Home  >  Articles

Subscribe to receive more articles like this: Print/digital | Webfeed (RSS)

As we move into the research and/or manufacturing aspects of nanotechnology, there is an ever-expanding landscape of facilities in the public and private sector being constructed to support these efforts. Watching and reviewing some of these projects, we have recognized an alarming absence of due diligence concerning the systems and components being installed to support them. Everyone understands the need for some level of controlled environment. This is especially true in the area of contamination control. What seems to be missing is that many of the installations did not take into account the need to review materials of construction for the basic components being used in these facilities.

Introduction
Perhaps, from its title, you already wrote this article off as another attempt to break down the concept of airborne molecular contamination (AMC) into a few simple discussions and clichés, and closing with the tidy “Top 10” lists of the materials to avoid and covet in cleanroom design and construction. This sort of catch-all approach makes us feel more at ease and in control. However, AMC should not be taken this lightly within a short magazine article, at design during the 60% review stage, or in the field when process tools are being installed. The discussion of AMC is not just a discussion of materials selection, but a discussion of ownership, conveyance, and interpretation of materials selection during design and through to construction. Certainly, AMC abatement techniques exist that may be applied to an existing problem in an existing facility or laboratory. However, this article discusses techniques to address AMC at the most opportune time—during the design and construction of these buildings.

Background
Nanotechnology development has arrived at an opportune time with respect to design and construction of manufacturing facilities. Designers, engineers, and constructors of semiconductor facilities have ridden the steep part of the learning curve to a near level plateau over the past 20 years as the industry has developed. Although AMC is a frequent topic of conversation these days due to shrinking chip geometries, it has been on the minds of design professionals and their clients for some time. These professionals honed their abilities in cleanroom design and particle contamination control, learned from their mistakes, and even developed rules of thumb along the way to expedite design. Nanotechnology research labs and eventual manufacturing facilities should be able to benefit from this broad knowledge base, circumvent the mistakes of the past, and hopefully not have to reinvent the wheel.

Accelerated schedules, shrinking budgets, and a short-term perspective on research opportunities (e.g. shrinking chip geometries in the near future) can shift focus away from nebulous problems such as AMC. An argument often made is the lack of concern for high yields in research. Many times, the yield of a single specimen is all that is necessary in studying it. However, when research focus shifts to an order-of-magnitude reduction with processes reaching into <100 nanometer line widths, yields are no longer considered; the lab offers no flexibility for AMC-control retrofit and the only solution is to tear out the lab and start over. Clearly, for small labs, this could be an option to reduce initial cost, but it can also delay generation changes and cost far more than one might anticipate.

In order to avoid poor choices in defining your present and future needs, it is wise to apply historical knowledge, and to make sure everyone involved in a project understands who to ask when questions arise on what you need to consider in your planning. For now, let’s review the tools we have at hand to tackle AMC.

Fundamentals
The fundamentals of AMC have been previously written about in this publication and others, and do not require laborious rehashing here. I assume that you are aware of AMC types, its sources and effects, and control and filtration possibilities. This knowledge is well documented. However, every story needs conflict. Here, the conflict is in the interpretation, distribution, and use of this knowledge.

The AMC cleanliness classification defined by SEMI F21 is deceptively simple and seemingly analogous to the ISO-14644 particle cleanliness classification. (Note: ISO–14644 has replaced FED STD 209E, which has been sunsetted by the IEST.) These similarities and simplifications hide the underlying complexity that is AMC. SEMI presents the various classes in terms of parts per trillion-molar of gas-phase particles in the clean environment. These benchmarks allow a constructed cleanroom at-rest or in-operation to be classified with the measurement of air stream or sample surfaces (e.g. witness wafers).

As with particle contamination classification, AMC classes may only be achieved and proven after the facility is constructed. Certainly, a design/ build contractor may present a package that includes clean spaces operating at ISO Class 5, for example. But these designs are based primarily on previous experience— certain filter coverage and airflow velocities typically achieve certain cleanroom classes (with proper balancing). The classification cannot be proven until cleanroom certification contractors are brought in to “bless” the space. This is one of the few analogies with AMC cleanliness classifications. That is, there exists no “recipe” of construction materials that can be combined to create a MA-100 space, per SEMI F21-1995.

