Silicon Bilayer Research Articles

Formation and microstructure of various thin film chromium silicide phases

The formation of the various chromium silicide phases, as predicted by the Cr-Si phase diagram for bulk materials, was studied when thin Cr/a-Si (where a-Si is amorphous silicon) bilayers were annealed in situ in a transmission electron microscope using a limited supply of a-Si. These results are compared with the silicide formed when unlimited a-Si or single-crystal silicon was used. The specimens were analysed using electron micrographs and diffraction patterns. The detailed studies of the bilayers were preceded by studies of single layers of chromium and a-Si. The chromium single layers proved to be continuous for films as thin as approximately 2 nm. The as-deposited films with thickness in excess of about 5 nm were crystalline with b.c.c. structure and comprised very small grains. A phase change occured at about 450°C from the b.c.c to a simple cubic lattice structure. The silicon single layers were completely amorphous in their as-deposited state and crystallized around 600°C. Very large grains formed. The self-supporting Cr/a-Si bilayers with a typical total thickness of about 50 nm, where the relative film thickness were adjusted to yield Cr:a-Si atomic ratios of 1:2, 1:1, 5:3 and 3:1, were prepared by sequential electron gun vacuum deposition of chromium and silicon onto photoresist-covered glass slides. In the early stages of phase formation, when both unreacted chromium and silicon were present, the CrSi 2 phase was formed at about 450°C for all of these specimens. This was the end phase for the 1:2 ratio specimen. For all the other specimens the metal- rich Cr 5Si 3 phase was next to grow at about 550°C. For the 5:3 ratio specimen this was the end phase. Upon further heating of specimens with a Cr:a-Si atomic ration of 3:1, a more chromium-rich phase of Cr 3Si was formed at about 650°C. In the 1:1 ratio specimen, however, the next and end phase observed was CrSi, also growing at about 650°C. The end phase was thus determined by the availability of chromium and silicon during the reactions and could be predicted from the phase diagram. When using an unlimited supply of a-Si (or of single-crystal silicon), instead of limiting it to the thickness necessary for the predetermined ratios mentioned above, the only phase that ever formed was CrSi 2. The grain sizes observed in the various final phase specimens were as follows: CrSi 2, 25 nm; Cr 3Si, 40 nm; Cr 5Si 3, 50 nm; CrSi, 100 nm.

Selective patterning of single-crystal GaAs/Ge structures on Si substrates by molecular beam epitaxy

Patterned single-crystal GaAs/Ge layers are grown on Si substrates by molecular beam epitaxy deposition through a bilayer mask. The ability to pattern device-quality GaAs structures on Si allows for integration of GaAs and Si devices on a monolithic chip. In this paper we describe the experimental details of a process that could be used to fabricate such circuits. The Si substrate, before growth, is patterned using a bilayer silicon nitride/silicon dioxide mask and conventional photolithographic techniques. During growth, in areas where the Si substrate is exposed, the GaAs/Ge layers grow epitaxially. After growth, a chemical etch is used that dissolves the mask, retaining the single-crystal GaAs/Ge pattern. The bilayer mask yields good pattern replication and prevents diffusion of Ga and As into the underlying Si devices, as shown by secondary ion mass spectroscopy data. The bilayer mask also aids in the lift-off process where thick GaAs/Ge layers are required and 100% lift-off yield is mandatory. Sample morphology and crystallinity are examined by scanning electron microscopy and reflection high-energy electron diffraction.

Localized electron states in a random stacking of silicon bilayers

The electronic states of randomly stacked Si bilayers are studied using a first-principles linear-combination-of-atomic-orbitals method. States at the conduction-band edge are found to be all highly localized. At the top of the valence band there is a boundary separating strongly localized states from those which are quasidelocalized. A possible experimental test is suggested.