Low-Temperature Magnetic Behavior of Perovskite Compounds

The isostructural perovskite compounds PbFe1/2Ta1/2O3 and PbFe1/2Nb1/2O3 have been known for long time, and they are part of the important class of ...
0 downloads 0 Views 88KB Size
J. Phys. Chem. B 2005, 109, 22967-22970

22967

Low-Temperature Magnetic Behavior of Perovskite Compounds PbFe1/2Ta1/2O3 and PbFe1/2Nb1/2O3 Andrea Falqui,*,† Nathascia Lampis,‡ Alessandra Geddo-Lehmann,‡ and Gabriella Pinna† Dipartimento di Scienze Chimiche and Dipartimento di Fisica, UniVersita` degli, Studi di Cagliari, S.P. Monserrato-Sestu km 0.700, I-09042 Monserrato, (Cagliari), Italy ReceiVed: September 9, 2005; In Final Form: October 9, 2005

The isostructural perovskite compounds PbFe1/2Ta1/2O3 and PbFe1/2Nb1/2O3 have been known for long time, and they are part of the important class of materials called multiferroic, where ferroelasticity, ferroelectricity, and ferromagnetism coexist. In the literature regarding PbFe1/2Ta1/2O3 and PbFe1/2Nb1/2O3, an “anomaly” of their low-temperature magnetic behavior has not always been reported. Moreover, both the origin of this behavior, and the cause for which it was not always observed, were never completely explained. In this paper, the magnetic behavior of the two compounds at low temperature has been extensively studied and explained as the occurring of a spin-glasslike transition.

Introduction PbFe1/2Ta1/2O3 (PFT) and PbFe1/2Nb1/2O3 (PFN) are compounds of the family of Pb-based complex perovskites, with the general formula AB′xB′′1-xO3, obtained from the simple perovskite ABO3 by partial substitution of the B cation. The simple cubic perovskite structure ABO3 consists of a small cation B at the center of an octahedron of oxygen anions and a large cation A at the corners of the unit cell. The ferroelectric ordering of ABO3 compounds with diamagnetic (d0) B cations is well-known; below the Curie temperature a structural transition toward a lower symmetry phase takes place, with an off-center shift of B. The spontaneous polarization arises largely because of the electric dipole moment created by this shift. When a magnetic (dn) cation is introduced in the octahedral B site, a number of novel effects may appear.1 Magnetic interactions give rise to a new class of materials, the so-called multiferroic magnetoelectrics, where ferroelectric (or antiferroelectric) and ferromagnetic (or antiferromagnetic) ordering coexist at low temperature. The physics of such materials is, then, strongly influenced by the underlying chemical distribution of B cations in the octahedral sublattice of the perovskite structure. PFT and PFN are among the first perovskites in which magnetic (d5) Fe3+ and nonmagnetic (d0) Nb5+ or Ta5+ share the B site of the simple perovskite structure. Accurate structural studies on both compounds2-5 have shown a sequence of ferroelastic and ferroelectric symmetry changes, defined as cubic (Pm3hm) f tetragonal (P4mm) f monoclinic (Cm), with Fe and Ta or Nb randomly distributed over the octahedral sites in all three phases. As they can be obtained only in the long-range octahedral disordered form, PFT and PFN are unique among the Pb2B′3+B′′5+O6 compounds, in which, usually, the ordering degree of B species can be varied during synthesis. However, the possible presence of nanoscale short-range ordered regions is still under debate for both compounds. For instance, the * Corresponding author. Tel.: 0039 070 6754379. Fax. 0039 070 6754388. E-mail: [email protected]. † Dipartimento di Scienze Chimiche. ‡ Dipartimento di Fisica.

