Simulation and testing of pollutant dispersion during preventive maintenance in a cleanroom

Sulfur hexafluoride (SF₆) of >99.9% purity was artificially released to simulate the emission sources in the etching-thin film area of a working cleanroom in a semiconductor fab at the rate of 492 g/h. Three mobile Fourier transform infrared spectrometers (FTIRs, detection limit: 10 ppb) were used simultaneously to measure the real time SF₆ concentrations at different locations of the cleanroom. A three-dimensional numerical model was also used to predict the unsteady gas concentration distribution and the results were compared with the experimental data. Due to high dilution of the pollutant in the cleanroom, it is found that the current gas sensors may not be sensitive enough and a better monitoring system and strategy is needed to protect workers from injury and to ensure good product yield. After comparison with the validated numerical results, the well-mixed model is found to predict the peak pollutant concentrations within a reasonable range which is 0.34–1.33 times the experimental values except when the monitored distance is very close to the release point. The well-mixed model is shown to be capable of predicting a reasonable attainable maximum concentration once a pollutant leaks in the cleanroom.