Specific heat of GdNi1 − xCux compounds

Solid State Communications, Vol. 89, No. 4, pp. 389-392, 1994 Copyright © 1994 Elsevier Science Ltd Printed in Great Britain. All rights reserved 0038 1098/94 $6.00 + .00

Pergamon

0038-1098(93)E0016-Q SPECIFIC H E A T OF GdNil_xCux C O M P O U N D S J.A. Blanco Depto. de Fisica, Universidad de Oviedo, 33007 Oviedo, Spain J.C. Gomez Sal and J. Rodriguez Fernandez D C I T T Y M (Materiales) Facultad de Ciencias, Santander 39005, Spain M. Castro and R. Burriel ICMA (CSIC) Universidad de Zaragoza, 50009 Zaragoza, Spain and D. Gignoux and D. Schmitt Laboratoire Louis Neel, CNRS, BP 166, 38042 Grenoble Cedex 9, France

(Received 8 February 1993; in revised form 23 September 1993 by P. Burlet)

The results of specific heat measurements on the pseudobinary compounds GdNil_xCux ( x = - 0 , 0.3 and 0.6) are reported. The shape of the magnetic contribution to the specific heat and the value of the discontinuity at the ordering temperature are correlated with the type of magnetic structures involved showing significant differences between ferromagnetic x = 0 and x = 0 . 3 compounds and the helimagnetic GdNi0.4Cu0. 6. The relationship between the magnetic specific heat and the magnetic electrical resistivity is also stressed. T H E I N T E R M E T A L L I C RNil _xCux orthorhombic compounds have been widely studied in order to determine the origin of the change from a ferromagnetic behavior for x = 0 to an antiferromagnetic one, which appears when the Cu content is larger than 35 %. The magnetic and electrical properties of the GdNil_xCux compounds were recently investigated by means of magnetization, resistivity and neutron diffraction experiments [1, 2]. GdNi (CrBtype structure) and GdNi0.7Cu0.3 (FeB-type structure) present a simple collinear ferromagnetic structure, with the moments along the b direction whereas GdNi0.nCu0. 6 (FeB-type structure) orders antiferromagnetically with a propagation vector Q equal or very close to (0, 0, 1/4). Two kinds of magnetic arrangements could give account for the observed magnetic intensities. The first one is a helical structure with the magnetic moments lying in the (ab) plane. The second one is an amplitude modulated structure with the moments in a particular direction of the (ab) plane. Due to the large neutron absorption cross section

for gadolinium at usual wavelengths (1-2/~), it is necessary to work with A = 0 . 5 A (D4B diffractometer on the hot source at the ILL). This leads to spectra with low resolution which make it quite difficult to distinguish between helical or amplitude modulated structures. However, in the case of Kramers ions, such as Gd 3+, the modulated structure must evolve to an antiphase one at low temperatures giving rise to peaks associated with 3Q, 5Q higher order harmonics, which have not been observed in the GdNi0.4Cu0.6 diffraction pattern at 4.2 K, within our experimental accuracy. In particular, the first reflection of 3Q would appear at 20 = 3.9 ° with the same intensity as the (1 0 1)- located at 20 = 5.4 ° which is the first one observed. These facts strongly support the helimagnetic structure as the most likely one. The aim of the present work is to analyze the specific heat measurements of GdNil_xCux compounds and correlate these data with the type of magnetic structure of each compound. Specific heat capacity measurements were carried

389

SPECIFIC H E A T OF GdNil xCux C O M P O U N D S

390 60 GdNi /

~_......~..~_ ~

...

Z o 40

~

20

//\ /

(GdNi)tamce

The main difficulty is to properly estimate the lattice contribution of the magnetic compounds, since it cannot be obtained directly. For this purpose, we have taken as the lattice contribution of the magnetic compound that of the non-magnetic one after multiplying the temperature values by a scaling factor rscale given by [6]

oo(RmX. rscal e --

Z o

E

Vol. 89, No. 4

] GdNi C u

OD(RtmXnYp)

= Ira(MR,)3~2 + n(Mx)3/2 +p(Mv)3/2- ,/3 Lrn(MR) 3/2 + n(Mx) 3/2 + p(Mr) 3/2 '

A \\• =0.3

40

~ x

=0.6

20 (GdNiL.xcux)lallicc 0

l_,.~i~7~_~

0

20

±

- -

40 60 80 Temperature ( K )

