- Email: [email protected]
Magnetic behaviour of the new intermetallic compound Ce2Pd3Ge5
ELSEVIER
Physica B 230-232 (1997) 250-252
Magnetic ordering of the heavy-fermion alloy CeCu5.1Ag0.9 O. S t o c k e r t , A. N e u b e r t , H . v . L r h n e y s e n * Physikalisches lnstitut, Universitgit Karlsruhe, D-76128 Karlsruhe, Germany
Abstract The thermodynamic and transport properties of single-crystalline CeCus.lAg0.9in magnetic fields are investigated. Measurements of the specific heat, magnetocaloric effect, DC magnetization and electrical resistivity yield a magnetic (B, T) phase diagram with two antiferromagnetic phases for magnetic fields applied parallel to the easy c-axis. The results are compared to those of CeCusAu. Keywords: CeCus.lAg0.9; Heavy electron system; Magnetic order; Thermodynamic properties
1. Introduction CeCu6 is a heavy-fermion system which does not show any long-range magnetic order down to 20 mK. However short-range antiferromagnetic and incommensurate correlated fluctuations have been observed by inelastic neutron scattering at low temperatures [1]. While intense investigations have been undertaken on alloys with Au - this leads to the pseudobinary compound CeCu6_xAux which orders antiferromagnetically for Au concentrations x > 0.1 [2] - only a few experiments on CeCu6 alloyed with Ag have been reported [3-5]. CeCu6_xAgx exhibits some type of magnetic order for x > 0.08 [3], presumably antiferromagnetic, too. The Nrel temperature, TN, increases linearly with Ag concentration up to x - 0.9 and was determined to be TN ~ 800mK for CeCus.1Ag0.9 [5]. However, a second magnetically ordered phase was detected beyond x = 0.9 by two peaks in the specific heat [4, 5]. It was suggested that phase separation is responsible for the complicated behavior above x = 0.9 [4, 5]. All these experiments were performed on polycrystals. Here we report on measurements on single-crystalline CeCus.lAg0.9 in * Corresponding author.
an attempt to overcome the two-phase problems of the polycrystals and to investigate the anisotropy of the system.
2. Experimental The CeCus.1Ag0.9 single crystal was grown by the Czochralski technique from the constituents (Ce: m4N, Cu and Ag: m5N). The orthorhombic Pnma structure of CeCu6 at room temperature remains unchanged, with increased lattice constants a and c (2.2% and 2.3%, respectively) and a slightly decreased b (-0.1%). Alloying with Ag expands the unit cell more than doping with Au, e.g. the unit cell of CeCu5.1Ago.9 is 1% larger than the unit cell of CeCusAu. In contrast to CeCusAu [6] the Ag site could not be determined. The specific heat was measured using the heat-pulse technique. The magnetocaloric effect was determined by measuring the temperature change by changing the magnetic field in the same semiadiabatic setup. The magnetization measurements were performed using a Foner-type magnetometer. The electrical resistivity was measured with the four-wire AC method. A detailed discussion of the measuring techniques can
0921-4526/97/$17.00 Copyright © 1997 Elsevier ScienceB.V. All rights reserved PII S0921-4526(96)00668-0
O. Stockert et al. /Physica B 230-232 (1997) 250-252
be found elsewhere [2, 7]. The magnetic field was applied parallel to the c-axis if not noted otherwise.
3. Results and discussion Fig. 1 shows the temperature dependence of the susceptibility Z ~ M/B in a small magnetic field B = 0.1 T applied parallel to the three crystallographic axes. Obviously, the c-direction is the magnetic easy axis (as for CeCu6) and the broad maximum of go(T) indicates the onset of antiferromagnetic order with TN ~ 800 mK. For za(T), too, a maximum is found while Zb(T) is featureless. In the field dependence of Zc = dM/dB at T = 150 mK (inset of Fig. 1) two transitions are found, one of them (at ~ 1 T), arising from a discontinuity in M, might be of first order. The temperature dependence of the specific heat in various magnetic fields is displayed in Fig. 2 where the most prominent features are the sharp peaks associated with the magnetic ordering. The hyperfine contribution (visible as an upturn toward very low T in high fields) has not been subtracted. The zero-field specific heat agrees very well with the data for a x = 0.9 polycrystal [5]. In small magnetic fields a second feature, i.e. a kink, occurs at lower temperatures indicating a transition within the ordered state. This feature is, in fact, already present as a shoulder in the B = 0 data (see also Ref. [5]). Likewise, measurements of the magnetocaloric effect at T = 300 mK (not shown) reveal two transitions. The caloric measurements hence confirm the results of the magnetization measurements. In B = 6 T where the magnetic order is completely suppressed, a fit of the specific-heat data (not shown) with the resonance-level model yields a Kondo temperature TK = 3.2 K, close to the value for CeCusAu [7] but distinctly smaller than for CeCu6 [2]. In Fig. 3 the temperature dependence of the electrical resistivity p is shown for B = 0 and B = 6 T. The current flow is parallel to the b-axis. While the zero-field resistivity shows a maximum around T = 850 mK probably corresponding to the magnetic ordering, followed by a rapid decrease to lower temperatures due to coherent scattering, the 6T results show no maximum at low temperatures. The residual resistivity P0 = 73 tx~ cm is 25 times higher than in CeCu6 suggesting a considerable degree of atomic disorder because of the nonstoichiometric Ag substitution. Even
251
0.28 "~eCu5.1Ag~ 0.26 0.24 0.22 ,,~0.12 =~0.10 0.08
o.o4~;~=~l==~=~ttt~t~.~.r.,,,.
