Journal of Magnetism and Magnetic Materials 125 (1993) L23-L28 North-Holland

Letter to the Editor

Evidence for very garnet structure

H. Pascard

high magnetic permeability in the ferrimagnetic

Laboratoire des Solides Irradik, URA CNRS 1380, Ecole Polytechnique, 91128 Palaiseau Cgdex, France

J.M. Grenitche

Laboratoire de Physique des Mat&am, UR4 CNRS 807, Unioersith du Maine, 72017 Le Mans Gdex, France

Received 16 February 1993; in revised form 4 May 1993

We have discovered a new combination in chemical composition in garnets showing very high vaiues of initial magnetic permeabiIity. These garnets were found by using two magnetic phase transition phenomena in the same crystallographic structure: a spin reorientation transition and a ferrimagnetic paramagnetic transition. The critical temperatures are fixed by the introduction in the garnet Y3Fe,0,, of terbium ions x in dodecahedral sites and scandium ions y in octahedral sites, Values of permeability from IO3 to IO4 have been measured in a broad range of temperature.

Until now, very high values of magnetic per- meability (103-104) in a broad temperature range have been observed only in materials with the spine1 structure. Generally, the MnZn spine1 fer- rite is used either as polyc~stalline or as single. crystal. The high value of permeabili~ finds its origin in the compensation of the negative mag- netocrystalline anisotropy by a large positive anisotropy arising from bivalent iron ions. It is well known that the initial magnetic permeability measured on a polyc~stalline toroidal sample is proportional to &/K, XI?&, where nB is the resulting magnetization, K, the magnetocrys- talline anisotropy constant and D, either the mean grain size, or the mean distance between defects that could pin the domain walls. The first attempt to obtain high pe~eability in a structure other than spine1 was made by using the garnet structure [I]. The concept of weakening of K, when the temperature tends toward the Curie

Cor~es~ndence to: Dr. H. Pascard, Laboratoire des Solides IrradiBs, URA CNRS 1380, Ecofe Polytechnique, 91128 Palaiseau CBdex, France.

temperature (ferrimagnetic-paramagnetic transi- tion), was used. High values of permeability were obtained by introducing in Y,Fe,O,, non-mag- netic Sc3+ ions in octahedral sites leading simul- taneously to a decrease in K, and to an increase in the net magnetization as expected from the N6el theory. VaIues of permeability of about lo3 were obtained, in a narrow range (about 50 K) around the room temperature, against the MnZn ferrite that permits to reach a permeability of lo4 in a wide temperature range of about 200 K 121.

This paper relates the attempt we made to obtain very high permeability in the garnet struc- ture by using simultaneously in the same crystal- lographic structure: the concept of weakening of the anisotropy K, near the Curie temperature, at high temperature, and the concept of canceIla- tion of Fi, at a spin reorientation transition, at low temperature. The coexistence of the two phe- nomena, realized by two different kinds of substi- tutions, is hopefully possible in garnets of given composition. It is well known that a spin reorien- tation transition is observed in substituted gar- nets, by repIacing in YIG the yttrium ions in

0304-8853/93/$06.00 0 1993 - Elsevier Science Publishers B.V. All rights reserved

L24

-101 , , /

100 200 301

T (K) Fig. I. Schematic representation of temperature dependence of the anIsotropy K,. for the hypothetical composition

Y2 <Tb,, iFe, :Sc,, HO ,? (curve (c)J. calculated by the addition of the anisotropy K, due to the Th,, -; ions (curve-(u) from ref. [3]) and

due to the Fe, 2 ions (curve (h) from ref. [I]).

dodecahedral sites, partially or entirely by certain rare earth ions such as samarium, erbium, ter- bium. Let us consider the Y,-,Tb,Fe,O,, garnet where a spin reorientation transition between the (111) (negative K,) and (100) (positive K,) easy axis is found over a broad range of compositions. K, becomes positive at low temperature for ter- bium content x between 0.26 and 2.12 [3]. At 300

K, where K, weakly depends on the terbium content, the main contribution to the anisotropy comes from the iron ions, that is negative. At low temperatures, K, changes sign because the anisotropy is determined mainly by the rare earth sublattice whose contribution to K, is positive. So, the net K, vanishes at the temperature where the anisotropy components from terbium and iron

,b 6 - ‘\\

\ \ ---___ \

k_ \ z4 - I

I? 2- ‘.

_. ‘\ ..,‘\

‘\\ \\ I

0 *O” T(K)

400

Fig. 2. Schematic representation of temperature dependence of the magnetization H ,+ per formula unit. for the hypothetical

composition YZSTb,,,5FeJ.ZSc,~X0,? (curve Cc)). calculated from the magnetization of the composition Y,Fe,,Sc,,,O,, (curve (b)

from ref. [l]) and from the magnetization temperature dependence of the composition Y2 <Tb,, <Fe,012 (curve (a), from refs. [4]

and [S]).

H. Pascard, J.M. Grendche / Very high magnetic permeability in ferrimagnetic garnet L25

ions compensate each other and around that tem- perature the permeability, which is inversely pro- portional to K,, can become very high.