anomaly in the heat capacity observed in PFT at about 350 K was recently discussed6,7 and interpreted as arising from the formation of B-site ordered clusters of symmetry Fm3hm (elpasolite-type), in which Fe and Ta alternate over B sites along the three crystallographic directions. These clusters undergo structural distortion giving rise to polar nanodomains, which are responsible for the relaxor-ferroelectric features of the dielectric response. Signs of B-site ordered regions both in PFN and PFT are also found in their magnetic behavior. Several magnetic studies point out the existence of an antiferromagnetic transition in the two cases, with a Ne´el temperature in the range 143-170 K for PFN8-11 and 133-180 K for PFT.12,13 An undistorted antiferromagnetic G-type cell, with lattice parameters doubled with respect to the crystallographic B-site disordered Pm3hm structure, was also observed by neutron diffraction for PFN9,14 and PFT.15 Watanabe and Kohn16 in 1988 reported the occurrence of a weak ferromagnetic moment and linear magnetoelectric (ME) effect for PFN at 9 K; Brixel et al.17 observed a new magnetic transition in PFT at T ) 9 K, which was deemed analogous to that observed in PFN. Schmid,18 commenting on these findings, interpreted the magnetic anomaly at 9 K as due to a weak “long-range superexchange” taking place through long paths of the type Fe-O-Ta(Nb)-O-Fe, caused by local shortrange order at the B site (elpasolite-type clusters) and leading to a low-temperature antiferromagnetic transition. Transition temperatures of about 10 K are, in fact, typical of antiferromagnetic phases of ordered doubled perovskites A2B′B′′O6 with one magnetic and one nonmagnetic B cation.18,19 The influence of B-site cation distribution on the magnetic properties of PFT and PFN was also studied by first-principle calculations performed on different ordered supercells of the long-range disordered Pm3hm phases,20,21 confirming that a hypothetical elpasolite-type PFT and PFN should give rise to Ne´el temperatures significantly lower than those experimentally observed in real disordered materials, which are around 160 K. In this paper an investigation about the low-temperature (below 100 K) magnetic properties of fine powders of crystalline PFT and PFN is reported, carried out with the aim of understanding the unusual magnetic behavior. All data show

10.1021/jp0551014 CCC: $30.25 © 2005 American Chemical Society Published on Web 11/04/2005

22968 J. Phys. Chem. B, Vol. 109, No. 48, 2005

Falqui et al.

that the anomaly observed in the magnetic susceptibility at 9 K has to be ascribed to a spin-glasslike behavior. Experimental Section Single crystals of PFN and PFT were grown from hightemperature PbO solutions in sealed Pt crucibles using the flux method described in refs 22 and 23. Carefully chosen crystals free from PbO inclusions were ground to powders, which were then tested for purity by conventional X-ray diffraction. Measurements of static magnetic susceptibilities, hysteretic behavior, and time-dependence of remanent magnetization were performed on the two compounds on a Quantum Design MPMS SQUID magnetometer, equipped with a superconducting magnet producing fields up to 50 kOe. Zero-field-cooled (ZFC) susceptibilities were measured by cooling the samples in a zero magnetic field and then increasing the temperature in a static field of different strength, while field-cooled (FC) curves were obtained by cooling the samples in the same static fields. The field dependence of the magnetization was measured up to 50 kOe, at T ) 4.2 K. The time dependence of remanent magnetization of the samples cooled in a field of 5 kOe was recorded immediately after the field was turned off. Results and Discussion The temperature dependence of the FC (χFC) and ZFC (χZFC) susceptibilities of PFN and PFT was measured for different field strengths (500 Oe, 2 kOe, 5 kOe, 10 kOe, and 15 kOe) in the range 4.2 e T e 100 K. Examples of the behavior observed are given in Figure 1, parts a and b, which reports chosen χFC and χZFC curves corresponding to the fields of 500 Oe, 10 kOe, and 15 kOe for both samples in the thermal range 4.2-30 K. The susceptibility is dependent on the sample thermal history: a bifurcation between χZFC and χFC is apparent at an irreversibility temperature Tirr, which is the temperature value at which the irreversible susceptibility ∆χ (defined as ∆χ ) χFC - χZFC) becomes different from zero. In the figure, the black arrows indicate the Tirr temperature for each pair of ZFC and FC susceptibility curves reported. Besides, the nature of the magnetization changes with the field, as the temperature Tmax at which a maximum in χZFC is observed at low field (500 Oe) broadens at higher fields and Tmax shifts to lower values together with the corresponding Tirr. The Tmax values are determined by fitting the ZFC susceptibility curve in the region close to the experimentally observed maximum with a polynomial function. They are indicated by red arrows in Figure 1a,b. It can be observed that in the field range from 500 Oe to 15 kOe, the FC susceptibility increases monotonically with decreasing temperature without reaching a maximum. For a higher field strength, at H ) 30 kOe, the χZFC and χFC are completely superimposed. These results already suggest a spin-glasslike magnetic behavior.24 Figure 2, parts a and b, shows that both Tirr and Tmax, determined by χZFC and χFC curves measured at field of 500 Oe, 2 kOe, 5 kOe, 10 kOe, and 15 kOe, decrease with increasing magnetic fields, as expected for a spin-glasslike magnetic behavior. Specifically, it appears that the temperature Tirr, that indicates the onset of the freezing process, decreases as H increases according to Tirr ∝ H2/3 below about 30 K for both samples. This dependence corresponds to the so-called de Almeida-Thouless (AT) line.25 It marks the starting point of the many-valley structure of phase space leading to diverging relaxation times and to nonergodic behavior. The linear extrapolation of the AT lines of Tirr to H ) 0 (shown in the figure) provides a way of determining the spin-freezing transition

