Real Time Process Monitoring
A common approach to controlling contamination in high technologymanufacturing cleanrooms is the
continuous monitoring of particles. Either dedicated, discrete,“real-time” particle counters or a multi-port
pneumatic manifold system to sample multiple locations throughoutthe cleanroom can be used to accomplish
this goal.
These approaches will give a good indication of overall particlecontamination in the cleanroom but may be
difficult to correlate to product yields. As manufacturing moves togreater levels of automation, products are
becoming more and more isolated from the general cleanroom.Cleanroom particle counts, though important,
may not have correlation with product failure, so this data isbecoming less critical. The future is with process
monitoring. Obtaining the sample close to the product is the key.Close sampling allows for direct correlation to
hard defects, yield hits, and particle contamination.
The idea is to bring the monitoring location as close to theproduct as possible. It is often impossible to get
close because of physical limitations such as robotic movement,isolation technology, or human limitations.
Some different techniques to monitor effectively within aminienvironment will be discussed in a different paper.
In addition, when monitoring processes, the critical nature ofcontamination requires a fast response to events.
A few particles can have such a detrimental effect on yields thatit is not prudent to wait for human intervention.
How long does it take for a person to (a) notice an event, and (b)take corrective action? If individuals are left
to decide, then the human factor will come into play. Productionsupervisors may not allow a tool or line to
be shut down when contamination occurs because they feel pressureto keep throughput high. Conversely,
Quality Assurance may push to have immediate action taken. Theexample below illustrates how one company
harnesses hard data to determine shutdowns automatically.
A high tech manufacturer implemented a real time monitoring systemon its production lines. Its production staff
is able to track each product as it flows through the productcycle. At the same time, it is possible to see the
real time particle contamination at each tool. Should a defectoccur on their products, the first step is to analyze
the data for the production run. Looking back to the real time datawould show which process contributed to the
contamination. A spike would be seen at one tool or another.
Analysis was done on the data and compared to which products hadhard defects. A trigger limit was
determined based on the optimal yield. This goes far beyondtraditional classification limit settings based on
ISO 14464-1 or Federal Standard 209E cleanroom standards. Theseindustry standards are useful for the
design and construction of the general cleanroom but should not beused to setup limits for specific processes.
In the real world, each process and each product may requiredifferent tolerances for airborne particles.
Industry standards may serve as guides to initial settings but thedata—and ultimately the acceptable yield—
are the true guides to a final trigger limit. In the case at hand,production staff was able to apply Six Sigma
rules to establish optimal limits based on well-founded empiricaldata.
Their next step was to determine the time and cost of a productbuild cycle. Production staff determined that
they wanted to do an automatic shutdown of certain tools whencontamination limits were exceeded. This
avoided human intervention or denial of events. The real timemonitoring system was connected to their
manufacturing system, which tracked all parts throughout thefacility. Because a typical sample time for a
by Clark Anderson - VP of Applications Engineering-Asia Pacific,Lighthouse Worldwide Solutions
Lighthouse Worldwide Solutions 1
Clark Anderson graduated from Arizona State University with a BSEE.He has been involved in setting up contamination control systems atmajor
manufacturers since 1993 in Malaysia, Singapore, Thailand, Taiwan,Korea, Japan, and the USA. Currently, he is the Vice President ofApplications
Engineering for the Asia Pacific region for Lighthouse WorldwideSolutions (www.golighthouse.com). He hasconducted contamination control seminars
at Singapore Polytechnic and The Asian Institute of Technology inBangkok Thailand. He can be reached at clarka@golighthouse.com or(510) 438-0500
Lighthouse Worldwide Solutions 2
particle counter is anywhere from 1 to 10 minutes, a data pointrepresents the sum of all contamination during
that time. Thus, the real time system enabled this manufacturer toidentify and flag all parts produced during
the 10 minutes immediately preceding the trigger event underanalysis.
This process has enabled a drastic reduction in the Defective PartsPer Million. Many factors contributed
to the overall reduction, but the most critical factor wasimplementation of real time process monitoring of
contamination.