Laser cleaning is one of the new promising dry cleaning techniques considered by semiconductor companies to replace wet cleans in the near future. A dry laser cleaning tool was tested that uses an inert gas jet to remove particles lifted off by the action of a DUV excimer laser. A model was developed to simulate the cleaning process and analyze the influence of experimental parameters on laser cleaning efficiency. The best cleaning efficiencies obtained with 1.0 μm SiO2, ∼0.3 μm Si3N4, and 0.3 μm SiO2 particles deposited on Si wafers were 84±8%, 33±4%, and 12±7%, respectively. This is in qualitative agreement with theoretical calculations showing the existence of a size threshold for the removal of nonabsorbing particles by dry laser cleaning. Among the process parameters tested to optimize the process efficiency, fluence showed the highest influence on removal efficiency, before the number of laser pulses and the laser repetition rate. The use of high fluences was limited by the damaging of the wafer surface, which was not homogeneous on a macroscopic scale. The optimum number of laser pulses per unit area depended on the type of particle. The laser repetition rate had no significant influence on cleaning efficiency and can be used to reduce process time. The influence of capillary condensation on the process was demonstrated by the higher removal efficiency of 0.3 μm SiO2 and Si3N4 particles, 88±6% and 78%, respectively, upon exposure of wafers to air saturated with moisture prior to laser processing. This was attributed to the explosive evaporation of capillary condensed water, similar to the mechanism proposed for liquid assisted laser cleaning.

No abstract

No abstract

No abstract

No abstract

No abstract

The successful implementation of the new high-k dielectrics and metal gate materials for future technology nodes requires a fundamental understanding of their wet etch and cleaning properties. In this paper, wet processing key challenges for introduction of these materials in different integration schemes will be presented.

No abstract