Learn tsmap
Not every wafer is a neat circle of positioned dies with one result each. This lesson covers the awkward cases: wafers that are mostly empty, wafers with no positions, dies probed more than once, and data the tester recorded wrongly.
A multi-project wafer shares one wafer between several products. Each reticle holds only a few of one product's dies, so the map is mostly empty and every die is tiny. This is MPW-LOT-08.
MPW-LOT-08
tsmap can tidy this up for you. When the occupied columns and rows repeat at a regular pitch, Overlays offers Compact layout and Layout diagnostics at the top.
In the compact layout, empty rows and columns are removed and each group of dies is outlined. Every die is still counted, so yield and findings do not change, and a die's tooltip keeps its original coordinates.
If a layout is not recognised as you expect, this window shows what the detector saw: counts and scores only, with no positions, bins or wafer names, so nothing about your data leaves with it. Use Copy or Save as file, then send the result to whoever supports tsmap.
In the lot COORDLESS-LOT-01, W01 has an X/Y position for every die. W02 is mixed: some dies have results but no X/Y position. W03 has results for its dies but no X/Y positions at all, so there is nowhere to draw them on a map.
COORDLESS-LOT-01
A mixed wafer draws the dies that have positions, so the map is true to what is known. A footer counts the rest; click it to see them.
With no positions there is no map to draw, so the card shows a summary instead: a bin breakdown in the bin colour modes, or a histogram when colouring by a test value. View die list shows every die.
The die list has a row for every die, with its position (blank here), site, bins and every test value. Export CSV saves it, and View chart switches back.
Layout findings (edge ring, quadrants, clusters) use only positioned dies, because they need to know where a die is. Yield, bin counts and test statistics include every die, so no result is lost.
More in the user guide
A die is sometimes probed again after it fails, to see whether the failure was real. Every probe is a result, but the die needs only one.
RETEST-LOT-07 has:
RETEST-LOT-07
When a lot has dies probed more than once, the Overview gains a Re-probed dies card: the share of dies re-probed in each ring of each wafer.
Look for where the retests concentrate.
The probe you choose changes the yield.
A result the tester marked as replacing an earlier one always counts, whichever choice you make.
The chosen result is the die on the map and in the yield, findings, Insights, tables and exports, so every figure agrees. RETEST-LOT-07 yields 86.1% with the last result, 79.2% with the first, 91.2% with the best and 75.8% with the worst.
If one file holds the same wafer more than once, a column that tells the passes apart (a temperature, say) makes each pass its own map, and the log names the column. With nothing to tell them apart the repeats are treated as retests.
Sites are a different kind of repeat, and are not retests.
Subdivide file by this column makes one wafer per value of a column, for files that pack several wafers with no wafer column. Never use it for sites: they share one wafer.
A reticle is the group of dies exposed in one stepper shot. Defects that repeat in every field point at the mask, so it is worth telling tsmap the reticle's size. Files don't record it, so enter its width and height in dies. Set the corner die only if a field doesn't start at (0, 0).
Here the field is 4 dies wide and 3 high. Apply takes effect at once. Clear removes the reticle. The reticle applies to every wafer and is remembered.
Overlays ▾ → Reticle grid hides or shows it. Plots gain two fields: Reticle cell, a die's place inside the field, and Reticle shot, which exposure it came from.
tsmap works out the wafer's size from its dies. When there are too few dies to do that, enter the diameter.
With 300 mm and 10 mm entered, dies within 10 mm of the edge are left out of yield and drawn dimmed. Both values are remembered.
Sometimes a number in the data is not a measurement.
Instead of validity limits, you can choose the spec limits or the test limits, or switch the exclusion Off.
The tooltip on a grey die names the excluded value, and the Summary panel's Test Values table shows an Excl. column. Bins and the tester's pass or fail are unchanged, so yield is the same.
The last lesson covers automation.
12 · Automation and integration →
The command line, links, and definitions files.
