Pico-nanosecond serial dual-beam laser cleaning of Al2O3 particles on silicon wafer surfaces

Abstract

Particle contaminants on the surfaces of semiconductor devices can significantly affect their electrical performance. The advantages of picosecond laser beams for non-thermal dispersion and agglomeration of large particles are combined with the thermal expansion effect of nanosecond laser beams to remove smaller particles. A sequential method using picosecond and nanosecond lasers was used to remove Al₂O₃ particle contaminants from the surfaces of silicon wafers. Numerical simulation models were established for different laser pulse widths to quantify the laser cleaning force, enabling the prediction of the cleaning threshold for Al₂O₃ particles and the damage threshold of the silicon substrate, followed by experimental validation. The results show that the cleaning thresholds for 1-μm Al₂O₃ particles are 0.4934 J/cm² and 5.38 J/cm² using picosecond and nanosecond lasers, respectively, while the damage thresholds for the silicon substrate are 0.5702 J/cm² and 6.68 J/cm², respectively. As the laser energy density increased, particle removal from the silicon substrate surface initially increased and then decreased. In addition, sequential dual-beam cleaning outperformed single-beam laser cleaning in terms of the particle removal rate, surface roughness, and surface micromorphology. The pico-nanosecond dual-beam sequential mode provides better cleaning results than the nano-picosecond sequential mode, achieving a particle removal rate of up to 97.8%, surface roughness Sz as low as 0.027 μm, and a smoother, more uniform surface microstructure. After cleaning, the silicon wafer surface closely resembled the original silicon surface.