抄録
Deposition of aerosol particles on semiconductor wafers is an important problem in manufacturing of integrated circuits. A model for predicting particle deposition in unidirectional flow cleanrooms is presented in this paper. Firstly, air velocity near a wafer surface was measured, and it was confirmed that the airflow in unidirectional flow cleanrooms is turbulent. Then the vicinity of a wall in turbulent flow is divided into turbulent zone, turbulent boundary layer, eddy diffusion sublayer and particle diffusion sublayer, and the eddy diffusivities in each zone are discussed. In the turbulent zone, the eddy diffusivity DT may be regarded as independent of the distance from the wall. Both the air velocity and the particle concentration have constant values. By Prandtl's theory, DT is proportional to the distance y from the wall in the turbulent boundary layer. The dominant machanism of particle transfer is turbulent diffusion. Within the eddy diffusion sublayer, the transfer of momentum is dominated by viscous forces, and the effect of weak turbulent fluctuations can be neglected. However, the situation is quite different for particle diffusion. Even weak fluctuations contribute significantly to particle transfer. DT is proportional to y3 in this sublayer. In the particle diffusion sublayer extremely close to the wall, the eddy fluctuations are so small that diffusion occurs principally by Brownian diffusion, i. E., within this sublayer, DT is much less than Brownian diffusivity D. According to the above-mentioned diffusivities, a model for predicting particle deposition in turbulence is proposed, including the effects of electrostatics and sedimentation. It is concluded as follows: 1) The fluctuation velocities near the wafer surface are about 10-15 % of time-averaged air velocities. 2) The deposition velocity obtained by the present model is about 10 times greater than that of the model in laminar flow for small uncharged particles. 3) The theoretical calculations by the present model agree well with experimental data.
