ONSET OF COLLECTIVE MAGNETIC BEHAVIOUR IN MAGNETICALLY DILUTED INSULATING Eu++ COMPOUNDS

(1)

HAL Id: jpa-00216651

https://hal.archives-ouvertes.fr/jpa-00216651

Submitted on 1 Jan 1976

HAL is a multi-disciplinary open access

archive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

ONSET OF COLLECTIVE MAGNETIC BEHAVIOUR

IN MAGNETICALLY DILUTED INSULATING Eu++

COMPOUNDS

H. Maletta, G. Crecelius

To cite this version:

(2)

JOURNAL DE PHYSIQUE Colloque C6, supplkment au no 12, Tome 37, De'cembre 1976, page C6.-645

ONSET OF COLLECTIVE MAGNETIC BEHAVIOUR IN MAGNETICALLY

DILUTED INSULATING

Eu+

+

COMPOUNDS

H. MALETTA and G. CRECELIUS

Institut fur Festk&perforschung, KFA Julich, 517 Jiilich, W.-Germany

R6sumB.

-

En utilisant l'effet Mossbauer du noyau l5lEu nous avons etudie I'influence de la tempkrature sur les champs magnetiques hyperfins pour des series de solutions diluks de E U I - ~ S ~ ~ S et E~l-~Sr,S04. Pour une dilution suffisante du syst6me magnktique nous observons un comporte- ment superparamagnktique que nous pensons &re typique pour un debut de magnetisme collectif. Abstract. - Using the Mossbauer of the 151Eu nucleus we have studied the temperature dependence of the magnetic hyperfine fields in the dilution series E U ~ - ~ S ~ ~ S and Eu1-~Sr+S04. For sufficient dilution of the magnetic system we observe a superparamagnetic behaviour which we consider to be typical for the onset of collective magnetism.

1. Introduction.

-

Recently appreciable effort has been undertaken for an understanding of the dilute magnetic limit. The simplest case should be the study of a single magnetic impurity in a diama- gnetic matrix. In metals those questions lead to the well-known Kondo or spin-fluctuation problems [I]. The next step led to the study of systems with a finite concentration of magnetic moments where one is concerned with impurity-impurity interactions. In addition, those systems directly exhibit the onset of collective magnetic behaviour and are nowadays referred to as spinglasses and mictomagnets 121.

Until now all those phenomena have been studied in metallic alloys, a fact which has the great advantage that one may use a wellknown mechanism describing the magnetic coupling in metallic systems, i. e. the Ruderman-Kittel-Kasuya-Yosida-interaction [3] via the polarization of the s-like conduction electrons. On the other hand the study of metallic systems intro- duces severe problems which are connected with the question of solubility of two metals with each other. Looking through the phase-diagrams of binary metallic systems one soon realizes that only a small number of metals forms continuous solute solution, whereas in most cases there exists a very limited solubility range. This has the consequence that one experiences large difficulties which are connected with metallurgical clustering. In all those systems one is far away from an understanding of the bonding mechanism for metallic systems whose solubility is limited or even vanishes. On the other hand the RKKY interaction is a poor approximation, too, because all those systems studied are binary transition metal systems whose Fermi surface is far from being a sphere.

Thus after a careful inspection most metallic systems appear not to be the best choice for a study

of the onset of collective magnetism. We have therefore decided to look for systems with welldefined chemical and magnetic interactions, namely insulating crystals with a varying degree of covalent and ionic bonds. There exists a large number of insulating mixed crystals forming a continuous dilution series without any change in the lattice structure and without much change in lattice constant and in chemical bond.

2. Experiments and results.

-

For our experiments we have chosen the concentrated magnetic systems EuS and EuSO,. The magnetic dilution in both systems is achieved by isostructural SrS and SrSO, with the lattice constant differing by 1

%

in the case of the Eu,-,Sr,S series, whereas X-ray analysis cannot distinguish between EuS0, and SrSO,. We have used the Mossbauer effect of the 21.7 keV y-line of the l5'Eu nucleus to study the temperature dependence of the magnetic hyperfine fields in the dilution series Eu, - ,Sr,S and Eu, -,Sr,SO,.

Pure EuS orders ferromagnetically below the Curie temperature of Tc = 16.6 K . Adding SrS the ordering

temperature is lowered to Tc = 2.9 K in E U ~ . ~ S ~ ~ . ~ S as has been established by magnetization measu-

rements [4]. With further increasing SrS content we were not able to detect a spontaneous magnetization above 2.0 K

-

the lowest measuring temperature possible in our magnetization equipment. We have thus determined the temperature at which the splitting of the Mossbauer spectrum vanishes which is identical with the ferromagnetic Curie temperature for lower SrS concentrations. In figure 1 the concentration dependence of these critical temperatures in the series Eu, -,Sr,S is shown.

