A process for the growth of single-crystal (epitaxial) multilayer films has been developed at the Naval Surface Warfare Center, Dahlgren Division. This process is predicated on the preparation of a compliant interfacial "template" layer of atomic dimensions that can overcome large lattice mismatches. The process can be adopted for the fabrication of integrated electro-optic sensors/receivers, and for new thin-film materials such as the III-nitrides.

Single-crystalline thin films are of technological importance in modern electro-optics (E-O) and electronics because they are the real estate upon which circuit elements, detectors, sensors, and emitters such as light-emitting diodes and diode lasers are built. These devices are usually built on wafers about 0.015-in. thick. Even so, the materials being actively used occupy only a small thickness near the surface of the substrate. The rest of the wafer is used simply to provide mechanical support. Therefore a successful thin-film technology can provide substantial savings in materials and processing costs.

An example of the Heteroepitaxial Structure is schematically illustrated in the figure. The circles represent different atoms of various sizes and in different crystalline structures. The bottom four rows represent a silicon substrate, the middle single row an interfacial "template" containing barium atoms, and those at the top a barium fluoride thin film. In this heterostructure, the silicon and the barium fluoride retain their normal crystalline structure. Strains arising from differences (15 percent) in lattice spacing between the BaFl and the Si, exaggerated by the arrows, is taken up by the barium layer at the interface. Without the intervening template layer, lattice mismatches of more than one or two percent will adversely affect the crystal structure of the thin film. The dielectric strength of the BaFl thin film is close to that of the bulk crystal.

In addition, a successful thin-film technology will allow rapid development of new electronic and E-O devices by passing over the more expensive approach of bulk crystal development. Furthermore, with a multilayer thin film, the desirable properties of each of the layers can be utilized in a single integrated package. Multifunction devices can be made during the manufacturing process, complete with packaging; assembly of discrete components into functional units can thus be bypassed completely. However, in order for the materials to perform efficiently, the films must be as defect-free as possible, so that device performance will not be degraded.

In the conventional approach to the deposition of multilayer single-crystalline thin films, the lattice spacing the distance between atoms in a crystal between different materials must be closely matched so that there is a regular transition in atomic arrangement from one material to another. Otherwise, the bonding between atoms across the interface of the components will be irregular and weak. As a result, the films can peel off, crack, or contain a large density of crystalline defects. With such an approach, multilayer films are limited to cases where the lattice mismatch is on the order of 1 or 2 percent at the maximum. This constraint of close lattice matching limits the combination of materials that can be mated together, and therefore limits the diversity of devices that can be achieved. The ability to fuse together highly lattice-mismatched materials can open up a multitude of possibilities for device engineers.

An important shortcoming in the conventional approach to heteroepitaxial growth is that the chemical interaction, i.e., the bonding between component materials, is usually ignored. Thus, an important factor in the film deposition process is not exploited to advantage. The Dahlgren group has adopted a contrary approach in which the chemical interaction is taken into account. In addition, this interaction is capitalized on further to create a structure favorable to the subsequent growth of single-crystalline films on the substrates. The basic idea is similar to the use of an "atomic glue," which can bond with a variety of materials. An important criterion for this glue, for applications in the making of multilayer crystalline films, is that it must be compliant. That means it must be easily deformable in the lateral direction so that strains resulting from the mismatch of atomic spacings will be accommodated by the glue, allowing the deposited material to adopt its natural lattice spacing.

This new method for making epitaxial films was developed by the molecular beam epitaxy (MBE) process, in conjunction with in-situ surface analysis and ex-situ film characterization methods. In the MBE process, single-crystal substrates are placed in a vacuum chamber where they are exposed to a beam (or beams) of atoms or molecules evaporated from heated sources. The thin film is formed on the substrate surface when the incident atoms coalesce into crystals. The method developed has yielded highly reproducible results. In fact, recipes have been generated for the preparation of a variety of thin-film combinations. These include barium fluoride on silicon and on gallium arsenide, lead telluride and cadmium telluride on silicon, and a gallium arsenide/barium fluoride/gallium arsenide sandwich. The lattice mismatch in these combinations goes as high as 19 percent. Mismatch of this magnitude was previously considered fatal for epitaxial growth using the conventional approach to heteroepitaxial thin films.

The technology developed here is generic, and thus has wide application potential. The multilayer films are now being used as substrate materials for the making of gallium and aluminum nitride films. This technology is being transitioned to small business to make monolithic low-cost infrared focal plane arrays for applications in surveillance and temperature/fire detection.

This work was carried out at the Naval Surface Warfare Center, Dahlgren Division, Systems Research and Technology Department, Dahlgren, VA 22448. Interested persons should contact Mary Lacey, Department Head, (540) 653-8535; fax (540) 653-4930. Inquiries concerning patent rights should be addressed to the Patent Counsel, NSWCDD, Dahlgren, VA 22448.