IEEE TRANSACTIONS ON MAGNETICS, VOL. 25, NO. 5, SEPTEMBER

MAGNETIC PROPERTIES OF GdBa, Cu, 0, CRYSTALS IN

M. Guillot' , J.L. Tholence** , A. Marchand*** , M. Potel**** , P. Gougeon**** , H. Noel**** and J . c . ~evet**** * Service National des Champs Intenses,

CNRS, 38042 Grenoble Cedex, France. ** Centre de Recherches sur les Tres Basses

Temperatures, CNRS, BP 166 X, 38042 Grenoble Cedex, France.

*** Laboratoire Louis Neel, CNRS, 38042 Grenoble Cedex, France.

**** Laboratoire de Chimie Minerale B, Uni- versite de Rennes, CNRS, UA 254, Avenue du General Leclerc, 45000 Rennes, France

HIGH MAGNETIC FIELD

I

---I- -- 4.2K '

10 "-.

Abstract : On single crystals of GdBa2Cu30, the magnetic properties were investigated in magnetic field of up to 20 T applied parallel to the c axis in the 1.7-300 K temperature range. In the supercon- ducting state, the field and temperature dependences of the critical current densities were deduced from the hysteresis of the half-cycle using Bean's critical state model. The Gd3+ paramagnetic moment was then studied ; above about 20 K, the M (H) isotherms were found to be given, at different temperatures, by the Brillouin function of the free Gd3+ ion. Below 20 K, the average magnetization does not obey the Brillouin law. The normal-state susceptibility was described by :he free ion Curie-Weiss law.

Introduction There has been an explosive growth of

interest in (Y)-Ba-Cu-0 systems soon after the discovery of high T, superconductivity, since it became evident that the potential of appli- cations was made available in the liquid nitrogen range. The superconducting transition temperature T, is not modified upon substitu- ting the tlmagneticlt rare earth ions to the Y ions in contrast to conventional superconduc- tors [l]. Because of the large structural anisotropy of the YBCO superconductors many intrinsic physical properties, i.e. critical fields, characteristic lengths, magnetization, etc, were observed to be highly anisotropic [ 21. Furthermore in RE Ba? Cu, 07-6 (RE : Nd, Hot Dy, Er, ...) an additionnal anisotropy originates from crystalline electric field (CEF) effects [3]: To avoid additionnal complications of arbitrarily oriented sintered ceramics, experimental studies on single crystals appear to be the best choice although the available single crystals are rather small. In this paper we present magnetization data obtained in high magnetic fields on GdBa, Cu, 0, single crystals.

GdBa, Cus 0, single crystals were grown using a mineralization method. The stoichoimetric composition of Gd,O, , BaCO, and CuO was heated for 12 hours at a temperature just lower than the peritectic decomposition 1040°

C, slowly cooled down to 900' C at a rate of 5' C/h and finally cooled slowly to ambient temperature at a rate of approximately 50" C per hour. The crystals were selected after crushing large polycrystallized blocks and were annealed during 12 hours at 450' C under oxygen flow. The single crystallity of the samples was checked using X-ray diffraction. The typical dimensions of the samples were of about 2 x 2.5 mm in the basal plane and 2 mm in the c direction and the mass of the biggest sinqle crystal was equal to 55 mg.

Emerimental

001 8-9464/89/0900-321

1989 3215

The magnetization !M) measurements were performed in the Service National des Champs Intenses in a continuous magnetic field of up to 2 0 tesla in the 4.2 - 20 K temperature range. On the other hand, M measurements were carried out in a magnetic field limited to 6.5 tesla in the 4.2 - 300 K temperature range to study especially the normal-state . magnetic susceptibility. Both experiments were using the extraction method described in [4] with the field applied along the c-axis and the experimental accuracies of M, T, H are respectively of the order of 2, 1 and 1 %. The crystal was first, in absence of any magnetic field, heated to about 130 K (above Tc) then cooled (or heated) to the desired temperature where half of the total hysteresis loop was measured after choosing a field step of 0.2 T and a field variation rate of 0.05 T/s.

Results and Discussion In the superconducting state , the

isothermal magnetization curves are shown in Fig. 1-2. Whatever the temperature, the magnetization is strongly hysteretic : this effect is due to the superconducting screening currents associated with flux pinning. Above about 10 K, the shape of the curves suggests an addition of a reversible component to this hysteretic part. It is noticeable that in polycrystalline GdBa, Cu, 0, specimens, such a behaviour was still observed at liquid helium temperature [5]. Negative values of M were observed below a field which is strongly temperature decreasing passing from 12 teslas at 20 K to less than 1 T at 8 0 K.

