polymer composites

polymer composites

ARTICLE IN PRESS Physica B 403 (2008) 1813–1818 www.elsevier.com/locate/physb Influence of polyparaphenylene on the magnetotransport of manganite/pol...

435KB Sizes 3 Downloads 53 Views

ARTICLE IN PRESS

Physica B 403 (2008) 1813–1818 www.elsevier.com/locate/physb

Influence of polyparaphenylene on the magnetotransport of manganite/polymer composites L.K. Gila,, E. Bacab, O. Mora´nc, C. Quinayasd, G. Bolan˜osd a

Departamento de Quı´mica, Universidad del Valle, A.A. 25360 Cali, Colombia Grupo de Ingenierı´a de Nuevos Materiales, Departamento de Fı´sica, Universidad del Valle, A.A. 25360 Cali, Colombia c Laboratorio de Materiales Cera´micos y Vı´treos, Departamento de Fı´sica, Universidad Nacional de Colombia, Sede Medellı´n, A.A. 568, Medellı´n, Colombia d Laboratorio de bajas Temperaturas, Departamento de Fı´sica, Universidad del Cauca, Popaya´n, Colombia b

Received 9 December 2006; received in revised form 2 May 2007; accepted 12 October 2007

Abstract The ferromagnetic, polycrystalline manganese oxide La0.7Sr0.3MnO3 (LSMO) and the conjugated polymer polyparaphenylene (PPP) were prepared by conventional solid-state reaction processing in air and synthesis by cationic polarization of benzene, respectively. The so-obtained compounds were individually characterized after their electrical, magnetic, structural or thermal properties by using conventional analysis methods. By mixing the pre-prepared LSMO and PPP powders, [LSMO]1x[PPP]x (x stands for the weigh fraction of PPP) two-phase manganite/polymer granular composites were obtained and their microstructural, transport and magnetic features carefully investigated. The resistance of composites calcined at 400 1C for 1 h in air atmosphere increased with x on the whole temperature range. In turn, these samples displayed a pronounced low-field magnetoresistance (LFMR) effect within a wide temperature range. Spin-polarized tunneling of conduction electrons across interfaces or grain boundaries was invoked as the dominant mechanism controlling this effect. The addition of the polymer may modify the LSMO grain surfaces adjusting the conduction carrier’s spindependent tunneling distance, which sensitively depends on the nature of the interface or on interface-related disorder. Concretely, the PPP should provoke spin disorder at the LSMO interface. r 2007 Elsevier B.V. All rights reserved. PACS: 72.80.Tm; 75.47.Lx; 7547.Gk Keywords: Composite materials; Manganites; Magnetoresistance.

1. Introduction Generally speaking, the phenomenon of colossal magnetoresistance (CMR) in the lanthanum manganite perovskite is obtained only in high magnetic fields (41 T) and within a narrow temperature region around the Curie temperature (TC) [1]. Nevertheless, most designed magnetoresistance (MR) devices operate under magnetic fields of less than several hundreds of Oe and/or within a wide temperature range [2]. Thus, reducing the magnetic field scale needed to achieve large MR in manganites and increasing the operating temperature has been the goal of a Corresponding author. Tel.: +57 2 3104097; fax: +57 2 3393237.

E-mail address: [email protected] (L.K. Gil). 0921-4526/$ - see front matter r 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.physb.2007.10.011

number of research groups. Recently, advances in the thin film deposition technique has allowed the fabrication multilayer magnetic tunnel junctions (MTJ), which displayed large low-field MR at low temperatures [3]. However, low-field MR in a wide temperature range has been observed in polycrystalline compounds, presumably due to grain boundary effects. Actually, spin-polarized intergrain tunneling of conduction electrons, taking place across either grain boundaries or barrier layers, has been claimed as the main mechanisms behind the low-field magnetoresistance (LFMR) effect in the polycrystalline specimens [4]. On the other hand, it is known that spinpolarized intergrain tunneling depends on the grain size and the properties of the intergranular material, which constructs the intergrain barrier [5]. When no nonmagnetic

