Contamination control in liquid chemical distribution systems
Newer processes present the challenge of minimizing overall contamination, but allowing larger volumetric flow while letting the process maintain a blended suspension.
By Scott A. Sturm and Penny M. Van Sickle, Fluoroware Inc.
The challenges of processing 300-mm wafers and creating devices with 0.18-micron line widths and smaller are here. The materials composition and components comprising bulk chemical distribution systems (BCDS) can affect the contamination of the chemicals reaching the point of use. Therefore, issues surrounding particle contribution, ion contribution and system cleanability must be quantified and accessible to customers. Traditional microelectronics process chemicals serve as the baseline for much of our experience; however, relatively newer processes such as CMP slurry present the slightly different challenge of not only minimizing overall contamination, but allowing larger volumetric flow while letting the process maintain a blended suspension.
Looking beyond the microelectronics horizon, the biopharmaceutical industry also employs BCDS to deliver process chemicals and ultrapure water. This industry is not as concerned with the contaminants traditionally seen within the semiconductor world, but is very concerned about bacterial growth, system rinseability and potential cross contamination.
To date, requirements have tended to be largely customer driven (both end user and OEM). Whether one is serving the microelectronics or biopharmaceutical industry, the onus has been placed on the component supplier to quantify the raw materials and the specific components used within a larger delivery system. Data derived from studies over recent years serve as the platform as requirements become tighter and tighter.
Fluoropolymers have been the general material of choice for BCDS in the semiconductor industry. However, there are differences between the fluoropolymers traditionally used within these systems. Perfluoroalkoxy (PFA), MFA, fluorinated ethylene propylene (FEP), polytetrafluoroethylene (PTFE) and polyvinylidene fluoride (PVDF) are all used to some degree depending on the application or process. Fluoropolymers such as PFA, MFA, FEP and PVDF are melt processable and can be used to manufacture components by way of injection, extrusion or rotational molding. PTFE is not melt processable but is usually sintered into block or rod form. It is then machined with metal-based tools into its final shape.
Among these fluoropolymers, there are differences with respect to chemical resistance in processes involving sulfuric acid or some of the newer blends of photoresist strippers. Extractable levels of fluorides vary widely between the different fluoropolymers: PVDF has higher levels of fluoride compared to PFA or MFA, while some grades of PFA are better than others.1
Differences in fluoropolymers extend beyond fluoride levels to other ionic extraction levels. These levels can vary not only from material, but also from the method of manufacture. Material handling also can contribute to these differences.2 One study found a large difference in ionic extraction from piping made from the same material that was processed by two different manufacturers.3
In the biopharmaceutical industry, 316L stainless steel has been the material of choice for process chemicals and even some ultrapure water lines. A recent study indicates that specific Food and Drug Administration (FDA)-approved grades of PFA show favorable results with respect to biofilm (bacteria) adhesion and hydrophobicity. 4
Once the system material has been chosen, it is up to the component supplier to manufacture product that contributes the least amount of contamination to the overall system.
Chemical containers. Contamination of chemical from the transport container has been an increasing concern since chemical manufacturers can produce liquids that are virtually pristine and end users are beginning to specify parts per quadrillion (PPQ) chemical. Chemical can be transported in tanker trucks, rotationally molded intermediate bulk containers (IBCs), sheet-lined containers or 55-gallon drums. Container size can be an important factor for the BCDS. See Table 1 for the respective advantages of the large and small containers.
One common container size is a 55-gallon drum made of either PFA or polyethylene. As these containers are so prevalent, improving the drum is a primary concern. Improvements tend to focus on ion extraction, organic extraction and particle levels after transportation.
One study focused on evaluating extraction levels in PFA and PE drums after up to a year of storage. Containers were filled with hydrogen peroxide, hydrofluoric acid, nitric acid, sulfuric acid or hydrochloric acid. This study found that, even after six to 11 months of storage, cation increases in the chemical were quite low.5 A study of IBCs performed by Texas Instruments found similar results.6
U.S. Department of Transportation (DOT) regulations require the inclusion of ultraviolet (UV) stabilizers for polyethylene materials that are used to transport chemicals on highways within the United States. These stabilizers are required to keep the polyethylene from cross-linking and becoming embrittled when exposed to UV light from the sun. These UV stabilizers are organic substances that extract over time and can contaminate the chemical housed within the drum.
A new style of blow-molded polyethylene drum has three layers of material. The outer layer contains the DOT-required UV stabilizers, while the inner layer is pristine polyethylene. Testing of this new triple-layer drum has yielded far lower organic levels extracting into chemicals. As this drum is made with newer equipment and a more efficiently controlled process, the particle levels in the triple-layer drum are also much lower than that of older single-layer drums. See Figure 1 for a comparison.7
Valves. Within a BCDS, there are numerous valves to direct and meter flow to point-of-use sites. Unlike most other components within a system, valves have moving parts and thus have the opportunity to generate particles. The wear of valve parts, along with the valve`s overall reliability, needs to be quantified. To ensure that valves specified for BCDS will perform for an extended length of time without generating numerous particles, valves are tested throughout their design to ensure low particle release.
See Figures 2 and 3 for particle evaluations of several types of valves. For these studies, valves were automatically cycled in a liquid system. At various points during the cycle, particle release was quantified by actuating the valves 20 times in 30 to 45 seconds and then measuring particle levels using a Particle Measuring System`s HSLS M65 liquid particle counter.8, 9, 10
Figure 2 focuses on the first 5,000 valve cycles. It is during this initial cycling that valves clean up to the background concentration of the liquid used for testing. Note the low level of particle released from pinch valves even during the first 100 cycles. This low initial level is due to the easy rinseability of a valve with virtually no particle entrapment areas in the flow path. This flow path has led to the use of these valves in slurry systems, which have large flow needs and require fast and complete rinsing when changing slurry formulations.
