Discussion

The presence of diffuse streaks, associated with, but distinct from the satellite reflections, was shown in Chapter 3 to be an intrinsic feature of the reciprocal space pattern of Bi-2212. Their origin, however, was not clear. The x-ray scattering results presented in this chapter have revealed new information about the nature of these streaks and, most importantly, have shown them to be strongly affected by certain annealing treatments. This observation in combination with the compositional analysis of the crystals makes possible some definite judgements about the origin of the diffuse streaks.

As was made clear in Chapter 3, their diffuse character indicates them to be associated with some structural feature which must be closely linked to the modulation within the basal a-b plane and which disrupts its body-centred symmetry, but which is poorly correlated between layers along the c axis. The transformation observed in the streaks of the B1 sample indicates that the annealing treatment has brought about an ordering of this structural feature along the c axis which results in an even more distinct violation of the symmetry. The annealing treatment of the A1 sample, however, brought about little or no change in the streaks. The compositional analysis identifies the primary differences between the A and B samples to be the low Sr, and slightly higher Bi content of the B samples, along with a greater oxygen content. An annealing induced ordering of either cation substitutions (possibly Bi on Sr sites) which are not present in the A1 sample, or an ordering related to the additional oxygen content of the B1 sample could therefore be potentially responsible for the observed changes. It is notable, however, that the Oxford crystal, which has a similar Bi stoichiometry to that of B1, has a diffuse streak closer instead to that of the A1 crystal in intensity and shape, and also that both the A1 and Oxford crystals have similar oxygen contents and T$_c$ values. This suggests that the larger oxygen content of the B1 crystal is the more significant parameter.

In as-grown crystals a large excess of oxygen will most reasonably be incorporated in a highly disordered manner; this is evidenced in STM images of oxygen rich BiO layers for instance [148]. If oxygen is indeed at the root of the streaks, this lack of coherence in the oxygen arrangement would explain the very flattened shape of the diffuse streak observed in the B1 sample before annealing. Upon annealing, the subsequent ordering may be dependent upon the extent to which oxygen content has also changed. The ordering could involve either the excess oxygen atoms themselves or vacancies created after oxygen loss. The brief ten minute annealing treatment of the B1 sample is, on the basis of the results of the many other annealing experiments discussed at the start of this chapter, too short a time to have resulted in a substantial change in the bulk oxygen content of the sample, particularly considering the sample's large dimensions. The effect of this brief annealing may then be rather to facilitate the establishment of a more ordered oxygen distribution, or to allow a slight oxygen loss which then facilitates the ordering of the remaining excess of oxygen. Perhaps this latter explanation is the most reasonable picture. A further potential experiment would be to investigate the effect of annealing at some lower temperature where oxygen loss will be more definitely curtailed. Low-temperature annealing of Y-123, for instance, has been found necessary to achieve the most ordered oxygen states [154]. The post-anneal quenching may also have been a factor in achieving the ordered state, and the effect of slow-cooling could also be investigated. The behaviour of the A1 sample in this picture is understandable if the oxygen annealing leads to oxygen incorporation at other vacant sites, and not to the additional oxygen positions responsible for the streaks. This is sensible because sample A1 starts from a much lower oxygen content, and even after the annealing the change in T$_c$ indicates it's oxygen content remains lower than B1.

Figure 5.10: A schematic view of the Bi and Cu bi-layers in the [1 0 0] projection illustrating the symmetry of the modulation. The large circles indicate positions which when altered would disturb the body-centred arrangement (outlined).

A schematic representation of the modulated structure in the [1 0 0] direction in Figure 5.10 illustrates how a violation of the body-centred ordering of the modulation, necessary to be consistent with the streaks, may be brought about (the large circles in the figure). The locations also correspond to areas of greatest separation between neighbouring Bi-O planes, and they would therefore be favourable locations for the incorporation of additional oxygen. The effect of the annealing is to allow the oxygen to migrate to this more favoured position, no c axis interaction is required between the oxygen to achieve the increased correlation, the guiding potential is provided by the modulation, and they merely align themselves within this. The readjustments of the local structure which may then result could include a closer bonding of the adjacent BiO$_2$ layers due to the linking oxygen, and would reduce the amplitude of the modulated displacements along the c axis. This effect would account for the reduction in satellite intensity observed to accompany the ordering of the streaks. If this ordering and associated change in the amplitude of the modulation are anything other than coincidental with the increase in T$_c$, which also resulted from the annealing, then it could be effected through a change in the tilt of the CuO$_5$ octahedra. The tilt angle is known to be an important quantity in determining superconducting properties [155], and would certainly alter along with the modulation. The flattening of the corrugation of the CuO$_2$ layers could also be a factor, and it is interesting to note that a change in the ${\bf b}$ axis was also observed to accompany this, in agreement with the model proposed at the close of Chapter 4. If there was indeed little change in oxygen content, the suggested model of ordering could explain the observed increase in T$_c$ of around 3K by these means. The ordering may also bring about an increase in the inter-layer coupling. The c axis resistivity $\rho_c$ in Bi-2212 is believed sensitive to the coupling and is known to be highly dependent upon annealing treatment and to vary considerably between samples. An ordering of the sort suggested here could explain the relief of interlayer distortion previously suggested by Yoo [133] to explain the change in $\rho_c$ due to annealing.

The suggestion that structural change induced by annealing has been responsible for influencing superconductivity and T$_c$ has been made by other experiments when it was believed that no suitable change in oxygen content to effect the carrier concentration could have occurred. Two already mentioned in the review [141,111] could perhaps be attributed to underestimates in the degree of oxygen evolution. A result which is harder to explain in this way is that obtained by Wu [113] who found that the T$_c$ of single crystals could be raised to around 86K by annealing at 550$^o$C but that the change was independent of both annealing period and, most striking of all, the atmosphere (be it oxygen or nitrogen); it was therefore concluded only a structural modification, and not changing oxygen content could be responsible. It has similarly been found in previous annealing studies of the Warwick samples of the B type used here, that it is difficult to effect T$_c$ any further than the initial rise to 86K, even with greatly extended periods of annealing. Evidence of a different kind comes from Mossbauer studies of Fe doped Bi-2212 by Lin [156]. The technique probes the local structure of the Fe atoms substituted onto the Cu sites and is sensitive to the distortions of the CuO$_2$ layer by the modulation. Vacuum annealing was found to induce changes in these local distortions, and this was in association with the rise of T$_c$ to its maximum value. It is always a problem to disentangle the individual effect of structure from that of carrier concentration upon the superconductivity, and very rarely are transport measurements combined with an explicit study of the affects upon the structure. But these interpretations are in agreement with the results presented here. The speculation needs now to be confirmed by careful measurement of oxygen content both before and after the annealing, and perhaps in combination with Hall measurements, to allow a definite conclusion to be made.

The model proposed here cannot be conclusively established by the data available, but it does appear the most viable picture compatible with the current information. Alternative structural modifications can be imagined which might be responsible, such as excess oxygen incorporation into the Ca layer perhaps, or the ordering of cation substitutions in the Ca or Sr layers. However, these alternative models must also be able to account for the associated changes of the modulation amplitude, and it is not so easy to visualise how the ordering of interstitial oxygen in the Ca layer for instance, would be so closely tied to the modulation period. Further studies are clearly warranted.