Introduction

Oxygen content has long been known to be influential to the electrical properties of perovskite oxides [5]. In the cuprates, the importance of oxygen content was established by the earliest of experiments. The annealing of $\rm {La_{2-x}Ba_xCuO_4}$ in vacuum, for instance, was found to raise the onset of T$_c$ by as much as 20K from the initial as-grown value [4]. Even the non superconducting $\rm {La_2CuO_4}$ can be rendered a superconductor by high pressure oxygen treatment [127]. But the role played by oxygen is far from simple. The superconducting transition temperature, T$_c$, is a complex function of oxygen content, cation composition, and crystal structure. The carrier concentration, $n_s$, is the critical quantity in maximising T$_c$, and it is as a means to controlling the carrier concentration that oxygen content derives its primary influence. Along with cation dopants such as in La$_{2-x}$Sr$_x$CuO$_4$, or similar variation in cation stoichiometry within a stability field (an important feature in the Bi-systems), the variation in oxygen content has proved an important means to control $n_s$, and has been a key to investigating the high-T$_c$ mechanism.

The relationship between T$_c$ and carrier concentration is a relatively straightforward one, and is illustrated in Figure 5.1; its nature has been found to be universal amongst the cuprates [128,129,130]. On the other hand, the relationship between carrier concentration, oxygen content, and structure is not so straightforward. A useful simplification, made by way of explanation, is to consider all cuprates as a charge reservoir separated by a number of CuO$_2$ layers. Although the carrier concentration is regulated by the charge reservoir, Zhang [129] has then shown that the optimum value of $n_s$ can be considered as a universal parameter related only to the number of CuO$_2$ layers. Thus thereby explaining the universal T$_c$-$n_s$ relationship of Figure 5.1. Yet the maximum transition temperature achievable, T$_c^{max}$, which corresponds to the optimum doping of the CuO$_2$ planes, does vary from structure to structure. This is specific because it is determined by other factors, principally involving the immediate structural environment around the CuO$_2$ planes, and may involve the interplanar coupling between CuO$_2$ layers and the charge reservoirs. Outside this simple model, the details of the mechanism of the charge transfer process, for instance, tend to be specific to each of the cuprate families. So this means changes in T$_c^{max}$ may be brought about, possibly independent of $n_s$, by structural rearrangements within the charge reservoir layers.

Figure 5.1: A schematic of the universal T$_c$-$n_s$ relationship determined for all p-type high-T$_c$ cuprates (after Zhang). The precise shape of the curve is varyingly expressed by different authors as having a flat maximum or as a negative parabola.

The effect of oxygen incorporation within a structure can therefore be twofold, controlling the carrier concentration, and accommodating structural readjustments within the charge reservoir. Both hold their own influence over T$_c$, and the mechanism by which each does so may be different. This is demonstrated most clearly by the correlations observed in Y-123 between a plateau in the T$_c$-$n_s$ curve and the ordering of oxygen vacancy superstructures in the CuO(reservoir) layers (as described in Chapter 2); over the range of oxygen content, not all of the additional carriers contributed are realised as an increased $n_s$ in the CuO$_2$ planes, but remain instead localised at sites within the CuO chains.

A remarkable level of understanding of the interplay between oxygen chain ordering and transport properties has emerged in the Y-123 structures [17]. The same, unfortunately, cannot be said of the other cuprate families, for which understanding has sadly trailed. This is especially true of the Bi-systems, due to the increased structural complexity of the charge reservoir brought about by the presence of the incommensurate modulation. Excess oxygen incorporated into the BiO layer every 4-5 unit cells by the modulation is an oft-cited claim, one which is all too frequently used as a blanket explanation for results of studies on a whole variety of structural, superconducting, and normal state properties (and can be found stated in almost all of the references in the following review, Section 5.2). When fully occupied it follows from this assumption that an oxygen content per formula unit of 8.2 (i.e. $\delta$=0.2) can be accounted for, compared to a value of 8.0 for the ideal unmodulated structure. As has already been discussed in Chapter 3, extra oxygen has indeed been successfully evidenced by neutron diffraction experiments [88,85]. However, as was made clear in Table 3.1 and Figure 3.4, the real structure may be substantially more complicated than is assumed in this single extra oxygen picture, allowing for a considerable range in oxygen variation and involving a number of potential sites. Furthermore, as will be made clear in Section 5.2 of this chapter, the results of many experiments which measure the variation in oxygen content due to annealing suggest that a range of variation in oxygen of $\Delta \delta>0.2$ may be possible. All of this suggests that the currently accepted assumptions are at best naive, and that the true degree of subtlety in the oxygen-structure relationship has yet to be discovered.

The failure to describe other structures as correctly as Y-123 is an important problem. It has been pointed out by Sendyka [131] that the many experiments which suggest subtle structural changes linked to the superconducting mechanism in the vicinity of T$_c$ (and discussed further in Chapter 6) cannot be interpreted conclusively because of the lack of understanding of local oxygen structure. The question of flux pinning in the Bi-systems is another problem of intense current interest. Two regimes of pinning are believed to exist: below about 30K in the flux lattice state where oxygen vacancies in the CuO$_2$ plane are believed to dominate [132], and above 30K in the 2D pancake vortex state where more extended defects are required to account for the observed pinning of the pancakes [120]. The lack of knowledge about the influence of oxygen incorporation upon micro-structure has severely hampered the understanding of these pinning mechanisms, and is also of importance in improving processing techniques to maximise $J_c$. Also, theories describing normal state properties are dependent upon this question [133]. The development of a better understanding of the oxygen-structure-T$_c$ relationship in Bi-2212 is a desirable end for all these reasons.

The experiments in this chapter attempt to discern changes in diffuse x-ray scattering which may be associated with changes in oxygen content or ordering. Although neutron scattering is more suitable for this purpose, the ability to observe oxygen changes with x-rays has been well demonstrated in Y-123 [134]. A summary will first be presented of the experimental information available to date which relates the behaviour of transport properties with oxygen content and annealing treatments, and in so doing the inadequacies of the current understanding are highlighted. The experiment itself starts by characterising three as-grown crystals using EPMA, AC-susceptibility, and x-ray scattering, and identifying differences in the diffuse scattering which appear to correlate with differences in oxygen content. The relevance of these differences is then confirmed by subsequent measurements made after ex situ high temperature annealing treatments in either nitrogen or oxygen atmospheres.