ARTICLE IN PRESS 1814

L.K. Gil et al. / Physica B 403 (2008) 1813–1818

material, which may act as a potential barrier between ferromagnetic grains, exists in granular perovskites, the interface between neighboring grains should be taken into account as a barrier. Considering the amorphous character of the lattice structure in the surface along with the high sensitivity of the magnetic configuration of such materials to their structure, it is expected that the magnetic configuration in the grain surface is more chaotic than that in the core [6]. Additionally, the large number of dangling bonds or noncoordination atoms existing in the surfaces may conduce to a weaker coupling among the magnetic ions in the surfaces than that in the cores. Thus, the double-exchange interaction, used to explain the conductive behavior of ferromagnetic manganites, could be weaker in the surfaces than in the cores. In this way, the surfaces between neighboring grains, together with the intergrain distance, may play the part of the potential barrier. The barrier layer may be adjusted by altering the size of the ferromagnetic grain, introducing artificial boundaries or defects, or diluting the ferromagnetic grains with a nonmagnetic insulator [7]. These procedures will significantly influence the tunnel process, and hence enhanced MR will be expected. Indeed, enhanced MR has been reported for some ferromagnet/insulator-type two-phase manganite-based composites, such as LSMO/ CeO2, LCMO/SrTiO3, and LSMO/glass, the second phase in this case being an insulating oxide [8–10]. On the other hand, polymeric materials such as polyparaphenylene (PPP) have displayed high thermal and chemical stability as well as good crystallization [11]. Electrically, the PPP displays good insulating properties, permitting its use as a transport barrier in a manganite matrix. Theoretically, it is argued that the addition of PPP provokes magnetic disorder at the LSMO surface, which may enhance the spin-polarized intergrain tunneling of conduction electrons and hence improved MR should be expected [4]. In this work, a careful study of the structural, morphological and magnetotransport properties of [LSMO]1x[PPP]x granular composites is presented.

in air atmosphere. The pellets were then slowly cooled down to room temperature. Such thermal procedure prevents possible interfacial diffusion between the magnetic oxides, but allows a proper connection between adjacent manganite and PPP particles. The polymer stability within the annealed composites was also confirmed by FTIR measurements. The average grain size of LSMO was estimated by means of the Scherrer’ formula by measuring the full-width halfmaximum (FWHM) of XRD patterns. The micrograph was probed by scanning electronic microscope (SEM). Magnetization measurements M(T) and M(H) were carried out in a He cryostat equipped with a superconducting quantum interference device (SQUID) magnetometer (Quantum design) and a superconducting solenoid (75.5 T). Magnetotransport measurements in the temperature range 20–300 K were performed using a standard fourprobe technique in a He-closed-cycle cryostat equipped with a normal 1.5 T electromagnet. In order to provide for good electrical contact during the measuring, especially at low temperatures, four small, thin gold pads were evaporated on the surface of the samples. The MR values were calculated as [r (m0H; T)r(m0H ¼ 0 T; T)]/ r(m0H ¼ 0 T; T)]. 3. Results and discussion The FTIR spectra of the [LSMO]0.8[PPP]0.2 composite (see Fig. 1) show that the main absorption band occurs at 804 cm1. This band is characteristic of para substitution and it is attributable to the C–H out-of-plane vibrations. The other bands centered at 998 and 1476 cm1 are attributable to the C–H in-plane vibrations of a parasubstituted benzene ring and C–C skeletal in-plane vibration, respectively [11]. It has been established that the wavenumber of the C–H out-of-plane vibration band is tightly related to the degree of polymerization (n).

100

2. Experiment

99 T (%)

A standard solid-state reaction was employed to prepare the LSMO powder. Chemically pure precursor oxides were mixed with acetone in an agate mortar, grounded and finally sintered at 1000 1C in air atmosphere for 64 h. The phase of the samples was determined by X-ray diffraction measurements (XRD) using a two-circle diffractometer with Cu Ka radiation (l ¼ 1.5406 nm) for standard y/2y scans on symmetric reflections. In turn, a Kovacic method through cation polymerization of benzene was used to synthesize the PPP powder. The para configuration of the PPP was confirmed by infrared spectra (FTIR) measurements (FTIR-8400 spectrometer, Shimadso). [LSMO]1x [PPP]x composites were obtained by first mixing the preliminarily synthesized phases LSMO and PPP and then firing the 6 mm in diameter pressed pellets at 400 1C for 1 h

98

97 3000

2800

2000 18001600140012001000 800 600  (cm-1)

Fig. 1. FTIR spectra of an [LSMO]0.7[PPP]0.3 composite thermally treated to 400 1C.