The first application of magnetization measurements is in terms of the Bean model [6] to allow a determination of the critical current densities which are dependent on both sample size and geometry. For a square parallepiped, the critical current density J, 55 simply given by :

30 d J, = - IM(H+) - M(H- ) I (1)

where M(H+ ) and M(H- ) represent the magneti- zations measured in increasing and decreasing field respectively and d the sample dimension (square section) in the plane perpendicular to +he field direction. J, , M(H+ ) , M(H- ) and d are expressed in A/cm2, e.m.u/cm3 and in cm respectively. Furthermore for the lowest field values, J, was taken given by - M(H- ) .

The calculated J (H) data are shown in Figs. 3 and 4. At liquid helium the critical

60 d

? 3 e m

3 0 r

-1 0 0 5 10 15 20

Ej.g.1 Magnetization versvs external magnetic field

.5$01.00@1989 IEEE

3216

current density is first a decreasing function of the field dropping rapidly at low and intermediary field (6 T) and much more slowly in higher field to be a factor of two smaller than the zero field value around 20 T. At 30 K, the J, ( H ) decrease is strong since in 6 T the initial current density is divided by a factor 3. Above 40 K a more or less pronoun- ced shoulder or a minimum is clearly observed. Close to T, (T = 80 K) the critical current density is small but not equal to zero.

It may be noted that, in zero applied field, the J, values of ErBa,Cu,O,-, was 1.13 106/cm2 at 10 K [3] : however our GdBazCu,O data obtained on a single crystal are stirb ten times larger than in polycrystalline specimens [ 5 ] . Finally it appears that the critical current density infered from the hysteresis of the M ( H ) isotherms presents zero field values that range from 1.4 lo5 at liquid helium temperature to about 4 x lo3 A/cm at liquid nitrogen temperature : in this last temperature range, applied fields of up to 7 T are not sufficient to eliminate the critical current density.

The paramagnetic magnetization was immedia- tly determined by means of the expression :

M = 0 . 5 ( M ( H + ) + M ( H - ) ) (2)

The calculated M ( H ) data per Gd3’ versus the applied field H are shown in Fig. 5 at different temperatures. The Gd3+ exchange interactions are very small since the antiferromagnetic Nee1 temperature- is about 1 K [7] : furthermore because of absence of CEF effects, the experimental data can be compared to the Brillouin function of the free Gd3+ ion :

Fig.2 Magnetic half cycle at T < Tc

I I I ‘ C 5 10 15 20

Fig.3 Critical current density JE versus external field deduced from Fig.1.

Fig.4 Critical current density Jc versus applied magnetic field.

where the Lande g-factor and the absolute saturation magnetization, M O , are equal to 2 and 7vb respectively. At 40 K, the experimen- tal curves were very well fitted by this Brillouin function where H was taken rigou- rously equal to the applied field (Fig. 5 ) , and that within our experimental accuracy on M of the order of k 4 %.

At 20 K, the quality of the fitting dete- riorates ; nevertheless if the effective field introduced in the Brillouin function is taken to be equal to 90 % of the external field only a reasonable agreement between calculated and experimental values were obtained (Fig. 5). At 4.2 K, a large discrepancy appears between the free ion calculated variation and the M ( H ) data. Such a behaviour cannot be explained by classical demagnetizing field corrections but originates, probably, from a superposition of a reversible diamagnetic (superconducting) contribution as found in YBa,Cu,07 [8] , this contribution was not substracted from M. In

3217

polycrystals, the same difference between the M(H) data and the Brillouin function was already observed and the g-factor was considered as a Ilscaling factor" (5). This procedure is not satisfying since above T, the reciprocal susceptibility is found to be given by a quasi-perfect Curie Weiss law (see next paragraph).

In the normal state (T>90 K), the magneti- zation curves were found to be linearly field dependent. The reciprocal magnetic susceptibi- lity is described by a simple Curie Weiss law where the Curie Constant is very close (+ 3 % ) from the value calculated for the free ~;r.-i,~ ysing the Hund's rules ground state

properties. A similar conclusion was previously found in polycrystalline materials [5]. The fact that the paramagnetic Curie temperature was found to be equal to zero confirms the absence of both CEF effects and the weakness of the magnetic ordering. Furthermore the absence of the Van Vleck susceptibility contribution allows to deduce that our sample does not contain a slight amount of impurity phase.

