t r a t e d in Fig. 2a) for b o t h s a m p l e s , w e r e fitted

IS (mm/s)

(QS) (mm/s)

%

0.52 (2) 1.32(1 0.52(1 1.38(2 0.52(8 1.15(3 0.45(5 1.29(1 0.53(4 0.523 1.21 0.546 1.318 0.538

0.60 2.97 0.8 2.15(6) 0.58(6) 2.34(5) 0.56(9) 2.74(1) 0.78(6) 0.612 2.82 0.633 2.84 0.758

100 40 60 3 97 88 12 91 9 lO0 30.8 53.3

w i t h a d i s c r e t e d i s t r i b u t i o n o f h y p e r f i n e field P(Hhf) (the h i s t o g r a m is p l o t t e d in Fig. 2b), a s s u m i n g a c o m m o n v a l u e for the i s o m e r shift as well as of the q u a d r u p o l a r shift in all c o m p o n e n t s ; the m e a n v a l u e s of the i s o m e r shift a n d o f the h y p e r f i n e field were estimated at 0.43 m m / s a n d 47.0 T, respectively.

P. Vaqueiro et al. / Journal o f Magnetism and Magnetic Materials 169 (1997) 56-68

i

(a)

e6

i

-io

Velocity. (ram/s)

25 (b)

* •

61

,13

,15

if)

55

HyperfineField(T)

Fig. 2. (a) M6ssbauer spectrum at 4.2 K of sample A2. (b) Hyperfine field distribution of the spectrum of Fig. 2a.

The distribution of the hyperfine field is asymmetrical with two maxima located at 45.0 and 51.6T. One can note that a mean value of the hyperfine field of 45.0 T has been obtained in the case of the a m o r p h o u s Y I G prepared by twin roller quenching technique [13], a value that agrees with our first maximum; the value of the second maximum could be attributed to the presence of magnetite. These results are in good agreement with those deduced from X R D patterns of these samples,

which exhibit, on the one hand, a high intensity and broad band and, on the other, low intensity and broad peaks; they may be attributed to a predominant amorphous-like phase and a crystallized phase identified as magnetite or maghemite, respectively. F r o m T G A measurements the existence of magnetite was established, consistent with M6ssbauer and X R D results. By analyzing the shape of the distributions P(A), the values of the ratio q = ( A 2 ) / ( A ) 2 are close to 1.28 [14], consistent

62

P. ~21queiro et al. /Journal c~fMagnetism and Magnetic Materials 169 (1997) 56 68

with expected value for d = 2, dimensional parameters used by Czjzek [15] in random packing of hard spheres. This last spectrum, recorded at 4.2 K, was also fitted according to the procedure proposed by Eibschutz, which consist in using six independent Gaussian distributions of natural width Lorentzian lines. In Table 4, the mean positions of the lines and the half-widths at half-height in the case of our sample are compared to those estimated for an amorphous Y | G prepared by twin roller quenching technique [13]. The disparities observed on the line positions as well as on the linewidths are consistent with two different amorphous states, originated from the methods of synthesis. These both results at 300 and 4.2 K obtained on these samples are consistent with an amorphous-like behaviour, accounting for X-ray diffraction and DTA results. Concerning the amorphous-like phase, the present results are consistent with those obtained on Fe203 amorphous particles; indeed, it was found that the hyperfine parameters are similar to those of amorphous iron oxides systems prepared by a conventional technique [16]. The Massbauer spectra recorded at 300 K on the samples annealed at 973 K, exhibit a rather complex hyperfine structure: an asymmetrical doublet is clearly observed in the low velocity range, and also magnetic sextets appear with resolved lines. The quadrupolar doublet may be attributed to an amorphous-like YIG phase, whereas the second contribution can be decomposed into three magnetic components characteristic of crystallized YIG.

