Exhibiting at SEMICON Japan 2026 (Dec 9–11)? We are now accepting advertising inquiries from exhibitors.Ads for SEMICON Japan 2026Learn more

Comparison · Test and reliability

Wafer test vs final test: what is the difference?

Most chips pass through both steps. Wafer test protects packaging yield and cost; final test protects the customer. The split of test content between them, and whether further steps such as burn-in or system-level test are added, depends on the product, its packaging and the quality level demanded.

The short answer


The difference is when and on what the test runs. Wafer test, also called wafer sort or circuit probe, contacts each die on the finished wafer with a probe card so bad dies are identified before money is spent packaging them. Final test runs on packaged devices, using automatic test equipment (ATE) and a handler, to confirm that the finished product meets its specification before shipment.

Side-by-side comparison


AspectWafer testFinal test
TimingAfter front-end fabrication, before dicing and packagingAfter assembly and packaging, before shipment
Device under testBare dies still on the waferIndividual packaged parts
Main purposeRemove defective dies early and map yieldVerify the finished product meets its data sheet
Contact methodProbe card needles or MEMS probes touching die pads or bumpsTest socket contacting package leads or balls
EquipmentWafer prober plus ATE and probe cardATE plus test handler and sockets
Typical contentDC parametrics, functional and scan tests, memory repair analysisFull functional and speed tests, multiple temperatures, binning by performance
TemperatureHot or cold chuck controls wafer temperatureHandler or chamber controls package temperature
Defects caughtFab defectsFab defects that escaped plus assembly defects
OutputWafer map marking good dies for assemblyPass/fail and speed or grade bins for each part

How wafer test works


A wafer prober loads the wafer onto a chuck, aligns it optically and steps it so that the probe card touches one or several dies at a time. The probe card links each pad to tester channels in the ATE, which applies test patterns and measures responses. Results are stored as a wafer map that tells the assembly line which dies to pick.

For memory, wafer test also identifies failing cells so that redundant rows and columns can be fused in. Because packaging is costly, removing bad dies at this point avoids wasting package materials and assembly time on them.

How final test works


After assembly, a test handler picks packaged parts, brings them to the required temperature and presses them into sockets on the test head. The ATE runs the full test programme, often at more than one temperature, and the handler sorts each part into bins: pass grades by speed or power, or fail categories.

Final test catches problems introduced during dicing, die attach, bonding and moulding, as well as fab defects that wafer test could not detect, for example because some high-speed interfaces are hard to test through probe needles.

Final test is also where a part is checked in the form the customer will use it, including package-level effects such as the parasitics of bond wires or bumps and the thermal path through the package. For this reason, test limits at final test are usually set against the published data sheet, while wafer test limits can be tuned to balance escapes against overkill of good dies.

Known good die and extra test steps


In multi-die packages such as chiplet designs and HBM, a single bad die can scrap an expensive assembly. This pushes more test coverage to wafer level to deliver known good dies, and in some cases wafer-level burn-in is used to remove early-life failures before stacking.

At the other end, complex processors increasingly add system-level test after final test, running real software on the part to catch failures that structured ATE patterns miss. Burn-in may also sit between assembly and final test for automotive and other high-reliability products.

FAQ


Which comes first?

Wafer test comes first, then dicing and packaging, then final test. Some products add burn-in before final test or system-level test after it.

Can a product skip wafer test?

Some low-cost, high-yield products use reduced wafer test and rely on final test, because cheap packages lose little when a bad die is assembled. Expensive packages and multi-die products need thorough wafer test.

Is the same ATE used for both steps?

Often the same tester platform is used, docked either to a wafer prober or to a package handler. The interface hardware differs: a probe card for wafers and load boards with sockets for packaged parts.

Where do Japanese suppliers fit?

Japan is strong in test equipment. Advantest supplies ATE, Tokyo Electron and Tokyo Seimitsu (Accretech) supply wafer probers, and Micronics Japan supplies probe cards.

Terms compared


Japanese companies involved


More test and reliability comparisons


Sources


  1. Tokyo Electron: Precio series wafer probers (Japanese)
  2. PubMed Central: 50 GHz Four-Port Coupling-Reduced Probe Card Utilizing Pogo Pins Housed in Custom Metallic Socket
  3. Tokyo Electron nanotec museum: Semiconductor glossary, burn-in (Japanese)

More on this topic


Semiconductor comparisons

Test and reliability terms

Last updated: November 2, 2026