Structural, magnetic, and transport properties of the giant magnetoresistive perovskites

,


I. INTRODUCTION
2][3][4] The basic magnetic and structural properties of R 1Ϫx A x MnO 3 (R: rare earth, A: Ca, Ba, Sr͒ were widely studied in the past. 5,6Both end members are antiferromagnetic insulators, 7 but doped compounds (0.2рxр0.4) are ferromagnetic and metallic 3 at low temperatures.This property was explained within the framework of the doubleexchange interaction by Zener 8 but recently, it has been claimed that an additional mechanism such as a Jahn-Teller distortion may be required to explain the large magnetoresistance effects. 9espite the exhaustive study of the effects of the rareearth replacement in these manganites ͑see, for instance, Refs.10-12͒, very little is known about the influence of the substitution at Mn sites with other elements.The influence of Fe substitution 13 on Sm-based compounds and the effects of Al substitution on Pr-based compounds 14 have recently been reported.We report here a thorough study on the effects of the replacement of Mn by Al in the archetypal giant magnetoresistive La 2/3 Ca 1/3 MnO 3 .The structural, magnetic, and electrical properties of this compound are well known.It undergoes a paraferromagnetic transition at T c Ϸ260 K.At this temperature an insulator-metal transition is observed.The highest magnetoresistance ratios are obtained around this temperature. 15In this compound two phenomena seem to be important: a charge-carrier localization process above T c ͑and suppressed at T c ), 16,17 whose nature is still a matter of discussion, 18 and the double-exchange interaction that causes the ferromagnetic ordering.The charge localization process above T c is characterized by an anomalous volume thermal expansion 16,17 and the existence of short-range magnetic order far above T c . 17 better understanding of the underlying physical mechanisms can be achieved by destabilizing the Mn sublattice.We have chosen Al to make the substitution because it has no magnetic moment and its atomic radius is smaller than the Mn one.Consequently, a rapid decrease in the intensity of the ferromagnetic interaction is expected and the localization process could also change.All this should be reflected in the magnetoresistance ratios.Moreover LaAlO 3 is a well-known perovskite and consequently, La 2/3 Ca 1/3 Mn 1Ϫx Al x O 3 solid solutions are expected in a wide range of concentration.

II. EXPERIMENT
La 2/3 Ca 1/3 Mn 1Ϫx Al x O 3 (xϭ0, 0.01, 0.03, 0.05, 0.1, 0.15, 0.2͒ samples were prepared by standard ceramic procedures.Stoichiometric proportions of La 2 O 3 , CaCO 3 , Al 2 O 3 , and MnCO 3 with nominal purities not less than 99.99% were mixed and heated in air at 950 °C for 12 h.After grinding they were pressed into bars and sintered in air at 1250 °C for 48 h and 1400 °C for 12 h with intermediate grindings.One selected sample (xϭ0.2͒ was also sintered in an oxygen current flow at 1000 °C and, at several temperatures ͑800-1000 °C͒ under an oxygen pressure of 200 bars.
The oxygen content was analyzed using standard redox titration with KMnO 4 and Mohr's salt.
Step-scanned powder diffraction patterns were collected at room temperature using a D-Max Rigaku system with a rotating anode.The diffractometer was operated at 100 mA and 40 kV with anode of Cu and a graphite monochromator was used to select the Cu K␣ 1,2 radiation.
Magnetic measurements were carried out between 5 and 300 K by using a commercial Quantum Design superconducting quantum interference device magnetometer.The re-sistance ͑magnetoresistance͒ measurements were performed by standard four-probe method.A superconducting coil was used to produce steady magnetic fields up to 12 T and the magnetic field was applied parallel to the current.
Neutron-diffraction measurements were performed at the high-flux reactor placed at the Institut Laue-Langevin.Two different instruments were used.For accurate refinement of the lattice parameters and the magnetic structure at low temperature, measurements were carried out on D2B, using a wavelength of 1.594 Å.For tracing the temperature dependence of the structural and magnetic properties, the instrument D1B was used.Using a wavelength of 2.52 Å, we studied the angular range 2.5°-82.5°attemperatures ranging from 1.5 to 320 K.