The values of the hyperfine field are smaller than those typically observed and the linewidth of the tetrahedral component is strongly broadened (F ~< 0.8). Let us remember that the XRD patterns exhibit peaks characteristic of crystalline YIG and a broad band attributed to the presence of remaining amorphous-like YIG phase, consistent with the M6ssbauer spectra. From Fig. 3, one can conclude that the larger the treatment time is, the larger the percentage of crystallized YIG. The spectra recorded on the A and B samples treated at 1073, 1173 and 1273 K, evidence clearly only crystalline YIG. Nevertheless, they present some singularities: all the spectra were first fitted with three magnetic components, two octahedral and one tetrahedral, since crystalline YIG exhibit three positions for Fe 3+. Whatever the temperature, the values of the hyperfine field remain smaller than those encountered for the bulk, and the linewidths values are high, even at 4.2 K. For all the samples, the linewidth of tetrahedral component is larger than those of octahedral components, whereas the population of tetrahedral site is larger than the expected value (60%) in well-crystallized Y I G [17]. With the annealing temperature as well as the treatment time of the samples, the values of the hyperfine field increase and those of the linewidth and of the population of tetrahedral sites decrease, tending to those of the bulk. Introducing three tetrahedral components during the fitting procedure to take into account the broadening of the component corresponding to the tetrahedral iron

Table 4 Line positions and half-widths at half-height (w) for the M6ssbauer spectrum measured at 4.2 K of the sample sample A2 and the amorphous YIG prepared by twin roller quenching technique [12] Lines

1 2 3 4 5 6 Note:

IS (mm/sl

w (ram/s)

samplc A2

roller quenching

sample A2

roller quenching

- 7.60 4.37 1.24 I. 19 4.36 7.62

7.308 4.168 - 1.168 1.180 4.169 7.307

0.73 0.49 0.29 0.28 0.52 0.80

0.601 0.541 0.458 0.468 0.572 0.703

The origin of these spectra correspond to an isomer shift of 0.44 mm/s with respect to iron metal.

P. Vaqueiro et al. /'Journal o]'Magnetism and Magnetic Materials 169 (1997) 56-68

63

Table 5 Linewidth of the tetrahedral component of the sample A8

i~ (~ " i

: ~*J G'~i 12hours

Temperature (K)

F (mm/s)

300

0.71(2)

77 4.2

i,!;

8 hours

*

: ..'

2 hours

i!!r

0.61(2) 0.59(1)

Table 6 Hyperfine fields of the sample A8 and the ceramic standard Sample

Temperature (K)

Hf oc, (T)

H'f oc, (T)

Hf tet (T)

Ceramic A8 Ceramic A8

300 300 4.2 4.2

48.0(3) 47.0(2) 54.3(3) 54.0(2)

49.7(3) 48.5(2) 56.1(3) 55.2(2)

39.8(3) 39.0(1) 47.5(3) 47.2(1)

"iI

"t I

I

i

Velocity (mm/s) Fig. 3. M6ssbauer spectra measured at room temperature of samples B treated at 973 K.

sites, the spectra are better fitted and the populations of tetrahedral sites are close to 60%. These tetrahedral components differ on the values of the quadrupolar shift, having a positive, a negative and a zero value, respectively, and are consistent with structural distortions on the tetrahedral units. The temperature dependence of the linewidth of the tetrahedral component (when the fitting procedure involves one tetrahedral sextet) is listed in Table 5, in the case of sample A8. Linewidths of the sextets corresponding to the octahedral iron sites take always values around 0.28-0.32mm/s. At 4.2 K the hyperfine field values for the crystallized samples are lower than those estimated on bulk, but the difference is smaller than it was at room temperature. One can expect the occurrence of

thermal fluctuations of the magnetization, essentially due to the smaller particles or/and relaxing superficial layer (Table 6). Spectra were recorded over the range 77-573 K for samples B8 and B l l . They were annealed at 1073K during 2 h and at 1173K during 12 h, respectively, and in consequence the mean size of the particles is smaller for B8 sample (~90nm) than for B l l ( ~ 3 2 0 n m ) . The MSssbauer spectra of the two samples recorded between 77 and 4 5 0 K are similar, and their hyperfine parameters values agree with those of crystalline YIG, with the particularities previously described. But with the approach to the Curie temperature, above 450 K, the hyperfine field values decrease faster than those of ceramic YIG, and also B8 values are smaller than B l l (see spectra recorded at 473 and 523 K in Fig. 4). The lowering of hyperfine field values may be attributed to superparamagnetic effects, originating from the smaller particles, or to the presence of boundaries which exhibit structural disorder, consistent with a reduction of magnetic interactions. Such a feature is confirmed by the temperature dependence of the linewidth of the ferromagnetic resonance [18].

