Getting started
- Upload a defect file (CSV, TSV, or XLSX) at the top of the sidebar.
- Set Wafer Size, Die Size, and Notch Direction to match your data — before or after uploading, both work.
- Hover a die on the wafer map for its coordinate. Clicking a wafer die highlights its entire set of defects; clicking a dot on the die defect map highlights just that one defect. Ctrl/Cmd/Shift-click, or -drag a box, on either map to add more to the selection; click empty space to clear it.
- Use Wafer Center and GDS Origin Offset (with their auto-calibrate icons) if the map doesn't line up with your data.
- Every plot (wafer map, die defect map, Pareto chart, Trend chart) has a full-view icon in its header to inspect it at a larger size in a modal.
- The Column, Group by, and Date column pickers are type-to-search — click one and start typing to filter instead of scrolling through every column in a wide file; arrow keys + Enter to confirm, Escape to cancel.
- The Pareto Chart below the two maps counts how often each value of any column occurs (pick it from the Column dropdown) — sorted most to least common, with a cumulative-% line and an 80% "vital few" cutoff. It always covers every loaded defect, independent of whatever's selected above. Optionally pick a second column with Group by to split each bar into a color-coded breakdown (up to 8 named values plus a shared "Other"); a legend and the per-bar list both spell out the exact numbers, so color is never the only way to read it.
- The Trend Chart beside it counts defect rows per date, for any column whose name contains "date" — pick it from the Date column dropdown. A bare 4-digit value is read as a
YYWW work-week code (e.g. 2405 = 2024, week 5); anything else is parsed as a normal date and collapsed to its calendar day. Only dates/weeks that actually appear in the file are plotted (gaps aren't filled in as zero); rows with a blank or unparseable value are excluded and noted below the chart.
File format
CSV, TSV or XLSX (.xlsx/.xls). Required columns (case-insensitive, flexible naming):
die_X / X coordinate — grid index, ≥ 0die_Y / Y coordinate — grid index, ≥ 0GDS-X — µm, absolute, origin at the die's own cornerGDS-Y — µm, absolute, origin at the die's own corner
Optional: ECID / CQC number — shown in
hover tooltips on both maps.
Any other extra columns are kept and shown in defect tooltips. A
row with a blank die_X/die_Y is dropped
entirely; a row with a valid die but blank
GDS-X/GDS-Y still counts on the wafer
map but isn't plotted on the die defect map.
Terminology
A file can have more than one column with "date" in its name, each
meaning something different along a typical fab-to-field
timeline — worth knowing which is which before picking one for the
Trend Chart:
- Fab date — when the die itself was manufactured at the fab.
- Date code — when the device was assembled/packaged. Normally on or after the fab date, but that ordering isn't guaranteed if the assembly used leftover (older) die stock.
- Return date — when the part came back for CQC analysis. Typically around six months after the date code, if the return cycle is relatively quick.
Die defect map plots each defect's physical
location from failure-analysis results, showing which circuit or
which physical layer is affected. Wafer map shows
the defect distribution across the whole wafer — useful for
spotting any commonality in the wafer's physical layout, and worth
comparing against the fab's own wafer defect map (if available) to
see whether a pattern originated at the fab or downstream.