conclude some remarks concerning the J, values will be proposed. In GdBa, Cu30, , J, is much smaller than for other single crystals we have studied. This is illustrated on Fig. 6 where the remanent M, magnetization is plotted as a function of the sample perimeter perpendicular to the field, as was previously made by Sulpice et al. (ref. 9). However, and on the contrary of ref. 9, we have eliminated the MS values obtained either after a succession of magnetization jumps (it is strongly reduced in that case) or after a large jump after which the magnetization is strongly reduced (and the sample broken ! ) . Some critical current densities corresponding to a proportionnality of M, with p in the Bean model are also shown in the case of small crystals grown by the flux method. One immediately notices that the smaller crystals Y grown by the mineralisation process also lie on the 2 x 106A/cm2 , and are of comparable quality. However for the biggest samples, smaller critical current densities are often observed. Several hypotheses to explain this phenomena can be envisaged : i) As assumed in ref. 9 the currents are con- fined inside domains having intrinsic geome- trical boundaries of the order of 300 pm.

Conclusion To

, M h

Y

Y

26

Gd 055 i: perimeter

0 0.2 0.4 0.6 OBcm I I 0

Fig.6 The remanent magnetization versus the sample perimeter for our YBa,Cu,O, (=) [ 8 ] , Er Baz Cu3 0, (v) [ 3 ] , HoBa,Cu,O, (0) [ 3 ] and GdBa,Cu,O, single crys- tals ; YBazCu,07 data ( 0 + '$) es quoted in ref [ 9 1 ; weight of each sample in mg. ii) The effect of Gd on the critical current density is stronger than this of Ho or Er ions. iii) The oxygenation of large single crystals takes time, even if these large crystals have a porous structure. (J.Y. Henry [lo] mentionned that an annealing time of more than three weeks is necessary to achieve an oxygen concentration of 6.93 in the all volume of several nun3 crystals). Moreover the annealing temperatures can be different for REBa,Cu307 than for YBaz Cu30, . iv) If one assumes that the critical current density is large enough (say lo7 A/cm2 as in thin films) a complete screening of the external field (10-20 T) could be achieved in the center of large crystals. However in that case the hysteresis cycle would be quite different from the one observed at 4 . 2 K.

A poor oxygenation of the center of such large crystals (hypothesis NI) appears as the most reasonable explanation, since it affects bulk properties (high field magnetization, specific heat, sound velocity ...) without changing the llsurfacell effects (Tc , resisti- vity, diamagnetic shielding...). We have undertaken a more complete study of these oxygenation effects [ll].

References 1- K.N. Yang, Y. Dalichaouch, J.M. Ferreira, B.W. Lee, J . J . Neumeier, M.S. Torikachoili, H. Zhou, M.B. Maple and R.R. Hake. Solid State Commun. 6 3 , 515 , (1987) and references cited therein. 2- 0. Laborde, P. Monceau, M. Potel, P. Gougeon, Padiou, J.C. Levet and H. Noel. Physica C. 153 , 1467, (1988) , 3 - M. Guillot, J.L. Tholence, A . Marchand, G. Chouteau, M. Potel, P. Gougeon, H. Noel, and J.C. Levet. Presented of 2nd Int. Conf. on High Field Magnetism - Leuven Belgium (July 1988) . 4- J.C. Picoche, M. Guillot and A. Marchand 2nd Int. Conf. on High Field Magnetism Leuven Belgium (July 1988) . 5- H. Zhou, C.L. Seaman, Y. Dalichaouch, B.W. Lee, K.N. Yang, R.R. Hake, M.B. Maple, R.P. Guertin and M.V. Kuric. Physica C 152, 321 (1988) . 6 - C.P. Bean. Phys. Rev. Lett. 8 , 250 (1962) . 7- B.D. Dunlap, M. Slaski, 2 . Sungaila, D.G. Hinks, K. Zhang, C. Segre, S.K. Malik and E.E. A l p . Phys. Rev. B37, 592 (1988) . 8- M. Guillot, M. Potel, P. Gougeon, H. Noel, J . C . Levet, G. Chouteau and J.L. Tholence Phys. Lett. 127 , 363 (1988) . 9 - A . Sulpice, P. Lejay, R. Tournier and J . Chaussy. Europhys. Lett. 7, 36f; -(1988).

Fig. 5 Gd3+ ion magnetization versus applied magnetic 10- J.Y, H ~ ~ ~ ~ , private communication, field. The lines correspond to the Brillouin function 11- J.Y. H ~ ~ ~ ~ , M. J . L . Tholence, H. ~ ~ ~ 1 , B, (J=7/2, g-2) to be published.