P. Vaqueiro et al, /Journal of Magnetism and Magnetic Materials 169 (1997) 56-68

64

K

473 K

573 K

15

I

I

I

I

I

-10

-5

0

5

10

573 K

15 -15

I

I

I

I

I

-I0

-5

(I

5

10

Vch~city (ram/s)

Velocity 0nm/s)

Sample B1 1

Sample B8

15

Fig. 4. M6ssbauer spectra of samples B8 (annealed at 1073 K, 2 h) and B11 (annealed at 1173 K, 12 h).

3.3. Magnetic measurements As illustrated for the A samples in Fig. 5, the saturation magnetization at room temperature increases with the annealing temperature, reaching values very close to the bulk value of 26 emu/g for the crystallized samples, but always remaining lower than the bulk value. Samples annealed below 1000 K, which contain mainly an amorphous-like phase, exhibit low values of the saturation magnetization. Let us remember that hyperfine fields for these samples increase also with annealing temperature and treatment time, tending to bulk values, which is in good agreement with the saturation magnetization behaviour. The values of the coercive field also increase with the annealing temper-

ature, reaching a maximum value for the samples treated at 1173 K. However, at higher annealing temperatures, the coercivity decreases due to the increase in particle size. In Fig. 6a and Fig. 6b one can observe the dependence, at room temperature, of the coercivity and the squareness, SQ, defined as Mr~Ms where Mr is the remanence magnetization and Ms the saturation magnetization, for the crystallized A series samples versus the average size of the particles estimated from XRD, using Scherrer's description. In this plot only samples annealed above 1000 K are presented where no other phase other than crystallized YIG was detected. Depending on the annealing temperature and treatment time, the obtained YIG particles have a different average

P. Vaqueiro et al. / Journal of Magnetism and Magnetic Materials 169 (1997) 56-68

65

25 .

-83

2560

//

15-

.

10-

2O 5Q-0

I

833

7133

l

I

I

I

I

830

930

1033

11Q3

1333

0

1..333

Annealing temperature (K) Fig. 5. Saturation magnetization M~ and coercive force, Hf, measured at room temperature for samples A treated during 2 h at different temperatures.

size, in the range 50 700 nm. A coercitivity maximum occurs at an average size of about 200 nm; this value is related to the existence of different magnetic processes in the particles: so, particles smaller than 200 nm are expected to be single domain and the reversion of the magnetization takes place by thermally activated and homogeneous rotation processes, whereas in larger particles the reversion of the magnetization is associated to inhomogeneous rotation processes or to domain wall displacements in multidomain particles [19]. Measurements of saturation magnetization versus temperature were carried out on the samples B8 and B l l as well as on the ceramic sample (see Fig. 7). Both B8 and B11 samples have very similar values of saturation magnetization and of Curie temperature, but they remain smaller than those characteristic of ceramic garnets. A dependence of Curie temperature with size is reported for MnFe204 nanoparticles [20]; however for our YIG particles no dependence was observed. These samples were also studied with M6ssbauer spectrometry up to the Curie temperature, and differences between hyperfine field values were observed above 450 K. As discussed previously, such behaviour can

be attributed to superparamagnetic effects or to the presence of boundaries.

4. Discussion

YIG particles prepared by annealing below 973 K exhibit XRD patterns typical of an amorphous structure, and the values of the hyperfine parameters are quite consistent with those reported for amorphous YIG, but no crystallization process is evidenced from DTA measurements. The absence of exothermic peak on DTA curve would indicate that the particles are crystalline before the treatment, inconsistent with XRD and M6ssbauer data. Let us recall briefly some differences which were evidenced from the hyperfine parameters values and from the temperature of magnetic ordering on various amorphous iron garnets, according to the synthesis method. The hydroxides coprecipitation procedure leads to amorphous variety with rather low magnetic ordering temperatures; the structure was modelled as a random packing based on tetrahedral and octahedral ferric units [21]. Glassy iron garnets prepared by twin roller quenching

66

P. Vaqueiro et al. / Journal o f Magnetism and Magnetic Materials 169 (1997) 56 68

60.~

50

.

