H. S. Yang, Guanyuan Yu, Z. J. Chen
Physics Department, University of Science and Technology of China, Hefei, Anhui 230026, P. R. China
B. C. Crooker
Physics Department, Purdue University, West Lafayette, IN 47907, U.S.A.
B. Y. Qi, Y. H. Zhang
Laboratory of Structure Research, University of Science and Technology of China, Hefei, Anhui 230026, P. R. China

The temperature dependence of the normal state dc susceptibility of La2-xSrxCu1-yFeyO4(x=0.13, 0.15, 0.17 and y=0.0, 0.002, 0.004, 0.006) polycrystalline samples in different fields (H=10, 30 and 50 KOe) and the relationships between superconducting transition temperature and Fe doping content have been studied. The experimental results show that the temperature dependence of normal state dc susceptibility presents board peak behavior for the undoped sample (x=0.0). With the increase of Fe content, a Curie-Weiss contribution appears in the normal state susceptibility, and the transition temperature changes with the change of Fe content. It indicates that the carrier concentration is the most important fact for the superconductivity. For the Fe doped samples, the normal state dc susceptibility can be well expressed as x=a+bT+c(T-T0). The term of a+bT is the Pauli paramagnetic contribution of Cu dx2-y2 -Op band. The Curie constant increases while the constant term decreases with the increase of field for the same Fe doped sample. We think that there exists a narrow unfilled band at Fermi level ( dx2 upper Hubbard band) which is the origin of the Curie paramagnetization. Our experimental results also indicate that there is a close relationship between superconductivity and the band structure.

I. Introduction
Recently people agree that the properties of Cu-O band are very important to the high Tc superconductivity and the properties of normal state. However, all attempts to explain the origin of the high Tc superconductivity and the anomalous properties of normal state have not been agreed so far. Many experimental and theoretic works have been devoted to it. Anderson[1] first points out the importance of electronic correlation. Subsequently, it has been extended to many models, most of which are based on the single Cu dx2-y2 -Op band (flat house model[2]). Kamimura first suggested the two-story house model[3] that is based on the two bands of Cu dx2 band and Cu dx2-y2 -Op band interacting by Hund's coupling as well as electron correlation[4]. The origin of the high Tc and some of the properties in the normal state can also be well explained by spin fluctuation model of two-dimensional itinerant electron system used in weak interactions[4]. There are lots of work to be done in the theoretical research. On the other hand, experimentalists have to design some experiments to distinguish which model is valid to explain the origin of the high Tc superconductivity and the anomalous properties in the normal states.
In this paper, we report our dc magnetization measurements for the Fe doped La2-xSrxCu1-yFeyO4 system in different magnetic fields in the temperature region between 4.2K and room temperature. The experimental results show that in the undoped sample the relationship between the dc magnetization and the temperature is broad peak behavior. With the increase of the Fe content, the relationship between the dc magnetization and the temperature changes rapidly from broad peak behavior to Curie-Weiss behavior. At the same time, the transition temperature changes with the change of Fe content. It clearly indicates that the carrier concentration is the most important fact for superconductivity. As to the same Fe doped sample, when increasing field the Curie constant c is increased while the constant term a is decreased. We think there exits a narrow unfilled band at Fermi level ( dx2 upper Hubbard band) and a Cu dx2-y2 -Op band. The Curie behavior comes from the contribution of narrow band, while the Pauli paramagnetic behavior comes from the contribution of broad band and narrow band. Under different fields, the position is different and the electrons will be redistributed. The above experimental results indicate that the narrow band behavior near the Fermi level must be taken into account when we discuss the high Tc superconductivity.