Figure 1. ZFC and FC susceptibility curves obtained with a static magnetic field of 500 Oe, 10 kOe, and 15 kOe for PFN (a) and PFT (b). The black arrows indicate the values corresponding to the Tirr temperature, the red ones the values corresponding to the the Tmax temperature.

temperature Tf. The use of the AT lines is frequently reported in the literature for the determination of the transition temperature of various and different frustrated magnetic systems that show spin-glass behavior (see, for instance, refs 26 and 27). By this method, values of Tf equal to (20 ( 2) K for PFT and to (28 ( 2) K for PFN were obtained. Figure 2 also shows that the same AT behavior is followed by the values of Tmax of ZFC susceptibility curves. Figure 3, parts a and b, show the field dependence of magnetization (up to a 50 kOe magnetic field) of PFN and PFT, respectively, at the preset temperature of 4.2 K. It is apparent that not even at 4.2 K is any sign of saturation present in either sample. It can also be observed that (i) the M versus H curves approach linearity; (ii) the magnetization displays a small hysteresis only in the low-field region. In particular, coercive field Hc values of (460 ( 5) Oe for PFN and of (332 ( 5) Oe for PFT were found. All these features are typical of spinglasslike phases. To obtain an estimation of the saturation magnetization value (Ms) of the samples, the high field M values of PFT and PFN versus 1/H were plotted, similarly to that reported in ref 28. The Ms value was then extrapolated by fitting the curves with a spline function and considering the value corresponding to 1/H ) 0. Values for Ms of 3.84 emu/g for PFN and of 2.93 emu/g for PFT were obtained. Another important piece of evidence of the spin-glasslike behavior is the time dependence of the remanent magnetization

Low-Temperature Magnetic Behavior of PFT and PFN

J. Phys. Chem. B, Vol. 109, No. 48, 2005 22969

Figure 3. M(H) loops for PFN (a) and PFT (b). Low field ranges are reported in the insets.

Figure 2. Field dependence of the transition temperatures Tirr, as determined by ZFC and FC susceptibility curves, showing the De Almeida-Thouless lines for PFN (a) and PFT (b) (black curves). Field dependence of Tmax of ZFC susceptibilities (red points).

below Tf when the external magnetic field is turned off after crossing Tf (field-cooling effect). At this point, the remanent magnetization typically relaxes following a stretched exponential function of the time t,

M(t) ) M(0) exp[Rt(1-n)] where M(0) is the magnetization value at the beginning of the relaxing process. Theoretical predictions29 give 1 - n ) 0.33, which is often found in typical spin-glass compounds. PFN and PFT samples were cooled to different temperatures under a field of 5 kOe; in Figure 4, parts a and b, the time dependence of the remanent magnetization at T ) 15 K is reported. By a least-squares fitting of the curves a value of (1 - n) was obtained, which confirms the spin-glasslike behavior of the two compounds in the low-temperature regime. Below 30 K in the case of PFN sample, (1 - n) is almost constant and equal to (0.33 ( 0.01); in the case of PFT, (1 - n) reaches the value of (0.39 ( 0.03) below 22 K. Some differences between PFT and PFN are evident. First, Tf and Tmax of PFN are higher than those of PFT. The ratio between the PFN and PFT Tf values is very close to 1.4. The same value of 1.4 is found for the ratio between the coercive field values of the two samples at 4.2 K. Finally, the ratio between the extrapolated Ms values of the two samples at 4.2 K is about 1.3.

