Equipment

Inspection equipment

The tools that find and classify defects on the wafer — where metrology measures dimensions, inspection hunts for the particles, scratches, and pattern flaws that kill chips.

In short

Inspection equipment scans wafers to find defects — particles, scratches, pattern flaws — and classifies them. Where metrology measures how big or thick things are, inspection asks 'is anything wrong, where, and what kind?' It is central to finding and fixing yield problems.

How does inspection close the control loop?

Why it matters

A single well-placed defect can kill a die. Inspection catches defects early, locates them, and — by classifying and counting them — reveals which process step is causing yield loss, so it can be fixed.

Beginner intuition

Metrology measures the things you meant to make; inspection looks for the things you didn't — a stray particle, a scratch, a missing or bridged pattern.

Finding a defect is only half the job: knowing what kind it is and where it came from is what lets you stop it happening again.

Where it fits in manufacturing

After key steps and throughout the flow — often catching defects on sample wafers so a drifting or contaminating tool is found before it spoils many wafers.

How it works

What goes in

A processed wafer
An inspection recipe defining sensitivity and the area to scan

What happens inside

Scan: the wafer surface or pattern is scanned to detect anything that differs from the expected pattern or a clean surface.
Detect & locate: candidate defects are found and their positions recorded on a defect map.
Classify & count: defects are sorted into types and tallied — often with higher-magnification review — to point at a root cause.

What comes out

Defect maps and classified defect counts that reveal where and how yield is being lost, feeding process monitoring and improvement.

Major subsystems

Imaging / scanning system

Scans the wafer to detect deviations (optical or other techniques).

Defect detection engine

Compares against the expected pattern or neighbouring dies to flag defects.

Review & classification

Higher-resolution review that categorizes defect types.

Defect-map & data system

Records locations and counts for analysis and trending.

Process parameters that matter

Sensitivity: The smallest or most subtle defect the tool can reliably catch.
Throughput vs coverage: How much area is scanned, how fast — more coverage catches more but costs time.
Capture & classification accuracy: How reliably real defects are found and correctly typed, versus false counts.

A note on numbers

Achievable sensitivity, throughput, and coverage are technique- and vendor-dependent — treat any figures elsewhere as examples, not universal specs.

What determines performance

Inspection balances sensitivity against speed and false counts. Too insensitive and killer defects slip through; too sensitive and engineers drown in nuisance flags. Good classification — sorting real, yield-relevant defects from harmless ones — is as important as raw detection.

Inspection is the sensing half of process monitoring: a rising count of a particular defect type points straight at the tool or step responsible, driving a targeted fix. It is the same control loop as metrology — measure, compare, adjust, run again — applied to defects rather than dimensions.

Common issues

What can go wrong:

Missed (undetected) killer defects → yield loss discovered too late
Too many nuisance or false defects → wasted review effort
Misclassification → the wrong root cause chased
Sparse sampling → excursions caught late

Yield implications

Inspection is a primary driver of yield improvement: by finding, classifying, and trending defects, it turns yield loss from a mystery into a list of fixable causes — and the faster the loop runs, the faster yield climbs.

Manufacturing implications

Inspection defines the defect-monitoring strategy; its sensitivity and sampling determine how quickly a contaminating or drifting tool (an 'excursion') is caught and contained.