High-resolution photoemission spectroscopy for the layered antiferromagnetic (La_1−zNd_z)_0.46Sr_0.54MnO₃†

High-resolution HeI photoemission spectroscopy (UPS), Mn 2p–3d resonant photoemission spectroscopy (RPES) and Mn 2p X-ray absorption spectroscopy (XAS) have been performed to investigate the electronic structure and its effect on the electrical resistivity in (La_1−zNd_z)_0.46Sr_0.54MnO₃ (z = 0, 0,2, 0.6 and 1.0). It was found that in the UPS and RPES spectra the Fermi edge persisted over the temperature range 15 ≤ T ≤ 340 K regardless of the magnetic structure or the composition of the samples. The density of states at the Fermi level [N(E_F)] in the samples where 0 ≤ z ≤ 0.6 was increased drastically at the Curie temperature (T_C) with decreasing temperature, but essentially kept unchanged across the Néel temperature (T_N). A fairly large reduction at T_C and a small increase at T_N in the electrical resistivity with decreasing temperature are found to be well accounted for in terms of the temperature dependence of N(E_F). The presence of a finite N(E_F) in the insulating Nd_0.46Sr_0.54MnO₃ was also found. Thus the origin of the insulating behavior in this sample can be regarded as the Anderson localization associated with the small density of states and the chemical disorder between Nd and Sr.

1. Introduction

The cubic perovskite manganites are known to be characterized by the variety of peculiar electron transport properties, such as the colossal magnetoresistance (Jin et al., 1994) associated with the magnetic-field-induced metal–insulator (MI) transition. Mn atoms in the perovskite structure provide not only itinerant electrons but also localized magnetic moments. The influence of the localized magnetic moments on the itinerant electrons is so strong that the electron transport properties of the perovskite manganites vary drastically depending on the variation in arrangement of the localized magnetic moments; the metallic conduction is in general observed for both the ferromagnetic phase and so-called A-type layered antiferromagnetic phase, while insulating behaviors were reported for the paramagnetic and other antiferromagnetic phases (Tokura et al., 1994; Urushibara et al., 1995; Akimoto et al., 1998; Moritomo et al., 1998).

The ferromagnetic phase with metallic behavior transforms into the insulating phase accompanied by the magnetic phase transition with varying temperature or external magnetic field. This peculiar MI transition had been believed to be explained in terms of the double-exchange (DE) interaction (Anderson & Hasegawa, 1955; de Gennes, 1960). Recently, however, it was pointed out that the DE interaction alone cannot fully account for the phase transition between the ferromagnetic metal and the paramagnetic insulator (Millis et al., 1995, 1996; Saitoh et al., 2000). In the framework of the DE theory, the relation between the metallic conductivity and the magnetic ordering is explained as follows. The hopping probability of the itinerant electrons from one site to another is formulated by t = t₀ cos(θ/2), where t₀ and θ indicate the bare hopping probability and the angle between two localized magnetic moments in the nearest-neighbor atomic sites, respectively. If these localized magnetic moments are ferromagnetically ordered, the hopping probability shows the maximum value with θ = 0°, while it is fairly reduced in the array of disordered magnetic moments and reaches zero in the array of antiferromagnetically ordered magnetic moments with θ = 180°. The reduced hopping probability leads to a reduction in the Fermi velocity. However, no increase in the electrical resistivity can be caused by it as long as a single band is concerned, because a simultaneously introduced increase in the density of states near E_F compensates the increase of the resistivity due to the reduced Fermi velocity. Thus, within the DE model, the resistivity variation cannot appear except when the phase transition between the antiferromagnetic phase and other phases occurs.