II. Sample Preparation and Experimental Details.
The polycrystalline samples of La2-xSrxCu1-yFeyO4 (x=0.13, 0.15, 0.17 and y=0.0, 0.002, 0.004, 0.006) were prepared by the standard solid state reaction method[1]. High-purity (99.9%) La2O3,SrCO3, CuO, and Fe2O3 powers were well grounded and mixed, and the mixture was then put in an Al2O3 crucible and heat treated in air. The mixture was kept at 900¡æ for 24 hr, and was well grounded. In order to obtain high quality single phase samples, the above procedure was repeated for three times. Then the powders were pressed into pellets with the 13mm of diameter and 1.2mm of thickness, and sintered at 1025¡æ for 48 hr, followed by annealing at 525¡æ for 48 hr.
The dc magnetization measurements were carried out in a commercial cryogenic superconducting quantum interference device magnetometer (SUSCON 600 made by Cryogenic company). The superconducting transition was measured at 0.03 T.
The magnetization was measured in magnetic fields of 1 T and 3 T as a function of temperature in the range of 4.2-300K. The scan length of the sample holder is 3 cm. The position of the sample was centered before every measurement. In order to minimize the experimental errors and interference signals, we take the mean of six scans. In order to minimize the end effect of the sample holder, the sample was arranged apart from the end by 60 mm. The sample was stuck on the center of the thread nest. Since the signal of the sample is weak, the signal of the sample holder should be taken into account. All the data of the samples presented later has been correlated with the data of the sample holder.

III. Experimental Results and Discussion
Figures 1, 2, and 3 show the superconducting transition procedure on the character of temperature dependence of the reduced dc magnetization (i.e. M/H) of La2-xSrxCu1-yFeyO4 (x=0.13, 0.15, 0.17 and y=0.0, 0.002, 0.004, 0.006) in the magnetic field of 30 Oe. As to the samples with x=0.13 and 0.17, the superconducting transition temperature decreases with the increase of Fe content. But as to the sample with x=0.15, the transition temperature increases from 31.0 K to 33.6 K by small Fe doping (y=0.002) and decreases with futher increasing Fe doping content. Since the 3d electrons of Fe doping are less than those of Cu and it is magnetic dopant, there are two important roles when Fe displaces Cu. First, it will change the carrier concentration that leads to the change of transition temperature. Secondly, the increase of magnetic dopant density will lead to the decrease of transition temperature. The experimental results indicate that the change of carrier concentration is the most important result of the Tc change. As to the sample with y=0.15, when the Fe doping y>0.002 the decrease of transition temperature is due to the combined contribution of above two facts. The increase of Tc while y<0.002 is due to the lack of O in this sample.
The temperature dependence of the reduced dc magnetization in various magnetic fields (H= 10 KOe and 30 KOe) is shown in Fig. 4, 5, and 6. These figures show that in the undoped sample, the relationship between the dc magnetization in the normal state and the temperature is broad peak behavior, and does not change obviously with the change of applied field. The spin fluctuation of Cu dx2-y2 Op can not explain it because the spin fluctuation will be suppressed by applying high field. It indicates the spin fluctuation model of two-dimensional itinerant electron system used in weak interactions[4] is unsuccessful as to our experimental results.
We think this board peak behavior is the Pauli paramagnetic contribution of the Cu dx2-y2 -Op band that is narrower than the p band. When applying magnetic field, the bands parallel and antiparallel to the magnetic direction will separate, and the electrons in these bands will be redistributed. Since it is a board band, the redistribution will not lead to obvious change of the density of states at Fermi level, and it will not lead to obvious change of board peak behavior. With the increase of Fe content, the board peak behavior is suppressed and the Curie-Weiss contribution is increased. As to the samples with the same Fe content, the Curie-Weiss contribution increases with the increase of field, which indicates the behavior of Fe foreign ion in these samples is not the behavior of independent magnetic ion. For all the Fe doped samples, the normal state dc susceptibility can be well fitted as SYMBOL 99 \f "Symbol"=a+bT+c/(T-T0), and the data parameters a, b, c and T0 are listed in Tab.1a, 1b, and 1c for x=0.13, 0.15, and 0.17, separately. These experimental results indicate there exists a narrow band at Fermi level ( dx2 upper Hubbard band). Without the Fe impurity, it is just below the Fermi level. The Cu dx2 band is fully occupied, and it makes no contribution to the normal state dc magnetic susceptibility. The only contribution is the Pauli paramagnetic contribution of the Cu dx2-y2 -OpSYMBOL 115 \f "Symbol" band. When Fe is doped, the carrier concentration decreases and the Cu dx2 Hubbard band moves to higher energy level. Holes are produced in the Cu dx2 Hubbard band. So the partly occupied Cu Hubbard band makes contribution to the Curie-Weiss term. When increasing Fe content, more holes are produced in the Cu dx2 Hubbard band and the Curie-Weiss contribution increases. As to the sample with the same Fe content, when applying magnetic field the upper and lower Hubbard bands of Cu dx2 are shifted to higher and lower energy level respectively. Then there are more holes in the Cu dx2 Hubbard band and the excess magnetic moment will increase subsequently. When the electrons in the Cu dx2 upper Hubbard band is moved to the Cu dx2-y2 -Op band, the Pauli paramagnetic contribution will decrease (the term a decreases, refer to Table 1a, 1b, and 1c.) because the Fermi level is near the top of band. Our experimental results are well explained by the above model.
In figure 6 M/H(T) exhibits slight bending at Tt =160 K. The following characteristic features are observed: as the magnetic field increases the temperature Tt increases slightly. As the Fe content y increases the Tt moves to higher temperature and the bending is boarder. The same slight bendings are also shown in figures 4 and 5. The only difference is that the bending is not so obviously. We can also observe that as the Sr content x decreases the Tt of the sample with same Fe content y also decreases.
While analysis the slight bending behavior obtained, we notice the similar work[5]-[8]. We think the slight bending due to the structural phase transitions (SPT) from high-temperature tetragonal (HTT) to low-temperature orthorhombic (LTO) phase. The structural phase transition is coupled to the electronic structure via the Cu dx2-y2 -Op band. As the bands formed by the Cu 3d and O 2p orbital, are sensitive to the SPT, the M/H(T) should also depend on the SPT. And secondly, a change in the electronic structure, by doping with Sr, leads to a change in the temperature of the SPT. The additional charge due to the doping decreases the equilibrium distance of Cu and O. The HTT phase is therefore stabilized, and the reduction of Tt can then be understood.

