Evaluation of Teflon filters for high-flow and low-differential pressure applications

Gas Filtration

The type of filter media can have a profound effect on outgassing rates. Additionally, the manner in which a filter is manufactured, assembled and packaged can influence the final performance

by Armando Colorado and Jim Snow

The establishment of new chemical vapor deposition, etch and rapid thermal processes, and the incorporation of 300 mm equipment and procedures into the semiconductor manufacturing environment have been accompanied by new gases and more challenging process conditions.

Click here to enlarge image

Many of these recently introduced gases have lower vapor pressures, while the flow rates of others have increased dramatically. The advent of these new fluids and gases and their concomitant problematical physical properties and usage conditions necessitate the need for new contamination control solutions. The higher flow rates and reduction in vapor pressures actually present a similar requirement for gas filters-lower pressure drop across the filter element.

Click here to enlarge image

Gas filtration requirements will not be changing for the next several years, according to the 2001 update of the International Technology Roadmap for Semiconductors. Particle cleanliness needs in bulk and specialty gases will remain at less than 0.1 and 2 particles per liter, respectively. End users can select gas filters to meet this prerequisite in a variety of sizes, end connections, flow rates and filter media.

Click here to enlarge image

Facility hookup and valve manifold box applications call for filters that can accommodate relatively high flow rates. The delivery of low vapor pressure gases and vaporized liquids requires filters that have reduced pressure drops to prevent condensation of the fluid within the membrane.

Click here to enlarge image

Manufacturing advancements over the last several years have improved the overall cleanliness and performance of gas filters. In addition to improvements in the membrane manufacturing process, the final processing and packaging can influence an out-of-package filter's performance.

Click here to enlarge image

There are advantages in using metal face-sealed endcaps as opposed to other packaging schemes with respect to out-of-package trace impurity outgassing performance. The benefits afforded by manufacturing and final packaging improvements to the overall performance-out-of-package cleanliness (moisture, oxygen, hydrocarbons, particles), pressure drop and internal surface quality-of a newly introduced Teflon filter are contrasted with those of another current high-flow Teflon filter.

Cleanliness tests
Teflon filters, manufactured under separate conditions, were evaluated with regards to product cleanliness and performance. Manufacturer specifications for the two filters are provided in Table 1.

Click here to enlarge image

Volatile and particle cleanliness tests were performed on filters that were purchased and tested immediately after removal from the packaging. Pressure drop measurements were performed after the cleanliness tests. Volatile outgassing measurements were performed at a flow rate of one standard liter per minute using an Extrel Attospec-1 atmospheric pressure ionization mass spectrometer (APIMS) with parts-per-trillion detection limits for moisture, oxygen and hydrocarbons. Particle shed measurements were conducted at 68 slpm with a Thermo-Systems Inc. (TSI; Shoreview, MN) Model 3760A condensation nuclear counter (CNC) with particle detection capabilities to more than 0.014 micron.

The appropriate industry standard test methods were followed to gauge filter performance:

  • SEMASPEC 90120393B-STD Test Method for Determination of Filter Flow Pressure Drop Curves for Gas Distribution Components
  • SEMASPEC 90120396B-STD Test Method for Determination of Total Hydrocarbon Contribution by Gas Distribution System Components
  • SEMASPEC 90120397B-STD Test Method for Determination of Moisture Contribution by Gas Distribution System Components
  • SEMASPEC 90120398B-STD Test Method for Determination of Oxygen Contribution by Gas Distribution System Components
  • SEMASPEC 90120401B-STD Test Method for Scanning Electron Microscopy (SEM) Analysis of Metallic Surface Condition for Gas Distribution System Components
  • SEMASPEC 93021511A-STD Test Method for Determination of Particle Contribution by Filters in Gas Distribution Systems
  • International Organization for Standardization (ISO; Geneva) 4288 Geometrical Products Specification (GPS) (Profile method and rules and procedures for the assessment of surface texture)

The cleanliness of a gas component can be affected not only by the cleanliness of the subassemblies and the manufacturing process, but also by the final packaging. To reinforce this point, four different packaging methods for the filters from “Manufacturer A” were evaluated and compared to a baseline as shown in Table 2.

Outgassing testing
The moisture outgassing results are shown in Table 3 and Figure 1. While a plastic cap (p) provided protection to the face seal connection and threads, there was no benefit in regards to moisture outgassing; that is, both filters demonstrated similar high moisture spikes and required relatively lengthy drydown times.

