Susceptibility and Mössbauer studies of magnetic rare earth-iron-silicides

Susceptibility and Mössbauer studies of magnetic rare earth-iron-silicides

H.F. Braun et al. /Properties of RE2Fe$is I 0 Gd2Fe$i5 . ErzFe3Si5 - d Dy2FeaSi5 I ’ d 0’ 0’ /' 1.2 - compounds . 119 I POWDER I .,...

435KB Sizes 2 Downloads 19 Views

H.F. Braun et al. /Properties of RE2Fe$is

I

0 Gd2Fe$i5

.

ErzFe3Si5

- d

Dy2FeaSi5

I

’ d

0’ 0’

/'

1.2 -

compounds

.

119

I

POWDER

I

.,d

. XII Xl

q

': l.Oe as i

b g

0.0 0 Ho2FesSi5

2

0.0;

I

TEMPEAATURE,K

0.0;

I

I

I

Fig. 3. Inverse molar susceptibility lines represent Jeast square tits.

I

Fig. 2. Inverse molar susceptibility of polycrystalline RE2Fe& compounds, measured with the field parallel to the growth direction.

between clearly negative temperatures around -10 K and very small values. The susceptibility is anisotropic. We denote the I

00

0

I

7

0

0

O%J&0

\ I I I

TbzFe3Si5

\

Xl

O

XII

.

Xl

Er2FesSi5

O

XII .

1 \ \ t,

*\r

I

I

I

IO

I5

Fig. 4. Low temperature

susceptibility

l

0

I

I

I

I

0 TEMPERATURE,

5

IO

15

of polycrystalline

l

l

_:?L-__~--.-

5

of TbZFe&.

300

The dashed

direction with high susceptibility as the magnetic “easy” direction, the direction perpendicular to it as magnetic “hard” (figs. 2 and 3). Below about 60 K, the inverse susceptibility deviates from the CurieWeiss law (fig. 3). Just above the ordering temperature, the slope of the inverse susceptibility is different for the two orientations (fig. 4). “Effective mo-

b

$ :

200

300

100 200 TEMPEflATURE,K

I

I

I

100

K Tb2Fe&

and Er2FeaSi5

in both orientations.

120

H.F. Braun et al. /Properties of RE2Fe$iS

Table 1 Effective magnetic moment and pammagnetic Weiss temperature (parallel crystallization direction) of REz Fe& compounds

compounds

98

100

RE2Fes Si5

98

100 cd

l-b DY

Ho Er

1.8 f

9.6 10.4 10.4 9.7

1.9 9.7 10.6 10.6 9.6

0.1 * 0.2 f 0.1 * 0.1 f 0.1

-15.4 f 1.8 f -11.4 f 0.3 f -8.0 f

0.6 3.5 0.4 0.1 0.3

98

98

9 0 _

ments” calculated in this range, assuming CurieWeiss behavior, deviate strongly from those reported in table 1.

g

v, ul 5

!: a

3.2. Miissbauer spectroscopy

E 9

The “Fe Mijssbauer spectra of all the investigated compounds taken at 300 K are shown in fig. 5. The lines are relatively broad and ln the cases of Scs Fes Sis and LusFesSis clearly show quadrupole splitting. As the Fe atoms are distributed over two types of crystallographic sites in the ratio 1 : 2 one could expect at least two slightly different values of the isomer shift and quadrupole splitting. However, best fits to all the spectra were obtained using just two lines with Lorentzian profiles. The use of two quadrupole split pairs of Lorentzian lines did not lead to any

Table 2 Isomer shift IS, quadrupole splitting QS and linewidth LW at 300 K of M2FesSis compounds a) Composition (M)

LW (mm s-l )

IS (mm s-l )

QS (mm s-l )

SC Y Sm Cd Tb DY Ho EI Lu

0.33 0.36 0.32 0.36 0.33 0.35 0.38 0.35 0.35

0.19 0.24 0.25 0.24 0.23 0.23 0.22 0.21 0.22

0.37 (3) 0.25 (6) 0.26 (9) 0.25 (3) 0.25 (2) 0.23 (8) 0.24 (3) 0.21 (8) 0.31 (1)

(7) (1) (8) (1) (3) (2) (8) (7) (9)

(3) (4) (3) (5) (8) (2) (6) (9) (2)

a) The figure in parentheses indicates the error in the last digit.

