Few will argue that choice is a good thing. Whether it’s as personal as the physicians handling your care or as complex as the suppliers with whom you do business, choice matters.

We value choice because it gives us options; it drives us to contemplate various outcomes and new possibilities. In the filtration business, the recent round of consolidations has changed the market landscape, creating a new set of choices for microelectronics manufacturers to consider.

In this exercise, it’s worth asking: what do you want from your filtration company today?  You need great technology, yes – that’s a given. You also need a supplier capable of supporting you no matter where you’re located. From Shanghai to Hillsboro to Dresden, global reach is important.  A great filtration company – the smart choice -- will provide this and more.

At Pall Microelectronics, for example, our goal is to provide customers with the flexibility and variety they need to achieve their manufacturing goals. We deliver this flexibility and variety with Total Fluid Management, our strategy to integrate properly selected filtration and separation equipment and services into a production process that yields the highest efficiency at the lowest cost for our customer.

The power of materials engineering
At the heart of everything we offer is our proprietary membrane technology, which is based on core, in-house materials engineering know-how and expertise. We have over 200 permutations of membranes, and that list continues to expand with each customer interaction.

What this in-house capability means to customers is that Pall can respond quickly and easily to even the most obscure customer need, unlike other suppliers that require a minimum level of critical mass before they can justify the return-on-investment mathematics.

Partnering for the customers’ benefit 
Pall's Total Fluid Management program consists of a wide range of filtration products, advanced technologies and services to improve system operation and increase productivity. It also may include products, technologies and services from other suppliers, because we believe that no single company excels in every area of the microelectronics manufacturing process.

When Pall provides a complete system (whether filtration and purification or a wastewater treatment system) we select only the best components in the market. Our hands are not tied to using just those components with a Pall label. This is very different than the mindset that seeks to align a train of single-sourced components; these arrangements rarely lead to a best-in-class solution.

To this end, Pall maintains alliances and connections with a number of market leaders, including:

GE – an alliance on total water management solutions Matheson Tri-Gas – an alliance to manufacture and market gas purification technologies Asahi – an alliance on ultra- and microfiltration crossflow technologies

Our alliances enable us to deliver products and TFM solutions that are optimal in every sense of the word. We’re better prepared to meet individual customer needs than if we attempted to provide all the answers ourselves. Why? Because we have choices in our partners and in the technology options we offer to customers.

In matters of filtration and purification, having broad reach across the process stream is important, but it’s only part of the answer. In order to help customers achieve their goals, suppliers also must afford them choices in technology and products.

Many filter materials are available for use in the filtration of high-purity semiconductor gases. Pall Microelectronics recommends the use of PTFE, stainless steel, nickel and ceramics media for these applications. With so many options, users often ask, “Which filter medium is the best for my application?” To help users make better, more informed decisions, we present a look at some of the criteria for medium selection (and how each affects the filter material and performance):

Compatibility
Maximum Operating Temperature for Filter Media and Hardware
Particle Removal Efficiency
Cleanliness
Preconditioning

Compatibility. The gas to be filtered must be compatible with the filter medium, cartridge hardware and housing. Where applicable, filter cartridge hardware and sealing materials (i.e., gasket or o-ring) must also be considered. Users often neglect to give these items appropriate consideration when looking at compatibility.  

An extensive compatibility guide is available in Pall’s product catalog or on our Web site at http://www.pall.com/microe_26234.asp#26325.

General Compatibility Guidelines for Commonly Used Semiconductor Gases
Carbon Monoxide (CO)	Nickel is not compatible.
Ammonia (NH3)	Do not use an assembly that has an FEP (Fluorinated Ethylene Propylene)-encapsulated fluoroelastomer o-ring. The NH3 could permeate the o-ring and attack the fluoroelastomer, which could lead to o-ring degradation and bypass.
Fluorinated Gases	Ceramic medium is not compatible with hydrofluoric acid (HF) or tungsten hexafluoride (WF6). Pall should be consulted on the use of ceramic in other fluorinated gases.
Hardware	PTFE filter cartridges are available in a variety of hardware material with PFA (Perfluoroalkoxy) and polypropylene being the most common. Cartridge hardware is an especially important consideration when using oxygen. O2 is compatible with PFA, but will oxidize polypropylene, causing it become brittle and degrade over time.

Maximum Operating Temperature for Filter Media and Hardware. For ambient temperature gases, PTFE, stainless steel, nickel and ceramics are all acceptable.