Now, by using the concept of single ion anisotropy additivity, it is possible to predict quantitatively the temperature dependence of K,. Fig. 1 shows the calculated variation of K, versus temperature for the hypothetical composition Y,,,Tb,,Fe,.,Sc,,O,, obtained by the addition of two contributions to K, as follows:

- a negative contribution due to the iron sub- lattice in which scandium ions are introduced to weaken the anisotropy at temperatures higher than 300 K (curve (bl corresponding to the for- mula Y,Fe,,,Sc,,,O,, with the values of K, from ref. [ll).

- a positive contribution due to the terbium ions introduced in the yttrium sublattice to com- pensate the negative contribution at tempera- tures lower than 300 K (curve (a) corresponding to the formula Y,,Tb,,,Fe,O,, with the values of K, from ref. [3]).

The calculated resulting anisotropy K, (curve (c)l shows that it should be possible to realize one garnet material whose anisotropy would be very small in a large range of temperature between the spin reorientation transition and the Curie temperature.

In the same way, it is possible to predict the temperature dependence of the magnetization

eta, by using the Neel theory of ferrimagnetism based on the additivity of the sublattices’ magne- tizations. In fig. 2, we show the calculated tem- perature dependence of the resulting magnetiza- tion ~ln from:

- the contribution due to the iron sublattice in which the scandium ions in octahedral site tend to increase n, with respect to Y,Fe,0,2 (curve (b) corresponding to the composition Y3Fe4,2 Sc,,O,, with the values of n, from ref. [ll);

- the contribution from the terbium sublattice leads to a decrease of n, with respect to YIG (curve (a) corresponding to the composition Y,,STb,,,Fe,O,, deduced from refs. [4] and [5]). The curve (cl in fig. 2 shows the behaviour of the resulting magnetization ng that could be ex- pected for the hypothetical composition Y2.5Tb0.5

Fc,zScO.sOrZ. To check the possibility of obtaining very high

magnetic permeabilities, we chose to study two different compositions close to the hypothetical

one Yz.STbO.sFe&%.sO r2, namely YzAT~R~ Fe,,,,Sc,,,,O,, (x = 0.4; y = 0.75) and Y,.,Tb,,s Fe,,,Sc,,,O,, (x = 0.8; y = 0.85). The prepara- tion conditions of polycrystalline samples are the following. We start from Y203, Tb,O,, Fe,O, and Sc,O, of high purity (5N), taken in stoichio- metric quantities corresponding to the final for- mula. Then, these oxides are mixed and milled by a specific process that does not affect the purity

0

T (K) Fig. 3. Experimental temperature dependence of the magnetization nB for per formula unit for two different substituted garnets:

curve (a): Y2.6Tb0,4Fe4.2SSc0.75012; curve (b): Y2.2Tbo.sFe,,,,Sco.ss0,~.

LX I-I Pascard, J.M. Grentche / Very high magnetic permeuhility in ferrimagnetic garner

[6]. Several samples of toroidal shape were pre- pared using a compacting pressure in the range 1 ton/cm* to 10 tons/cm* and were sintered at temperatures in the range 1450-1530°C for a duration varying from 50 to 100 hours. Polycrys- tals of high crystalline quality were obtained. The samples show a brittle aspect and a high mechan- ical hardness. High density, within l/1000 of that of the theoretical one, was obtained. Crystallite sizes up to two hundreds microns were obtained. The X-ray diffraction pattern, characterized by very narrow peaks, confirmed the garnet struc- ture. We have plotted in fig. 3 the temperature dependence of the magnetization ng measured by using a vibrating magnetometer, in applied magnetic fields up to 1.7 T. For both composi- tions, a very good agreement is obtained with the predicted temperature dependence (compare curve (c) in fig. 2 and curves (a> and (b) in fig. 3).

Let us examine now the results concerning the initial permeability measured on toroidal sam- ples, using the transformer method, at a fre- quency of 1 kHz. The field applied through the primary winding was lower than 1 mOe. The secondary voltage, proportional to the permeabil- ity, was recorded as a function of temperature. The sample was slowly cooled from the Curie

temperature down to 77 K. The exact value of the permeability is measured, using an inductance bridge, at 300 K. Fig. 4 shows the results obtained for the temperature dependence of the perme- ability CL’ for the two compositions:

Y,.,~,,.,Fe,.,sSc,,.,sO r2 (cur= (a)) fig. 4) >

Y2.2~,,.,Fe,,sSc,,.,s0,, @WC W l fig. 4).