100 99 98 100 98 100 99 100

2

98

i

100 98

I

I

I

I

-3

-2

-1 RELATIVE

0

I

I

1

2

VELOCITY,

II 3

mmlsec

Fig. 5. Mijssbauer spectra of Sc2FesSis, YsFesSis and REZFesSis at room temperature.

improvement. The values of the isomer shift IS, quadrupole splitting QS and line widths LW are given in table 2. Our values of IS and QS for Sc2Fe3Sis are in good agreement with those obtained by Cashion et al. [4]. For Gd2 Fe3Sis and Tb2 Fe3 Sis which are antiferromagnetically ordered below 8.6 and 10.4 K, respectively, Miissbauer spectra of “Fe were taken at 4.2 K (fig. 6). No significant changes from the room temperature spectra were observed. There is no evidence for any magnetic splitting due to the magnetic ordering. This is consistent with the results of the susceptibility measurements which indicate that only

HF. Braun et al. [Properties of RE2Fe&

compounds

121

the applied field was smaller than the experimental error of 2%. 97 100

4. Discussion

98 96 9L 92 90 60 86

1 -3

-2

-1

0

RELATIVE

1

VELOCITY,

2 mmlsec

Fig. 6. Mijssbauer spectra of GdaFesSis and TbsFesSis at 4.2 K.

the moments of the RE-ions order antiferromagnetically. In order to try to find some evidence either for an influence of the onset of the magnetic order on the Fe atoms or for the existence of paramagnetic Fe ions below TN, a magnetic field of 50 kG was applied at 4.2 K, parallel to the direction of the gamma rays. The MGssbauer spectrum of Tba FesSis shown in fig. 7 exhibits the splitting characteristic for this geometry of the experiment. As previously reported for Scs FesSis , Dya Fes Sis and Er2 Fes Sis [4,5], the observed splitting of TbsFesSis was almost solely due to the occurrence of Zeeman levels in the applied field. The difference between the measured value of the “Fe hyperfme field and

-7

-6

-5

-L

-3

-2

RELATIVE

-1

0

1

VELOCITY,

23L567 mmlsec

Fig. 7. MGssbauer spectrum of TbsFesSis at 4.2 K in an external field of 50 kOe.

d 3

Since the REsFesSis compounds are metallic, the rare earth magnetic moments will interact via the RKKY interaction. Under the assumption that all these compounds exhibit the same type of antiferromagnetic order and that the values of k~ and the distances between RE-atoms do not change significantly with the nature of the RE, one can expect that the magnetic ordering temperatures should vary as the de Gennes factor (g - 1)2J(J t 1). However, the ordering temperatures do not agree well with this factor (fig. l), indicating that either the preceding assumption is only approximately true or the magnetic order in the REs Fe, Sis compounds may not be explained by the RKKY interaction alone. Effects of the crystalline electric field may also account for this deviation. On the other hand, the short RE-RE distances in these compounds of about 3.7 A as compared to 5-6 A in the ternary rhodium borides or Chevrel phases suggest that direct dipolar interactions may not be negligible. The isotropic portion of the dipole-dipole interaction energy can be estimated making use of a fictitious ordering temperature, 2’~) due to the dipole-dipole interaction only, according to n

where the,sum is taken over four nearest RE neighbors with p,, the effective free ion moment and r, the nearest neighbor distance. The calculated TM’s are shown in fig. 1 by a dashed line. Despite the simplification, the calculated TM’Sare of the same order of magnitude as the measured ordering temperatures. This indicates that dipolar interactions may be important in these compounds even though they alone cannot explain the observed trend in the Neel temperatures. The importance of dipolar interactions for the ferro- or antiferromagnetic ordering in other,classes of ternary compounds has recently been pointed out [7]. The effective magnetic moments in table 1 are in

122

H.F. Braun et al. /Properties of RE2Fe$iS

‘good agreement with the free ion moments of the rare earth ions, indicating no contribution from the Fe sublattice. This is confirmed by the Mossbauer results which indicate an effective moment smaller than 7 X lO-‘~.(u per Fe ion in an external field of 50 kOe at 4.2 K in the case of Gdz FesSis and TbzFe&. Below 60 K, the susceptibility of TbaFesSis is lower than expected from an extrapolation of the Curie-Weiss behavior at higher temperatures. This is consistent with the existence of a low-lying crystal field component of lower effective moment whose relative population increases with decreasing temperature, leading to an overall lower susceptibility. The very small amount of impurity phases present in this particular specimen effectively precludes their influence on the susceptibility measurements. In a tetragonal material, the susceptibility tensor has two independent components, XIIand a, parallel and perpendicular to the tetragonal axis. Thus, since the growth direction of the RE2 Fes Sis compounds corresponds to the crystallographic [OOl] direction, our measurements essentially determine the two components of the susceptibility tensor. How closely these measurements reflect the intrinsic susceptibility of the material depends on the degree of alignment of the individual crystallites and the accuracy of the demagnetization correction. For this reason, we forego a detailed analysis of the anisotropy in terms of the crystal field parameters. We point out, however, that for the compounds with Gd-Ho, the “easy” direction is parallel to the growth direction, while it is perpendicular to the growth direction for the Er compound. In this, the susceptibility follows the single ion anisotropy which changes sign between Ho and Er. Misalignment of the crystallites could also be responsible for the temperature dependence of the susceptibility in the “hard” orientation below the Nobeltemperature. According to mean field theory for a simple antiferromagnet, the susceptibility in the hard direction is temperature independent in the ordered state, while the susceptibility in the easy direction decreases to zero. Qualitatively, this is observed in the REa Fes Sis compounds (fig. 4). In the case of Tb2FeaSis a change in slope is observed for x/j at 7 K, possibly indicative of a change in the nature of ordering. In fact,,neutron diffraction on this compound [S] shows the appearance of a noncommensurate

