By Hank Hogan
Although contamination control professionals are waging a battle against AMC on many fronts, a look at three cleanroom trends reveals that new weapons are still needed in the fight to control a vicious, invisible foe
Across the myriad industries that utilize cleanrooms, chemical spills, fires, floods or other unexpected catastrophes are bound to happen. These events will undoubtedly impact processing and product yield for a period of time, from days to weeks to months; and those consequences can continue—thanks to airborne molecular contamination (AMC)—long after the “substance” has literally hit the fan.
According to Mark Camenzind, a senior technical advisor with AMC analysis experts Air Liquide-Balazs Analytical Services (Fremont, Calif.), an adverse event can put contaminants in AMC controlling carbon filters or in particle-removing HEPA filters. These could be from the escaping chemical itself or from an interaction between the chemical and the cleanroom environment.
“They can then slowly evaporate over time and keep adding some contaminant level to the cleanroom,” says Camenzind.
That's just one example of how AMCs can trigger subtle havoc with cleanroom product. Making sure that AMC-driven problems don't lead to further events requires the use of evolving contamination control technologies and techniques. A look at three cleanroom trends—minienvironments, short wavelength light sources and improved test and measurement—reveals what weapons are still needed in the fight to control AMC and what's currently being done in the trenches.
One key difference between a 300-mm wafer cleanroom and those of previous generations is the use of minienvironments and front-opening unified pods (FOUPs). Together, these shield the wafers inside from contaminants in the ballroom cleanroom.
The consensus among industry players International Sematech (Austin, Texas), IBM Corp. (Armonk, N.Y.), Intel Corp. (Santa Clara, Calif.), Infineon Technologies AG (Munich, Germany) and others is that this arrangement works well to reduce or eliminate particulate problems.
The same can't be said for AMCs.
FOUPs and minienvironments can use and do have particle filters. However, they typically don't have anything—not yet, at least—to remove molecular contamination. The air in these enclosed spaces is in intimate contact with product. Any AMC inside does not get diluted in larger volume of air found in the cleanroom. Thus, the materials in or near such enclosures can create a problem.
“Offgassing is going to be playing more of a role for floor tiles, walls and everything else,” predicts Billy Jones, cleanroom support group leader at Infineon's 200-mm wafer fab in Richmond, Va.
Jones reports that a 300-mm production facility has already experienced and corrected this problem. Floor tiles that met existing specifications for outgassing were found to be a risk when placed near the air intake of specific minienvironments. The company upgraded the tiles and eliminated the source.
There are several proposed solutions for this problem. Since FOUPs and minienvironments already have particle filters, AMC trapping carbon and other filters could be added to remove or reduce molecular contaminants.
Another approach would be to carefully balance the pressure within an enclosure with that of the outside cleanroom. Then, when the wafers are moved from one to the other, there wouldn't be a rush of air to stir up or perhaps carry in contaminants. The enclosures could also be purged with an inert and relatively AMC-free gas.
However, these approaches have their drawbacks. Purging could involve moving a great deal of dry gas through a small area, leading to a build-up of charge and electrostatic issues. There could also be a problem if ventilation faltered and the oxygen inside a cleanroom was displaced by the flow of purge gas from all the enclosures.
The adding of filters and balancing of pressure would involve maintaining and tracking the efficiency of AMC abatement devices for the 300 or so individual process tools and minienvironments in a typical wafer fab. This is in contrast to the dozen or fewer carbon filters needed in the makeup air path to decontaminate the cleanroom as a whole. Such molecular filtering could also be extended to the recirculation air path. Chris Muller, technical services manager at AMC filter maker Purafil, Inc. (Doraville, Ga.), thinks that this bigger picture approach may be the way to attack the problem.
“Where does the minienvironment get its air from but from the facility itself or directly from the outside in some cases?” asks Muller.
The dark side of light
The next trend, short-wavelength light sources in lithography, is also primarily a semiconductor issue; however, the AMC concerns here also impact telecommunications and other cleanroom industries—all that's needed is for the power density of a light source to be high enough.
Purafil's OnGuard ERM “reactivity monitor” is designed to measure the effects of different contaminants, such as acids or bases, in total, without attempting to pin each down to a specific type or source.
Current state-of-the-art semiconductor manufacturing lithography is transitioning from 248-nm laser light sources down to 193-nm sources. Plans call for the next generation to be at 157 nm. Research and development is already laying the groundwork for this to become a manufacturing reality in a few years.
Next in line are extreme ultraviolet light sources at 13.2 nm. This is actively being pursued by Sematech and other organizations. This downward path is not without its problems, one of which is molecular contamination.
Extraction's new long term sampling device, Vantara, monitors optics contamination in 193-nm and 157-nm lithography.
“There are a lot of concerns when you get down to sub-193-nm lithography with respect to AMCs. That's where it's going to be a really significant issue,” remarks Chris Long, an IBM cleanroom contamination control engineer.
The problem in this case does not involve the product directly; rather, it's centered on the optics and beam steering in the lithography tool.
According to Nigel Farrar of Cymer Inc. (San Diego, Calif.), a leading supplier of the excimer lasers and optical modules that power semiconductor lithographic tools, the company carefully selects the materials in those modules and seals them against the outside world. The tool makers, for their part, use purges and other methods to ensure that the air inside the lithographic tools is free of contaminants.
All of this is done so that the expensive and very precise optics in the system aren't rendered useless by fogging and film deposition. The problem, says Farrar, who works directly with the chief technology officer, is that the beam must pass through the tool before it strikes the wafer. Along the way, the beam has the power to initiate wanted—and unwanted—chemical reactions. The light can catalyze the deposition of materials out of what appears to be thin air. And that ability grows stronger as the wavelength grows shorter.
