Issue



Guidelines and considerations for emerging AMC control


01/01/2004







Molecular contamination has made its way into the cleanroom management vernacular; now it's time to put a smart control system into action. Here's how it's done.

BY CHRIS MULLER

It's safe to say that airborne molecular contamination (AMC) control has become fully integrated into the cleanroom environmental management requirements of high-tech manufacturing facilities.

Cleaning the outside air being used for ventilation and pressurization, removing tramp and fugitive emissions from recirculation airstreams, and cleaning process emission and exhaust streams are just some of the many applications for this vital control. And just as there are a wide variety of AMC control applications in the cleanroom, there seems to be an equal number of control options available to cleanroom managers.

But with the overwhelming number of items to be considered, how is the contamination control engineer supposed to make the proper choice for a particular application? Here are the key issues to understand and steps you need to take as you prepare to establish a successful AMC control.

AMC classifications

The first consideration in the selection of an AMC control system should be the type and quantities of AMC to be controlled. There can be dozens of contaminants present in a makeup or recirculation airstream, but there may only be relatively few that require close control.

Contaminant classifications can help to group similar contaminant types and help make the selection of the proper control system more straightforward.

SEMI Standard F-21-1102 classifies AMC in cleanrooms by their chemical properties—acids, bases, condensables, or dopants—providing a way to characterize the environment by groups of materials that could have similar effects on an exposed wafer. The purpose of this standard is to classify cleanrooms with respect to their molecular (non-particulate) contaminant levels. The classifications are defined as:

  • Acid: A corrosive material whose chemical reaction is that of an electron acceptor.
  • Base: A corrosive material whose chemical reaction is that of an electron donor.
  • Condensable: A chemical substance capable of condensation on a clean surface (excluding water).
  • Dopant: A chemical element that modifies the electrical properties of a semiconductive material.

Another way to characterize AMC is based on the potential effects to exposed personnel. In this light, we could generally classify AMC as being toxic, corrosive, irritating or odorous in order of decreasing severity.

Given recent publicity concerning personnel exposure to potentially toxic and hazardous materials in semiconductor manufacturing, AMC control systems must also be designed to protect people as well as products. Definitions related to personnel protection are:

  • Toxic: Substances that can cause damage to living tissue, impairment of the central nervous system or, in extreme cases, death;
  • Corrosive: Substances that are likely to cause deterioration or damage to the interior of a building or its contents;
  • Irritant: Substances that can cause discomfort, and potentially permanent damage, to an exposed person;
  • Odorous: Substances that primarily affect the sense of smell.

Although none of these classifications are definitive, they can at least provide the AMC control system designer with an indication of the types of contaminants that may be present and levels to which they must be controlled.

But even with a classification system in place, it doesn't always guarantee a quick and easy selection of the most effective and economical AMC control system. The classification, however, helps determine whether absolute contaminant control is required or whether a system with a fractional efficiency could be employed.

Specifying an AMC control system

Manufacturers have become much more sophisticated in their knowledge and understanding of AMC and its effects in the cleanroom. They have a better understanding of where AMC control should be applied and why, and as their knowledge of AMC-related problems has increased, so too has their expectation for an AMC control system.

In fact, some manufacturers' concerns about the proper selection of a control system have become so acute that it's being reflected in their control specifications. One manufacturer may call for a minimum of 90-percent removal of target contaminants, while another will set AMC control limits at 1 ppb or less. Still another may require that the system must last a minimum of one year between filter changeouts. As strict as these may seem, there are current specifications that require all three of the above criteria to be met.

Compounding the situation, some manufacturers insist on trying to find a "one filter fits all" solution for AMC control, demanding that a single filter should meet all of their control criteria for all contaminants of concern. But chlorine requires one type of filter, ammonia another, and organic compounds yet another. The type of filters/systems used for toxic gases would not be the same for an odor control.

So, are you looking for efficiency or service life? Do you need absolute control of one contaminant or relative control of a group of contaminants? These are only some of the questions that an AMC control system designer must consider, even though his customer may not understand the implications:

• Specification by removal efficiency. The removal efficiency of an AMC control system can be considered as the fraction of a single contaminant or group of contaminants that are removed either by physical or chemical means.

Many manufacturers are able to provide test data for their systems that show removal efficiency over time. But this testing has been performed almost exclusively under accelerated conditions using high contaminant challenge concentrations that can be up to three to four orders of magnitude higher than what would be expected under actual use conditions.

Although realistic extrapolations of filter efficiency can be provided, many manufacturers are now requiring low-level gas challenge testing for efficiency ratings of the filters (see Figure 1). This is the best way to gauge filter performance; however, testing of this sort is more complicated, takes a long time to complete, and is much more expensive to perform.


Low-level gas challenge testing is the best way to gauge filter performance, but also the most expensive.
Click here to enlarge image

One of the main problems in trying to provide an efficiency rating for a particular contaminant/filter combination is that any rating has to have a time component specified along with it. If a specification calls for a 90-percent minimum removal efficiency, the end user is expecting that to hold true over the life of the filter.

Most tests, however, are only run long enough to provide an "initial" efficiency, and the test is stopped after one hour, eight hours, 24 hours, and so on. All this does is provide information on how well the filter may work under a specific contaminant load, but does not provide or predict an indication of performance beyond the stated test period.