compounds

magnetic phase at TN which is gradually replaced around 7 K by a magnetically ordered phase commensurate with the crystallographic lattice. A change in the nature of ordering has also been observed in .NdRh,B4 by specific heat [9] and neutron diffraction [lo]. The anomaly of the Nobeltemperatures at Tb (fig. 1) may be directly related with the appearance of these noncommensurate/commensurate magnetic phases. The “lower ordering temperature” of the Tb compound at about 7 K is in good accord with the TN ‘s of its neighbors in the REa Fea Sis -series, the compounds with Gd and Dy, for which no anomalies were found in the measured susceptibilities below their respective ordering temperatures. The results of the Mijssbauer study show that the Fe atoms are nonmagnetic in all the investigated compounds. There is no significant paramagnetic contribution to the “‘Fe hyperfine field below TN in the case of the antiferromagnetic REa Fe, Sis compounds. There is neither Fe 3d-band magnetism nor Fe-CurieWeiss paramagnetism. Further evidence for the nonmagnetic character of Fe is provided by the positive values of the isomer shift reaching 0.2-0.25 mm s-‘. The positive increase of the isomer shift as compared to the value of the metallic Fe means a decrease of the density of the s-electron at the Fe-nucleus. This can be due either to a decrease in the number or to an enhanced screening of the 4s electrons. Similar changes of the isomer shift were observed in the case of FeSi amorphous alloys as for instance Fe.&?+,, [ 1l] where, as in our case, Fe is nonmagnetic. The diamagnetism (or Pauli paramagnetism) of Fe in REa Fea Sis shows that the electronic structure of these compounds does not allow the intra-atomic exchange alone to create localized paramagnetic Fe moments. The onset of a kind of 3d-band magnetism as a result of Fe-Fe exchange interactions could hardly be expected in these compounds since the Fe-Fe distances are relatively large (2.64-2.67 A) and the number of Fe nearest neighbors small (2 for the 4d and 8h sites; Fe in 8h has also one next nearest Fe neighbor at 3.82 A). However, the crystallographic data lends support to strong Fe-Si interactions. The Fe-Si distances for the two types of Fe atoms in REsFeaSis are rather short, 2.28 W and 2.30-2.36 A [ 1,121 as compared to 2.40 A which represents the sum of the Goldschmidt radii of Fe and Si. Fe-Si electron

H.F. Braun et al. /Properties of RE2FesSiS compounds

hybridization is therefore likely, magnetically destabilizing the Fe atoms and thus giving rise to two identical subbands with spin up and spin down and strong d-character. Despite the fact that we have no direct evidence for it, the role of the Fe 3d electrons seems to be essential to the occurrence of superconductivity in the M2 Fe3 Sis compounds, especially in view of the existence of isotypic compounds with Ru and OS which are not superconducting [3,12]. In the Fe-compounds, the bonding of Fe with Si probably leads to the formation of d-bands without any band magnetism, i.e. with Pauli paramagnetism, and to a high density of states at the Fermi level, thereby favoring the appearance of superconductivity.

References [l] 0.1. Bodak, B.Ya. Kotur, V.I. Yarovets and E.I. Gladyshevskii, Sov. Phys. Cryst. 22 (1977) 217. [2] H.F. Braun, Phys. Lett. 75A (1980) 386.

123

[3] H.F. Braun, F. Acker and C.U. Segre, Bull. Am. Phys. Sot. 25 (1980) 232. [4] J.D. Cashion, G.K. Shenoy, D. Niarchos, P.J. Viccaro and C.M. Falco, Phys. Lett. 79A (1980) 454. [5] J.D. Cashion, G.K. Shenoy, D. Niarchos, P.J. Viccaro, A.T. Aldred and C.M. FaIco, J. Appl. Phys. 52 (1981) 2180. [6] E.I. Gladyshevskii, B.Ya. Kotur, 0.1. Bodak and V.P. Skvorchuk, Dopovidi Akad. Nauk Ukr. RSR Ser. A (1977) (8) 751. [7] M. Redi and P.W. Anderson, Proc. Natl. Acad. Sci. USA 78 (1981) 27. [ 81 A.R. Moodenbaugh, D.E. Cox and H.F. Braun, Bull. Am. Phys. Sot. 26 (1981) 468, and to be published. [9] H.C. Hamaker, L.D. Woolf, H.B. MacKay, Z. Fisk and M.B. Maple, Solid State Commun. 31 (1979) 139. [lo] C.F. Majkrzak, D.E. Cox, G. Shirane, H.A. Mook, H.C. Hamaker, H.B. MacKay, Z. Fisk and M.B. Maple, to be published. [ 111 K. Yamakawa and F.E. Fujita, J. de Phys. 40 (1979) c2-101. [ 121 H.F. Braun, in: Ter,nary Superconductors, eds. G.K. Shenoy, B.D. Dunlap and F.Y. Fradin (Elsevier, NorthHolland, Amsterdam, 1981) p. 225.