~

100

120

Fig. 1. Total specific heat in GdNil _xCu~ (x = 0, 0.3 and 0.6), LaNi and YNi. The hatched line corresponds to renormalized LaNi and YNi data, taken as the lattice contribution for GdNi and GdNil xCux (x = 0.3 and 0.6), respectively. out at the Instituto de Ciencia de Materiales de Arag6n, on a commercial a.c. calorimeter [3] between 4.2 and 150 K. Thin slab shape samples 0.5 mm thick and around 7 mg weight were used for the measurements. The frequency used for the heating wave was 2 Hz. The relative heat capacity values obtained with the a.c. technique were scaled to absolute values with DSC measurements performed on a PerkinElmer DSC7 in the 130-150 K temperature range. In Fig. 1 the total specific heat of GdNil_xCux compounds (x = 0, 0.3 and 0.6) is shown. A welldefined A-type anomaly can be observed, leading to the ordering temperatures 7",. = 71.5 and 70.2 K and T u = 65.5K for x = 0, 0.3 and 0.6, respectively. These ordering temperatures have been taken at the inflexion point above the maximum of Cmag curve. GdNi specific heat reported by other authors [4, 5] presents similar A-type anomaly leading to the same T c. We have also measured the specific heat of the isomorphous non-magnetic compounds LaNi (CrBtype structure) and YNi (FeB-type structure). At low temperatures, the experimental variations have been fitted with a Debye law, leading to Debye temperatures 0o = 172 and 228K, for LaNi and YNi, respectively, which are quite close to those obtained from resistivity measurements [2].

where Oo(RmXnYp) and OD(R~Xn Yp) are the Debye temperatures of the magnetic and non-magnetic compounds; MR, MR,, Mx and Mr being the molar mass of R, R', X and Y atoms, respectively. LaNi has been taken as the non-magnetic reference for GdNi because they have the same CrB structure, while for GdNi0.7Cu0. 3 and GdNi0.4Cu0. 6 the reference is YNi with FeB structure• The rscaje values used were (GdNi) = 0•95 and rscale (GdNi0.TCu0.3) = rscale (Gd Nio.4Cu0.6) = 0.81. In Fig. 2, the corresponding magnetic specific heats Cmag of GdNil xCux compounds are presented. The maximum values of Cmag at the A-anomalies are 22, 24 and 15.5JKmo1-1 for x = 0 , 0.3 and 0.6, respectively. These quantities depend on how the structural lattice contributions are estimated; in particular, using the criterion explained above that contribution for the case of the FeB type compounds (x = 0.3 and 0.6) is clearly overestimated (see Fig. 1). The Cmagshapes of the x ~ 0 and 0.3 compounds are quite similar and they are also reminiscent of that of

~, x=0 GdNi

1-x

Cu

. ]

x

•

721 i

y=CI 6

•

X06EM

15 10

WE 5

0

o.5

T/Tc

1

1.5

Fig. 2. Magnetic contribution to the specific heat in GdNil xCux. The solid line corresponds to the theoretical variation for Equal Moment systems (EM).

Vol. 89, No. 4

SPECIFIC H E A T OF GdNil_xCux C O M P O U N D S

GdCu2 Si2 [7], with a total Cmag height of 18 J K m o l - t, having an equal moment simple antiferromagnetic structure. At low temperatures, a hump appears related to a Schottky-like anomaly in the ordered state involving the ( 2 J + 1)-fold degenerate multiplet [7]. The position and shape of this hump relative to Tu or To, depend on temperature through the thermal variation of the exchange field. In all the cases the humps appear to be quite similar to each other. Above Tc (henceforth the critical temperature), a noticeable magnetic entropy persists up to at least 80 K, revealing the existence of magnetic fluctuations. The magnetic entropy Smag (see Fig. 3) can be obtained by integrating Cmag/ T starting from the lowest investigated temperature. At 75 K, Smag reaches approximately the value of 14.5JKmol H for each compound, after including the missing entropy below the lowest investigated temperatures, which is estimated as about 0.5 J Kmo1-1 . This is noticeably lower than the theoretical value for lmol of Gd 3+ ions, i.e., R l n 8 = 17.3JKmol H and could be related to the overestimation of the lattice contribution commented before. However, it is worth noting the differences observed between the shapes below T~ of the magnetic entropies of x = 0 . 6 compound and the other ones (Fig. 3). This interesting fact will be considered later on in the text. In Fig. 2 we also compare the experimental Cmag variations with that expected for equal moment structures (such as ferromagnetic, simple antiferromagnetic and helimagnetic structures) within the mean field approximation. From this simple model the value of the discontinuity at the ordering temperature is 20.2 J Kmol H for J = 7/2, which is far from those obtained experimentally, specially in the x = 0.6 case (15.5 J KmolH). These differences could be attributed to the misestimation of the lattice contribution but mostly to the spin fluctuations which have not been considered in the theoretical calculations. This last feature is revealed if we 20 15

GdNil-xCUx

o~

x= 0.6~.~,~

~1 0 . . ~

X

~

/

X=0.3 x=0

0

....

0

20

• . . . . . .