Above experimental results can be described as:
1) The narrow Hubbard band near the Fermi level is responsible for the anomaly field dependence of dc magnetization in the normal state. There is a close relationship between superconductivity and the band structure.
2) The peculiarities of magnetic characteristics of La2-xSrxCu1-yFeyO4 in the vicinity of Tt are associated with a structure phase transition via the Cu-O band.

Reference
[1] P. W. Anderson, Science, 235(1987),1165.
[2] D. Yoshioka, J. Phys. Soc. Jpn. 58(1989)1516.
[3] H. Kamimura, Int. J Mod. Phys. B1(1988), 699.
[4] T. Moriya, Y. Takahashi and K. Ueda, J. Phys. Soc. Jap. , 59(1990),2905.
[5] M. Braden, P. Schweiss, G. Heger, W. Reichardt, Z.Fisk, K. Gamayunov, I. Tanaka, and H. Kojima, Physica C, 223(1994),396.
[6] R.M. Fleming, B. Batlogg, R.J. Cava, and E.A. Rietman, Phys. Rev., 35 (1987)7191.
[7] T. Atake, H. Kawaji, K. Yoshimura, T. Uemura and K. Kosuge, Physica C, 185-189(1991),1413.
[8] D.A. Wright, J.P. Emerson, B.F. Woodfield, R.A. Fisher, N.E. Phillips,M.K. Crawford, and E.M. McCarron, Physica C, 185-189(1991),1387.