wafer sort / probe

Wafer sort (wafer probe) is the **electrical testing of every die on a wafer** before dicing, identifying good dies and defective dies so that only known-good dies proceed to the expensive packaging step. **How It Works** **Step 1**: Processed wafer loaded onto the prober (automated wafer handling system). **Step 2**: Thousands of tiny **probe needles** (or probe card contacts) are aligned to the die's bond pads and lowered to make electrical contact. **Step 3**: The **tester** sends test patterns through the probes, measuring the die's electrical response. **Step 4**: Each die is classified as pass (good), fail (bad), or binned by performance level. **Step 5**: Results stored in a **wafer map**—a visual representation showing good/bad die locations. **Step 6**: Defective dies are marked (ink dot or electronic map) for exclusion during die pick. **Probe Card Technology** • **Cantilever probes**: Traditional bent wire probes. Good for I/O testing. Limited pin count (~1,000 pins) • **MEMS probes**: Micro-fabricated probes for fine-pitch pads. Higher pin count and better planarity • **Vertical probes**: Straight vertical needles for very fine pitch (< 50μm pad pitch) • **Probe cards cost**: $50K-500K+ depending on technology. Must match every product's pad layout **What Wafer Sort Tests** **Continuity/shorts**: Verify all I/O connections work and no shorts exist. **Leakage**: Measure standby current (IDDQ) to catch defective transistors. **Functional**: Run logic patterns to verify the circuit operates correctly. **Parametric**: Measure speed, voltage margins, and analog specifications. **Binning**: Determine frequency/power grade for each die. **Why Test Before Dicing?** Packaging costs **$1-20 per die**. Testing before dicing avoids wasting packaging resources on defective dies. For expensive packages (flip-chip, 3D stacking), this savings is critical. The concept of **Known Good Die (KGD)** is essential for chiplet and 3D stacking applications where one bad die ruins an entire multi-die package.