Figure 3 focuses on the first 1,000,000 valve cycles. Data on the long-term cycling of diaphragm poppet valves are from a Sematech study performed in 1994. As shown in this graph, the diaphragm poppet valves, pinch valves and 3-way valves all survived for 1,000,000 cycles without failure or particle generation increases. The Sematech test ended after the first 1,000,000 cycles without experiencing any valve failures. Testing on 3-way valves ended after 2,140,000 cycles when an increase in particle release was observed. Testing on the pinch valves ended after 3,192,000 cycles when 1 of the 8 pinch valves in the test failed.
Tubing and piping connections. Tubing and piping connections (NPTs, flares, pipe welds, etc.) probably amount to the largest number of single components in the BCDS. Particle and chemical entrapment by tubing and piping connections within a BCDS can lead to chemical contamination at the point-of-use. Entrapment by the connector can act as a bacterial growth site in a deionized (DI) water system, increasing rinse-down time of the system and exposing the chemical to particle bursts.
Thus, the entrapment potential of connector designs must be quantified. However, quantification using particle counting has proven to be problematic. This has forced a novel approach to quantifying entrapment potential. In one study10, connectors were exposed to sulfuric acid for one hour to allow the acid to penetrate entrapment areas. The test connector was then placed in a rinse assembly attached to a conductivity meter and rinsed with DI water at 2 liters/min until the DI reached 16.7 megohm-cm (0.06 microSiemens). The connector was then impacted with a pneumatic apparatus, which vibrated the connector, forcing entrapped sulfuric acid into the flowing DI water. The acid burst would increase the DI conductivity (and reduce the resistivity). The cycle of impacts continued until no more conductivity excursions were noted.
See Table 2 for rinse times on the five connector types tested and Figure 4 for an example of the conductivity chart. As shown in Table 2, some connectors offer ample entrapment areas that can affect a fluid stream hours after initial rinsing. For some of the connectors with large entrapment areas, it was not uncommon to have the conductivity rise to 100 microS (resistivity falling to 0.01 megohm-cm) even after four hours of rinsing and impacting.
As noted, the material as well as the numerous components that comprise a BCDS will have an effect on contamination of the chemical reaching the points of use. Care must be taken in selecting the system material, proper transport and storage containers, valve configurations and line connections. As the industry moves toward even finer linewidths, these considerations assume greater importance. The expertise of a component supplier becomes broader in scope as material and manufacturing competency become of equal importance to offering application-specific products.
There is no doubt that industries will continue to refine processes, but they also will continue to mature. With maturation comes greater emphasis on overall cost of ownership and a need for more statistically driven reliability data. Component suppliers will need to develop this type of appropriate data to support their materials and finished products. In addition, the industry will devote increasingly greater attention to contamination reductions. CR
Scott A. Sturm is the microelectronics marketing manager for the critical fluid management business at Fluoroware Inc. (Chaska, MN). He holds a bachelor`s degree in communication arts from the University of Wisconsin.
Penny M. Van Sickle is a senior contamination control engineer in the technology research group at Fluoroware Inc. (Chaska, MN) and serves as the vice general chair for the 1999 Annual Technical Meeting of the IEST. She holds a bachelor`s degree in material science and engineering from the University of Minnesota and is the author or co-author of over 10 technical articles.
1. J. Goodman and S. Andrews, “Fluoride Contamination from Fluoropolymers in Semiconductor Manufacture,” Solid State Technology, July 1990.
2. J. Goodman and P. Van Sickle, “Extracting Ionic Contaminants from PFA Polymeric Materials,” Microcontamination, November 1991.
3. Ultraclean Fluorocarbon Resin Working Group, “PFA Immersion Test,” Proceedings of the UCA 6th Workshop on ULSI Ultra Clean Technology, Tokyo, pp. 167-222, July 1990.
4. F. Hyde, M. Alberg, and K. Smith, “Comparison of Fluorinated Polymers Against Stainless Steel, Glass and Polypropylene in Microbial Biofilm Adherence and Removal,” Journal of Industrial Microbiology and Biotechnology, (1997) 19, 142-149.
5. W. Buttner, K. Cardinal, T. Talasek, and P. Van Sickle, “Cation Evaluation of PFA and Polyethylene Drums in Chemical Service,” Proceedings of the Bulk Chemical Distribution Workshop, SEMICON West, 1995.
6. T. Talasek, L. Hall, et al, “Determination of Leachable Metallic Impurities from Semiconductor Packaging Material,” Proceedings of the Microcontamination Conference, 1994, pp. 362-368.
7. Fluoroware Technical Update “Particle Evaluation of 200 Liter Drums,” June 1998.
8. D. Grant, W. Kelly, “Evaluation of Valves Used for Delivering Semiconductor Process Chemicals,” SEMATECH Technology Transfer Document #94112615A-XFR, November 30, 1994.
9. P. Van Sickle, “Particle Concentration of Valves During Initial Cycling,” Proceedings of the Institute of the Environmental Sciences 1996 Annual Technical Meeting, pp. 71-75.
10. M. Alberg, “Particle Generation and Control in Tubing and Piping Connection Design,” Particles on Surfaces: Detection, Adhesion and Removal, K.L. Mittal (ed) 1995.
Figure 4. Conductivity test of particle and chemical entrapment by tubing and piping connections.
Table 1. Advantages of large and small intermediate bulk containers.