Figure 4. M(t) isothermal relaxation curves measured after FC from 220 K down to 15 K, in an external field of 5 kOe for PFN (a) and PFT (b).

All these results point out the different interaction strength that sets up among the magnetic Fe ions in the lattices of the two compounds and gives origin to the spin-glasslike behavior, with the magnetic interactions in PFN stronger than those in PFT. The results presented here confirm the essential role of

22970 J. Phys. Chem. B, Vol. 109, No. 48, 2005 the two non-magnetic Nb and Ta cations, that occupy the B-site together with the Fe atoms, in determining the physical properties of the compounds. In particular, the structural2-5 and ferroelectric3,4 transition temperatures, as well as the magnetic ones reported here, are all shifted toward higher values in the case of PFN, displaying ratios in the range of 1.4-1.6. Considering such a characteristic ratio found between different physical parameters as transition temperatures and magnetic data, one could first think it to be correlated with the electronic properties of the two compounds. However, first-principle density functional calculations on PFT and PFN,20,21 using local spin density approximation with on-site Coulomb interaction U (LSDA+U), reveal that there is not any substantial difference in the electronic properties of the two compounds. The spin-glasslike transition herewith observed in PFT and PFN is then likely to result from the structural disorder in B-site occupancy of Fe cations. Different local atomic arrangements around a central magnetic Fe ion result in a distribution of the exchange constants. A similar situation was observed in the Srbased B-site disordered perovskite SrFe1/2Nb1/2O3,30 in which, however, the low-temperature spin-glass behavior was the only observed magnetic phenomenon; in contrast, PFT and PFN undergo, in addition, a long-range antiferromagnetic transition at higher temperature, connected with the simple-cubic magnetic lattice formed by the unique octahedral site of the long-range Pm3hm symmetry, with a short exchange path (Fe/Nb)-O-(Fe/ Nb) and the analogous one for the Ta case. A competition between ferromagnetic and antiferromagnetic interactions can also play a role in the low-temperature magnetic behavior of PFT and PFN. In fact, in the presence of elpasolitetype ordered clusters, their face-centered-cubic magnetic sublattice is known to give rise to geometric magnetic frustration. Indeed, as recently reviewed,31 spin-glasslike magnetic behavior has been indicated as one of the possibilities in long-range ordered elpasolite-type double perovskites. Finally, looking at the results given above, an explanation may be advanced on why the “magnetic anomaly” of PFT and PFN at 9 K has not always been reported in the literature regarding the magnetic properties of the two compounds. In fact, the anomaly consists of the maximum which arises in the susceptibility determined after cooling in zero magnetic field. When it was experimentally found, it was often interpreted as a transition to an antiferromagnetic state. On the other hand, when the susceptibility happened to be measured after or during cooling in a nonzero field, being in such a case the FC curve monotonically increasing with decreasing temperature, no effect could be observed and no transition was then reported. Conclusions The most remarkable result of this work consists of a new study and interpretation of the low-temperature magnetic behavior of PbFe1/2Ta1/2O3 and PbFe1/2Nb1/2O3, compounds that show several similar physical features due to the same longrange disordered crystalline structure. All the results point out