3O

20

I

I

].03

~

)

1

I

l

I

300

-1(30

500

600

700

Avera.~ size (nm)

(a) (2-10

0.35-

0.30II ~SY

0.~'

1

0.20I

0 (b)

103

I

200

I

I

I

I

300

403

5o0

603

y_ [

7Oo

A v e r a ~ Size (nm)

Fig. 6. (a) Coercive field at room temperature versus the average size of the particles (estimated from XRD measurements). (b) Dependence of the squareness at room temperature on average particle size for the A series crystallized samples.

technique exhibit slightly higher m a g n e t i c freezing t e m p e r a t u r e : a d i s t r i b u t i o n of ferric ion c o o r d i n a tion centered a r o u n d 5 has been suggested [12]. A m o r p h o u s b e h a v i o u r was evidenced after h e a v y ion i r r a d i a t i o n a p p l i e d on crystalline i r o n garnet:

the M 6 s s b a u e r d a t a m a y suggest the presence of iron l o c a t e d in o c t a h e d r a l , p y r a m i d a l a n d t e t r a h e d ral units [-22]. N e u t r o n diffraction studies of a m o r p h o u s Y I G o b t a i n e d with fast n e u t r o n irrad i a t i o n s h o w that local cationic o r d e r i n g occurs;

P. Vaqueiro et al. /Journal of Magnetism and Magnetic Materials 169 (1997) 56 68

25.

=..,,.1. e_

.

"L'm-.=

-----

Ceramic

--o

138

67

20.

15.

E 10'

I

I

I

I

I

I

300

350

400

450

5(13

550

Temperature (Ix-') Fig. 7. Saturation magnetization versus temperature of B8 and B11 samples and ceramic YIG.

it is concluded that the coordination polyhedra remain typical of the garnet structure [23], which is in agreement with M6ssbauer data [8]. According to the different results previously mentioned, one can suggest that the citrate gel process leads to nanoparticles with an amorphous structure. The first annealing treatments will give structural relaxation with the onset of crystallization: the amorphous-crystalline transformation gives a nanostructured behaviour to the particles for which the atomic weight of boundaries remains important. For grains with an average size of 10nm if the thickness of the grain boundaries is around 2 nm, the volume fraction of boundaries is estimated at 30%. Analogous behaviours have been observed in the case of nanoalloys prepared by mechanical alloying [24]. Such a description is consistent with the presence of broad Bragg peaks on the X-ray diffraction pattern, and with the absence of exothermic peak on the DTA curve. Further annealing treatments favour first the growth of the nanocrystallites within the particles, and the aggregation of nanoparticles, and then the crystalline behaviour of the particles.

Samples treated above 973 K exhibit crystalline structure, according to X-ray diffraction and MiSssbauer data; nevertheless, the value of the linewidth of the magnetic components on MSssbauer spectra decreases when the annealing temperature increases, and tends to the normal value observed in well-crystallized YIG. Such a feature is consistent with the DC magnetic measurements and can be explained as a progressive increasing in size of the crystallites within the particles and of their crystalline quality. Furthermore, in these samples, the linewidth of the sextet attributed to the tetrahedral iron site is always larger than those of octahedral components. This widening in the tetrahedral component lines may be originated by various deviations of the direction of magnetization from (1 1 1) [-25] resulting from structural distortions which are stronger in the tetrahedral than in the octahedral iron sites. Indeed, as the magnetization lies along (1 1 1) direction in crystalline iron garnets, all the tetrahedral sites have their symmetry axes at an angle of 55 ° from the direction of magnetization, are equivalent and a single sextet occurs, whereas structural distortions of the tetrahedral iron sites originate line broadening.

