Edge chipping and burr detection after wafer dicing
The diced wafer is scanned under a camera on its dicing frame. Cutting texture is distinguished from chips or cracks by deep-learning anomaly analysis; defective die locations are written to a map.
Defective die coordinates are written to the wafer map; the following picking station skips these die locations. Chip and crack regions are individually marked on the display.
- 01 Load
- 02 Scan
- 03 Analyse
- 04 Map
The frame seats on the scanning stage.
- Die position
- —
- Flagged regions
- —
Scenes are illustrative simulations; they are not customer references or performance guarantees.
Need
Normal dicing produces irregular edge texture that can resemble small chips and microcracks. Reflective die surfaces can hide a defect depending on lighting direction.
Solution approach
A linear stage carries dicing streets under the overhead camera. Opposing low-angle lights emphasise edge scattering; a deep-learning anomaly map identifies regions departing from normal cut texture. Locations are recorded against the die grid.
Process steps
- 01
Load
The frame seats on the scanning stage.
- 02
Scan
Dicing streets are imaged in dark field.
- 03
Analyse
Edges departing from cut texture are located.
- 04
Map
Defective die locations are written to the picking map.
Related applications
Let us assess your application together
We test your sample in our laboratory and report on resolution, optics and lighting selection.