“The contamination problem is worse at shorter wavelengths just because they're more energetic photons,” explains Farrar.
To counter this future problem, tool makers and semiconductor manufacturers may well turn to and, in turn, improve upon existing AMC control technology. In order to enhance overall lithographic performance, for example, semiconductor manufacturers have long used AMC abatement systems on their photoresist dispense and develop equipment.
For the most part, these have consisted of filters that specifically gobble up ammonia and other amine sources that interfere with the development of chemically amplified photoresists.
Extraction Systems Inc. (Franklin, Mass.) specializes in the analysis and control of such lithographic AMCs. The company's filtration media allow it to produce products with filters in series instead of using filters in a parallel approach.
John Higley, vice president of sales and marketing at Extraction, says his company's filters are constructed in such a way as to not need a large pressure drop across the media, and that allows placement in series. That has several advantages but one of them has more to do with monitoring the situation than any active filtering.
“What that really means is you can measure in between cells instead of just measuring upstream and downstream,” notes Higley.
He added that such information enables performance to be more accurately monitored and for filter changes to be done on a cost-effective schedule.
Which leads us to the broadest trend and need in AMC control: measurement. It's a problem that confronts any cleanroom concerned with airborne molecular contaminants.
“The biggest amount of work needing to be done is in metrology equipment in terms of having real-time facility monitoring systems doing multiple locations,” says Joe O'Sullivan, microcontamination and cleanroom systems program manager at Intel's strategic facilities technology development group.
O'Sullivan would like to get real-time data that fingerprints molecular contaminants to specific types and levels. Such monitoring would be beneficial in assessing the effectiveness of AMC abatement. It could also track molecular contaminant levels in the makeup and recirculation air.
Test methods and cleanroom designations are not standardized across industries and national boundaries. There are, however, attempts underway to hammer out techniques and approaches applicable to all cleanroom industries (See “AMC standards near completion”).
Another issue is that airborne molecular contamination involves airborne chemistry and multiple possible sources of the same contaminant. The nature of the source determines which test method or instrument will work the best for that particular source. That makes the problem of tracking and identifying AMCs difficult.
There are some ways around some of these issues, however. Purafil, for example, has created what it calls “reactivity monitors.” These measure the effects of different contaminants, such as acids or bases, in total, without attempting to pin each down to a specific type or source. According to the company, the monitor has been used successfully by semiconductor companies to reveal AMC problems related to those contaminants.
Extraction Systems also makes measurement systems, as does Molecular Analytics Inc. (Sparks, Md.). Those from Extraction Systems are aimed at monitoring lithographic areas. On the other hand, Molecular Analytics' AirSentry is designed to monitor entire facilities for multiple trace contaminants. The company's IonPro is a smaller and more portable version of the same detection technology.
Beyond these technologies, however, there are other measurement schemes being considered.
Fred Lakhani, a senior member of the technical staff at International Sematech, thinks that testing ultimately has to tie into the product itself. With the right kind of test structures and methods, for example, semiconductor manufacturers will be able to start tying trace elements in the air and consumables to their impact on yield.
If done correctly, this could result in the improved AMC monitoring and ultimately enhanced AMC control. Such structures and methods do not yet exist, but plans are underway that may eventually bring them into being.
“There will be activities down the road that will generate those types of methods and structures,” says Lakhani.
Hank Hogan is a special correspondent for CleanRooms magazine based in Austin, Texas. He can be reached at hank@ hankhogan.com.
AMC standards near completion
SAN JOSE, Calif.—While particle levels for cleanroom classification are well established, the same can't be said for AMCs.
Semiconductor Equipment and Materials International (SEMI), the trade group based here, has issued the F21-95 standard that covers molecular acids, bases, condensables and dopants for the semiconductor industry. However, critics say the SEMI standard overlooks metals and is not applicable to other industries.
In turn, Working Group 8 of the Geneva-based International Organization for Standardization (ISO) 209 technical committee has been working on its own document. Allyson Hartzell, senior staff scientist in manufacturing engineering at Analog Devices Inc. (Norwood, Mass.), is the United States' delegate to the group and notes that the ISO document has taken a different course than the SEMI assessment.
“We tried to make our classification scheme as broad as possible so that not only can it include semiconductors but it can also include other industries that are affected by this, such as thin film head manufacturers, biotech or anybody that has a cleanroom and is making a product,” says Hartzell.
She adds that the proposed standard currently includes a section on references, another on terms and definitions and a third that discusses the classification scheme. This is based on the SEMI approach but is more complex in that it also deals with concentrations as well as species types or total.
The document also calls for a demonstration of compliance and has various annexes covering typical contaminating chemicals and substances, typical methods of measurement and analysis, and test methods and examples.
The standard is in review by the various national ISO committees and there are plans for a revision that takes all returned comments into account.
According to Hartzell, the review and revise process will not be over before 2004.
The Institute of Environmental Sciences and Technology (IEST; Rolling Meadows, Ill.) serves as Secretariat for the ISO 209 technical committee and currently has two working groups wrestling with AMC.
The first, Working Group 31, is focusing on the materials side, or source, of the problem. The second, Working Group 35, is looking into chemical filtration, or the removal technology for AMCs.
Both are concerned with testing. In one case, it's the testing needed to measure how much of a problem, if any, a particular material will pose due to outgassing. In the other, it's the testing to measure how well a given chemical filter actually works.
According to Steve Paddon, a mechanical engineer at Portland, Ore.-based IDC who serves as secretary for both groups, Working Group 31's document will be done this May. The second will take a bit longer.