New thinking on cleanroom air-flow modeling
08/01/2003
By Robert P. Donovan
Over the years, most semiconductor cleanrooms have been built with their air filters mounted in the ceiling—a configuration initially referred to as a vertical laminar flow (VLF) installation.
 In the airflow and velocities depicted here, the floor return gives rise to a small recirculation zone in front of the solid bench, while the sidewall return does not. |
null
This description has long since been recognized as a euphemism, and probably accurate only for a cleanroom at rest with essentially 100 percent coverage of the ceiling and floor with filters and return flow vents respectively, housing only objects that introduce no turbulence into the airflow from ceiling to floor.
 With a shortened curtain (EF), the results are reversed—the sidewall return gives rise to a recirculation flow adjacent to EF, while the floor return design does not. |
null
The introduction of people and manufacturing equipment will invariably agitate the streamlines and introduce turbulence so that the cleanroom no longer operates in this ideal VLF state.
The impossibility of achieving "ideal VLF" in a working cleanroom led the authors of recent major cleanroom standards to change the description of the flow conditions within the conventional, ceiling-fed cleanroom from VLF to unidirectional, implying that, while streamlines may no longer be laminar, the average of the air mass does continuously move from the ceiling entry ports to the floor return vents.
Even this unidirectional property is not always realized in practical working cleanrooms. People or product motion within the cleanroom space can create eddies and recirculation zones that upset unidirectional flow. Stationary objects (equipment, furniture, hoods) typically positioned for production convenience or efficiency have an impact on the cleanroom's airflow characteristics, which may degrade cleanroom performance.
Cleanrooms not only isolate contaminant-sensitive products from outside ambient air contaminants, but also use the filtered airflow to sweep out contaminants generated within the cleanroom by people, processes or operating equipment—internally generated contaminants now usually viewed as being the major sources of cleanroom contamination. Recirculation flow loops, inadvertently introduced into the cleanroom airflow, can trap some of these internally generated contaminants and prevent their removal from the cleanroom by airflow convection.
Happily, even relatively modest analytical techniques can help guide sizing, positioning and airflow velocity decisions to minimize or avoid such undesirable recirculation loops. A number of publications illustrate the usefulness of even simple two-dimensional numerical modeling of cleanroom airflow, based on the conservation of airflow mass and momentum—classical principles of fluid mechanics.1, 2
For example, Figure 1 compares modeled airflow streamlines in and adjacent to a solid bench shielded by a curtain (EF) when located in a cleanroom having: a.) floor return vents; and b.) sidewall return vents.3
For the airflow velocities and dimensions chosen here, the floor return gives rise to a small recirculation zone in front of the solid bench (HI), while the sidewall return does not. Particles entrapped in this recirculation zone will not be convected out of the cleanroom by the airflow and must be removed by some other mechanism, such as gravity, diffusion or electrostatics.
If, however, the curtain (EF) is shortened, as in Figure 2, the results are reversed. The sidewall return now gives rise to a recirculation flow adjacent to EF, while the floor return design does not.
Equipment dimensions, shape and placement, airflow-entry velocities as well as cleanroom design features and configurations can all alter cleanroom airflow paths. Basic fluid dynamic modeling can predict many of these effects, even when carried out on a PC.4
I suspect that most cleanroom owners/users don't worry much about the fluid dynamics in their cleanrooms or carry out much flow modeling; however, recently released software can accurately model more complicated and realistic scenarios.
 |
Robert P. Donovan is a process engineer assigned to the Sandia National Laboratories and a monthly columnist for CleanRooms magazine. He can be reached at: [email protected]
References
- Shanmugavelu, I., T. H. Kuehn and B. Y. H. Liu, "Numerical Simulation of Flow Fields in Cleanrooms", 1987 Proceedings of the IES, pp 298-303
- Kuehn, T. H., V. A. Marple, H. Han, D. Liu, I. Shanmugavelu and S. W. Youssef, "Comparison of Measured and Predicted Airflow Patterns in a Cleanroom", 1988 Proceedings of the IES, pp 331-336
- Yamamoto, T. "Airflow Modeling and Particle Control by Vertical Laminar Flow", Chap 18 in Particle Control for Semiconductor Manufacturing, R. P. Donovan, editor, Marcel Dekker, 1990
- Busnaina, A. A., "Modeling of Cleanrooms on the IBM Personal Computer", 1987 Proceedings of the IES, pp 292-297
See Bob Donovan present S-5 Ultrapure Water Systems: Components, Monitoring and Conservation at CleanRooms West 2003 in Anaheim, Calif., on Tuesday, September 23 at 1:00 pm. Go to cleanrooms.com. for more information on attending CleanRooms West 2003.