~[
0.02~ ~ 0
1
2
3
T (K)
Fig. 1. Magnetization M of CeCusAAg0.9 divided by the magnetic field B versus temperature T for the three crystallographic directions in B = 0.1 T. The inset shows dM/dB versus B for T = 1 5 0 m K and Bllc.
4 t''''l
,
,
,
CeCus,AOo9
, illf
I
]
o 0.5 • 0.8
" .... 0.05
i
(°t>]
0 0.1
0.01
,
,
,
, ,,a,I
0.1
, 1
, 4
T (K) Fig. 2. Specific heat C of CeCus.lAg0.9 versus temperature T for B II c. With increasing B each data set is shifted downward by a factor of 0.5.
in the nominally stoichiometric compound CeCusAu we found P0 ~ 30 ~tf~cm [8]. The zero-field resistivity above TN decreases towards higher temperatures as expected for the Kondo effect and passes through a minimum at T ~ 130 K. It is not clear whether the resistivity rise above this minimum is caused by scattering from crystal-field excitations because a well-defined maximum as found for CeCusAu [8] is absent.
252
O. Stockert et al. / Physica B 230-232 (1997) 250-252
120
3
,
,
,
,
i
,
,
,
,
CeCus.IAg0.a
110
"
•
Bllc
100 0
E o
/
90
"K
0~00000
100
200
r (K)
0
0
0
0
O.
2A F O - - ' ~ ~ o
Pi
/r A
70 3a=~__ ~1 301"
A"
aA
At` a
A a
"1
~_
0
B (T) _1" oO l a 6 q . . . .
0
8
0.5
1
1.5 T (K)
2
2.5
Fig. 3. Temperature dependence of the electrical resistivity of CeCus.zAgo.9 for B IIc. The inset displays the resistivity at higher temperatures in zero field.
The low-temperature properties of CeCu5.1Ago.9 can be summarized in a magnetic (B, T) phase diagram (Fig. 4) with three different phases for B parallel to the c-axis. The paramagnetic phase is followed by two antiferromagnetic phases, the one at low temperatures and high magnetic fields probably being a spin-flop phase. In contrast to CeCusAu [7] where three magnetic phases occur, only two ordered phases in CeCus.~Ago.9 are detected. It is unknown if the different behavior is due to the nonstoichiometry of CeCus.1Ago.9. Although the Nrel temperature in CeCus.lAgo.9 is three times smaller than in CeCusAu the critical magnetic fields are of comparable magnitude. The smaller TN contrasts with the larger unitcell volume, showing that the differences between both systems are not due to a simple volume effect. Like in CeCusAu only one antiferromagnetic phase is detected for B II a. Neutron scattering experiments, which up to now have not been performed on the CeCu6_xAgx system, are underway to determine the magnetic structure of CeCus.lAgo.9.
0
•
,
, o,ko_
,
0.5
T (K) Fig. 4. Magnetic phase diagram of CeCusAAgo.9 for B II c as determined from specific heat (closed circles), magnetization (open circles), and magnetocaloric effect (triangles).
Acknowledgements We thank A. Sidorenko for his help with the hightemperature resistivity measurements. This work was supported by the Deutsche Forschungsgemeinschaft.
References [1] J. Rossat-Mignod et al., J. Magn. Magn. Mater. 76&77 (1988) 376. [2] H.G. Schlager et al., J. Low Temp. Phys. 90 (1993) 181; A. Schrrder et al., Physica B 199&200 (1994) 47. [3] A.K. Gangopadhyay et al., Phys. Rev. B 38 (1988) 2603. [4] G. Fmunberger et al., Phys. Rev. B 40 (1989) 4735. [5] A.K. Gangopadhyay et al., J. Magn. Magn. Mater• 103 (1992) 267. [6] M. Ruck et al., Acta Crystallogr. B49 (1993) 936. [7] C. Paschke et al., J. Low Temp. Phys. 97 (1994) 229. [8] C. Roth et al., unpublished results.
