Analytical Techniques for Microelectronic Device Development and Failure Mode Diagnostics

We report on the onset of reverse bias diode leakage due to CVD selective tungsten deposition on (a) progressively shallow n+(As)/p diodes made by implantation into silicon and (b) n+(As or P)/p or p+(BF2)/n diodes made by implantation into CoSi2. Tungsten, nominally 1000Angstrom thick, was deposited in a hot wall tubular reactor using either a one-step or a two-step deposition sequence at 290 degreesC. N+/p junctions, made by implanting arsenic into silicon, show an increase in leakage population after tungsten deposition, as the junction depth decreases for both the one-step and two-step deposition processes. The one-step process was found to be more applicable to shallow n+ (As)/p junctions than the two-step process. The junction depth at which a majority of the diodes showed high leakage was approximately equal 1500Angstrom for the one-step process compared to approximately equal 2500Angstrom for the two-step process. Cobalt siliciding the junctions does not alter the conclusions obtained from non-silicided junctions since n+(As)/p junctions which are approximately equal 1000Angstrom below an approximately equal 800Angstrom thick CoSi2 also exhibit junction leakage after tungsten deposition with either the one-step or the two-step processes. In contrast to arsenic, shallow (is equal to or less than 1000Angstrom below CoSi2) junctions made by outdiffusion of phosphorus or boron (BF2) implanted into CoSi2 do not show significant changes in leakage population after tungsten deposition. Junction leakage measurements, and their temperature dependence, will be correlated with transmission electron micrographs.