STP1340: Silicon Lifetime Degradation as a Function of Wet Chemical Cleans and Chemical Purity

    Linn, J
    Harris Semiconductor, Palm Bay, FL

    Rouse, G
    Harris Semiconductor, Palm Bay, FL

    Wereb, W
    Harris Semiconductor, Palm Bay, FL

    Leggett, R
    Harris Semiconductor, Palm Bay, FL

    Slasor, S
    Harris Semiconductor, Palm Bay, FL

    Lowry, R
    Harris Semiconductor, Palm Bay, FL

    Cameron, R
    Harris Semiconductor, Palm Bay, FL

    Canfield, R
    Harris Semiconductor, Palm Bay, FL

    Pages: 11    Published: Jan 1998


    Abstract

    Diffusion length data from silicon wafers exposed to various standard semiconductor process cleans and wet cleaning process tools will be presented. Data collected with an ElectroLYtic Metal Analysis Tool, or ELYMAT, will show the effects on silicon lifetimes of SC1, SC2 and SC1+SC2 combined cleans, as well as the cumulative effects which result from multiple cleaning cycles. Results from a variety of tools widely used for wafer cleaning will be presented. Data indicate that SC1 cleaning alone prior to thermal treatment can lead to extensive lifetime degradation. The data also suggest that lifetimes decrease as the number of SC1 + SC2 cleaning cycles increases. Along with the effects on silicon lifetimes of the standard wet cleans, number of clean cycles and specific tools will be data comparing different purity grades of chemicals. Higher purity grades of process chemicals are shown to favorably affect lifetimes in many but not all cleaning chemistries. ELYMAT analysis can be used to help determine which chemical grades are best suited for specific device processing requirements and to justify the added costs of ultra pure chemicals for processing certain product types which benefit from high silicon lifetimes.

    Keywords:

    silicon lifetime, ELYMAT, wet chemical process cleaning, process chemical purity


    Paper ID: STP15712S

    Committee/Subcommittee: F01.10

    DOI: 10.1520/STP15712S


    CrossRef ASTM International is a member of CrossRef.