Nanoscale Domain Control in Multiferroic BiFeO3 Thin Films

With an ever-expanding demand for data storage, transducers, and microelectromechanical (MEMS) systems applications, materials with superior ferroelectric and piezoelectric responses are of great interest. The lead zirconate titanate (PZT) family of materials has served as the cornerstone for such applications up until now. A critical drawback of this material, however, is the presence of lead and the recent concerns about the toxicity of lead-containing devices. Recently, the lead-free ferroelectric BiFeO3 (BFO) has attracted a great deal of attention because of its superior thin-film ferroelectric properties, which are comparable to those of the tetragonal, Ti-rich PZT system; therefore, BFO provides an alternate choice as a “green” ferro/piezoelectric material. Another advantage of BFO is its high ferroelectric Curie temperature (Tc = 850 °C in single crystals), [3,4] which enables it to be used reliably at high temperatures. The ferroelectric domain structure of epitaxial BFO films are typically discussed in the context of the crystallographic model of Kubel and Schmid; however, by suppressing other structural variants in BFO, we can obtain periodic domain structures that may open additional application opportunities for this material. Ferroelectrics with periodic domain structures are of great interest for applications in photonic devices and nanolithography. Such a periodic polarization could be obtained by applying an external electric field while utilizing lithographically defined electrodes or by a direct writing process. To obtain sub-micrometer feature sizes, however, domain engineering using a scanning force microscope with an appropriate bias voltage must be used to fabricate the patterned domain structures. Unfortunately, this method works only on small areas and is limited by its slow scanning rate. Theoretical models predict the feasibility of controlling the domain architecture in thin films through suitable control over the heteroepitaxial constraints. In the case of BFO thin films, we have found that such a control is indeed possible, mainly through control over the growth of the underlying SrRuO3 electrode. Using this approach, we demonstrate the growth of highly ordered 1D ferroelectric domains in 120 nm thick BFO films. On the (001)C perovskite surface there are eight possible ferroelectric polarization directions corresponding to four structural variants of the rhombohedral ferroelectric thin film. (For simplicity, the c and o subscripts refer to the pseudocubic structures for BFO and orthorhombic structures of SrRuO3 (SRO) and DyScO3(110)O (DSO), respectively.) Domain patterns can develop with either {100}C or {101}C boundaries for (001)C-oriented rhombohedral films. [12] In both cases, the individual domains in the patterns are energetically degenerate and thus equal-width stripe patterns are theoretically predicted. When the spontaneous polarization is included in the analysis, the {100}C boundary patterns have no normal component of the net polarization, whereas the {101}C boundary patterns correspond to the fully poled state. The formation of domain patterns leads to the release of elastic energy at the expense of increased interfacial energy associated with the domain boundaries. Therefore, four possible polarization variants still exist when one examines large areas of the sample. If control over the ferroelectric domain structure is desired, one has to recourse to other approaches. In our work, we have used the constraints imposed by heteroepitaxy as well as filmgrowth mechanisms, shown in Figure 1a, to create long-range order in the domain structure of BFO. First, we use the fact that on the (110)O surface the DSO lattice is extremely closely matched to that of SRO. Further, the small structural anisotropy in DSO is used to pin the structure of the SRO layer C O M M U N IC A IO N S