Physica B 230-232 (1997) 250-252
Magnetic ordering of the heavy-fermion alloy CeCu5.1Ag0.9 O. S t o c k e r t , A. N e u b e r t , H . v . L r h n e y s e n * Physikalisches lnstitut, Universitgit Karlsruhe, D-76128 Karlsruhe, Germany
Abstract The thermodynamic and transport properties of single-crystalline CeCus.lAg0.9in magnetic fields are investigated. Measurements of the specific heat, magnetocaloric effect, DC magnetization and electrical resistivity yield a magnetic (B, T) phase diagram with two antiferromagnetic phases for magnetic fields applied parallel to the easy c-axis. The results are compared to those of CeCusAu. Keywords: CeCus.lAg0.9; Heavy electron system; Magnetic order; Thermodynamic properties
1. Introduction CeCu6 is a heavy-fermion system which does not show any long-range magnetic order down to 20 mK. However short-range antiferromagnetic and incommensurate correlated fluctuations have been observed by inelastic neutron scattering at low temperatures [1]. While intense investigations have been undertaken on alloys with Au - this leads to the pseudobinary compound CeCu6_xAux which orders antiferromagnetically for Au concentrations x > 0.1 [2] - only a few experiments on CeCu6 alloyed with Ag have been reported [3-5]. CeCu6_xAgx exhibits some type of magnetic order for x > 0.08 [3], presumably antiferromagnetic, too. The Nrel temperature, TN, increases linearly with Ag concentration up to x - 0.9 and was determined to be TN ~ 800mK for CeCus.1Ag0.9 [5]. However, a second magnetically ordered phase was detected beyond x = 0.9 by two peaks in the specific heat [4, 5]. It was suggested that phase separation is responsible for the complicated behavior above x = 0.9 [4, 5]. All these experiments were performed on polycrystals. Here we report on measurements on single-crystalline CeCus.lAg0.9 in * Corresponding author.
an attempt to overcome the two-phase problems of the polycrystals and to investigate the anisotropy of the system.
2. Experimental The CeCus.1Ag0.9 single crystal was grown by the Czochralski technique from the constituents (Ce: m4N, Cu and Ag: m5N). The orthorhombic Pnma structure of CeCu6 at room temperature remains unchanged, with increased lattice constants a and c (2.2% and 2.3%, respectively) and a slightly decreased b (-0.1%). Alloying with Ag expands the unit cell more than doping with Au, e.g. the unit cell of CeCu5.1Ago.9 is 1% larger than the unit cell of CeCusAu. In contrast to CeCusAu [6] the Ag site could not be determined. The specific heat was measured using the heat-pulse technique. The magnetocaloric effect was determined by measuring the temperature change by changing the magnetic field in the same semiadiabatic setup. The magnetization measurements were performed using a Foner-type magnetometer. The electrical resistivity was measured with the four-wire AC method. A detailed discussion of the measuring techniques can
0921-4526/97/$17.00 Copyright © 1997 Elsevier ScienceB.V. All rights reserved PII S0921-4526(96)00668-0
O. Stockert et al. /Physica B 230-232 (1997) 250-252
be found elsewhere [2, 7]. The magnetic field was applied parallel to the c-axis if not noted otherwise.
3. Results and discussion Fig. 1 shows the temperature dependence of the susceptibility Z ~ M/B in a small magnetic field B = 0.1 T applied parallel to the three crystallographic axes. Obviously, the c-direction is the magnetic easy axis (as for CeCu6) and the broad maximum of go(T) indicates the onset of antiferromagnetic order with TN ~ 800 mK. For za(T), too, a maximum is found while Zb(T) is featureless. In the field dependence of Zc = dM/dB at T = 150 mK (inset of Fig. 1) two transitions are found, one of them (at ~ 1 T), arising from a discontinuity in M, might be of first order. The temperature dependence of the specific heat in various magnetic fields is displayed in Fig. 2 where the most prominent features are the sharp peaks associated with the magnetic ordering. The hyperfine contribution (visible as an upturn toward very low T in high fields) has not been subtracted. The zero-field specific heat agrees very well with the data for a x = 0.9 polycrystal [5]. In small magnetic fields a second feature, i.e. a kink, occurs at lower temperatures indicating a transition within the ordered state. This feature is, in fact, already present as a shoulder in the B = 0 data (see also Ref. [5]). Likewise, measurements of the magnetocaloric effect at T = 300 mK (not shown) reveal two transitions. The caloric measurements hence confirm the results of the magnetization measurements. In B = 6 T where the magnetic order is completely suppressed, a fit of the specific-heat data (not shown) with the resonance-level model yields a Kondo temperature TK = 3.2 K, close to the value for CeCusAu [7] but distinctly smaller than for CeCu6 [2]. In Fig. 3 the temperature dependence of the electrical resistivity p is shown for B = 0 and B = 6 T. The current flow is parallel to the b-axis. While the zero-field resistivity shows a maximum around T = 850 mK probably corresponding to the magnetic ordering, followed by a rapid decrease to lower temperatures due to coherent scattering, the 6T results show no maximum at low temperatures. The residual resistivity P0 = 73 tx~ cm is 25 times higher than in CeCu6 suggesting a considerable degree of atomic disorder because of the nonstoichiometric Ag substitution. Even
251
0.28 "~eCu5.1Ag~ 0.26 0.24 0.22 ,,~0.12 =~0.10 0.08
o.o4~;~=~l==~=~ttt~t~.~.r.,,,.
