Evaluation of rapid thermal processing systems for use in CMOS fabrication

Abstract

The role of rapid thermal processes (RTP) is evolving to meet the demands of next generation complementary metal oxide semiconductors technology. The purpose of this paper is to describe three modern RTP systems and evaluate them in terms of their process capability on control wafers, throughput, and cost of ownership. Tool A is a RTP system equipped with wafer rotation and multiple pyrometer based temperature control system to optimize within wafer uniformity during soak and spike anneals. Tool A has 1–2 °C of run to run temperature variation. Tool A also has 2–3 °C of within wafer variation for soak anneals as shown by arsenic and boron implanted control wafers. During emissivity independence testing, Tool A showed higher sheet resistivity values for low backside emissivity test wafers than high emissivity wafers. Emissivity independence testing for Tool A and Tool B (ripple pyrometry based tool) shows similar results as shown by difference of about 2 Ω/□ between high and low emissivity wafers on soak and spike anneal processes. Tool B shows about 1–2 and 7–9 °C of within wafer temperature variation for soak and spike anneal respectively. Tool C (hot liner based tool) is feasible to be used for older technologies at about 0.25 μm as it has a wider process window for temperature control and higher within wafer uniformity. This stems from the fact that the pyrometer views the hot liner and deduces wafer temperature. Run to run variation for this system ranges about 5–7 °C and within wafer temperature variation is about 8–10 °C. Tool A has the highest cost of ownership (almost four times as much as of Tool B). Tool A has two and half times the throughput as of Tool B. Tool B and Tool A are similar in process capability study as shown by sheet resistivity based control wafers. Tool A has an additive advantage of detecting “bare” silicon temperature even at low temperature region of 400–700 °C due to small operating wavelength of its pyrometer. Tool B's ripple pyrometer does not sense the “bare” silicon wafer radiation until it reaches 700 °C. Tool B has throughput of about 30–40 wafers/h for standard soak anneals. Tool C is most cost effective tool but does not meet requirements for manufacturing technologies beyond 0.25 μm. Its capacity is about 25 wafers/h for soak anneals.