But we’re getting ahead of ourselves; the place isn’t even built yet! And this methodology still somewhat oversimplifies AMC. With particle contamination, a particle is a particle is a particle. Set the flow, install filters, and balance the cleanroom = it’s clean. With AMC contamination, besides the fact that molecules are smaller (10-1000 times) than particles, cleanliness characterization involves the additional variable of chemical reaction. Different processes use different chemicals, which results in a multitude of possible chemical reactions. It is, therefore, difficult to exclude any one material for cleanroom construction because one client’s process may be sensitive while another client’s process may not.

CFD modeling can predict the path of contamination sources in lab workspaces. Figure illustrates a "plume" of contamination traveling toward a worker at a flow hood from an adjacent desk space via the air flows of a ISO Class 8 environment

The Initiative
So, we’re left with a dilemma: we can’t make a global list to fall back on, so what must we do? Delegate!

Ownership: One thing we can be almost certain of is that the clients know their processes best. They know the types of chemicals in use and the reactions that occur in their processes; they know how their existing facilities operate; and they are aware of typical product yield rates to expect and the sort of occurrences that can upset this optimal yield. It is advisable that the client appoint a person with this expertise to oversee product approval, testing, and application regarding compliance with AMC criteria. Defined in SEMI F21 as the AMC Controller, these specialists can accept or reject materials proposed for installation in the cleanroom or in areas where such materials come into contact with air serving the cleanroom. These designees are also the client’s link to outside sources of AMC information. Independent laboratories such as Balazs Analytical Services advertise a wealth of previous experience and previous experiments on file.

Conveyance: The engineer produces designs and specifications to convey the client’s requirements. Typical engineering/design groups can design cleanroom air management schemes that optimize existing AMC control technologies. Specially qualified teams have airflow modeling analysts at their disposal to simulate the designed facility in operation using computational fluid dynamics (CFD, discussed in a following section). Furthermore, these engineering groups typically have significant experience to draw on, and a database of construction material outgas tests serves to complement the client’s knowledge.

Interpretation: Product manufacturers and contractors have the obligation to furnish and install materials and finishes and to employ assembly techniques that achieve the AMC performance criteria established for the project. The contractor gains approval for procuring the cleanroom construction materials by submitting laboratory tests by an independent laboratory, confirming that the product does not outgas beyond the limits defined by the client, and does not outgas process-critical or target compounds. However—and this is also where the AMC Controller’s expertise can be invaluable—the contractor may also gain material approval without testing, if the client or engineer has been successful in using certain materials in previous experience.

It would be ideal if the client had an AMC expert on staff, but this is simply not realistic. Product manufacturers know their products and how they perform their intended functions, but they may not know off-gassing levels. The engineer cannot be expected to understand the complete range of chemical interaction possibilities between materials and a particular production process. The AMC Controller does not necessarily have to be AMC savvy either. In fact, there are probably only a select few individuals in the world today who have this complete knowledge, and who are able to convey it and apply it properly.

Although the average designer/ client cannot be expected to understand all the complex interactions of AMC and the built environment, they need remember only one question: “Will this material be used in the cleanroom or anywhere where the cleanroom air stream will be present?”

This one question gives everyone involved in the construction material procurement process the power. A “yes” to this question initiates the AMC portion of the submittal process. This could add time to project schedules and possibly material procurement lead times, but just think of the alternative. A hypothetical situation is as follows:
A client’s construction contractor hires a painting crew to prime the roof truss steel exposed in the ceiling of the cleanroom interstitial plenum. The contractor substitutes the normal epoxy primer for a new paint system with improved adhesion offered by the same manufacturer. Although the contractor has confidence in the new product, the foreman diligently informs the contractor of the substitution. The contractor has heard of AMC and roughly understands its implications, but has worked with this manufacturer’s products before without issues. The problem is discovered well into fit-up with the client’s initial test wafers failing QA/QC after leaving the lithography process. It turns out that this new paint offgasses unusually high amounts of amines which are known for attacking the lithography process in semiconductor facilities.

So, you ask, how would the paint crew know that the paint emits amines, or, for that matter, what amines are? This, of course, is not the painting crew’s job. The scenario would have played differently had the contractor been aware of the gravity of the question. Armed with this question, the contractor refers back to the AMC Controller who then can call on his or her own expertise and experience, or can call on the expertise of others.

The Toolbox
The methodology described above has shown success in minimizing off-gassing due to cleanroom materials in the past. But can other steps be taken to minimize AMC during the design stages of a project? First we must ask if we’ve minimized AMC from cleanroom materials, what are the other sources? As previous articles in this publication have observed, AMC sources include the exterior ambient air (entering through makeup-air handling units), along with processes and personnel inside the cleanroom, particularly during maintenance procedures. As an example, in this case, one solution to outside contamination may be as simple as activated carbon filtration.

