ShowDaily Day1 - Flipbook - Page 17
sponsored by
Spotfire
continued from page 16
Software Group, is at SEMICON West in Booth M6
in the Smart Manufacturing Pavilion (North Hall),
where it is showing visual semiconductor analytics and
AI-augmented workflows for yield engineers.
Every engineer in the fab
Yield engineers are Spotfire’s core users because they
work with the widest range of data, Hopper said. But
the platform is now used across the manufacturing
enterprise: process engineers and technology integration
teams developing and troubleshooting etch, thin film and
lithography processes, and product engineers who own
a chip’s specifications and often need to compare test
results with simulation data from EDA tools.
The customer base spans the industry’s business models.
Fabless companies tend to have more product engineers,
foundries more process and yield engineers, and IDMs
have all of them. In his September Semiconductor Digest
cover story, Hopper noted that the world’s leading flash
makers, largest foundries and dominant HBM suppliers
have been Spotfire customers for years.
The problem with powerful tools
The fab is full of specialized, highly capable software
for EDA, yield management, defect management, fault
detection and classification (FDC), metrology and more.
The trouble, Hopper said, is that the most important
problems cut across those domains.
“Almost all of the important decisions need to understand the relationship between metrology and electrical
test, or between FDC and metrology, or between defect
and sort data,” he said. “Those highly specialized,
single-purpose tools actually create barriers to crossdomain problem solving.”
Spotfire positions itself as a decision layer above those
tools, blending data of different types so engineers can
find trends, correlations and outliers, and reach root
cause faster and with more confidence. In one example
from the September article, an analog and mixed-signal
device maker traced a cluster of unusual die to a single
process step, where one of two tools was running at
several times the median of the other (FIGURE 1). The
cause was a handler chuck that had
Continued on page 18
Figure 1. A Spotfire analysis lets a process engineer drill down from SPC excursions to spatial wafer signatures,
correlating data distributions against spec limits and Western Electric Rule violations. Pulling in tool signals,
maintenance logs, and factory environmental data enables rapid root-cause analysis. Source: Spotfire.
www.semiconductordigest.com
Tuesday, October 13 | 17