• If the operating temperature exceeds 120°C, the sealing mechanisms become the key considerations. Above 120°C, for high-purity semiconductor gases Pall recommends an all-metal welded construction, with stainless steel or nickel media.
• In addition, if you are using a filter cartridge with polypropylene hardware, the maximum allowable temperature is 80°C.
• For elevated temperature applications, nickel and stainless can be operated as high as 426°C.

Particle Removal Efficiency. All of Pall’s UHP gas filters offer 99.9999999% reduction of particles greater than 0.003 microns for the filter’s rated flow. Please consult Pall’s product literature for specific flow conditions and efficiencies.

Cleanliness. Since a filter’s job is to remove contaminants, it should not be contributing them to a process stream. While all components contribute minimal amounts of contaminants such as moisture, oxygen and hydrocarbons during start-up, filters are of particular concern due to their high surface area. It is known that polymeric filters have a greater tendency to adsorb and desorb moisture, causing longer dry down times and longer times to recover from unexpected moisture spikes.  Ceramics are better than PTFE, but still take longer to dry down than metal filters.  Stainless steel and nickel filters are the media of choice for processes that are particularly sensitive to moisture.

Third party testing has shown Pall’s fiber metal medium to have exceptional dry-down and gas displacement characteristics. Details of the testing can be found at
http://www.pall.com/microe_2646.asp.

Preconditioning. For those users who are particularly interested in rapid system start-up, Pall offers many filters that are preconditioned as a standard or as an option. Preconditioned filters are subjected to an ultra-high purity gas purge prior to final packaging, which is performed in a clean room environment. Filters that are preconditioned will contribute < 10 ppb of moisture, total hydrocarbons and oxygen as well as < 1 particle / ft³ (or m³) to the process stream background. 

Pall’s product list at the following link helps users determine which products offer preconditioning as a standard or as an option:  http://www.pall.com/microe_product_list.asp?groupby=market&sortby=headline&category=Gas+Filtration#Gas_Filtration
  
In summary, there are many factors that need to be considered when selecting the appropriate filter material for a specific high-purity gas application. Careful consideration of these items should help users make a selection best suited for the targeted application.

Additional reference material including technical and field reports can be found at http://www.pall.com/microe_2632.asp?level0=2.

The need for product standards is ever present and necessary in the semiconductor industry, but many times the presence of an industry-recognized standard is assumed when none exists.

Standards for filter ratings are an example. Filtration is an empirical science. A variety of factors can influence filter performance in a specific application. Expected results may vary to the extent that filter-testing conditions differ from the application conditions. In practice, a filter made by a responsible and ethical manufacturer that stands behind its performance claims with documented evidence will likely satisfy the most discriminating user. It is caveat emptor, however: users must be keenly aware of claims that are vague, incomplete or even slanted to confuse or misrepresent expected performance.

In a similar fashion, there are no recognized standards for thermoplastic pressure vessels in the microelectronics industry. Although the procedures established for thermoplastic pipe are well suited for simple pressure vessels, different manufacturers can apply different criteria to determine ratings. Safety concerns clearly drive the need for conservatism in rating thermoplastic pressure vessels and reputable manufacturers abide by this requirement. In the absence of a defining standard, makers of such pressure vessels typically build differing levels of safety into their rating methods. As with filter retention ratings, it’s important for the user to ask suppliers for information about their rating methods and safety factor philosophy in order to make fair comparisons and informed buying decisions.

Design Considerations
Designing a thermoplastic pressure vessel starts by matching the desired temperature, pressure and resistance requirements of the application to the mechanical, thermal and chemical properties of the polymer selected. A variety of tests are used to confirm suitability of design, including tensile testing, creep testing and fatigue testing.

The final product service ratings of maximum operating temperature and pressure are ultimately based upon destructive product testing with samples tested to point-of-failure conditions. Minimum testing requirements include short-term burst tests, creep strain tests with measurements taken over time at each stress level and long-term creep rupture tests, including test temperature parameters that are no lower than the highest anticipated service temperature. Additional tests may be required based on the intended application, such as cyclic (fatigue) testing, impact testing and fluid compatibility.

Safety Factor
The variations in material properties and processing, including test methods, can cause variations in test results. At a minimum, the mean and standard deviation for all test series must be calculated. From the tests, a reasonable probability of failure can be made and a reasonable sample size determined to provide accurate statistical data - the larger the sample size, the higher the confidence level.