ARTICLE IN PRESS L.K. Gil et al. / Physica B 403 (2008) 1813–1818

By increasing n, the band shifts to a lower wavenumber. Thus, the value of n may be estimated according to the wavenumber [11]. In the present case, the value of n for PPP was found to be between 10 and 25. Apart from this, the positions of these typical absorption bands of PPP in the composite were also observed in pure PPP, suggesting that no reaction between the two compounds took place at the chosen annealing temperature (400 1C). The XRD patterns of the [LSMO]1x[PPP]x composites with x ¼ 0, 0.1 and 0.3 are displayed in Fig. 2. A pure single perovskite phase, indexed by rhombohedral lattice symmetry, is observed on this XRD spectrum for the LSMO parent calcined at 1000 1C. The perovskite phase of LSMO is maintained for all the x weight fractions of PPP considered, which would be indicative of the coexistence of the two phases in the composite. From the XRD peaks, the average grain size (D) of the LSMO was calculated using the Scherrer’ formula [12] D ¼ 0.89l/b cos y, where l corresponds to the X-ray wavelength employed and y to the diffraction angle of the most intense peak. The parameter b is defined as b 2 ¼ b2mb2s , bm and bs being

Intensity (a.u.)

x = 0.3

x = 0.1

x=0 20

40

60

80

100

2θ [°] Fig. 2. X-ray diffraction patterns (y–2y scans) for [LSMO]1x[PPP]x composites with x ¼ 0, 0.1, and 0.3, respectively.

1815

the experimental FWHM value of the studied sample and the FWHM of a standard silicon sample, respectively. The as-calculated D for LSMO amounted to 20 nm. Direct evidence for the coexistence of the two phases in the composites is further given by a SEM micrograph, as shown in Fig. 3. The interfaces between the constituent materials within the [LSMO]0.6[PPP]0.4 composite can clearly be distinguished by brightness contrast. The dark and bright parts on the picture correspond to the PPP and LSMO, respectively [see Fig. 3(a)]. Note that the distribution of PPP powder within the ferromagnetic matrix does not display a recognizable order. Actually, the PPP crystallites form congregations, the size of which varies between hundreds of nanometers and a few microns. Thus, although the PPP particles separate the LSMO grains inhomogeneously, as shown by the micrographs, it may be assumed from the macroscopic point of few that the PPP particles are well distributed within the ferromagnetic matrices. A magnified image of an LSMO sector [see Fig. 3(b)] shows the closely piled-up LSMO crystallites. The polyhedral form of these crystallites and the generated grain boundaries are clearly distinguishable on this picture. Despite the inhomogeneous separation between the LSMO grains caused by the polymer, it may not be thin enough to act as an affective tunneling barrier. Nevertheless, this special material does modify the surfaces between neighboring ferromagnetic grains, leading to enhanced boundary effects in the composites. The temperature-dependent magnetization for composites with x ¼ 0, 0.1 and 0.3, measured in an applied magnetic field of 200 Oe, is plotted in Fig. 4. A constant temperature TC360 K was determined for the three studied samples by taking the first derivative dM/dT (see inset of Fig. 4). The invariability of TC with increasing concentration of PPP confirms the nonmagnetic character of this compound on the whole measuring temperature range, the LSMO particles being those that account for the strong magnetic response highlighted by the composites. As for the strong variation in M of the concerned composites at low temperatures, it might be ascribed to fluctuations in the surface spin state at the grain boundaries. It has been argued that in micrometric particles, the surface spins, contribution to the magnetization should be only a minor fraction of the bulk

Fig. 3. (a) SEM micrograph of an [LSMO]0.6[PPP]0.4 composite. (b) Magnified SEM image of the LSMO region.

ARTICLE IN PRESS L.K. Gil et al. / Physica B 403 (2008) 1813–1818

1816 0.6

x=0 x = 0.1

0.4 M (μB/Mn)

x = 0.3

dM/dT

0.00 0.2

-0.01 -0.02 100

0.0

μ0H = 0.02 T 0

100

200 300 T (K)

200

400

300

400

T (K)

Fig. 4. Magnetization as function of temperature of [LSMO]1x[PPP]x composites with x ¼ 0, 0.1, and 0.3 cooled in magnetic fields of 0.02 T. Inset: TC determined from the first derivative dM/dT.

