STRUCTURE OF THE NEW 1201 LEAD CUPRATE SUPERCONDUCTOR

The structure ofthe newly discovered “1201” lead cuprate superconductor (Pb, Cu) (Sr, L~)#ZUO,-~, with T,=27.5 K at onset and a shielding fraction 38% at 5 K and 10 Oe, has been determined using neutron powder diffraction. The structure is similar to the other 1201 materials TlBa, zLa,,,CuO, and T10,5PbOSSr2CuOS (where the former superconductors with a r, of 52 K and the latter is not superconducting), belonging to the space group P4/mmm. The doping scheme in (Pb, Cu)(Sr, La)zCuOS_,r com-bines the doping scheme used in these two compounds, in that both the Tl and Sr sites are doped. The starting stoichiometry. the refined scale factors for the impurity phases and the refined site occupancies for oxygen suggests that the stoichiometry (relative to Cu) of the superconducting phase is Pb o.60Cuo.40Sr,.08Lao.92Cu04.96. Calculation of the average hole concentration in these compounds from charge summation is difftcuh with these compounds because the TI/Pb-0 layers provide polarizable charge reservoirs that can participate in substantial covalent bonding and because of the probable mixed-valent nature of TI and Pb. Nevertheless, bond valence sums calculated for the Cu ions in the CuO, layers for the three 1201 cuprates do provide a correlation with the values of T, or the absence of superconductivity.


Introduction
In the homologous series of single-thallium-layer cuprate superconductors [ 11, represented by Tl-BaKa,-,CU,O~,,+~ ( n = 1 to 5 ) , it has been possible to replace half the thallium with lead an all the barium with strontium to give compounds of composition (T10.,Pbo5)Sr,Ca,_ ,CU~O~~+~ [2-41. The structure essentially consists of Ba (or Sr)-Cu-(Ca)-0 oxygen deficient perovskite slabs which are interconnected by a rock-salt like A0 layer, where A is Tl or (Tl, Pb). The great range of available cations that can be doped into these compounds offers us the opportunity to study the relationship between crystal structure, hole doping and superconductivity. Superconductivity in these compounds is believed to originate from the presence of holes in the CuOz sheets and the average oxidation state of the CuOz sheets can be controlled by changing the chemistry of these materials. The mechanisms for hole doping include cation substitution [ 5 1, cation vacancies [ 6 1, insertion of oxygen [7], and internal redox mechanisms of the type T13++2e+T1'+ [8].
We have been studying the structure and super-conductivity of the first member of this series, with n = 1, generally referred to as the " 1201" compounds. TlBazCu05 does not superconduct because it is overdoped with too high a hole concentration in the CuOz sheets. Thus, superconductivity can be induced through the partial substitution of trivalent rare earth ions for Ba*+ and TlBa,.2La0.8Cu0s is a superconductor with a T,= 52 K [ 91. However, partially doping TISrzCuO, in the Tl-0 planes with Pb4+ to what is expected to be nearly the same hole concentration does not produce superconductivity [ IO]. Presumably, the sort planar Cu-0 bond contrains the Cu02 sheets to too high a hole concentration de This article reports the synthesis and structural characterization, using neutron powder diffraction, of (Pb, Cu) (Sr, L~),CUO~-~ in order to better understand its structure, particularly the Cu-0 bonds lengths, the cation ratios in the mixed cation Pb/Cu and Sr/La sites, and the oxygen stoichiometry. These results allow us to correlate the structure and estimated hole concentration with superconductivity.

Experimental procedure and data analysis
A 25 g sample was prepared from starting powders PbO, SrC03, La203 and CuO, so as to give the metal composition Pb0,6SrLaCu1.505-6, using the procedure outlined by Adachi et al. As shown in fig. 1, resistance measurements indicated that the sample is superconducting with a T, (onset) of 27.5 K and a T, (zero resistance) of 23.5 K. The broadening of the lower half of the transition and lower T, (zero) probably arises from imperfections at the grain boundaries [ 12 1. Susceptibility measurements, the results of which are shown in fig. 2, also gave a T, (onset) of 27.5 K, a shielding fraction of 38% and a Meissner fraction of 16%, the latter two at 6 K.