It shows that all the effects that were expected appear:

_ the first peak, at low temperature. corre- sponding to the spin reorientation transition TSR whose position in temperature is fixed by the terbium content x (TSR = 180 K for x.,.~ = 0.4 (curve (a)) and TSR = 220 K for x-rh = 0.8 (curve (b));

_ the second peak, at high temperature, corre- sponding to the Hopkinson peak observed near the Curie temperature. The position of this peak is fixed by the scandium content 4’: T,,op = 370 K for y,, = 0.75 (curve (a)) and TElop = 320 K for ysc = 0.85 (curve (b));

_ the expected effect on the possibility to obtain very high permeability appears in the tem- perature range limited by the two peaks: perme- ability values of about lo3 to 10’ are observed in

c

Fig. 4. Experimental temperature dependence of the initial magnetic permeability: curve (a): for a golycrystalline sample of substituted garnet Y, ,Tb,,.,Fe, 25 SC 07s0,2 with D, about 100 k; curve (b): for a polyctystalline sample of substituted garnet

Y,,Tb”sFe,.,sSc,.ssO,, with 4, about 10 k; curve Cc): for a spine1 single crystal sample of composition Mn,,,,Zn,,,,Fez ,,O,

from ref. [2].

lo+ , , / , , , 200 300 400

T (K)

H. Fascard, .I. M. C&e&he / Very high magnetic per~eu~ifi~ in ~err~agnetjc garnet I.27

.a octahedra I sites A n tetrahedral sites

0 200 400

T (K) Fig. 5. Experimental temperature dependence of the hyperfine field mean values in octahedral and tetrahedral sites, deduced from “Fe Mijssbauer absorption spectra: curves 61): for a garnet Y,Fe,O,,; curves (b): for a substituted garnet Y2,hTb0_4Fe4,25Sc0,75012.

a broad range of temperature around the room temperature: a range of 200 K for xTb = 0.4 and ys, = 0.75, and a range of 120 K for xTb = 0.8 and ysC = 0.85. For comparison, we have also given in the same figure a representative result for the case of the spine1 structure, obtained with

Mn,,,Zn,.,, Fe,,,,O, single crystal [2]. These first results, from only two compositions, show that it is possible to increase still the permeability $ by adjusting the three very well defined parameters: the terbium content x, the scandium content y and the mean grain size 0, [6]. Further, as compared to the spine1 structure, the garnet has several advantages, such as that there are no bivalent iron ions and no unoccupied sites, the preparation conditions are very simple, the sam- ples are reproducible and the surface is very stable chemically. Also, the resistivity values of the garnets are orders of magnitude higher than that of MnZn ferrite: eddy current losses are nil.

Finally, the existence of a spin reorientation transition has been well established by 57Fe Miissbauer spectroscopy which detects changes in the thermal evolution of the hyperfine field. It is well known that the magnetic internal field inter- acting with the nuclear dipole moment will split the nuclear state according to the Zeeman the- ory. The main result is reported in fig. 5 (the details will be published elsewhere [7-J>, where the temperature dependence of the hyperfine field

for the composition Y2,6Tb0.4Fe4,25Sc0,750,2 is compared to that of Y,Fe,O,,. One can see that the substituted garnet exhibits two tetrahedral and one octahedral iron sites while for the YIG one has two octahedral and one tetrahedral iron sites. Also, one can observe clearly that, in (b), the temperature dependence of the hyperfine field is characterized, for the three iron sites, by a discontinuity around 200 K, which corresponds to the spin reorientation transition temperature for the composition.

In conclusion, a new combination in chemica1 composition in garnets, with the formula Y3_x Tb,Fe5_,Sc,0,~ is found:

- polyc~stalline sampies of good crystalline quality are obtained, as shown by mechanical aspect and X-ray diffraction pattern;

- magnetization measurements show that the terbium ions enter dodecahedral sites and the scandium ions in octahedral sites;

- MCissbauer and permeability spectra versus temperature give evidence for the existence of a spin reorientation transition at low temperature, related to the terbium ions content x;

- very high values of the initial magnetic per- meability $ are observed (lo3 to lo41 in a broad range of temperatures (of the order of 200 K), which is comparabte with the values obtained for MnZn ferrite, a reference material in the field of high permeability;

L28 H. Pascard, J.M. Grentche / Very high magnetic permeability in ferrimagnetic garnet

- it is possible to increase the permeability p’ by adjusting the parameters or,,, ysC and the mean grain size D,.

The experimental results are well explained by using the sublattice magnetization NCel theory and the single ion anisotropy theory.

References

[l] J.R. Cunningham, J. Appl. Phys. 36 (1965) 2491.

[2] D. Stoppels, J. Appl. Phys. 52 (1981) 2433.

[3] K.P. Belov, A.K. Gapeev. R.Z. Levitin. AS. Markosyan

and Y.F. Popov. Zh. Eksp. Tear. Fiz. 68 ( 1975) 741 (Sov.

Phys. JETP 41 (197% 117).

We thank N. Lorenzelli (Laboratoire des Solides Irradies, Palaiseau) for the X-ray diffrac- tion data, M. Leblanc (URA 449, Le Mans) for the magnetization measurements and R. Krish- nan (Laboratoire de Magnetisme et Materiaux Magnetiques, Bellevue) for helpful discussions in the field of garnets.

[4] R. Pauthenet. J. Appl. Phys. 29 (19.58) 253.

[5] E.E. Anderson. J.R. Cunningham and G.E. McDuffie,

Phvs. Rev. 116 (1959) 624.

[6] A.-Globus, C.R. Acad. Sci. Paris 257 (1963) 1752.

[7] J.M. Grenkche and H. Pascard (to be published).