Is Gage R&R with measurement data possible in a Semiconductor Fab?

[rigolef]

In our semiconductor Fab we're using a CD-SEM to measure some critical dimensions (down to .35µm). Up to now we did a gage R&R on a yearly basis on some specific products. The problem is, that this tool is used to measure a lot of different products, which all have their specific problems. Next to that, every product is measured several times during the process flow, with every time a different layer underneath. Because it's also time consuming for the operators, we can't do a gage R&R for all the products and for all the steps.
Is anyone aware of a method to use the process measurements to verify the reproducibility and repeatability of the tool ?
We have a standard wafer, which we measure on a weekly basis, but again this doesn't tell us everything about the tool.

Thanks.

 

[vpandit]

Re: Gage R&R with measurement data possible? Semiconductor Fab

Hi ,
I will attempt to provide an answer as I am also currently trying to work on a similar process for doing a GR&R exercise for our SEM's that measure CD. In your situation since you have multiple products I would recommend doing separate GR&R exercises(one each) for several products with a range of products that may cover the entire range of measurement that the SEM is being used currently. This is time consuming but at least if done once can give you a more accurate assessment.
Here are some general guidelines to do a GR&R
· Step 1 :Calibrate System/Gauge to be studied.
· Step2: Establish or ensure the stability of the system is acceptable.
· Step 3: ensure Study is being conducted in actual Measurement environment
· Step 3 : Design GR&R study to include Trained operators
· Step 4 : Ensure sample of parts being used are representing measurement range the gauge or system will be used for measurement.
· Step 5: Ensure all measurement data can be free from recording errors.
· Step 6 : Ensure samples are all requiring the same measurement Tolerance/Specification.( preferably of a single product)

· Step 7:Number of preferred samples 30 per study
· Step 8:sue Historic St Deviation if available from larger sample sizes.
· Step 9: Ensure randomized runs of partsduring measurement for the study .
· Step10 : Conduct study and Use Minitab for evaluation /analysis of the data.