Neutron-powder-diffraction data were collected at 300 K on the high intensity powder diffractometer (HIPD) at the Manuel Lujan, Jr. Neutron Scattering Center (LANSCE) at the Los Alamos National Laboratory. Data were collected in four detector banks (at approximately + 153" and ? 90" ) for approximately 1 h at an average proton current of approximately 80 PA. The structural models were relined using the Rietveld refinement code developed by Larson and Von Dreele [ 13 1. The data taken was indexed with the tetragonal space group P4/mmm. The structural parameters refined for (Pb, Cu) (Sr, La)2Cu05--6 include the lattice constants, atomic positions, anisotropic thermal parameters, and site occupancy. As in the structures for other 1201 compounds, atoms in the Pb/Cu-03 plane displayed large in-plane thermal parameters, and displacements in the [ 1001 and the [ 1 lo] directions were used to model the disordering of Pb/Cu and 03 in the u-b plane. 03 displacements of the type (x, l/ 2, 0) gave values of weighted profile agreement factors, Rwp, that were considerably lower (by 0.33-0.35%) than displacements of the type (x, x, 0). For Pb/Cu, R,, for the various displacement models dif- Measurements were performed in an applied field of 10 Oe and demagnetization effects were accounted for in calculating the percent diamagnetism. fered less, this time with the (x, 0, 0) positions favored over the (x, x, 0) positions by 0.02-0.04%. As a consequence, the displacement sites (x, 0, 0) and (x, l/2,0), with isotropic thermal parameters, were used to model the atomic positions for Pb/Cu and 03, respectively. In addition to the structural parameters shown in table 1, background, scale factor, absorption, extinction and anisotropic strain were also fitted.
The nonstoichiometric starting composition suggests that impurities would be present. In fact, we found that PbO, CuO and Laz-,SrXCuO~ were all present in addition to the parent (Pb, Cu) (Sr, La)zCuO,-J: these were included as second, third and fourth phases in the refinement. Since these impurity phases were only present at relatively low levels (at 1.2, 2.0, and 5.8 wt.% for PbO, CuO and Table 1 Structural parameters for Pbo soCu~,40Sr,.,,Laos2CuO~-~ at 300 K. Space group P4/mmm: Laz-,SrXCuOb, respectively, we fixed the isotropic thermal parameters for the cations and oxygen ions at 0.010 and 0.015 A and refined only the scale factors, lattice contants and atomic positions for these three phases. The initial stoichiometry and the refined phase fractions of these impurities suggested that the Pb/ Cu and Sr/La ratios for the two mixed cation sites in (Pb, Cu)(Sr, La)ZCu05--6must be 1.50and 1.17, respectively, which are to be compared with the expected ratios of 1.5 and 1 .O, respectively. These values were then used without further refinement for the relative site occupancies for Pb/Cu and Sr/La in the refinement of the structure for Pbo.60CUo.40Sr,.osLao.~~CuO~-~.
During the initial refinements for oxygen site occupancy, the 01 site occupancy would refine to values slightly greater than unity and was consequently set 1 .O and not further relined. The site occupancies for 02 and 03 refined to a values of 0.970(4) and 0.249(4), respectively. If we take into account the multiplicity of 2 for the 01 and 02 sites and 4 for the displaced 03 site, this then gives a value of 6= 0.04.
Portions of the data collected with the + 153 O detector bank and the fits from the structural refinement are shown in fig. 3. The results for the struc- tural refinements are given in table 1 and an ORTEP plot [ 141 of the structure is shown in fig. 4.

Results and discussion
The refined lattice constants [a= 3.78608 (4) and c=8.65623( 18) A] differ from those for the undoped parent compound TISrzCuOs [a = 3.7344 ( 5) and c=9.007 (1) 8,] in that a is longer and c is shorter. Thus the substitution of the smaller Laf3 for Sr+' and the smaller Pb+4/Cu+2 for T1+3 leads to an expected decrease in c but a surprising increase in a. Comparison of the lattice constants with those for superconducting TIBa,,,La,.JZuOS [a=3.8479( 3) and c= 9.0909( 6) A] and nonsuperconducting T1,,Pb,,~Sr2CuOs [a=3.7309 (3) andc=8.9$83(8) A] show that c is shorter than either but that a lies in between.   (2) recent study, where a titrimetric technique was applied to the double thallium layer cuprates T12Ba2Can-ICu,02n+4 (n = 1 and 2) to determine both the Tl and the oxygen contents [ 181; these studies suggested that Tl must be mixed-valent if the formal oxidation state of Cu is greater than 2, as would be expected for a hole superconductor. If Tl is indeed mixed valent, then the hole concentrations for Tlz-.,Cd,BazCuOb+d and TIBa,.zLaO.&uOS calculated from charge summation may be too small, and likewise for Tlo.,PbosSrzCuOs if both Tl and Pb, or either, are mixed valent.
