Its not the software its ZAP!

01/01/1999

It`s not the software, it`s . . . ZAP!

There is growing evidence that as many as 75% of system lockups and other glitches attributed to "software bugs," particularly for mature systems used in wafer fabrication, are caused by static discharge in a wafer fab. What is even more significant, a lockup at a given system may be caused by an electrostatic discharge (ESD) at a system "on the other side of the room."

According to Larry B. Levit, director of technology development at Ion Systems, "An ESD event makes a spark creating electromagnetic interference (EMI) that propagates through the air like a radio wave and can make it to a system`s microprocessors. This ESD-caused EMI can also find its way into the fab`s lighting, for example, where it travels unattenuated as a conducted emission through wires and into power boxes. It causes problems in other systems in the fab that may not be in close proximity to the site of the ESD event. With today`s excellent high frequency PC-board technology, these signals go everywhere."

One would first question the lack of correlation between the relatively high frequency of ESD events and lower frequency (and randomness) of system lockups. "It`s not just the occurrence of an ESD event. The microprocessor has to be exposed to the propagated emission, either radiation or conduction, at the exact nanosecond that it`s doing something critical, like an instruction fetch," say Levit. "Once in a while, the timing is perfect, however, and the resulting lockup is attributed to `software bugs,`" he says.

Similarly, Levit and others at Ion Systems have documented the role ESD has on mask defects. "Our research has shown that whenever reticles or their cassettes are handled, the reticles become electrically charged. When discharges occur, they literally melt chrome lines," says Levit. He explains that until linewidths dropped below 0.5-?m, this problem was difficult to detect because the actual damaging events were few and far between. But the effect over time adds up significantly. In addition, the subtlety of this problem is becoming more pronounced as linewidths get smaller.

Levit revealed that tests at Macronix in Hsinchu, Taiwan, placed an ionizer in the load-unload chamber of a KLA reticle-inspection system. Measured ESD events that caused reticle damage dropped from 7/month to 3/month with the ionizer on, a five standard deviation effect. "Even though this is a dramatic result, it was done with only one tool. We believe that the remaining lockups represent ESD events happening on other tools," he says.

Research done by Ion Systems is so convincing that the company is guaranteeing, with its properly installed ionization systems, a 50% reduction in reticle repair cost or a 5% increase in photolithography throughput, the latter coming from a reduction in downtime, interrupts, and mean time between assists. "We believe we can make these tools run more efficiently in fabs," says Levit.

Ion Systems recently donated ionization equipment - a series of ceiling ionizers and bar units for equipment minienvironments - to SEMATECH. These will be used in the consortium`s deep-UV lithography program to further document the effectiveness of ESD control on masks. Dan Holiday, lithography engineering manager at SEMATECH, says, "Ionization increases man-ufacturing efficiency and will help us ramp up for production DUV processing." - P.B.