68

P. Vaqueiro et al. / Journal of Magnetism and Magnetic Materials 169 (1997) 56 68

5. Conclusions These results show the potentiality of the citrate gel process in the preparation of YIG nanoparticles. Samples annealed above 973 K contain YIG crystallized particles and exhibit saturation magnetization values at room temperature in the range 23-25 emu/g, close to bulk value. Coercivity values depend on the average particle size and their maximum value is associated with the transition from homogeneous rotation processes to non-homogeneous rotation processes in single-domain particles. The presence of Fe z+ ions in the initial gel used to prepare the YIG nanoparticles was established from M6ssbauer measurements. After heat treatment of the gel below 973 K, samples contain magnetite and an amorphous-like YIG phase. Above 973 K, samples contain crystalline YIG nanoparticles which present a structural distortion affecting iron in tetrahedral sites. Depending on the annealing temperature, particles have a different average size and show changes in magnetic behaviour.

Acknowledgements P. Vaqueiro wishes to acknowledge partial financial support from XUGA209031395 and DGICYT PB94-1528, and thanks Xunta de Galicia and University of Santiago.

References [1] F. Bertaut, F. Forrat and C.R. Acad, Sci. Paris 242 (1956) 382. [2] D. Roy, R. Bhatnagar and D. Bahadur, J. Mater. Sci. 20 (1985) 157.

[3] V.K. Sankaranarayanan and N.S. Gajbhiye, J. Am. Ceram. Soc. 73 (1990) 1301. [4] K. Matsumoto, K. Yamaguchi and T. Fujii, J. Appl. Phys. 69 (1991) 5918. [5] J. Teillet and F. Varret, Mosfit program, unpublished. [6] P. Vaqueiro, M.P. Crosnier-Lopez and M.A. Ldpez Quintela, J. Solid State Chem. 126 (1996) 161. [7] P. Vaqueiro and M.A. Ldpez-Quintela, unpublished resuits. [8] J.M. Greneche, H. Pascard and J.R. Regnard, Solid State Commun. 65 (1988) 713. [9] D.N. Buchanan, J. Inorg. Nucl. Chem. 32 (1970) 3531. [10] E. Baggio-Saitovich, J.M. Friedt and J. Danon, J. Chem. Phys. 55 (1972) 1269. [11] B.C. Faust and R.G. Zapp, Environ. Sci. Technol. 27 (1993) 2517. [12] M. Eibschutz and M.E. Lines, Hyperfine Interactions 27 (1986) 47. [13] M. Eibschutz and M.E. Lines, Phys. Rev. B 26 (1982) 2288. [14] M.E. Lopez-Herrera, J.M. Greneche and F. Varret, Phys. Rev. B 28 (1983) 4944. [15] G. Czjzek, J. Fink, F. Gotz, H. Schmidt, J.M.D. Coey, J.P. Rebouillat and A. Lienard, Phys. Rev. B 23 (1981) 2513. [16] P. Ayyub, M. Multani, M. Barma, V.R. Palkar and R. Vijayaraghavan, J. Phys. C: Solid State Phys. 21 (1988) 2229. [17] Landolt-B6rnstein, Numerical Data and Functional Relationships in Science and Technology Ser. III, Vol. 12a. (Springer, Berlin, 1996). [18] F. Tesson and J.C. Fayet, unpublished results. [19] H. Kronmiiller, in: Supermagnets Hard Magnetic Materials, Eds. G.J. Long and F. Grandjean (Kluwer, Dordrecht, 1991) p. 461. [20] Z.X. Tang, C.M. Sorensen and K.J. Klabunde, Phys. Rev. Lett. 67 (1991) 3602. [21] W. Girnus, H. Beuthien, R. Priess and W. Gunser, J. Magn. Magn. Mater. 54-57 (1986) 225. [22] C. Houpert, Ph.D. Thesis, University of Caen, France (1989) unpublished. [23] Y.G. Chukalkin, V.R. Shtirz and B.N. Goshchitskii, Phys. Star. Sol. (a) 112 (1989) 161. [24] R.W. Siegel, J. Phys. Chem. Solids 55 (1994) 1097. [25] G. Ballestrino, P. Paroli and S. Geller, Phys. Rev. B 34 (1986) 8104.