The A-type layered antiferromagnetic phase has an ordered magnetic structure in which localized magnetic moments are ferromagnetically ordered in planes and those planes stack together antiferromagnetically. Akimoto et al. (1998) reported the temperature dependence of the electrical resistivity for La_0.46Sr_0.54MnO₃, that possesses the paramagnetic phase above T_C = 300 K, the ferromagnetic phase at T_N ≤ T ≤ T_C and the A-type antiferromagnetic phase below T_N = 200 K. Here, T_C and T_N indicate the Curie and Néel temperatures, respectively. In their report the electrical resistivity of this sample was decreased by about 40–50% of its magnitude at T_C and slightly increased at T_N with decreasing temperature. Obviously, these behaviors cannot be understood in terms of the DE interaction. Thus in order to investigate the factors causing these behaviors in the electrical resistivity, the determination of the electronic structure of the paramagnetic state, the ferromagnetic state and the A-type antiferromagnetic state is strongly required. It should be noted here that although many experiments with photoemission spectroscopy have already been reported for the ferromagnetic and paramagnetic phases (Saitoh et al., 1995, 1997; Park et al., 1996; Sarma et al., 1996; Chainani et al., 1993, 1997; Sekiyama et al., 1999), that for the A-type antiferromagnetic phase has so far not been reported.

In this paper we report the results of high-energy-resolution photoemission spectroscopy measurements for the (La_1−zNd_z)_0.46Sr_0.54MnO₃ (z = 0.0, 0.2, 0.6 and 1.0) cubic perovskite manganites. These samples, except for z = 1.0, possess two different magnetic transitions with varying temperature; the paramagnetic phase exists above T_C = 230–300 K, the ferromagnetic phase at T_N ≤ T ≤ T_C and the A-type antiferromagnetic phase below T_N = 200–230 K (Akimoto et al., 1998). Only Nd_0.46Sr_0.54MnO₃ possesses a direct transition from the paramagnetic phase into the A-type antiferromagnetic phase at T_N = 225 K without being in the ferromagnetic phase. The electronic structure of the A-type antiferromagnetic phase was investigated in comparison with that of the paramagnetic and ferromagnetic phases. The relation between the measured density of states at the Fermi level [N(E_F)] and the electrical resistivity in each compound will be discussed in detail.

2. Experimental

High-quality (La_1−zNd_z)_0.46Sr_0.54MnO₃ (z = 0.0, 0.2, 0.6 and 1.0) single crystals were grown by the floating-zone method. Details of the sample preparation have been reported elsewhere (Akimoto et al., 1998; Moritomo et al., 1998). The bulk-sensitive Mn 2p–3d resonant photoemission spectroscopy (RPES) measurements (Sekiyama et al., 2000) with incident photon energies of hυ = 639–641 eV were performed at beamline BL25SU at the third-generation 8 GeV synchrotron radiation facility SPring-8, Hyogo, Japan. The energy resolution of the measurements was about 80–120 meV, which was determined as the energy width of the intensity drop from 90% to 10% at the Fermi edge of the reference gold electrically contacting with the samples. The HeI photoemission spectroscopy (UPS) measurements with an energy resolution of 50 meV were also performed for (La_1−zNd_z)_0.46Sr_0.54MnO₃ (z = 0.0, 0.2, 0.6 and 1.0). All UPS and RPES measurements were carried out at temperatures between 15 K and 340 K. The sample surfaces were scraped with a diamond file just before the data accumulation at each temperature under a base pressure better than 10–11 torr. We intentionally finished each measurement within 60 min after the surface scraping to avoid the serious surface degradation reported by Saitoh et al. (1997). Mn 2p soft X-ray absorption spectroscopy (XAS) measurements were performed at BL19B at the Photon Factory, KEK, Japan. We used a bulk-sensitive (∼2000 Å in depth) fluorescence yield method for the present Mn 2p XAS measurement.

3. Results and discussions

The electronic structure of these cubic perovskite manganites has been frequently studied by means of photoemission spectroscopy measurements (Saitoh et al., 1995, 1997; Park et al., 1996; Sarma et al., 1996; Chainani et al., 1993, 1997; Sekiyama et al., 1999). The density of states near E_F in these reports was fairly small. Sometimes it was reported to be finite in the ferromagnetic metallic phases and to become essentially zero in the insulating phases. The small intensity near E_F prevented us from evaluating the gap width and the density of states at E_F. In order to overcome this difficulty, we employed bulk-sensitive and high-energy-resolution Mn 2p–3d RPES measurements, by which the density of states associated with the Mn 3d electron is so enhanced that the Fermi edges are expected to be clearly observed in the metallic states.

