Abstract
Plasma doping (PLAD) is increasingly applied in microelectronic device manufacture to produce high throughput, high fluence implants. In this medium energy ion scattering (MEIS) study of the PLAD process, Si(100) wafers were exposed to an As containing plasma while pulse biased negatively to 7 kV to cause (recoil) implantation and deposition of As. Quantitative MEIS depth profiling analysis in conjunction with energy spectrum simulation was applied to characterize the near-surface layer changes of the Si wafer following (i) the PLAD process, (ii) two types of chemical wet clean (oxidizing and nonoxidizing), and (iii) spike annealing in an N2 atmosphere. MEIS analysis showed that the PLAD process produced an intermixed As/Si layer, with a near-surface As content of ∼40% that decayed almost linearly to near-zero at a depth of ∼17 nm. This mixed As/Si layer was unstable in air and the initially recorded 1.2 nm thick oxide cap layer grew over a period of four months to 3.5 nm with a concurrent 25% As loss by sublimation. The application of the industry standard, oxidizing wet chemical clean removed the top As and concurrently produced a ∼14 nm thick Si oxide above the remaining implanted As profile, which matched the tail of original As implant profile. As depth profiles measured for the 7 kV PLAD process after a wet clean and spike annealing showed solid phase epitaxial regrowth of the disordered layer. A detailed comparison of the random and aligned MEIS spectra yielded depth profiles of substitutional As with a concentration in excess of 1 × 1021 As cm−3 over a depth greater than 10 nm. The retained dose of 1.35 × 1015 cm−2 represents a ∼70% increase in substitutional As compared to that recorded after a nonoxidizing clean. Such an alternative wet chemical clean, in which Si reoxidation did not occur, was applied to determine the depth of the mixed As/Si layer removed. Found to be 7 nm, the analysis indicated that the etching process ceased when the Si concentration reached 4 × 1022 cm−3. After spike annealing, part of the remaining As had segregated in a thin layer under a 1.6 nm thick surface oxide. The retained As dose in this case was 8 × 1014 cm−2, equivalent to a 1% As substitutional dopant concentration to a depth of ∼14 nm. Different substitutional As doses measured with MEIS were found to correlate closely with sheet resistance measurements, confirming that equating the substitutional As with the active As dose remains correct for these ultrashallow profiles, typically 10 nm deep.