Click here to enlarge image

One of the filters packaged with a moisture desiccant (d) provided the best result for all the ten filters tested; however, the other two desiccated filters did not perform much better compared to the baseline and much worse in the oxygen outgassing test. The concern over proper activation of the desiccant, potential particle cleanliness issues within the enclosed bag and the variability among the three filters tested eliminated this packaging scheme from further consideration.

Click here to enlarge image

The hard acrylic endcap (a) afforded better outgassing performance when compared to the baseline (b) result. The drydown curves were similar; however, the time to reach less than 10 ppb was quite different-one hour compared to 2.5 hours.

Click here to enlarge image

Figure 1 shows the performance of the best performing filter from each scenario, with the exception of the desiccated group. In this case, the one filter that performed the best was considered an anomaly, so the best performing filter from the other two was plotted. The filter packaged with the metal endcaps had the lowest initial moisture spike and quickest drydown curve. The consistency among the three metal-capped filters was also the best as supported by the low standard deviation among the three.

Most of the filters achieved less than
10 ppb O2 during oxygen outgassing within ten minutes after installation with a few exceptions as shown in Table 4 and Figure 2.

Click here to enlarge image

The baseline (b) and metal face seal-capped filters (m) were the best performers of the group. Plastic-capped (p) and desiccated filters (d) actually seemed to add oxygen to the effluent compared to the baseline. Performance of the hard acrylic-capped (a) filters was again variable, with one filter performing quite well and the other requiring 1.5 hours to reach less than 10 ppb.

Click here to enlarge image

Methane and carbon dioxide emissions were also monitored, but contributed negligible a mounts from all filters investigated. Based upon the low outgassing rates of moisture and oxygen from the metal face-sealed filters and the consistency among the three filters from this group, hydrocarbon outgassing was selected as the best and most reliable for the Teflon filters packaged with the metal endcaps from “Manufacturer A.”

Competitive results
The performance of the metal endcapped filters from “Manufacturer A” was compared with filters from “Manufacturer B.” Filters from “Manufacturer B” were supplied with similar hard acrylic endcaps as previously evaluated and double-bagged. Several units of each model were tested for out-of-package cleanliness and performance characteristics.

Click here to enlarge image

The moisture and oxygen outgassing tests were performed simultaneously on the APIMS. Separate, new filters were used for the particle shed testing. Pressure drop and surface roughness measurements were subsequently recorded.

Click here to enlarge image

From the evaluation of three filters each from the two manufacturers for out-of-package moisture drydown, two aspects are evident as shown in Figure 3. First, the three filters from “Manufacturer A” demonstrated good repeatability with an average drydown time to 10 ppb of 5 minutes. Second, the average drydown time of these filters was fifty times faster than the filters from “Manufacturer B.” One of the three “Manufacturer B” filters had noticeably poorer performance, requiring more than 10 hours to attain 10 ppb, while the other two reached 10 ppb within 48 and 67 minutes.

Click here to enlarge image

The poor repeatability of the “Manufacturer B” filters is similar to the acrylic endcap results previously reported in the packaging evaluation. If the results of the poorest performing Filter B2 are discounted, the “Manufacturer A” filters still dried down 10 times faster than those from “Manufacturer B.”

Click here to enlarge image

Oxygen outgassing results were collected at the same time from filters used in the moisture testing. The performance ranking between the two sets was similar as the moisture drydown result, that is, the outgassing rate from “Manufacturer A” filters were repeatable with an average time to less than 10 ppb O2 of 3.5 minutes. The same filter from “Manufacturer B” that had poor moisture outgassing performance (B2) also exhibited extensive oxygen outgassing. While the other two “Manufacturer B” filters required only one and 10 minutes to reach 10 ppb, Filter B2 required more than 10 hours.

Three new filters from each manufacturer were evaluated for out-of-package particle cleanliness with the results shown in Figures 5 and 6 and Table 5. The filters from “Manufacturer A” were noticeably cleaner and shed fewer particles than the three from “Manufacturer B.” In fact, none of the “Manufacturer B” filters met the product specification of less than one particle per standard cubic foot.

The repeatability among the three different filters from each supplier was also evident with “Manufacturer A” filters averaging one particle and “Manufacturer B” filters averaging 128 particles during the evaluation. More than 80 percent on average of the shed particles from “Manufacturer B” came during the impaction phase of the test.