4

M (μB/Mn)

2

x=0 x = 0.1 x = 0.3

0

-2 T=5K -4 -1.0

-0.5

0.5

0.0

1.0

μ0H (T) Fig. 5. Magnetic hysteresis loops of [LSMO]1x[PPP]x composites with x ¼ 0, 0.1, and 0.3 at T ¼ 5 K and m0H ¼ 1 T.

magnetization [8]. In this way, TC, as determined by the intrinsic intra-grain magnetization, should feature the same value for all the composites as it is effectively observed in Fig. 4 (inset). The very low measuring field (m0H ¼ 0.02 T) is strong enough to quantify TC, but it will in no way conduce to saturation of the bulk magnetic interactions. As the magnetic energy, in this case, is not much larger than the thermal one, the latter will play an important role in fluctuations of the chaotic magnetic configuration in the grain surface even at low temperatures. As the magnetic field strength is increased, the degree of spin disorder at the grain surface will be reduced and the magnetization will reach its bulk value (see Fig. 5). Typical magnetic hysteresis loops for [LSMO]1x[PPP]x samples with x ¼ 0, x ¼ 0.1, and x ¼ 0.3 at 5 K are depicted in Fig. 5. A sharp rise in M is observed in the low-field region (m0Ho0.5 T), which has been attributed to the spin-polarized tunneling between ferromagnetic grains [4]. As it will be discussed in the following section, the strong low-field variation in M corresponds to a strong increase in MR in the same magnetic field range. At magnetic field strengths m0HX0.5 T a pretty similar magnetic saturation value (MS) is obtained for the three considered samples, which is in good agreement with the theoretical value predicted for the bulk magnetization of LSMO. The pretty small differences in MS observed for the samples with concentrations x ¼ 0.1, x ¼ 0.2 could still reflect the weak influence of the surface magnetic disorder on the bulk magnetization at these relative high fields. As it will be discussed in the next section, the electrical resistivity measurements should sense this spin-state at the grain boundaries. In this sense, any effect modifying the surface magnetization should change Dr; particularly, the LFMR could originate from the field-suppressed magnetic disorder at the particle surface layer. The dependence of the zero-field resistivity on the temperature for various [LSMO]1x[PPP]x, composites is plotted in Fig. 6(a). An increased resistivity is observed in the whole temperature range when x varies from 0 to 0.5. Nevertheless, a conspicuous increase in this quantity takes

50 K

106 x = 0.5 x = 0.4 x = 0.3 x = 0.2 x = 0.1 x=0

104

 (Ω-cm)

 (Ω-cm)

106

102

77 K 300 K

104 102 100

100 0

100

200 T (K)

300

10-2

0.0

0.1

0.2

0.3

0.4

0.5

x (PPP)

Fig. 6. (a) Temperature dependence of the resistivity r(T) in zero field for samples with x ¼ 0, 0.1, 0.2, 0.3, 0.4, and 0.5. (b) Variation of the resistivity as a function of the PPP weigh fraction (x) of the same samples, obtained at T ¼ 300, 77, and 50 K.

ARTICLE IN PRESS L.K. Gil et al. / Physica B 403 (2008) 1813–1818

-25

50 K 77 K 300 K

-20

μ0H ~ 1 T

MR (%)

-15 -10 -5 0 0.0

0.1

0.2

0.3

0.4

0.5

x (PPP)

Fig. 7. Variation of the MR as a function of the PPP weigh fraction (x) of [LSMO]1x[PPP]x composites with x ¼ 0, 0.1, and 0.3, obtained at T ¼ 300, 77, and 50 K in a 1 T applied magnetic field.

0

x=0

77 K

x = 0.1 -5

x = 0.2 x = 0.3

MR (%)