For the EuS-rich samples including x = 0.6 the temperature dependence of the magnetic hyperfine splitting in the series Eul-,Sr,S is well described by a

(3)

C6-646 H. MALETTA AND G. CRECELIUS

Ob ' ' O 5

"

x-

' '----A 10 FIG. 2.

-

Typical lslEu Mossbauer spectra for the tempera-

EuS SrS ture dependence in the Eu-rich and Eu-poor region fo E U , - ~ S ~ ~ S with the examples of a) x = 0.5 at temperatures 1) 110 mK ;

FIG. 1.

-

Concentration dependence of the critical temperature 2) 1.4 K ; 3) 2.0 K ; and of b) x = 0.9 at 1) 46 mK ; 2) 420mK ;

To in the series E U ~ - ~ S ~ ~ S . For 0

<

x

<

0.5 Tcr is identical 3) 700 mK. with the ferromagnetic Curie temperature Te.

Brillouin function. For the more diluted samples we observe a quite different temperature dependence. Below temperatures of about 100 mK we always measure one single magnetically split spectrum. At elevated temperatures we do not as in the case of the concentrated samples observe a Brillouin function type decrease of the splitting with temperature, but the splitting remains nearly constant whereas the intensity of the spectrum decreases. Simultaneously an appreciable broadened line without structure appears superimposed on the split spectrum whose intensity increases at the cost of the split one.

This different type of behaviour is illustrated in figure 2 where we compare three spectra for x = 0.5 at different TIT,, with those for x = 0.9. Figure 3 gives the survey of the temperature dependence of the behaviour shown in figure 2b.

As is shown in figure 1 the critical temperature remains nearly independent on the SrS concentration for x 0.7. This does not mean, however, that the samples behave magnetically equivalent. This is demonstrated in figure 4 where the concentration dependence of the intensities of the split spectrum at

a

fixed temperature is shown.

To summarize these results we have observed a quite different temperature dependence of the magnetic hyperfine splittings in the series Eu,-,Sr,S in the EuS-rich region compared to the diluted region with x

>

0.7. To get more insight into this problem

it is best to look for a system which is even in the chemically concentrated regime magnetically weaker coupled. For unchanged magnetic moments the measure for the magnetic coupling is the ordering tempe- rature. In the work of Ehnholm et al. [5] the ordering

FIG. 3.

-

Temperature dependence of the intensities of the split and unsplit part of the lslEu Mossbauer spectra in Euo.lSro.09S. The upper limit of the temperature scale is defined by the temperature, at which the hyperfine field of the split spectrum is decreased to about 80 % of its saturation value.

O h ' a " ' * z '

0 5

x-

1.0

EuS SrS

FIG. 4.

-

Concentration dependence of the magnetically split part of the lslEu Mossbauer spectra in the Eu-poor region of

the series E ~ l - ~ s r ~ S at the k e d temperature of 650 mK.

temperature of EuS04 is given to be T, = 430 mK.

(4)

ONSET OF COLLECTIVE MAGNETIC BEHAVIOUR IN MAGNETICALLY _C6-647

Therefore we have performed a similar study using the series Eu,-,Sr,SO,. In this case as for the diluted Eu,-,Sr,S samples we were not able to detect spontaneous magnetization but only the critical temperature for vanishing hyperfine splitting in the Mossbauer spectra. The concentration dependence of these critical temperatures is shown in figure 5.

FIG. 5. - Concentration dependence of the critical temperature

Ter in the series E ~ l - ~ S r ~ S 0 4 .

The temperature dependence of the magnetic hyperfine splitting for the whole concentration range in Eu, -,SrXSO4 is the same as that described for the diluted samples in the series Eu, -,Sr,S. The behaviour corresponding to that depicted in figure 4 is shown in figure 6 .

FIG. 6.

-

Concentration dependence of the magnetically split

part of the 151Eu Mossbauer spectra in the series E u l - ~ S r ~ S 0 4 at the fixed temperature of 150 mK.

3. Discussion.

-

The interpretation of our MBss- bauer results principally suffers from the difficulty of an unique interpretation of the fact that a magnetically split Mossbauer spectrum is observed for a nuclear transition in a magnetic ion. In the paramagnetic state the necessary condition for the observation of a magnetic hyperfine splitting using the classical picture is that the Larmor precession period of the nuclear spin is short as compared to the life time of

the electronic spin state. In the magnetically ordered case which is commonly considered the quantization axis is defined by the existence of a finite magnetization which replaces the individual electronic spin state. Consequently the observation of a magnetically split Mossbauer spectrum is a necessary but no sufficient condition for the detection of magnetic order. So the situation is quite clear for the Eu rich samples of the series Eu,-,Sr,S where we observe both a net spontaneous magnetization and simultaneously one single hyperfine split Mossbauer spectrum whose temperature dependence is describable by a Brillouin function.