Fig. 1(a) shows photoemission spectra of La_0.46Sr_0.54MnO₃ taken with incident photon energies across the Mn 2p absorption edge. All spectra are normalized by the photon flux. It can be confirmed that the intensity of the valence-band spectra increases by more than ten times when the Mn 2p–3d resonance takes place. The enhanced intensity due to the Mn 2p–3d resonance allowed us to easily observe the density of states near E_F. We employed hυ = 640 eV as the incident photon energy for the RPES measurements to avoid the effect of the undesirable auger process. Even with this incident energy the spectrum intensity is enhanced by more than five times than that of the off-resonant spectrum measured with hυ = 637 eV. The high-resolution Mn 2p RPES spectra near E_F for (La_0.8Nd_0.2)_0.46­Sr_0.54MnO₃ in the paramagnetic phase at 300 K, the ferromagnetic phase at 250 K and the A-type layered antiferromagnetic phase at 150 K are shown in Fig. 1(b). The clearly observed intensity at E_F in all spectra indicates that (La_0.8Nd_0.2)_0.46Sr_0.54MnO₃ must be metallic regardless of the magnetic status involved. It should be noted that the spectrum intensity at E_F of the ferromagnetic phase and the A-type antiferromagnetic phase obviously shows a much higher value than in the paramagnetic phase. The electrical resistivity of (La_1−zNd_z)_0.46Sr_0.54MnO₃ (0.0 ≤ z ≤ 0.4) reported by Akimoto et al. (1998) decreases by about 50% in its magnitude when the paramagnetic phase transforms into the ferromagnetic phase at T_C but it is almost kept constant across T_N. This variation in the electrical resistivity must be caused by the variation in N(E_F) observed in the present RPES measurement.

Figure 1 (a) Photoemission spectra taken with incident photon energies across the Mn 2p absorption edge. Mn 2p–3d resonance taking place at hυ ≥ 639 eV allows us to clearly observe the intensity near EF. (b) Mn 2p–3d RPES spectra of (La0.8Nd0.2)0.46Sr0.54MnO3 perovskite manganite in the paramagnetic state at 300 K, the ferromagnetic state at 250 K and the A-type layered antiferromagnetic phase at 150 K measured with hυ = 640 eV. The vertical dashed line indicates the Fermi level (EF) determined by the reference gold. The density of states at EF is finite in all the spectra.

In order to gain further insight into the electronic structure near E_F we employed the HeI UPS measurement with a higher energy resolution of 50 meV. The observed spectra for (La_1−zNd_z)_0.46Sr_0.54MnO₃ (z = 0.0, 0.2, 0.6 and 1.0) at different temperatures are shown in Figs. 2(a)–2(d). Notably, all UPS spectra possess a steep Fermi edge, which shows much clearer structure than that reported for the ferromagnetic phase in the cubic perovskite manganite (Sarma et al., 1996, 1997; Sekiyama et al., 1999). The easiness in observing the Fermi edge with UPS measurements for the present samples most likely originates from the large O 2p component along with Mn 3d in the valence band near E_F.

Figure 2 Temperature dependence of the high-resolution HeI UPS spectra near the Fermi level for (La1−zNdz)0.46Sr0.54MnO3 for (a) z = 0, (b) z = 0.2, (c) z = 0.6 and (d) z = 1.0. The density of states at EF is finite in all the spectra. The vertical dashed line indicates EF determined by the reference gold. The electrical resistivity of (La1−zNdz)0.46Sr0.54MnO3 (z = 0.2, 0.4, 0.8 and 1.0) reported by Akimoto et al. (1998) is shown in (e).

The UPS spectra of (La_1−zNd_z)_0.46Sr_0.54MnO₃ with z = 0.0, 0.2 and 0.6 possess a fairly large enhancement in intensity near E_F at T_C and less obvious variation across T_N with decreasing temperature. The same tendency has already been mentioned above in the RPES measurements for z = 0.2. Therefore it is certain that N(E_F) associated with both the Mn 3d and O 2p electrons increases when the phase transition from the paramagnetic phase to the ferromagnetic phase takes place, and that this variation in N(E_F) leads to the reduction in the magnitude of the electrical resistivity.