place at x ¼ 0.1 and x ¼ 0.3 [see Fig. 6(b)]. Thus, the resistivity changes in two orders of magnitude for a variation in x from 0 to 0.1, which changes in four orders when x varies from 0.3 to 0.5. On the other hand, the ferromagnetic transition shifts to lower temperatures with increasing x as it is clearly displayed by the curves corresponding to x ¼ 0.2 and 0.3. For samples with x40.3 no ferromagnetic transition can be discerned on the r–T-curves and the composites behave rather as an insulator. Thus, the fraction x ¼ 0.3 seems to be the conduction threshold in the present case. The insulating state of PPP was confirmed by direct measurement of its electrical resistance. This value resulted to be too large and could not be quantified by the experimental set-up. In spite of the insulating regions in the composites created by the presence of PPP within the ferromagnetic matrix, the MR values for composites with x ¼ 0.1, 0.2 and 0.3 are clearly enhanced in comparison to those of the parent LSMO, even up to room temperature (see Fig. 7). The MR–x curves, recorded at various temperatures in an applied magnetic field of 1 T, indicate a maximal MR enhancement occurring at x ¼ 0.1 when Tp77 K and x ¼ 0.2 when T ¼ 300 K. These two optimal values for x are lower than the conduction threshold x ¼ 0.3. Other important aspect of the magnetoelectrical behavior of the composites has to do with the observed LFMR effect, especially at low temperatures. Magnetic field strengths m0Ho0.5 T provoke a steep rise in MR, as shown in Fig. 8 at 77 K. Such strong variation in MR at low fields corresponds to a similar variation in M in the same field range (see Fig. 5). The large negative MR at very low fields, associated with magnetic domain rotation at the grain boundaries, is a key feature observed when the negative MR of polycrystalline samples is dominated by spinpolarized tunneling between grains [4].

1817

-10

-15

-20 -1.0

-0.5

0.0

0.5

1.0

μ0H (T) Fig. 8. Field dependence of the magnetoresistance ratio MR(T) ¼ [r(m0H; T)–r(m0H ¼ 0 T; T)]/r(m0H ¼ 0 T; T) for x ¼ 0, 0.1, 0.2, and 0.3 at 77 K.

As it was previously discussed, findings such as the reduced ferromagnetic transition or the LFMR effect provide clear evidence of the determinant role played by the grain boundary effects in the magnetotransport properties of the composites. Certainly, the enhanced MR in the manganite composites has commonly been interpreted within the framework of spin-polarized tunneling between neighboring grains of manganite [8–10]. This tunneling takes place across interfaces or grains separated by an energy barrier that contains a magnetic term related to the magnetic disorder at each part of the barrier [5]. The embedment of PPP does not change the tunneling conduction model of LSMO, but influences the boundary effects. As revealed by the SEM micrographs, PPP congregations within the LSMO matrix are not thin enough for serving directly as tunneling barriers between the ferromagnetic grains. Nevertheless, this polymer gives rise to additional magnetic disorder at the LSMO surfaces. Concretely, the softening temperature of 400 1C of PPP [13] facilitates its diffusion across grain boundaries of LSMO, leading to their enhancement or more precisely to enhanced spin disorder at the surfaces of LSMO particles. This magnetic disorder promoted by PPP gives the crucial contribution to the MR enhancement. Randomly oriented moments of grains can be aligned by an external field, leading to a significant increase in the tunnel conductance, thereby reducing the resistivity of the granular system. Finally, it should be mentioned that in addition to the intergranular spin-polarized tunneling, the common mechanism of spin-dependent scattering of polarized electrons across grain boundaries, which serve as pinning centers for the magnetic domain walls, has also been suggested as origin of the LFMR [14]. Based on the data reported in this work, it is pretty difficult to discern between the two models. The main problem lies in that in both the scattering and the tunneling models, r is expected to be a maximum at the coercive field and decrease as the relative

ARTICLE IN PRESS 1818

L.K. Gil et al. / Physica B 403 (2008) 1813–1818

orientation of the magnetization between grains changes with the application of a field [15]. The MR hysteresis loops will therefore mirror the global magnetization, M, with the MR being proportional to (M/MS)2 [16]. Since both the scattering and tunneling models predict the same dependence on M, it is not possible to distinguish between the two, based on this fact alone. Thus, in order to clarify the source of MR at low fields, the current–voltage (I–V) characteristics at low temperatures must carefully be measured. For tunneling to be the dominant mechanism across the grain boundaries, strong non-ohmic I–V curves at high voltages should be expected. Although great numbers of experimental data are confusing, recent reports seem to confirm the spin-polarized tunneling hypothesis [17,18]. 4. Summary and conclusions The structural, morphological and magnetotransport properties of La0.7Sr0.3MnO3/polymer composites were carefully studied. The short thermal treatment at 400 1C, used to improve the connection between adjacent LSMO and PPP particles, has no appreciable effect on the stability of PPP. Thus, the magnetoelectrical characterization of the composites showed significantly enhanced MR even up to room temperature. The strength of the MR enhancement depends critically on the fraction of PPP added in the ferromagnetic matrix. The observed LFMR, related to the alignment of the magnetic moments of adjacent grains, strongly suggest that the negative MR of the polycrystalline samples is dominated by spin-polarized tunneling. This effect becomes increasingly important at low temperatures. The magnetic disorder provoked by the addition of PPP on the surfaces between neighboring grains may be responsible for the enhanced MR as compared to that of a pure manganite. Acknowledgments The authors wish to thank D. Ernst, D. Fuchs, and P. Adelmann, at the Research Center Karlsruhe (Federal