The oxidation state of the Pb ions in these kinds of compounds have been estimated from X-ray absorption near-edge spectroscopy (XANES) and from the Pb-03 bond lengths, but neither method is quantitative.
An earlier structural study of T10.5Pb0,5Sr,Ca,-,Ct.1,,0~,,+~ (n=2 and 3) made use of separate XANES measurements to determine the oxidation state of Pb [ 17 1. A comparison of the nearedge features at the PbL,,,edge for both superconductors with those for PbO and PbOz showed a strong similarly to the latter and suggested that the valence of Pb in the superconductors is predominantly $4. A similar study showed that the XANES spectrum for Tl in Tlz-,Cd,BazCuOh+B showed the pre-edge features characteristic of Tl+3 ; however, it was felt that this result does not rule out the existence of Tl+ '/ T1+3 mixed valence but instead shows that the average oxidation state of Tl does not change with Cd substitution [ 5 1. Sums of ionic radii [ 191 give an expected bond length of 2.59 8, for symmetric octahedrally coordinated Pb+*-O-* and 2.175 A for similarly coordinated Pb+4-0-2. CU+~ tends to form distorted octahedral complexes with oxygen, with four short bonds (about 1.93 A) and two long axial bonds (about 2.25 A). The lattice parameter c1 constrains the average length of the in-plane Pb/Cu-03 bond to 2.677 A, while the length of the axial Pb/ Cu-0 1 bonds is I .94 A. This results in four long and two short bonds for the Pb/Cu-01, 03 octahedra that differ substantially from the expected geometry for symmetric Pb06 octahedra or distorted Cu06 octahedra. This frustration of the optimum bonding and coordination for the oxygens is relieved by displacements of the constituents in Pb/Cu-03 planes to give four short Pb/Cu-0 bonds (two in-plane and two axial) and two long Pb/Cu-0 bonds. In fact, the displacements of both Pb/Cu and 03 from their ideal sites result in two short bonds at approximately 2.22 A and two long bonds at approximately 2.75 A in the Pb/Cu-03 plane. Thus neither the EXAFS results nor the lengths of the Pb/Cu-0 bonds sheds much light on the oxidation states of the Pb ions.
Since the mixed-valent nature of the Pb/Cu-03 layer makes it difficult to determine the hole concentration in the Cu-0 layers from charge summation, we instead try to estimate the hole doping from valence bond sums for the Cu02 sheets [ 20,2 11 Since all of te 1201 cuprates have polarizable and probably mixed-valent Pb/Cu-0, Tl-0 or Tl/Pb-0 layers, the calculation of hole concentrations based on summation of oxidation states is difficult. This is a reflection of the possible covalent bonding and the more complicated redox chemistry that is available with the Tl/Pb-0 layers in these systems. The length of the Cu-0 bonds may provide an alternate measure of the hole concentration in the Cu-0 planes, although internal strain in these crystal result in overestimates of the Cu valence. We find that the Cu-0 in-plane distance, which provides the dominant contribution to the valence sum, in Pb,&uO.,LaSrCuOS is 1.893 A, shorter than the equivalent bond in TIBa,.,LaO.CuOS (at 1.915 A) and longer than the Cu-0 bond in T10,5Pb,,$r2CuOs (at 1.865 A). The resulting valence sums for Cu in 140,  I  I  I  I  I  I  I   I-TPSCO   0   I  I  I  I   I   Tlh.&ao.&u05, PbO.SCu,,SLaSrCuOS and T1,sPb0,sSr2CuOs correlate well with the TCs of 52 and 27.5 K for the first two compounds, which are superconductors, and with the absence of superconductivity in the third compound. Thus the hole concentration in the Cu-0 sheets, calculated from the length of the in-plane and axial Cu-0 bonds, provide a useful correlation with superconductivity as noted earlier by de Leeuw et al. [ 23 1.