III. RESULTS AND DISCUSSION
X-ray powder diffraction showed clean single-phase patterns for all samples.The diffractograms can be indexed in the Pbnm spatial group, very common in perovskites orthorhombically distorted and in agreement with related compounds. 2,3Table I summarizes the relevant structural parameters obtained by Rietveld analysis of the diffractograms by using the FULLPROF program. 19The unit-cell volume and the average Mn-O distance decrease with increasing the Al content.Both results are consistent with the lower size of Al 3ϩ .However, this substitution produces minor changes in the angles of the MnO 6 octahedra, the average Mn-O-Mn angle remaining rather constant.Two effects can be expected by the partial replacement of Mn by Al in order to preserve the charge equilibrium: either an oxidation from Mn 3ϩ to Mn 4ϩ keeping the oxygen stoichiometry, or the formation of oxygen vacancies in the lattice.The first mechanism is preponderant at low Al concentrations as can be seen from the Mn 4ϩ percentage in Table I.Therefore, samples with xр0.05 are nominally stoichiometric.Nevertheless, the Mn 4ϩ percentage decreases for higher Al content and consequently, oxygen vacancies are produced for xу0.1.For in-stance, the actual formula for xϭ0.2 as measured by redox titration is La 2/3 Ca 1/3 Mn 0.8 Al 0.2 O 2.9Ϯ0.02, which is in agreement with the formula La 2/3 Ca 1/3 Mn 0.8 Al 0.2 O 2.93Ϯ0.02, obtained from Rietveld refinement of the neutron-diffraction data as can be seen below.One could think that this effect is not intrinsic but depending on the sample preparation.Several different methods of preparation have been used for the xϭ0.2 sample as has been described in the experimental section but in all of them, the value of Mn 4ϩ obtained by titration redox was the same within the experimental error even for samples annealed at different temperatures under high oxygen pressure ͑200 bars͒.
Figure 1 shows the thermal dependence of the ac magnetic susceptibility for all samples.For low Al content ͑up to xϭ0.05͒ a rapid decrease in T c is observed but the curves are rather similar to the undoped compound one.It is noteworthy that the T c is not as well defined as for the undoped compound due to a certain degree of disorder in the Mn sublattice, responsible for the ferromagnetic interaction, caused by the Al presence.The same effect of T c broadening occurs in the Al-substituted Pr manganites. 16However, this effect does FIG. 1. ac magnetic susceptibility as a function of temperature for all studied compounds.not prevent the Mn ions from being all ferromagnetically ordered at low temperature.The refinement of the lowtemperature neutron-diffraction spectra shows that at 4 K the magnetic moment per Mn ion is 3.4 B for xϭ0.05, which is close to the theoretical saturation magnetic moment of 3.6 B .This is important because it indicates that at low temperatures the samples with xр0.05 are magnetically homogeneous.The magnetic behavior of the compounds with xу0.1 is completely different.From the ac magnetic susceptibility, T c cannot be straightforwardly obtained as the curves are rather smooth and a cusp is developed at around 60 K that is more pronounced for greater Al content.From the susceptibility measurements alone, it is not possible to draw trusty conclusions and consequently, a microscopic study has been carried out by measuring neutron diffraction.
The thermodiffractograms at scattering angles between 3°and 83°and at temperatures ranging from 1.5 to 320 K are shown in Fig. 2 for the compounds with xϭ0.05 and 0.2.For xϭ0.05 the ferromagnetic ordering is clearly seen through the increase in intensity of the first Bragg nuclear peak at T c Ϸ200 K, in agreement with the ac susceptibility measurements.Another feature reminiscent of the xϭ0 compound is the observation of small-angle neutron scattering ͑SANS͒.In La 2/3 Ca 1/3 MnO 3 it was shown that as T c is approached from above, ferromagnetic clusters develop bringing about a strong contribution to the small-angle scattering. 17One can see that the same effect takes place for the xϭ0.05 compound.A clear SANS contribution appears above T c that is suppressed at T c , when the paraferromagnetic and insulator-metal transitions occur.