Republic of Germany) for their assistance with XRD- and SQUID-measurements. The authors wish also to thank F. Zuluaga (Laboratory for synthesis and mechanisms of reactions in organic chemistry at the Universidad del Valle) for valuable collaboration. One of the authors (O.M.) acknowledges the financial support of the German Service of Academic Interchange (DAAD). References [1] S. Jin, T.H. Tiefel, M. McCormack, R.A. Fastnacht, R. Ramesh, L.H. Chen, Science 264 (1994) 413. [2] G.C. Xiong, Q. Li, H.L. Ju, S.N. Mao, L. Senapati, X.X. Xi, R.L. Greene, T. Venkatesan, Appl. Phys. Lett. 66 (1995) 1427. [3] Y. Lu, X.W. Li, G.Q. Gong, G. Xiao, A. Gupta, L. Lecouer, J.Z. Sun, Y.Y. Wang, V.P. Dravid, Phys. Rev. B 54 (1996) R8357. [4] H.Y. Huang, S.W. Cheong, N.P. Ong, B. Batlog, Phys. Rev. Lett. 77 (1996) 2041. [5] N. Zhang, W. Ding, W. Zhong, D. Xing, Y. Du, Phys. Rev. B 56 (1997) 8138. [6] K. Sattler, J. Mu¨hlbach, E. Recknagel, Phys. Rev. Lett. 45 (1980) 821. [7] Y.-H. Huang, X. Chen, Z.-M. Wang, C.-S. Sheng Liao, C.-H. Yan, H.-W. Zhao, B.-G. Shen, J. Appl. Phys. 91 (2002) 7733. [8] L.I. Balcells, A.E. Carrillo, B. Martinez, J. Fontcuberta, Appl. Phys. Lett. 74 (1999) 4014; B. Martinez, L.l. Balcells, J. Fontcuberta, X. Obradors, C.H. Cohenca, R.F. Jardim, J. Appl. Phys. 83 (1998) 7058. [9] D.K. Petrov, L. Krusin-Elbaum, J.Z. Sun, C. Field, P.R. Duncombe, Appl. Phys. Lett. 75 (1999) 995. [10] S. Gupta, R. Ranjit, C. Mitra, P. Raychaudhuri, R. Pinto, Appl. Phys. Lett. 78 (2001) 362. [11] P. Kovacik, J. Oziomek, J. Org. Chem. 29 (1964) 100. [12] M.I. Mendelson, J. Am. Ceram. Soc. 52 (1969) 443. [13] C.-H. Yan, Y.-H. Huang, X. Chen, C.-S. Liao, Z.-M. Wang, J. Phys.: Condens. Matter 14 (2002) 9607. [14] A. Gupta, G.Q. Gong, Gang Xiao, P.R. Duncombe, P. Lecouer, P. Troilloud, Y.Y. Wang, V.P. Dravid, J.Z. Sun, Phys. Rev. B 54 (1996) 15629. [15] X.W. Li, A. Gupta, G. Xiao, G.Q. Gong, Appl. Phys. Lett. 71 (1997) 1124. [16] J.Q. Xiao, J.S. Jiang, C.L. Chien, Phys. Rev. Lett. 68 (1992) 3749. [17] L.l. Balcells, J. Fontcuberta, B. Martinez, X. Obradors, Phys. Rev. B 58 (1998) 14697. [18] P. Raychaudhuri, K. Sheshadri, P. Taneja, S. Bandyopadhyay, P. Ayyub, A.K. Nigam, R. Pinto, S. Chaudhary, S.B. Roy, Phys. Rev. B 59 (1999) 13919.