Figure 1 shows a high-resolution image of 2212 overgrowth on a SrTiO3 (001) substrate surface. SrTiO31 has a lattice constant of 3.936 Å, while that of 2212 in the c-axis direction is 30.7Å.[1,2] Examination of the image shows that the ratio of the interplanar spacing in the SrTiO3 to that in the epilayer is in rough agreement with the ratio of the lattice constants. In addition, one can see good continuity between planes of atoms in the SrTiO3 and in the overlayer, indicating a coherent interface.
The most remarkable feature of the image shown in Figure 1 is the lack of conformity of the 2212 film to substrate surface features. In the center of the image, there is an asperity on the surface of the substrate. The bright lines in the 2212 film are thought because of the image simulations (discussed further below) to be BiO planes. In the immediate vicinity of the defect, there is a slight amount of disorder in the BiO planes, but this disorder is healed on a very short lateral scale beyond which the BiO planes are straight and unbroken. In the vertical direction, the BiO planes are continuous at only two 2212 molecular layers above the substrate. Somehow the 2212 layer is able to accomodate the distortion caused by the substrate roughness within a single unit cell. The result is that after only two monolayers of 2212 growth, the surface of the film is significantly smoother than the substrate surface. Apparently the energetics of the 2212 growth are such that it is energetically costly to disrupt the bounding BiO planes.
The healing of defects at the 2212/SrTiO3 interface is in contrast to the propagation of roughness which is often observed in other epitaxial systems like Ge/Si[12] and GaAs/AlAs.[13] In some cases the distribution of defects on the surface of epitaxial films will be a replica of defects at the film-substrate interface. The shape of substrate-surface defects can even be propagated through the entirety of thick multilayer films.[13] Propagative roughness is minimized in multilayers meant for x-ray mirror applications by using an amorphous material as one of the constituents. For example, in Nb/Si and Mo/Si multilayers, TEM studies have shown that the crystalline transition metal layers have rough surfaces, but that the surfaces of the amorphous Si layer above them are considerably flatter.[14] Presumably healing occurs in the amorphous layer because the energetic cost of distorting local bonding to accomodate interface roughness is less than in the crystalline layers. The surprising aspect of the 2212 overgrowth on SrTiO3 is that the roughness of the substrate is healed within a crystalline layer rather than an amorphous one. The image suggests that there is less of an energetic penalty associated with disrupting the stacking of the BiO and CuO layers than with bending them. This assertion is consistent with the ability to grow BSSCO-family materials with a varying number of CuO and Ca layers.
The strong tendency in the BiSrCaCuO materials to form continuous, straight BiO planes likely contributes to the success of the ALL-MBE technique with these materials. The naturally occurring layering in the BiSrCaCuO superconductors may be favorable for techniques which sequentially deposit the constituents, as ALL-MBE does. Efforts to grow the less anisotropic YBa2Cu3O7 superconductors using a similar technique have been less successful.[15]
The non-conformal growth of the 2212 may mean that coverage of lithographically patterned steps in possible superconducting microcircuits may be a problem. On the other hand, the insensitivity of the 2212 growth to the quality of the substrate surface undoubtedly means that substrate preparation techniques need be less stringent in this materials system than in many others. For example, high-quality overgrowth of Fe on Ag substrates requires painstaking surface preparation because the vertical mismatch of the Fe and Ag lattice constants causes disruption of the first three to four Fe monolayers at atomic steps on the substrate.[16] This disturbance occurs despite the good in-plane lattice match of Fe and Ag. The short range of the disruption of 2212 growth at asperities on the SrTiO3 surface is further evidence of the inherently anisotropic nature of the 2212 compound.