~[
0.02~ ~ 0
1
2
3
T (K)
Fig. 1. Magnetization M of CeCusAAg0.9 divided by the magnetic field B versus temperature T for the three crystallographic directions in B = 0.1 T. The inset shows dM/dB versus B for T = 1 5 0 m K and Bllc.
4 t''''l
,
,
,
CeCus,AOo9
, illf
I
]
o 0.5 • 0.8
" .... 0.05
i
(°t>]
0 0.1
0.01
,
,
,
, ,,a,I
0.1
, 1
, 4
T (K) Fig. 2. Specific heat C of CeCus.lAg0.9 versus temperature T for B II c. With increasing B each data set is shifted downward by a factor of 0.5.
in the nominally stoichiometric compound CeCusAu we found P0 ~ 30 ~tf~cm [8]. The zero-field resistivity above TN decreases towards higher temperatures as expected for the Kondo effect and passes through a minimum at T ~ 130 K. It is not clear whether the resistivity rise above this minimum is caused by scattering from crystal-field excitations because a well-defined maximum as found for CeCusAu [8] is absent.
252
O. Stockert et al. / Physica B 230-232 (1997) 250-252
120
3
,
,
,
,
i
,
,
,
,
CeCus.IAg0.a
110
"
•
Bllc
100 0
E o
/
90
"K
0~00000
100
200
r (K)
0
0
0
0
O.
2A F O - - ' ~ ~ o
Pi
/r A
70 3a=~__ ~1 301"
A"
aA
At` a
A a
"1
~_
0
B (T) _1" oO l a 6 q . . . .
0
8
0.5
1
1.5 T (K)
2
2.5
Fig. 3. Temperature dependence of the electrical resistivity of CeCus.zAgo.9 for B IIc. The inset displays the resistivity at higher temperatures in zero field.
The low-temperature properties of CeCu5.1Ago.9 can be summarized in a magnetic (B, T) phase diagram (Fig. 4) with three different phases for B parallel to the c-axis. The paramagnetic phase is followed by two antiferromagnetic phases, the one at low temperatures and high magnetic fields probably being a spin-flop phase. In contrast to CeCusAu [7] where three magnetic phases occur, only two ordered phases in CeCus.~Ago.9 are detected. It is unknown if the different behavior is due to the nonstoichiometry of CeCus.1Ago.9. Although the Nrel temperature in CeCus.lAgo.9 is three times smaller than in CeCusAu the critical magnetic fields are of comparable magnitude. The smaller TN contrasts with the larger unitcell volume, showing that the differences between both systems are not due to a simple volume effect. Like in CeCusAu only one antiferromagnetic phase is detected for B II a. Neutron scattering experiments, which up to now have not been performed on the CeCu6_xAgx system, are underway to determine the magnetic structure of CeCus.lAgo.9.
0
•
,
, o,ko_
,
0.5
T (K) Fig. 4. Magnetic phase diagram of CeCusAAgo.9 for B II c as determined from specific heat (closed circles), magnetization (open circles), and magnetocaloric effect (triangles).
Acknowledgements We thank A. Sidorenko for his help with the hightemperature resistivity measurements. This work was supported by the Deutsche Forschungsgemeinschaft.
References [1] J. Rossat-Mignod et al., J. Magn. Magn. Mater. 76&77 (1988) 376. [2] H.G. Schlager et al., J. Low Temp. Phys. 90 (1993) 181; A. Schrrder et al., Physica B 199&200 (1994) 47. [3] A.K. Gangopadhyay et al., Phys. Rev. B 38 (1988) 2603. [4] G. Fmunberger et al., Phys. Rev. B 40 (1989) 4735. [5] A.K. Gangopadhyay et al., J. Magn. Magn. Mater• 103 (1992) 267. [6] M. Ruck et al., Acta Crystallogr. B49 (1993) 936. [7] C. Paschke et al., J. Low Temp. Phys. 97 (1994) 229. [8] C. Roth et al., unpublished results.