Lithography dataI'm putting together a dataset to use at a lithography equipment manufacturer and need to figure out what kind of data they might be collecting. Do they typically collect data on equipment that's installed at customer sites? Would they have access to it? I was told that the typical dataset has 10-20K observations and 30 or so columns. Rows correspond to parts (which I'm guessing means IC devices, although I suppose it could also be test structures). I was also told that the columns include failure mode, part, part type, supplier (what besides photoresist? coater manufacturer? Would other folks build subassemblies inside the steppers?). I suspect they may also be collecting inline and offline data, but I'm not sure how for installed equipment. Any input would be appreciated.
Re: Lithography dataHey, Don, do you know titles/roles of the folks you'll be working with? Given the row/column descriptions you've provided, I'm wondering if they may be referring to tool components or subassemblies. For steppers, this could be anything from bearings to motors to lenses, etc. Field Service Engineers would have access to process results data (CD's, thicknesses, uniformities, overlay, etc) and electrical data on test vehicles for monitoring performance. They would also have the tool recipe and process parameters that produced those results, but it's doubtful they would have access to electrical results on production die; so I'm not sure how I would interpret failure mode, part type, and supplier from their perspective.
Re: Lithography dataLitho has several levels of granularity in their data. They'll probably have access to all of it. They'll have things like exposure, wafer topology (wafers are never flat and the stepper - the part of the tool that exposes the wafer to print the circuits has to compensate). They'll also probably have parameters from the track part of the tool which coats the wafer with a photoactive chemical called resist. These would be things like the amount of resist dispensed, spin speeds on the coater, bake times and temperatures to harden the resist before and after printing. You would also probably have developing times, etc. Most likely the data set will have Critical Dimension measurements from electron microscopes before and after the final baking process (to harden the resist before it leaves litho). These would be things like widths and aspect ratios for lines and holes printed by the mask. You can also expect the dataset to contain information about what die the CD measurements came from and what field that die was in (a reticle - semi analog to a film negative - prints a field which can have multiple die). That would be a short list of what I'd expect to see... maybe someone else can chime in with some others.
Best,
M