When the electrical resistivity of metallic phases is discussed in terms of the Boltzmann transport equation, we should consider not only the electronic structure but also the scattering event, that is closely related to the ordering of the local magnetic moments. It is very important to know that a reduction in the electrical resistivity of about 10% in its magnitude occurs at T_N with decreasing temperature for Nd_0.46Sr_0.54MnO₃ while its UPS spectrum was almost kept unchanged across T_N. The reduction in the electrical resistivity with almost constant N(E_F) must be attributed to a reduction in the scattering probability due to the ordering of the local magnetic moments. Thus we conclude that the resistivity drop at T_C for the (La_1−zNd_z)_0.46Sr_0.54MnO₃ is introduced by the enhanced N(E_F) and the reduced scattering event, which cause reduction in the electrical resistivity of about 40–50% and less than 10% in its magnitude.

The electrical resistivity of Nd_0.46Sr_0.54MnO₃ was reported to diverge at low temperature (Akimoto et al., 1998). In sharp contrast with the divergence of the electrical resistivity, the UPS spectrum of this sample shows a clear Fermi edge at 15 K. Thus we can safely assign the insulating behavior in this sample to being caused by the Anderson localization, which is brought about by the small density of states coupled with the disorders introduced by the mixture of Nd and Sr. The other samples with z ≤ 0.6 do not show the Anderson localization, most likely because the relatively large density of states below T_C weakens the localization tendency of the conduction electrons.

The increase of the density of states in the vicinity of E_F for the samples with z ≤ 0.6 and its absence for z = 1.0 were also confirmed in the conduction-band structure. The Mn 2p XAS spectra of (La_1−zNd_z)_0.46Sr_0.54MnO₃ with z = 0.0 and 1.0 measured at room temperature and liquid-nitrogen temperature are shown in Fig. 3. A significant enhancement in N(E_F) with decreasing temperature could be observed only for La_0.46Sr_0.54MnO₃, while the intensity at E_F in Nd_0.46Sr_0.54MnO₃ was kept almost unchanged. These behaviors are consistent with the present RPES and UPS spectra. Unfortunately, since we were not able to control the temperature in the XAS measurement, the exact temperature where the significant enhancement in intensity at E_F occurs was unable to be determined. However, we believe that it took place at T_C with decreasing temperature.

Figure 3 Mn 2p X-ray absorption spectra for (a) La0.46Sr0.54MnO3 and (b) Nd0.46Sr0.54MnO3. The vertical dashed line indicates the absorption edge, which corresponds to EF in metallic samples. A large increase in the intensity just above EF with decreasing temperature was observed only for La0.46Sr0.54MnO3.

Let us now introduce the electronic structure commonly adopted for the cubic perovskite manganites to understand the origin of the variation in the density of states. A simple cluster-level configuration suggests that the valence band is composed of the bonding t2g, the antibonding t2g, the non-bonding O 2p, the bonding eg and the antibonding eg bands. Band structures calculated by LSDA and LSDA+U methods (Pickett & Singh, 1996; Madvedeva et al., 2001) seem roughly consistent with this configuration. The bonding and antibonding bands are constructed as a consequence of the hybridization between Mn 3d and O 2p electrons. The resulting electronic structure is schematically drawn in Fig. 4. Note here that each band can involve only one electron per unit structure rather than two because of the strong on-site Coulomb interaction U and/or the on-site exchange interaction J, which were reported typically to be 4–8 eV and 0.6–0.9 eV, respectively (Madvedeva et al., 2001; Mizokawa & Fujimori, 1995).

Figure 4 Schematic drawing of (a) the cluster level, (b) the band structure and (c) the density of states expected for the cubic perovskite manganites. AB, NB and B mean anti-bonding, non-bonding and bonding levels, respectively.

