(425d) Lowering Material and Energy Usage During Purging Ultra-High-Purity Gas Distribution Systems

The timely dry down of ultra-high-purity (UHP) gas distribution systems that are used to deliver process gases to point-of-use has historically been a challenge in semiconductor manufacturing. New technologies with shrinking geometries and with increased sensitivities to defects and trace contaminants, particularly moisture, make the problem more challenging with every new technology cycle. The cost and competition driven need to shorten new facility qualification schedules is in direct conflict with the basic physics and time dependency of moisture elimination from new piping and distribution system components due to the facts that moisture molecules dissociatively chemisorb on the surface of gas delivery pipes producing a tightly bound layer with a high activation energy of desorption. Upset events such as a contaminated delivery or catastrophic failure of a purifier can shut down a manufacturing line for weeks while the distribution system is decontaminated and dried down. Less ultra-pure gas and energy usage in the dry down or cleanup process directly satisfies the requirements of environmentally friendly semiconductor manufacturing processes.

These challenges underscore the need for thorough characterization of the physics and chemistry involved in system dry down, as well as understanding the most effective methods for accelerating the dry down in a high volume manufacturing environment. In order to optimize the purge process with minimum purge time and/or total costs that include purge gas and energy usage, a dynamic process model that couples gas phase transport and the surface interactions between moisture molecules and the components of gas distribution systems was developed. The process model is capable of characterizing the interaction mechanisms and can be used to predict the dry down behavior of large scale industrial gas distribution systems on which it is mostly unfeasible to run tests that may disturb the regular high volume manufacturing processes. The model was verified through the combination of theoretical analysis and experiments on small scale systems. The unknown parameters, such as rate constants of adsorption and desorption and surface sorption density, in the model were estimated by fitting model predictions to experimental data. The process model is scalable and applicable to various size systems once the fluid dynamics properties of the system are fully characterized.

With the assistance of the process model, it is easy to run a parametric study on the impacts of purge gas flow rate, purity, temperature, and system pressure on the dry down process of a large industrial scale system. Furthermore, during purging and cleaning processes at system start up, system recovery from disturbance or during the normal operation of gas distribution, the process model essentially can be applied as a distribution simulator that can be used to develop optimum purging recipes for minimum purge time and/or costs for an industrial scale gas distribution system.

The results of both experiments and model simulation indicated that the cleanup process is sensitive to the inherent properties of the gas delivery pipes, such as surface roughness, exposure concentration, and purge conditions. As compared to the conventional purge method with high purity and high flow rate of purge gas without heating the system and changing the system pressure, the results showed that different combinations of purge gas purity and flow rate at different purge stages can lead to the reduction of 37% of purge cost and 49% of ultra-pure purge gas while slight increase of purge time. On a small gas distribution system, zone heating and pressure cyclic purge were experimentally proven to be two effective methods to accelerate the dry down process. For instance, an elctropolished stainless steel pipe with 65 inch in length and 1.5 inch in O.D. that was originally exposed to 200 ppb of moisture at room temperature, the experimental data showed that ultra pure purge gas usage can be dramatically decreased by 55% and 35% by zone heating and cyclic purge, respectively, as compared to conventional purge. Model simulation showed further save on purge gas usage can be achieved on large scale systems by these two methods. Zone heating is a very straight forward approach to drive moisture out from short and straight gas delivery pipes, while pressure cyclic purge is a more practical method for a complex configuration system. In a pressure cyclic purge process, the number of cycles, the time duration of each cycle, the ratio of high pressure to low pressure and the rate of pressure change are the key factors affecting the effects of cyclic purge on the cleanup process.

The model developed in this study can be applied as a generic dry down simulator of large scale UHP gas distribution systems to seek optimum purge conditions with the purpose of reducing purge gas usage and energy and/or cleanup time.