Interior sources typically comprise the majority of AMC in a cleanroom in-operation. These sources are typically referred to as fugitive contamination by molecular migration. At first glance, these sources appear to be something that the facility must live with, and they may not be addressed during design. In fact, contamination due to process operations can be modeled virtually using computational fluid dynamics (CFD) modeling tools.

CFD is a mathematical modeling procedure whereby the fluid parameters of velocity, temperature, pressure, turbulence, and contaminant concentrations are calculated by solving the governing partial differential equations for fluid flow and heat transfer. These differential equations describe a three-dimensional viscous fluid flow field and cannot be solved analytically. The CFD approach is to transform the differential equations into a set of discrete algebraic equations and solve the algebraic equations by an iterative procedure. CFD has been continually validated since its inception by modeling many known fluid phenomena, and is considered a very useful tool for engineers and scientists involved in fluid flow problems across many industries. The degree of accuracy of a CFD model generally depends on the correct representation of boundary conditions, solution grid, and the level of transient characteristics. Boundary conditions are the set points at the boundaries of the model, such as wall temperature, flow characteristics at the face of a supply register, characteristics of contamination sources, the velocity profile of an approaching wind, and so on.

This tool allows for the characterization of overall cleanroom airflow patterns, pressurization, and temperature effects. CFD is also useful in nanotechnology spaces of lower ISO cleanroom classifications. Airflow patterns, eddy currents, and recirculation zones are more prominent in cleanrooms of lesser filter coverage and can be predicted reliably only with CFD. The general approach of the analysis is to improve the overall airflow patterns by minimizing recirculation zones that collect and transport contaminants; analysts can improve the space virtually to optimize air-change efficiency.

Additionally, virtual tracer gases can be included in the model to simulate release points of AMC sources. The path of this contamination can be tracked through the recirculation air stream, thus allowing capture methods to be selectively targeted. Contamination may also be modeled for intermittent functions such as maintenance.

Many nanotechnology environments are currently being designed with flexibility in mind. Initially such facility layouts are optimized for research tasks, possibly with mini- or micro- environments. As the technology is proven, however, the space may be modified for small production runs. Airflow modeling can complement the design of alternative layouts that incorporate minienvironments by optimizing each with respect to airflow, before ground is ever broken on a single layout.

The airflow model is only as accurate as the assumptions used to generate the model input. The services of a competent airflow-modeling group include not simply software operators, but CFD analysts who can properly translate system designs and configurations, or real scenarios, into accurately modeled structures in the virtual space. These professionals have significant experience also with interpretation and analysis of the output of CFD models. If you purchase the software, you get a license to learn and operate the software yourself. If you purchase the services of an airflow-modeling group, however, you acquire a professional service with years of experience in multiple applications using the CFD package as their “tool.”

Conclusion
The more traditional approaches to AMC seem to be reactionary. That is, response to molecular contamination issues occurs when the problem makes itself known either with poor yield or unexplained failures of experiments. In fact, several tools allow AMC issues to be tackled on the front-end of the design process. The design/build firm and the customer can work to properly delegate responsibilities with respect to AMC issues before the facility begins construction—improving the Clean Zone construction material selection process. Additionally, the engineer may also address AMC issues in a laboratory or facility yet in its infancy. CFD can virtually model contamination release scenarios; filtration may be strategically located and mini-environment configurations can be properly specified and utilized. While several options exist to control AMC after the facility is in operation, the good money is spent on AMC control when a new lab or facility retrofit is still on the drawing board.

References:
1 SEMI F21-1102

2 “Sizes and characteristics of airborne particles,” Fig. 18.1.1, Avallone, Baumeister, Marks’ Standard Handbook for Mechanical Engineers, 10th Ed., (1996), p. 18-9.

Steve Yellin is a Managing Director at Lockwood Greene, 14901 Quorum Drive, Suite 900, Dallas, TX 75022. He can be reached at 972-778-0239 or syellin@lg.com.

Keith Kibbee is with the Mechanical Department of CH2M HILL Industrial Design and Construction, 2020 SW 4th Ave, 3rd Floor, Portland, OR 97201. He can be reached at 503-224-6040 or keith.kibbee@idc-ch2m.com