Based on actual results, a safety factor is applied to various material properties and test results are published for the thermoplastic pressure vessel or filter housing. The minimum nominal safety factor (SF_m) should be at least two (2x). Thermoplastic pressure vessels manufactured by Pall have an additional safety factor multiplier added. This compound safety factor (CSF) is the minimum nominal factor of safety multiplied by a variable safety factor.

CSF = SF_m x SF_v

The variable safety factor (SF_v) is calculated based on the statistical significance of the data collected during testing, the parameters of which take into account the severity of service and material properties. The CSF is used to account for different levels of intended service, material properties and so on.

See the sidebar for an example of how the safety factor may be used to define a Maximum Service Pressure claim. Once set, that claim may influence the buying decision.

Megaplast™ G2 Filter Housing (Chemical Filtration)

Safety Factor Development, An Example A vessel is to be designed and rated for filtering ambient temperature hazardous liquid. The number of samples tested can be determined based on the statistical confidence level and the assurance levels that are established internally by the manufacturer. Assume that, SFm=2, SFv=2, then CSF=SFm x SFv, CSF=4 The Maximum Service Pressure rating is determined as the lesser of the short-term burst pressure, tested at Tsh and Tsl, divided by the compound safety factor, CSF, where: Tsh is the highest tested service temperature, and Tsl is the lowest tested service temperature.

Decipher the Claims
Using the example as a guide, if the lesser burst pressure is 420 psi, (pounds per square inch), or 29.5 kg/cm², then the maximum service pressure rating is 420 psi divided by 4, or 105 psi (7.4 kg/cm²). A less conservative approach using a minimum safety factor of 2 would result in a service pressure rating of 210 psi (14.8 kg/cm²).

So we may have a situation where one product data sheet rates the vessel at 210 psi (14.8 kg/cm²), call this data sheet “A,” and another data sheet, call it “B,” rates the vessel at 105 psi (7.4 kg/cm²).

Which vessel would you prefer to own and have installed in your facility – the one rated by data sheet “A” with the maximum service rating of 210 psi (14.8 kg/cm²), or the one rated by data sheet “B” with a maximum service rating of 105 psi (7.4 kg/cm²)? This is, of course, a trick question. In this case, the vessels are exactly the same; the data sheets (and rating methods) are different.

Now consider a case where the data sheets describe similar (looking) pressure vessels from different manufacturers. Manufacturer “A” uses a safety factor of 2x, i.e., one-half the burst pressure. Manufacturer “B” uses a safety factor of 4x, i.e., one-quarter the burst pressure. Manufacturer “A” rates the vessel at 110 psi, (7.7 kg/cm²), and manufacturer “B” rates the vessel at 80 psi, (5.6 kg/cm²).

Which vessel do you now choose, assuming both meet other purchasing requirements (price, delivery, dimensions, etc.)?  If you assumed that both vessels are rated using the same criteria, then the vessel from Manufacturer “A” has the higher pressure rating. But because of the different safety factors used to rate the vessels, Manufacturer B’s housing actually has the higher burst pressure, 320 psi (22.5 kg/cm²), vs. A’s 220 psi (15.5 kg/cm²). If both manufacturers used the same safety factor of 4x, then the pressure ratings would be A=55 psi (3.9 kg/cm²), and B=80 psi (5.6 kg/cm²).

Both vessels may indeed meet the application requirements and perform equally well in service. The point is that if the maximum pressure rating (or Service Temperature) is one of the factors important in the decision-making process, then the buyer should be aware of the fact that rating methods may vary from manufacturer to manufacturer. These methods should be known in order to make an informed choice.

Temperature vs. Allowable Pressure

Summary
In the absence of a microelectronics industry-recognized standard for rating thermoplastic pressure vessels, manufacturers may employ their own defined rating methods. Pall uses a Plastic Pressure Vessel Service Rating Procedure based on accepted statistical methods and established standards for thermoplastic pipe. Pall maintains an internal guideline that is based on a Compound Safety Factor that is typically 4x to 5x (i.e., one-fourth to one-fifth) the minimum short-term burst pressure value used for determination of thermoplastic pressure vessel service ratings.

More information about Pall’s Megaplast G2 housing can be found at: http://www.pall.com/datasheet_microe_26410.asp.

Footnotes:

1. Miller, Irwin and Freund, John E. “Probability and Statistics For Engineers, 2^nd Ed. 1977, Prentice Hall, Inc, NJ.
2. Pall Corporation, Megaplast® G2 Filter Housing data sheet Number H59, 2004.
3. Pall Corporation, Kleen-Change® Assembly data sheet Number A98a, 2004.

