CHARACTERIZATION STUDY ON THE BEHAVIOR OF N-DOPED RESISTIVITY BETWEEN SINGLE CRYSTALLINE AND POLY-CRYSTALLINE WAFERS

Abstract

Process Control (SPC) methodology was used to study the impacts of the behaviour of phosphorus on the resistivity uniformity on non-patterned and patterned wafers. Plackett-Burman design was applied to determine the most significant design variables to the resistivity and resistivity uniformity. Process temperature is the major factor that determines resistivity, variations of resistivity and the thickness of the doped oxide. The variation of the sheet resistance is high when the doping temperature below than 900°C for polycrystalline silicon wafers. Single crystalline silicon has low variation of the resistivity at 850°C-950°C. Besides, the flow rate of nitrogen and oxygen play certain roles on the resistivity and resistivity uniformity of single crystalline silicon and polycrystalline silicon. Not obvious difference of growth rate for the single crystalline silicon and polycrystalline silicon wafers under the experimental conditions was found. Central composite of response surface methodology was applied to determine the optimum fabrication parameters for POCl3 doping. The performance of the n-diffusion sheet resistance (RS_NP), pull-down polysilicon sheet resistance (RS_PL), resistance per contact n+ contact chain (R_NPCC), resistance per contact polysilicon chain (R_PLCC), delta_NP and delta_PL of the JETMOS device were examined to justify the impacts on the performance of the JETMOS when the design variables were changed. A possible better recipe was found at the temperature of 925°C. Substrate resistivity of the wafers has little effect on the performance of the parameters. A confirmation run was carried out to test how well the performance after implementation of the new doping recipe. Current recipe was still considered better than the confirmation run after considering the trade-off between the performance of n-well, junction depth and pull-down resistor. Statistical analysis on the experimental results was carried out using JMP 6.0.