An examination of the efficiency curves can provide some additional indication of future performance; however, it still cannot guarantee performance under actual-use conditions. Further, a contaminant that performs well by itself can perform poorly when gas mixtures are considered.

• Specification by contaminant limits. Some manufacturers have investigated specific AMC-related process problems well enough that they have been able to set specific control levels for one or more contaminants. Common control specifications call for less than 1 ppb of ammonia in lithography bays or less than 1 ppb of chlorine or hydrogen fluoride (HF) in metallization processes.

Regardless of the ambient levels of ammonia chlorine or HF, the AMC control system is expected to keep the controlled environment at or less than 1 ppb over the service life of the system. Depending on ambient levels, this could require a minimum working filter efficiency of 50 percent, 80 percent, 95 percent, and so on. If transient high-level episodes are probable, filter efficiencies greater than 99 percent may be required to maintain a 1 ppb level coming out of the AMC control system.

• Specification of service life. Cost-of-ownership is always a major consideration when applying AMC control and can unfortunately be the primary factor when the final purchasing decision is made. Budget cycles, production schedules, capital expenditures or simply a customer's preconceived notion of how long a filter should function before it must be replaced can be the basis for specifying filter life.

Testing to determine working removal capacities of a certain AMC control product for a specific contaminant follows the same basic protocol as for efficiency testing. Using a known contaminant concentration at a specified airflow (typically, the filter's maximum rated airflow) and running to a specific efficiency endpoint, an end user can calculate the amount of gas that has been removed.

The media's removal capacity is reported as a volumetric capacity (g/cc) or as a weight percent. This can then be used to estimate media consumption rates under a given set of conditions and provide an estimate of service life for a specific filter type.

But capacities have to be determined for each contaminant of concern in order to provide a total consumption rate for a particular application. This is not practical, however, due to the uncertainty of the total contaminant load, the nature of the contaminants in question and any associated safety concerns as well as the time and costs involved.

Using these observed removal capacities are useful in providing estimates of service life, and if a conservative approach is taken, these estimates can be valid tools for the determination of filter service life (see Table 1).

Click here to enlarge image

Ultimately, filter changeout schedules will come from experience. Actual-use conditions often dictate that filters in individual AMC control systems will need to be replaced at different intervals to maintain optimum performance.

This is not the answer that manufacturers want to hear because they want to be able to set maintenance schedules and budgets; however, the reality is that one cannot predict, much less guarantee, filter service life from testing on one or two target contaminants while not knowing the total contaminant load the system will actually see.

Final considerations

Although the specification of an AMC control system is a difficult undertaking, given the proper considerations, one can be successful in specifying an effective and economical solution for most applications.

An end user can't go into this process with preconceived notions about how a particular system will perform under a given set of conditions; rather, users should determine what they ultimately want to achieve by the installation of such a system.

Consider these five points:

  1. Does AMC pose an immediate health threat to cleanroom personnel or are they primarily an odor control issue? Personnel protection requires 100-percent control of the offending contaminants whereas odor control could be achieved with a system operating at an overall efficiency of 50 percent or less. Process protection requires very high performance, but that does not mean absolute control. Many specifications today call for removal efficiencies of 90 percent and this is realistic to expect.
  2. Accept filter performance testing for what it truly is—a comparison of different systems being offered for a particular application. Efficiency or capacity test data for a single contaminant can't be used as an absolute predictor of system performance. It can provide an indication of the relative performance differences between competitive systems.
  3. Request performance data for contaminant concentrations to be generated at or near expected use conditions. If system performance and/or filter life estimates are based on accelerated test results, consider limiting gas challenge concentrations to no more than two orders of magnitude above what would be expected under actual use conditions.
  4. Filter service life should not be a part of a performance specification. Filter life estimates should be required for all systems being considered; but again, these should be used as relative comparisons and not absolute values. There is no way to provide service life estimates without having performance data for the total contaminant load and not just a handful of target contaminants. Monitoring of filter performance using direct gas monitoring, reactivity monitoring, or by using grab samplers should be started along with system start-up, and should continue as long as the AMC control system is in place.
  5. Use a multiple stage AMC control system. Just as particulate control in cleanrooms requires several stages of filtration to achieve the desired cleanliness level in the protected space, the same should be considered for an AMC control system. A "prefilter" stage should be used to remove as much of the "junk" AMC as possible. This would help protect and preserve the final filter's ability to remove those contaminants of concern with good efficiency and capacity.

A properly designed, installed and maintained AMC control system can fairly easily achieve removal efficiencies greater than 90 percent for specific target contaminants. How long a system will meet specific performance criteria depends on the average and peak values of all contaminants, which must be considered in the final design of the AMC control system.

The optimum system will contain multiple stages of filtration to provide high initial and average removal efficiencies as well as an acceptable service life. It will involve higher front-end costs, but will ultimately lead to lower operating costs and a lower likelihood of AMC making its way into the cleanroom and into critical process areas.

CHRIS MULLER is technical services manager for Purafil, Inc., Doraville, GA. He has written and spoken extensively about AMC control and reactivity monitoring in cleanroom applications, and has published more than 50 papers and articles, as well as four handbooks. He is a senior member of the IEST and is a member of working groups WG-CC012 on cleanroom environments and WG-CC035 on design considerations for AMC filtration systems. He also contributes to WG-CC008 on gas-phase absorber cells. He can be reached at: [email protected]