Reference:

1. ASTM D1599-99e1 Standard Test Method for Resistance to Short-Time Hydraulic Failure of Plastic Pipe, Tubing and Fittings.
2. ASTM D1598 –02 Standard Test Method for Time-to-Failure of Plastic Pipe Under Constant Internal Pressure.
3. ASTM D 2837 –04 Standard Test Method for Obtaining Hydrostatic Design Basis for Thermoplastic Pipe Materials or Pressure Design Basis for Thermoplastic Pipe Product.
4. ASME Boiler and Pressure Vessel Code, Section VIII, Div. 1 & 2 (Non-mandatory Appendices) and Section X.
5. NSF Standard #53.
6. National Fluid Power Association Recommended Standard NFPA/T2.6.1 R1-1991, Second Edition, 9 May 1991.

Most companies that rely on high-quality water for manufacturing incorporate ultrafiltration (UF) modules in the main DI water loop. These modules are often the last filtration step before the water is used in a process, but they rarely are treated with the respect that their pivotal role in the ultra-pure water (UPW) system demands.

Pall’s Microza OLT series UF modules are exceptionally robust: testing reveals that even after five to six years of service Microza modules show no discernable (or only minor) alterations in strength and efficiency of the hollow fiber membrane. However, the hollow fibers are composed of polymeric membranes (polysulfone) that will, with time, be subject to fatigue as a result of slow oxidation, pressure and flow fluctuation, and frequency and type of sanitization. Hot water service will, of course, exacerbate these conditions.

Figure 1.  Cross section of Microza UF Hollow Fiber.

Although there is constant inline monitoring of particles, resistivity and TOC, it is a fallacy to assume that metrology will always detect the presence of one or two broken fibers before the consequences are apparent in manufacturing. Any problems may be subtle and difficult to pinpoint initially.  However, the effects of a couple of broken fibers should not be underestimated. First, because water will follow the path of least resistance, a disproportionate amount of unfiltered water will pass through the broken fiber into the permeate side of the module. Second, after multiple years in service, particulate matter accumulates on the upstream surface (see Figure 2).  This contamination can be swept readily downstream as the broken fiber is whipped back and forth.

Inner Membrane Surface (Downstream Side)

Outer Membrane Surface (Upstream Side)

How Do UF Modules Fail?
After roughly six years of service – and this can vary considerably depending on module operating conditions – fiber tensile strength deteriorates. The deterioration does not occur evenly throughout the hollow fiber bundle. Certain areas may be more susceptible initially than others, such as the feed and retentate ports near the potting resins. Through years of measuring fiber tensile strengths from modules used for varying lengths of time and conditions, along with those from actual broken fibers, and by comparing these measurements with freshly produced fibers, it was found that when the tensile strength dropped below 50% of the original value, the fibers were weakened significantly (see Figure 3). Any sudden system shock at this point could result in fiber breakage.

Figure 3.  Example of relative tensile strengths of fibers from modules after varying service life to that of new fibers.

Furthermore, before fibers weaken to the point where they break, membrane efficiency is often shown to be compromised. The molecular weight cut-off removal rating of a used module can be verified by challenging fibers with a 1% solution of a specific dextran mixture. When compared to the original value, efficiency levels of 10–25% less than that of a new module are frequently observed. This is attributed to a “loosening” of the membrane and often is accompanied by the finding that the module flux rate has actually increased during service life as opposed to decreasing due to the build up of surface contamination.

Another concern with keeping modules online for longer than six or seven years is the potential for extractables to be eluted from the accumulated surface contamination and passed downstream. One of the most common contaminants is ion exchange residue. Figure 4 illustrates the results of staining over eight-year-old fibers from a well-maintained UPW system with an acidic dye that binds to cationic material, such as anion exchange resin. Most notable is that the downstream (inner surface) part of the fiber is colored a deep orange, indicating that low molecular weight matter has leached through the membrane.

Figure 4.  Exposure of new and used fibers to Solar Orange Dye.

How to Determine a UF Module Change-Out Schedule
Experience has shown that under typical ambient temperature operating conditions, modules should be replaced after 5-7 years. Users can assess the condition of their modules by sending a module back to Pall Corporation for analysis for a nominal fee.  Since access to the fibers is essential, a replacement module would need to be acquired.

The module will be subjected to some or all of the tests in Table I.

Table I.  UF Module Analysis.

Further information about Pall’s Microza OLT series UF modules can be found at: http://www.pall.com/datasheet_MicroE_2801.asp.