40 60 80 Temperature ( K )

100

Fig. 3. Variation of the magnetic entropy of Gd Ni I _xCux compounds.

0.8

i

0.6 ~

391 I

0.6

0.4

•"~ 0.2

-0.2

0

~ I 50 100 Temperature (K)

150

Fig. 4. Temperature derivative of magnetic resistivity vs temperature for GdNil _xCux compounds. compare the x = 0.3 and 0.6 compounds. Both have the same crystallographic structure, leading to similar lattice contribution. Then the Cmagshapes must be the same in the mean field approximation framework (both have equal moments magnetic structures), however, they are quite different. This substantial observed fact could be ascribed to the existence near the critical temperature of spin fluctuations. Furthermore, it is worth noting, that while the dimension of the order parameter is one for the simple collinear ferromagnetic structure of GdNio.7Cuo. 3, the dimension is two for the helimagnetic GdNio.4Cuo. 6 case, being the mean field calculations too simplified for this situation. The importance of short range spin fluctuations in our compounds was signalled in [1], about the analysis of the magnetic electrical resistivity. The existence of a peak around Tc in the Pmag(T) curves is characteristic of such correlations [8]. It is worth mentioning that the magnetic entropy and the magnetic electrical resistivity reflect the same effects, and then Cmagand dpmag/dT correspond to the same physical phenomena. In order to illustrate this idea we show in Fig. 4 the temperature derivative of the magnetic resistivity for GdNi~_ xCux compounds. The general aspect of such curves are strongly reminiscent of that of Cmag (Fig. 2). In particular the anomaly in dpmag/dT near the critical temperature for the two ferromagnetic compounds is sharper than that of the helimagnetic one. At low temperatures in all the three compounds we observe a hump related to the way in which the 2 J + 1 degeneracy of Gd ions evolves with the ordered range, in a similar way as was commented above in the case of the specific heat. From the above mentioned features of the magnetic contributions of the specific heat and the electrical resistivity, we can ascertain that the short range spin correlations clearly dominate the behavior of these properties near the critical tempera-

392

SPECIFIC HEAT OF GdNil xCu x COMPOUNDS

ture. In particular, it is important to note that the different shapes found for the ferromagnetic GdNi and GdNi0.TCU0.3, and the helimagnetic GdNi0.4Cu0. 6 compound are mostly related to the differences in the spin fluctuations spectrum of such structures. Neutron scattering experiments on single crystals could give us more useful information about the nature of such short range correlations. Although the mean field calculations account quite well for other Gd based compounds [7], it seems to be a rough approximation in the present case due to the helimagnetic nature of the GdNi0.4Cu0. 6 structure and the existence of important spin fluctuations near T,.. Before concluding, it is interesting to mention that the close similarity between magnetic entropy and magnetic electrical resistivity and their relationship with the magnetic structures is in our opinion a general feature and has been found in other compounds as RGa2 [9] and RNi2Si 2 [10].

REFERENCES 1. 2.

3. 4. 5. 6. 7. 8. 9. 10.

Acknowledgement - - This work is supported by the Comision Interministerial de Ciencia y Tecnologia (grants MAT-90, 0877-C02-1 and MAT91-0923).

Vol. 89, No. 4

J.A. Blanco, D. Gignoux, J.C. G6mez Sal, J. Rodriguez Carvajal, J. Rodriguez Fern~indez & D, Schmitt, Physica B180-181, 100 (1992). J.A. Blanco, J.C. G6mez Sal, J. Rodriguez Fernfindez, D. Gignoux, D. Schmitt & J. Rodriguez Carvajal, J. Phys.: Condensed Matter 4, 8233 (1992). Model ACC-1VL, Sinku-Riko, Inc. Japan. K. Sato, Y. Isikawa, K. Mori & T. Miguzaki, J. Appl. Phys. 67, 5300 (1990). C.B. Zimm, W.F. Stewart, J.A. Barclay & C.K. Camperi, Advances in Cryogenic Engineering, Vol. 3, p. 741. Plenum, New York (1988). M. Bouvier, P. Lethuiller & D. Schmitt, Phys. Rev. B43, 13137 (1991). J.A. Blanco, D. Gignoux & D. Schmitt, Phys. Rev. B43, 13145 (1991). P.G. De Gennes & J.J. Friedel, Phys. Chem. Solids 4, 71 (1958). J.A. Blanco, D. Gignoux, J.C. G6mez Sal, J. Rodriguez Fern{mdez & D. Schmitt, J. Magn. Magn. Mater. 104-107, 1285 (1992). J.M. Barandiar~m, D. Gignoux, D. Schmitt, J.C. G6mez Sal, J. Rodriguez Fern~indez, P. Chieux & J. Schweizer, J. Magn. Magn. Mater. 73, 233 (1988).