Joydeep continues his talk…

Batch “fingerprinting”
Having a good batch structure in place is essential for PAT
It helps to identify opportunities by allowing online feedback of “atypical process performance” A 1% increase in yield, identified, is worth $2.7 million per year.
In the absence of a batch structure process does not know where it is–
It is difficult to determine what the “golden batch” is
If we do determine the “golden batch” almost impossible to determine deviations from it…
Safeguards revert to relying on operator knowledge of process
For realtime QA, the DCS is able to detect deviations from golden batch and contorl (administratively or feedback) the process to prevent danger of bad product/yield loss.
If the batch is bad, we used to send it to drain. Now we wait and stop the batch, check to see what the problem is, and if it can be corrected, and if not, then we dump the batch. Just doing this affords us another 1% of increase, or about $2 million per year.
Conclusion and Outlook
Utilization of PAT based on an automation concept and infrastructure
Robust application of inline analysis and controls for a WFI system
Development of an effective online analysis and reporting tool that allows
for batch ‘fingerprinting.”
Reduced manual testing
Realtime quality assurmance
increased process understanding
reduced cycle time
eventually parametric release of final product.