ATC at AMD
Fab managers depend on factory and tool performance-monitoring data. But sometimes the monthly reports don’t tell the whole story.
If most of the tools in a module are working at or above spec, there
still could be one or two laggards lurking behind the cover of the process’s high variability.
In a presentation at the ISMI Symposium on Manufacturing
Effectiveness, Thomas Quarg of AMD’s Fab 30 in Dresden talked about an innovative solution that identifies underperforming tools and helps improve tool productivity—advanced throughput control (ATC).
Integrated into AMD’s equipment-performance-monitoring framework, ATC uses a wide range of data and customized algorithms to scrutinize tool productivity.
According to Quarg, the “automated calculation of tool performance characteristics using the tool’s wafer event data increases accuracy as well as responsiveness.” He has seen big differences between static and dynamic tool models, noting “dynamic simulation increases the precision” of theoretical models.
A few case studies offered the most compelling evidence of ATC’s productivity enhancement potential. One study of a PVD tool revealed
a serious throughput crisis that recovered partially over a couple
of weeks and was eventually linked to a malfunctioning chamber pump. If ATC had been in place, the problem would have been identified within a few hours, preventing the loss of about 15% of the wafer outs from the tool.
Those are numbers any fab manager can appreciate.