In Fig. 2͑b͒ the thermodiffractogram for xϭ0.2 indicates that the characteristics of a unique transition temperature are not present.The intensity of the first Bragg nuclear peak, ͑1 1 0͒, as a function of temperature is plotted in Fig. 3.A magnetic contribution appears at 200 K and rises steadily down to low temperatures.Instead of having a typical ferromagnetic dependence on temperature as it occurs in xϭ0.05, for xϭ0.2 ͑and xϭ0.15͒ the magnetic intensity behaves in a rather unusual way.The Rietveld refinement at 4 K indicates that the magnetic moment per Mn ion ͑see Table II͒ is 2.5 B for xϭ0.15 and 1.85 B for xϭ0.2, far from the saturation magnetic moment.These results are in agreement with the ac susceptibility measurements and can be explained under the assumption that the compounds are magnetically inhomogeneous.This magnetic disorder has two sources: the presence of Al in the Mn sublattice and the spontaneous appearance of oxygen vacancies.The distribution of both of them in the lattice is plausible to be random in a first approximation.Statically some regions of the sample will be more populated with them than the rest.If one thinks in the xϭ0.2 compound, with a 20% of Al impurities and 3% of oxygen vacancies, the sample will be possibly very disordered from a structural point of view.This structural disorder must lead to high magnetic inhomogeneity.The ferromagnetic double-exchange mechanism is a process consisting of a Mn-Mn indirect interaction through the oxygen orbitals.Consequently, it will be very sensitive to the presence of Al impurities and oxygen vacancies.In regions of high concentration of them, the double-exchange will be thwarted.The scenario that comes up from the neutron measurements for xу0.1 is one where there is a continuous distribution of Curie temperatures from 200 K down and probably at the lowest temperature some regions are clusterlike spin-glass rather than ferromagnetic.Otherwise, the refined magnetic moment per Mn ion would not be so far from the expected value for saturation.It should be mentioned that regions with magnetic coherent length of the order of 10 3 Å will contribute in intensity to the magnetic scattering on Bragg peaks.Consequently, regions of much smaller magnetic coherent length will not give such contribution and should be consid- ered cluster-glass-type instead of ferromagnetic.In Fig. 4 the Rietveld refinement of the xϭ0.15 compound at 1.5 K is shown.Some parameters obtained from the fit for this compound and for the other compounds are listed in Table II.
The other feature of carrier localization above T c in La 2/3 Ca 1/3 MnO 3 is the existence of an anomalous volume thermal expansion with a jump at T c . 16,17In Fig. 5 the volume thermal expansion calculated with the lattice parameters obtained from the neutron-diffraction spectra fits is shown.A volume anomaly is detected at around T c for xϭ0.05 while it is no longer visible for xу0.1 because there is not a magnetic transition at a specific temperature.Correspondingly, the characteristic SANS intensity above T c has also disappeared ͓see Fig. 2͑b͔͒.Although D1B is not a high-resolution spectrometer for lattice parameters and the absolute values should be taken with care, qualitatively one can see the anomalous volume thermal expansion for xϭ0.05.
The resistivity as a function of temperature is plotted in Fig. 6.The results nicely reflect the microscopic behavior but under no circumstances should one draw microscopic conclusions from the resistivity measurements alone.Apparently  the curves are quite similar for compounds with xр0.15, with a maximum that shifts to lower temperatures as the Al content increases.However, as we have shown with the neutron measurements, the microscopic behavior is different.For xϭ0, it is well known that at the paraferromagnetic transition temperature, T c , a metal-insulator transition occurs simultaneously.The maximum of the curve for xϭ0.05 should be interpreted in the same way because all the hallmarks of this transition are present: volume anomaly at T c and SANS above T c .In this compound the maximum is rather smooth because of the T c broadening aforementioned.For the xϭ0.1 and xϭ0.15 compounds the maximum does not correspond to a true insulator-metal transition.The maximum occurs at a lower temperature than the onset of longrange magnetic order.In these compounds where the electrical and magnetic behavior is so tight, the magnetic inhomogeneity should alter the transport properties.As the ferromagnetic state should give metallic behavior and the paramagnetic ͑or spin-glass͒ state should give insulator behavior, the presence of magnetically inhomogeneous regions must bring about the coexistence of conductive and insulating regions within the sample.The important parameters are the local magnetic coherence length and the free mean path for the electron scattering in a spin-aligned region.If the electron moves in a region of magnetic coherence length smaller than this electronic free mean path, it will be additionally scattered.The macroscopic behavior will be a balance that depends on the extent of the conductive and insulating regions.Due to this, for xϭ0.2 a maximum is no longer seen and the sample remains insulator up to 50 K, where the resistivity is so high that it saturates our experimental setup.