Falqui et al. that the magnetic anomaly at 9 K, observed by some authors in the past, can be ascribed to a spin-glasslike transition. The ratios between the spin-freezing transition temperatures, between the low-temperature coercive fields, and between the extrapolated saturation magnetizations of PFN and PFT are all close to the same value. This fact is ascribed to the fundamental role carried out by the nonmagnetic Nb and Ta cations in influencing the interaction strength among the magnetic Fe ions that determine the observed magnetic behavior of the two compounds. References and Notes (1) Hill, N. A. J. Phys. Chem. B 2000, 104, 6694. (2) Bonny, V.; Bonin, M.; Sciau, Ph.; Schenk, K. J.; Chapuis, G. Solid State Commun. 1997, 102, 347. (3) Lampis, N.; Sciau, Ph.; Geddo-Lehmann, A. J. Phys.: Condens. Matter 1999, 11, 3489. (4) Lampis, N.; Sciau, Ph.; Geddo-Lehmann, A. J. Phys.: Condens. Matter 2000, 12, 2367. (5) Geddo-Lehmann, A.; Sciau, Ph. J. Phys.: Condens. Matter 1999, 11, 1235. (6) Gorev, M. V.; Flerov, I. N.; Sciau, Ph.; Bondarev, V. S.; GeddoLehmann, A. Ferroelectrics 2004, 307, 1. (7) Gorev, M. V.; Flerov, I. N.; Bondarev, V. S.; Sciau, Ph.; GeddoLehmann, A. Phys. Solid State 2004, 46, 521. (8) Bokov, V. A.; Myl’nikova, I. E.; Smolenskii, G. A. SoV. Phys. JETP 1962, 15, 447. (9) Pietrzak, J.; Maryanowska, A.; Leciejewicz, J. Phys. Status Solidi A 1981, 65, K79. (10) Howes, B.; Pelizzone, M.; Fischer, P.; Tabares-Munos, C.; Rivera, J. P.; Schimd, H. Ferroelectrics 1984, 54, 317. (11) Jang, H. M.; Kim, Su-C. J. Mater. Res. 1997, 12, 2117. (12) Shvorneva, L. I.; Venevtsev, Yu. N. SoV. Phys. JETP 1965, 22, 722. (13) Nomura, S.; Takabayashi, H.; Nakagawa, T. Jpn. J. Appl. Phys. 1968, 7, 600. (14) Ivanov, S. A.; Tellgren, R.; Rundlof, H.; Thomas, N. W.; Ananta, S. J. Phys.: Condens. Matter 2000, 12, 2393. (15) Ivanov, S. A.; Eriksson, S.; Thomas, N. W.; Tellgren, R.; Rundlof, H. J. Phys.: Condens. Matter 2001, 13, 25. (16) Watanabe, T.; Kohn, K. Phase Transitions 1989, 15, 57. (17) Brixel, W.; Rivera, J.-P.; Steiner, A.; Schmid, H. Ferroelectrics 1988, 79, 201. (18) Schmid, H. Ferroelectrics 1994, 162, 317. (19) Galasso, F. S. PeroVskite and High Tc Superconductors; Gordon and Breach: New York, 1990. (20) Lampis, N.; Franchini, C.; Satta, G.; Geddo-Lehmann, A.; Massidda, S. Phys. ReV. B 2004, 69, 064412. (21) Franchini, C. Private communication. (22) Brunskill, I. H.; Boutellier, R.; Depmeier, W.; Schmid, H.; Scheel, H. J. J. Cryst. Growth 1982, 56, 541. (23) Brixel, W.; Boutellier, R.; Schmid, H. J. Cryst. Growth 1987, 82, 396. (24) Spin Glasses and Random Fields; Young, A. P., Ed.; World Scientific: Singapore, 1997. (25) De Almeida, J. R. L.; Thouless, D. J. J. Phys. A 1978, 11, 983. (26) Chatterjee, S.; Nigam, A. K. Phys. ReV. B 2002, 66, 104403-1. (27) Martinez, B.; Obradors, X.; Balcells, Ll.; Rouanet, A.; Monty, C. Phys. ReV. Lett. 1998, 80, 181. (28) Ahmed, S. R.; Ogale, S. B.; Papaefthymiou, G. C.; Ramesh, R.; Kofinas, P. Appl. Phys. Lett. 2002, 80, 1616. (29) Me´zard, M.; Parisi, G.; Virasoro, M. A. Spin Glass Theory and Beyond; World Scientific: Singapore, 1987. (30) Rodriguez, R.; Fernandez, A.; Isalgue, A.; Rodriguez, J.; Labarta, A.; Tejada, J.; Obradorst, X. J. Phys. C: Solid State Phys. 1985, 18, L401. (31) Philipp, J. B.; Majewski, P.; Reisinger, D.; Geprags, S.; Opel, M.; Erb, A.; Alff, L. Acta Phys. Pol. A 2004, 105, 7.