These Teflon filters are designed for high-flow and low pressure drop applications. The pressure differential across the filters were measured at three different inlet pressures up to 1,000 slpm. Three filters with 1/4-inch male face-seal connections from each manufacturer were measured and the resulting pressure drops averaged. Performance of the filters were similar to the two manufacturers as shown in Figure 7. The shapes of the pressure drop curves were almost identical. Filters from “Manufacturer B” did exhibit slightly lower pressure drops at the different inlet pressures and flow rates.

The pressure drop study was repeated with one filter from each manufacturer with 1/2-inch male face-seal connections. In this case, the pressure drops were noticeably different with the filter from “Manufacturer A” exhibiting significantly lower pressure drops, especially at the higher flow rates.

The significant difference between these filters and the ones with the 1/4-inch connections is the diameter of the tubing connections to the housing. Both the 1/4- and 1/2-inch (specified as 3/8 inches in the product literature) male face-seal filters from “Manufacturer B” all utilized 1/4-inch tube stubs (0.24-inch ID) connecting the swivel face-seal fitting to the filter housing. “Manufacturer A” had a machined housing with integral male face-seal connection with an internal diameter of 0.21 inch and 0.39 inch for the 1/4-inch and 1/2-inch versions, respectively. It was assumed that the actual filter cartridge within each housing was the same between the two different form factors, so the pressure drop was due entirely to the flow restriction from the internal tube diameter.

Evaluating the results
Filter cleanliness performance can be influenced by several different attributes. The type of filter media-metal, ceramic or polymeric-can have a profound effect on outgassing rates.1 Additionally, the manner in which a filter is manufactured, assembled and packaged can influence the final performance.

In this study, we investigated the influence of final packaging. Most filter manufacturers include pre-baking and double bagging to minimize filter installation times. Results observed in this study showed that consistent, high-purity performance could be maintained by using metal endcaps. This type of protection helped keep the filter media dry and isolated from the influence of ambient conditions.

The performance of high-flow, low-pressure drop Teflon filters from two different manufacturers was subsequently evaluated. Both filters were constructed from Teflon membrane and should therefore have similar interactions with molecular impurities.

Product literature from “Manufacturer A” indicates that the total flow area had been minimized to enhance the downstream cleanliness. No specific information was available about total available filtration area of the filters. If the total available area was larger for the filters from “Manufacturer B,” then this could be one of the reasons the outgassing level was higher.

The filters from “Manufacturer B” also came packaged with the hard, acrylic-type endcaps. As shown in the packaging investigation, these types of endcaps did not provide the cleanliness levels found with the metal endcaps.

Filters are typically flushed with high volumes of gas after manufacturing to dislodge particles before final packaging. The filters from “Manufacturer B” shed numerous particles during the particle-shed test.

Particle removal efficiency was not investigated in this study. But based upon the number of particles shed from the “Manufacturer B” filters, these do not meet the International Technology Roadmap for Semiconductors (ITRS) guidelines of less than 0.1 particle per liter (2.8 particles per cubic foot).

In conclusion, the filters from “Manufacturer A” were cleaner in all respects. Due to the lower impurity outgassing rates, manufacturing start-up times would be quicker. The higher-quality surface finish and all-Teflon cartridge will enable better specialty gas compatibility. In fact, exposure of “Manufacturer A” filters to 15 percent ozone for 350 hours showed no degradation of the membrane or corrosion of the internal housing surface.2


Armando Colorado is a senior application scientist at Mykrolis Corp. (Allen, TX) where he currently manages the atmospheric pressure ionization mass spectrometry (APIMS) lab oratory. He can be reached at [email protected].

Jim Snow is a senior researcher at IMEC in Belgium developing novel cleaning processes on single-wafer wet cleaning tools. You can e-mail him at [email protected].

Acknowledgement
The authors would like to acknowledge Troy Scoggins for the surface roughness measurements and Kareem Vakhshoori for the particle counting and differential pressure measurements.

References

  1. “Is Your Gas Filter as Clean as You Think? Evaluation of UHP Gas Filters of Differing Membrane Types for Contamination,” Colorado, A. Vakhshoori, K. 12th Annual IEEE/SEMI Advanced Semiconductor Manufacturing Conferenence (ASMC) 2001, Munich, Germany, April 2001.
  2. Personal communication, Yasushi Ohyashiki, 2001.

POST A COMMENT

Easily post a comment below using your Linkedin, Twitter, Google or Facebook account. Comments won't automatically be posted to your social media accounts unless you select to share.