The situation is slightly more complicated for the diluted samples with 70

%

SrS content or more. In these samples we were not able to detect spontaneous magnetization because of experimental reasons. But we also did not find a single hyperfine pattern at least for intermediate temperatures. As shown in figure

Zb,

these spectra can be best fitted by a linear superposition of two spectra. These findings clearly support a model of superparamagnetic behaviour because relaxation of individual spins in an electronically homogeneous system only in a very poor approximation can be described by a linear superposition of two spectra. We feel that this possibility is ruled out by the accuracy of our experiments.

So if we remember our experimental findings these can be described by part of the nuclei

-

i. e. those which exhibit only a broadened line - being situated in a surrounding in which they experience rapid electronic relaxation, whereas the rest of the matrix is characterized by a mechanism which pro- duces a split Mossbauer spectrum. The only mechanism to perform this we can think of is the onset of collective magnetism in restricted areas of the sample in which the concentration of magnetic ions is increases due to statistical fluctuations. Our samples consequently are thought to consist of (( superparamagnetic ))

clusters which are separated by paramagnetic regions. Lowering of the temperature increases (( superpara-

magnetic )) clusters and vice versa. In fact, at lower

temperatures even magnetically more diluted regions may become magnetically coupled. The behaviour that we have found in our experiments and which is described in figure 3 and 4 could, qualitatively, follow from the suggested model. Chemical inhomo- geneity can be ruled out because in such a case the size of the clusters can hardly be imagined to be temperature dependent.

(5)

C6-648 H. MALETTA AND G. CRECELIUS

hyperfine splitting is describable by a Brillouin function 'with sufficient accuracy whereas on the other hand we already observe for a small temperature range the occurrence of an unsplit line.

Our model of superparamagnetic behaviour for the diluted Eu, -,Sr,S samples leads to difficulties in explaining our experimental results in pure EuSO, because in this case we are concerned with a chemically concentrated substance. The experimental results for pure EuS04 are nearly the same as for the Eu,,Sr,.,S sample. The only difference is the critical temperature to be 0.44 K in EuSO, instead of 1.9 K

in Eu,.,Sr,.,S. For such a small magnetic coupling

-

even dipole-dipole interaction may give an impor- tant contribution to the magnetic coupling in the Eu, -,Sr,SO, series - crystal imperfections may have the same effect as statistical fluctuations. On the other hand if dipole-dipole interaction would be the dominant part in the magnetic coupling, even in an ideal non-cubic lattice one might obtain spatial varying interactions.

More insight into the nature of the supposed

(( superparamagnetic )) behaviour may be obtained by

applying an external magnetic field in the temperature region where one observes the superposition of two spectra. The observation of the field dependence of the degree of polarization of the nuclear levels provides the same information as the bulk magnetization. Applying an external field we first removed the single- line-part of the spectrum and upon further increase of the external field up to 0.2 T complete polarization equivalent with a magnetically saturated sample is

achieved. This finding strongly supports our model. Details will be given in a subsequent communication. Finally we mention that our model is also supported by infrared luminescence measurements in Eu, -,Sr,S by Westerholt et al. [6]. They also proposed the

existence of spin clusters without long range magnetic order in the SrS-rich region.

4. Conclusion.

-

Our results indicate that in magnetically diluted systems a magnetically inhomo- geneous state may exist in a crystalIographicalIy homogeneous matrix. We claim to have observed a phenomenon in insulators which is typical for the onset of collective magnetism, leading to a (( superpa-

ramagnetic )) behaviour similar to that observed in

diluted metallic systems, nowadays called spinglasses.

Acknowledgments.

-

We are indebted to Dr. H. Pink from Siemens Forschungslaboratorium Miinchen for preparing the samples and to our collegue Dr. 5. Hauck for checking the Eu,~,Sr,SO, series by X-rays. We also thank Prof. Dr. W. Zinn for illuminating discussions.

Note added in proof. - The difficulties mentioned in the understanding of our findings in pure EuSO, may be overcome by the results of a study by R. B. GRIF-

mTHs, Phys. Rev. 176 (1968) 655 of the free energy of

magnetic dipoles coupled by dipole interaction only. Griffiths' results suggest a ferromagnetic alignment, but the direction of this alignment is different in different parts of the specimen. This is exactly, what we observe in EuS04.

References

[l] KONDO, J., Solid State Phys. (ed. M. Ehrenreich, F. Seitz, YOSIDA, K., Phys. Rev. 106 (1957) 893.

D. Turnbull, Acad. Press, New York) Vol. 23 (1969) 141 S A ~ , B., KOBLER, U. (private communication).

183. [5] EHNHOLM, G. J., KATILA, T. E., LOUNASMAA, 0. V., REI-

HEEGER, A. J., Solid State Phys. Val. 23 (1969) 283. VARI, P., KALVIUS, G. M., SHENOY, G . K., 2. Phys.

[2] MYDOSH, J. A., AZP Con$ Proc. 24 (1974) 131. 235 (1970) 289.

[3] RUDERMAN, M. A., K I ~ E L , C., Phys. Rev. 96 (1954) 99. [6] WESTERHOLT, K., METHFESSEL, S. (private communication).