The magnetoresistance curves for some selected samples are shown in Fig. 7.The magnetoresistance is defined here as MR͑%͒ϭ100ϫ͉R(Hϭ12 T͒ϪR(Hϭ0)͉/R(Hϭ12 T͒.It is observed that as the Al content increases, the MR ratio increases, but at lower temperatures, which is not suitable for technological applications at room temperature.For xϭ0.2 the MR ratios are very high and the effect should be ascribed to the effect of the magnetic field on the insulating regions.It was shown in ͑La 2/3 Tb 1/3 ) 2/3 Ca 1/3 MnO 3 , 12,20 which is cluster glass at low temperatures, that the magnetic field can change the resistivity in 11 orders of magnitude.In fact, it is like an insulator-metal transition.In the metallic regions the resis-tance comes from the misalignment of spins.The effect of the magnetic field is to align them and the expected magnetoresistance is much smaller.In Fig. 8͑a͒ the resistivity for xϭ0.2 at 0 and 12 T is shown.As can be seen, the residual resistivity at 12 T remains very high for xϭ0.2 sample.This was also observed in ͑La 2/3 Tb 1/3 ) 2/3 Ca 1/3 MnO 3 , 12,20 due to the nature of the cluster-glass state.We have also measured ac magnetic susceptibility at different frequencies to check if the spin-glass regions present in the xϭ0.other related series 21 that exhibit giant magnetoresistance: when the double-exchange interaction is weakened, the T c diminishes and the magnetoresistance ratios are enhanced but at lower temperatures, which is a drawback for technological applications.However, the microscopic mechanisms in both kinds of systems are different.When the substitution takes place in the rare-earth site with a smaller ion, the Mn-O-Mn angle decreases, disfavoring the transfer integral between Mn ions and consequently weakening the doubleexchange ferromagnetic mechanism.In those compounds the disorder that occurs in the rare-earth sites is a second-order correction on the magnetic homogeneity because the magnetism is governed by the Mn and O ions positions.On the other hand, when the Mn sublattice is destabilized by substituting Mn by nonmagnetic ions like Al, the compounds result in inhomogeneous magnetic systems.In the studied La 2/3 Ca 1/3 Mn 1Ϫx Al x O 3Ϫ␦ compounds, this disorder is enhanced by the presence of oxygen vacancies and at low temperatures there is coexistence of regions in the samples with very different magnetic coherence lengths.In this series we have found that for high Al contents the magnetoresistance appears not around T c ͑which is not well defined any more͒ but at low temperatures due to the effect of the magnetic field on the insulating regions ͑those with short magnetic coherence length͒.

IV. CONCLUSIONS
By replacing Mn by Al in the giant magnetoresistive perovskite La 2/3 Ca 1/3 MnO 3 , colossal magnetoresistance ratios appear at low temperture for high Al contents (xϭ0.2).For xр0.05, the properties of the undoped compound are retained except for a broadening of the Curie temperature.For xу0.1, the compounds become magnetically inhomogeneous as a result of the structural disorder caused in the Mn sublattice and the unexpected presence of intrinsic oxygen vacancies.Neutron-diffraction measurements show that a single T c does not exist but regions of very different magnetic coherence length.This brings about the coexistence of metallic and insulating regions at low temperature, the magnetoresistance being caused mainly by the effect of the magnetic field on the insulating regions.Once more the electrical properties of the manganese perovskites have been found to be a consequence of the microscopic structural and magnetic behavior.

FIG. 4 .
FIG. 4. Rietveld refinement of the spectra at Tϭ1.5 K of the xϭ0.15 compound.The main parameters obtained from this fit are listed in TableII.The difference between measured and calculated data is plotted below.