We observed a large change in the electronic structure near E_F across T_C for the samples with z ≤ 0.6. This strongly indicates that a fairly large variation takes place in the antibonding eg band. If this modification in the antibonding eg band is brought about by the structure distortion about the MnO₆ octahedron, the corresponding bonding eg band should also be greatly modified at the same temperature. This is in sharp contrast to the case under the influence of the DE interaction limited only upon the antibonding eg band. Thus we measured valence-band spectra for these compounds and investigated the variation in the valence-band structure below and above T_C. The temperature dependence of the UPS valence-band spectra for (La_1−zNd_z)_0.46Sr_0.54MnO₃ samples with z = 0.2 and 1.0 are shown in Fig. 5. Indeed, a large modification in the valence-band structure of the sample with z = 0.2 occurred at T_C in the energy range 4–6 eV where the bonding eg band most likely exists. Thus we conclude that the structure distortion occurs in the (La_1−zNd_z)_0.46Sr_0.54MnO₃ samples with 0.0 ≤ z ≤ 0.6 at T_C with decreasing temperature, and that it is responsible for the large enhancement in N(E_F) resulting in the reduction of the electrical resistivity. In sharp contrast to the large change in the spectral shape of the sample with z = 0.2 at 4–6 eV, the spectrum shape of the sample with z = 1.0 showed less significant temperature dependence. This result agrees well with the relatively small variation in the shape of the antibonding eg band near E_F for this sample, as shown in Figs. 2(d) and 3(b). Therefore the structure distortion in the sample of z = 1.0 is expected to be much smaller than that in the samples with z ≤ 0.6.

Figure 5 Temperature dependence of the HeI UPS valence band spectra for (a) (La0.8Nd0.2)0.46Sr0.54MnO3 and (b) Nd0.46Sr0.54MnO3. The backgrounds were subtracted by using the Shirley method, and the spectra were normalized by the maximum intensity. A large change in the shape of the spectrum at about 4–6 eV, which is indicated as the energy-range between the dotted lines, occurred across TC = 260 K only for (La0.8Nd0.2)0.46Sr0.54MnO3.

Two possible structure distortions, that can lead to a variation in their electronic structure, are known to exist for the perovskite manganites. The first one is the Jahn–Teller-type (JT-type) distortion brought about inside the MnO₆ octahedron. When the degree of the JT-type distortion changes, a complicated modification would be introduced in the valence-band structure, because both the relative energy of the bands and the ratio of Mn 3d components to O 2p components in each band are drastically changed with it. The second is the GdFeO₃-type distortion that contributes to the change in the one-electron bandwidth with altering Mn—O—Mn angle of the neighboring MnO₆ octahedra. These two structure distortions cooperatively or competitively contribute to the variation in the electronic structure at T_C. We speculate that the JT-type distortion rather than the GdFeO₃-type distortion dominantly affects the electron transport properties of the present samples because the significant changes in the whole electronic structure were observed in the photoemission spectra with varying temperature.

The structure parameters of Nd_0.45Sr_0.55MnO₃ below and above T_N have already been reported (Kawano et al., 1997), but the structure analysis on (La_1−zNd_z)_0.46Sr_0.54MnO₃ with 0.0 ≤ z ≤ 0.6 has not been reported. Once those data are obtained, the relation between the electronic structure and the atomic structure in the cubic perovskite manganites can be quantitatively and systematically discussed. These discussions will lead to a better understanding of the electron transport behavior in these perovskite manganites.

4. Conclusion

High-resolution HeI UPS, Mn 2p–3d RPES and Mn 2p XAS measurements have been performed for (La_1−zNd_z)_0.46Sr_0.54MnO₃ (z = 0.0, 0.2, 0.6 and 1.0), which possess phase transitions from the paramagnetic phase through the ferromagnetic phase into the A-type antiferromagnetic phase with decreasing temperature. Temperature dependence of the electrical resistivity is discussed in terms of the observed photoemission spectra. We found that the Fermi edge in (La_1−zNd_z)_0.46Sr_0.54MnO₃ (0.0 ≤ z ≤ 1.0) persisted over the temperature range 15 ≤ T ≤ 340 K. The temperature dependence of the electrical resistivity of these compounds was well accounted for by considering the combined effect of the varying density of states and scattering events associated with the ordering in the local magnetic moments. The former effect, which is related to the structure distortion, leads to about 50% variation at the maximum in the electrical resistivity. The latter is so weak that only less than 10% variation in the resistivity can be induced by it. The insulating behavior in Nd_0.46Sr_0.54MnO₃ at low temperature is attributed to the Anderson localization, most likely caused by the small density of states at E_F coupled with the chemical disorder between Nd and Sr.

Footnotes

†Presented at the `International Workshop on High-Resolution Photoemission Spectroscopy of Correlated Electron Systems' held at Osaka, Japan, in January 2002.