Often, contamination problems emanate from edge contact handlers and cassette slots, he adds. "This is a common method of contamination transfer and one that has been traditionally difficult to control," Lysaght says. Regarding the film stripper, Sparks says the BEAT tool is able to determine how well the stripper is removing film at the edge of the reclaimed wafer.

Experts working in this area have long recognized the need for accurate measurements in the edge exclusion and bevel areas, Sparks notes. Several standard analytical techniques are inadequate. These techniques are total reflection x-ray fluorescence (TXRF), time-of-flight secondary ion mass spectroscopy (TOF-SIMS), vapor-phase decomposition inductively coupled mass spectrometry (VPD-ICP-MS), and direct acid-drop-decomposition inductively coupled mass spectrometry (DADD-ICP-MS).

CENTERED: The circular vacuum chuck on the BEAT tool has an alignment jig that automaticlaly locates a wafer's center, the Sematech development team says. PHOTO COURTESY OF INTERNATIONAL SEMATECH

The brainstorming team drafted John Donahue, an equipment maintenance technician at Sematech, to help build a prototype. Sparks says Gondran found enough money in the lab budget to pay for construction of the BEAT. Meanwhile, Sparks met with Gary Donahue, John's brother, to work on actual construction of the tool. Gary Donahue works at Universal Engineering, the custom design and engineering firm in Lowell, MA, that made the prototype.

The BEAT tool works by supporting the wafer vertically, bringing an extraction solution up to the chosen depth on the wafer's edge, and rotating the edge slowly through the solution at approximately eight minutes per rotation. Finally, the solution is analyzed for trace metals by ICP-MS.

Asked how difficult it was to develop the analytical method, Sparks says, "Once we decided on a solution extraction technique, we knew it would be a feasible method as long as we could create a hydrophobic surface in the area we wanted to sample."

After that point, it became a matter of "engineering the design of the jig to support and rotate the wafer, which did take some effort. Gary Donahue at Universal Engineering took our ideas and sketches and turned them into a working device. There were no off-the-shelf parts used except for the stepper motor and vacuum chuck. Gary did the design and machining of the parts, which required a substantial time investment on his part."

Developing a way to swing the alignment jig so that the sample boat could be raised up to the wafer's edge proved to be Gary Donahue's greatest engineering feat, Sparks says. This function requires a high level of repeatability because the alignment of the wafer around the center of the BEAT's vacuum chuck "is critical for precise sampling."

Sparks and crew also discovered initially that control over the wafer's rotation was not precise enough. Michael Pendley from Sematech's calibration lab corrected the problem by building a variable-speed control box for the stepper motor. Sparks also had the original sample boats redesigned in order to improve the tool's precision. Molded ultrapure PFA was inserted into the boat's PTFE housing to avoid the negative effects of a meniscus by changing the dimensions of the boats.

As Sparks, Gondran, Lysaght, and Donahue reported at the SPIE conference, experimental results show the BEAT tool performing well compared with different standard analytical techniques. Broadly, the experiment consisted of using a dip method to contaminate wafers with copper, using nine-point TXRF to analyze the result, analyzing the bevel/edge with a 4-mm included edge, and analyzing the remainder of the wafers by VPD-ICP-MS. The team then remeasured by TXRF to examine the extraction efficiency of VPD.

The BEAT tool did not cause any contamination, Sparks says. At the SPIE presentation Sparks noted possible causes for higher copper measured at the control wafer levels. "From experiments since the time of that presentation we have determined that, while there is a background interference present, we are actually measuring copper that is at the edge of the wafer.

"These are new test wafers. Where does it come from?" Sparks asks. "I'm not really sure. We've never seen copper on test wafers before with TXRF or VPD-ICP-MS. The differences between those techniques and BEAT are that we are measuring only the edge of the wafer with BEAT, and we also collect our sample from both sides of the wafer."

Sparks and Lysaght assert that the BEAT analytical tool is important for FEOL manufacturing. At Sematech in particular, potential cross-contamination looms large. The consortium has "subsets of shared process tools between FEOL and BEOL" such as metrology and lithography systems, Sparks says. "We also process a variety of materials here, like copper versus aluminum interconnects, or high-k versus silicon dioxide gates, so we are being very cautious with potential cross-contamination."

Lysaght notes that front-end processing "involves transistor production and the most critical clean process," pregate dielectric deposition. In addition, the dopant activation temperatures "are sufficiently high enough to breach surface gettering and drive surface metals contamination into the wafer and potentially into the active area of transistors."

International Sematech is not pursuing a patent on the BEAT tool, Sparks says. "If anyone wants to build one, they can contact Universal Engineering." He welcomes inquiries from interested parties, adding, "I'd be happy to talk with them about our experiences."

---

Further information on the BEAT tool is available from Chris Sparks at chris.sparks@sematech.org. Universal Engineering's Web site is www.universalengineering.org.