Wafer metrology
Before a wafer lapping trial, agree where thickness will be measured, how the wafer will be oriented, which edge band is excluded, and which values determine acceptance. Put that information on one drawing and use the same convention for incoming and finished maps. The center thickness can document hitting a nominal target at one location. A map identifies whether areas away from that location still fall within the agreed limits and whether an apparent edge trend survives a repeat measurement.
Hisemi's wafer lapping and polishing machine page lists acceptable wafer sizes of 3, 4 and 6 inches for the model described. That is an equipment configuration reference, separate from a process capability claim for a particular material, finish or incoming profile. Submit the actual wafer type, dimensions and process targets to the supplier before the trial. Discuss which instrument and support condition will be used, what constitutes a valid map and how mapped results will be returned to the customer.

View the related product or series · Compare the second product or supplier reference
Agree on a wafer thickness map for lapping
Ask the Metrology and Process owners to identify the wafer origin, orientation feature and measurement direction. Decide how to record the flat or notch and whether the wafer is measured with the same face up before and after the trial. Show the origin and orientation on the map. An incoming instrument export and a post-lap export can look comparable while having opposite axes or a rotated origin. Give the operator a visual key so a trend near a marked feature can be found again.
Choose a point pattern that addresses the real engineering question. A center and a ring of evenly spaced points may be practical for a trial. Another project needs denser coverage or different edge positions. Agree where the sample will be measured and whether edge bands are excluded from the numerical acceptance calculation. Do not silently delete points after seeing them. Record values that were measured but excluded, together with the reason, so the reader can review the reporting boundary.
Agree the units and resolution to be shown. Keep the instrument model, method and verified status with the evidence. Include any fixture or support condition that can influence the reading. For trials across two facilities, establish a way to check instrument agreement on a reference sample before interpreting process changes. Each lab should supply raw readings and identify any recalculated values or rounding used in the final report.
Specify the calculation and run an illustrative check
TTV, as reported for an agreed point map, is the maximum valid thickness minus the minimum valid thickness using the stated method. A quoted nominal center reading does not tell you that spread. Meanwhile, thickness range describes a different question from where the highest or lowest reading appeared. A table without positions does not support root-cause work if a particular clocking position or fixture repeat is suspected.
Illustrative example: suppose five measurement points registered 150, 149, 151, 152 and 148 μm on an otherwise agreed example map. The sampled variation is 152 − 148 = 4 μm. If the middle value of 150 μm is the center measurement, the report should label it separately. It should also include the point coordinates and protocol. An unmeasured area could have a different value, and this example is not a machine capability claim or an allowable production limit.
| Trial record | Agreement to make in advance | Result to preserve |
|---|---|---|
| Map identity | Wafer ID, material, face, reference feature and coordinate direction | Marked drawing and export filename |
| Point set | Requested pattern and edge exclusion rule | All measured positions and excluded points |
| Instrument | Method, support condition, units and verified status | Instrument record and raw values |
| Approval | Nominal target, permitted variation and required evidence | Pass or hold by criterion with owner |
Run incoming and finished measurements as a matched pair
Label individual trial wafers before the process and connect each wafer to its material lot and current condition. Note if the incoming side has a special film, temporary carrier or other condition considered relevant by the process owner. Record any difference between incoming and outgoing support or instrument setup. If the prepared surface requires a different measurement method, obtain an agreed explanation of how the results should be compared and which result will control acceptance.
Ask for each finished wafer's map beside its incoming map. Use the approved report to calculate both the final thickness range and the material removed where corresponding points can be legitimately aligned. Mark points that cannot be compared. A tabular "before/after" column with the exact coordinates is better than a color-only map if engineering will be reviewing where a pattern moved. Keep the original export so a reviewer can check a suspected unit or indexing error.
When one edge position is outside the expected trend, repeat the appropriate measurement under the approved method. Check orientation, edge mark, sample support and any reading validity flags before linking the result to lapping. Record the repeat even if it moves toward the rest of the result. Inconsistent measurements are a reason to investigate. Replacing an original result without audit history makes the trial harder to reproduce and can frustrate an otherwise straightforward discussion of the machine or fixture.
Translate the report into a sample agreement
State which parts of the trial are a process development exercise and which constitute witnessed acceptance. On a development trial, leave enough flexibility for documented adjustment. For equipment acceptance, ensure the sample identity, revision, setup, operator, conditions and pass criteria match the agreement. Retain the list of unresolved items with an owner and a defined evidence requirement. Ask for exception handling if a wafer is broken, requires rework or can no longer be measured by the original method.
Procurement can then compare proposals on what will be shown. Receive the proposed plate, fixture and approved wafer range in the equipment line. Put maps, raw exports, sample retention and any witness event in a separate records line. That keeps the machine description distinct from process evidence. The resulting agreement should answer: Which wafer was measured? Which points counted? What conditions applied? Can the buyer's metrology team reproduce the comparison?
A sample measurement protocol becomes difficult to apply when the drawing names points that the instrument cannot reach. Check the agreed wafer TTV map points against a representative part before booking the machine. Have each lab walk its process from incoming wafer thickness map to the final export. One practical check is to ask a reviewer unfamiliar with the trial to find the same point on the physical sample and on the two maps. They should also be able to tell which points were left out under the edge exclusion agreement. If the reviewer cannot reconstruct the map, resolve the coordinate or labeling issue while the bench record is recent. For a wafer thickness map for lapping, use a version field for the map and lock it before the acceptance measurements. Later engineering investigation can use a new point set or a separately named map. Report it as additional evidence, with the revised sampling pattern available for comparison. This makes process troubleshooting possible while keeping the original witness agreement readable.
Agree on what the buyer will receive when a run needs a second pass. Some teams need a marked map after each pass even when only the final part will be released. Others need a single final export and a process summary with the history kept by the machining lab. Name the required handover and keep it distinct from the machining cycle count. This is a quotable deliverable and can affect equipment demonstration time.
Wafer measurement questions
Is a center measurement enough to approve a lapped wafer?
It can answer a nominal-center question. Add the agreed map when the decision involves variation across the surface. Check limits at the appropriate level and keep the center value separate from the calculated spread in the report.
Should both parties use the same edge exclusion?
Use an agreed boundary before comparison. Mark it in the drawing and write down which points count and why. A new edge exclusion changes the analysis and should be presented separately if the trial team needs both views.
Can orientation changes affect the conclusion?
Yes, particularly when matching an incoming and finished pattern or studying one edge region. Use a defined reference mark and coordinate convention. Include the marked photo or drawing if the instrument export cannot carry that orientation by itself.
What if a map includes an invalid point?
Identify the point and its reason for invalidation. Repeat it under the approved procedure where possible. Consult the trial rule before issuing a pass if the required point set is incomplete. Keep both the original and repeated record.
Should the report mix thickness and surface finish?
Report both when required, but identify the measurement method and status for each. A mapped thickness result does not establish roughness. Preserve separate values and acceptance rules for every property used in the sample decision.
Can another laboratory use the same approval sheet?
Send the map convention, raw results, instrument descriptions, support conditions and comparison limits. Verify which evidence the receiving lab can reproduce. Use a shared reference sample when engineering considers instrument correlation necessary.
Request a mapped lapping trial
Send Hisemi the wafer diameter, material, starting and target thickness, edge exclusion and required acceptance evidence. Ask which equipment configuration and sample report can be quoted.

