PFMEA Scope and Detection Controls

[QMSProgramManager]

I work for a semiconductor company and am currently responsible for creating/managing our FMEA standards. We are having some difficulty in determining whether detection controls (from downstream electrical test steps) should be included as a detection control in the upstream processes (i.e. Dry Etch) where we can typically only identify physical or mechanical effects. Overall manufacturing process flow is Fab, Probe, Assembly, Test...where each of those major process areas has their own PFMEA and Control Plan, and have their own multiple operations within their respective area.

Electrical effects can typically only be detected once the chip is fully manufactured...1st opportunity is Probe and then later in backend Test. The FMEA teams in the fab process do not want to accept a detection of 10 (since they cannot detect at their step & operation), and they believe their detection is the downstream process step. Note, it is important to realize that the downstream detection could be 10, 20, 30 days away from the fab process that caused the issue...thus at best, it is not an effective control for the fab process since it is so far away.

The approach we are trying to drive and enforce, is that if the detection control is from a downstream process step that is outside of the scope of the PFMEA (in this case, Fab Process FMEA), then that control should not be listed in that FMEA. Instead, the Fab PFMEA team should be identifying this issue to the Test PFMEA team or via cross-functional review, so that this electrical failure effect and detection requirement can be properly assessed in the Test PFMEA.

Our registrar has indicated that identifying downstream detection controls that are outside the scope of the "step caused" FMEA, is not acceptable...and I tend to agree.

The question:
What are the views on the correct approach, or better yet, the most effective approach?

 

[Zbigniew Huber]

Hello,

I my opinion:

Detection methods could be anything that can detect failure cause (in station) and failure mode (in station or post processing) to prevent shipment of defective product.

“In station” - in the process step where failure cause or failure mode occurred
“Post processing” – in subsequent processes in our factory (i.e. end of line functional test)

So the end of line test is valid detection method and I rank it:
Detection = 4 - if the tester locks defective product to prevent further processing
Detection = 5 – if the tester warns operator about failure, but does not lock the defective part.

Read AIAG FMEA Detection table – check “in station / at source” and “post processing” statements.

I see no reason to distinguish between test executed few minutes after manufacturing step or test executed 20 days later (as long as all products are tested/not omitted)

The purpose/function of the test process is to check product conformance with requirements and prevent shipment of defective products and this is a tool to decrease the manufacturing risk.

Limiting the detection to “in station controls” is in my opinion is an overkill and may decrease team open-minded thinking to detect defects, because:
- sometimes it is NOT possible to detect defects “in station”
- it may be very expensive to implement something “in station” and management will not agree, but it may be much cheaper to detect is later in the process.

Consider expanding FMEA scope to include all operations from receiving to shipping (and include testing) as the AIAG FMEA says.
Then the issue will be solved.

Best regards,
Zbigniew