J Park

STEP COVERAGE PREDICTIONS USING COMBINED REACTOR SCALE AND FEATURE SCALE MODELS FOR BLANKET TUNGSTEN LPCVD

A reactor scale model (RSM) for a stagnation point, single wafer reactor for blanket tungsten LPCVD is used to calculate concentrations at the wafer surface. These concentrations and the wafer temperature, which is assumed to be measurable, are needed to determine the local tungsten deposition rate on the wafer and local film conformality (step coverage) in features on patterned wafers. Two feature scale models (FSMs) are used to determine step coverages in infinite trenches which have rectangular initial cross sections and an aspect ratio of five, as a function of reactor operating conditions; 1. a continuum-like diffusion-reaction model (DRM) for simultaneous Knudsen diffusion and heterogeneous surface reactions, and 2. a flux based model which includes ballistic transport of molecules and heterogeneous surface reactions (BTRM).|The RSM establishes “boundary conditions” for the feature scale models, by providing the flux of each species to the local wafer surface. Step coverages predicted using the FSMs...

Stacking fault pyramid formation and energetics in silicon‐on‐insulator material formed by multiple cycles of oxygen implantation and annealing

The defect microstructure of silicon‐on‐insulator wafers produced by multiple cycles of oxygen implantation and annealing was studied with transmission electron microscopy. The dominant defects are stacking fault pyramids (SFPs), 30–100 nm wide, located at the upper buried oxide interface at a density of ∼106 cm−2. The defects are produced by the expansion and interaction of narrow stacking fault (NSF) ribbons pinned to residual precipitates in the top silicon layer. Consideration of the energetics of the transformation from a collection of four NSF ribbons to a single SFP indicates that the reaction is energetically favorable below a critical NSF length. Thus small defects are stable as SFPs while large defects are stable as NSF ribbons.

Defect formation mechanism causing increasing defect density during decreasing implant dose in low-dose SIMOX

Silicon-on-insulator material synthesized by oxygen implantation (SIMOX) is a leading candidate for advanced large scale integrated circuit applications due to thickness uniformity and moderate defect density. In the past few years, there has been a significant reduction of the defect density by optimizing processing conditions. Studies on defect formation mechanisms may suggest further modification of the processing conditions for both production cost and material quality. Recently, it was shown that through-thickness defects (TTDs) in high-dose SIMOX originated from as-implanted defects, dislocation half-loops (DHLs). On the other hand, a high density (/spl sim/10/sup 8/ cm/sup -2/) of defects in very low dose (0.25/spl times/10/sup 18/ cm/sup -2/) SIMOX has been observed by Nakashirna et al. (1993), but the origin of these defects has not been understood. In this paper we report on the effect of implant dose on defect formation mechanisms, and propose a defect formation mechanism in the very low-dose regime for the first time. It was found that the stacking faults generated during the initial stage of annealing are the origin of TTDs in the low-dose (<0.5/spl times/10/sup 18/ cm/sup -2/) regime, while as-implanted defects (DHL) are the origin of TTDs in the high-dose regime (>1.3/spl times/10/sup 18/ cm/sup -2/). Several approaches of process modification have been suggested for economical production of low-defect-density SIMOX.

Effect of single vs. multiple implant processing on defect types and densities in SIMOX

In this paper we describe the origin and characteristics of the defect structures in contemporary SIMOX and show how their densities are controlled by the processing method and conditions.<<ETX>>