Thickness Maps: Choose Locations That Can Reveal Edge and Center Differences

Measurement purpose

Define spatial thickness coverage that can distinguish the intended edge, interior and transition behavior on identified printed features.

Specimens and conditions

Geometry and stack

Use current artwork and a consistent physical thickness definition at every compared location.

Orientation

Preserve print direction, panel coordinates and specimen identity throughout acquisition.

Equipment and records required

• Thickness acquisition: Characterize the effective lateral footprint and retain actual measurement coordinates.
• Location mapping: Overlay planned and acquired regions with valid, missing and interpolated status distinguished.

Method sequence

1. Plan

   Select geometry-based location classes and the minimum feature behavior of interest.

   Record: Coordinate and footprint plan.

2. Acquire

   Measure the planned regions and retain missing-location reasons without silent substitution.

   Record: Located raw thickness dataset.

3. Compare

   Calculate paired spatial contrasts and state the unobserved coverage.

   Record: Edge-center differences and coverage statement.

Decision and uncertainty

Use the map only for contrasts and locations its footprint and selection can resolve; an observed minimum is not a guaranteed population minimum.

Account for lateral response, overlapping footprints, missed regions and physical layer differences as well as vertical measurement uncertainty.

The drawing and process owners define consequential locations; the inspection engineer verifies that the map covers them.

Traceable outputs

Choose the spatial contrast the map must reveal

Identify whether the decision concerns across-width thickness, along-print-direction variation, terminal neighborhoods or differences between repeated features. These questions require different locations. A single center line may describe longitudinal behavior while saying little about both edges. A dense scan of one feature does not establish panel-wide uniformity.

Keep the measured quantity fixed across positions. If one location is a film plateau above bare ceramic and another lies over a lower conductor, their raw surface heights are not directly comparable layer thicknesses. Resolve that physical definition first, then design the spatial plan. This page concerns where valid measurements are taken, not how a buried interface is reconstructed.

Create named location classes on the artwork

Mark interior plateaus, both lateral edges, ends, corners, conductor overlaps and any process-relevant transition. Use a drawing datum and specify offsets from actual features rather than only an instrument stage origin. Preserve the relationship to print direction and panel orientation so repeated patterns can be compared consistently.

Do not define every edge point at a fixed absolute distance if the feature widths vary greatly. The location rule may need both a physical minimum distance compatible with the instrument and a fraction of the local feature width. Document any location that cannot accommodate the required footprint instead of relocating it silently to a convenient center region.

Distinguish point spacing from measurement footprint

A measurement labeled with one coordinate represents a finite region or probe interaction. Record the scan width, averaging window and relevant lateral response. Two points placed close together may sample overlapping surface area and therefore do not provide independent evidence about two separate regions.

Conversely, a wide gap between narrow footprints leaves an unobserved interval. Interpolating across that interval creates a model, not another measurement. A smooth color map can make this distinction hard to see, so retain the raw location overlay and identify areas supported only by interpolation. Sampling density should follow the feature size that matters to the decision.

Use an explicit missed-feature example

Consider an illustrative ten millimeter-wide printed pad measured along five narrow traces at x positions of one, three, five, seven and nine millimeters. A localized depression between x equals 5.8 and 6.2 millimeters can lie between the traces. Adding more readings along the existing five traces will not necessarily reveal it because the missing coverage is across the width.

A targeted additional trace at six millimeters can investigate that region once a mechanism suggests it, but it does not retroactively make the original map complete. If the requirement is detection of features anywhere across the pad, choose a coverage strategy based on the minimum consequential feature width and the validated lateral response. These dimensions illustrate sampling geometry, not a universal inspection pitch.

Compare paired locations rather than unrelated averages

For an edge-versus-center study, pair locations on the same identified feature and processing state. Report the signed edge-minus-center difference separately for each side before averaging across specimens. Opposite edge effects can cancel in a pooled mean even when both are important to the application.

An illustrative feature with center thickness of 18 micrometers and left and right values of 22 and 14 micrometers has edge differences of positive four and negative four micrometers. The mean edge value equals the center, but the two edges differ by eight micrometers. Preserve that asymmetry rather than concluding that the edge and center are equivalent from their average alone.

Give invalid and inaccessible locations their own status

A missing optical signal, probe collision risk or unresolved layer boundary is not a zero-thickness result and not a conforming reading. Record why the planned location could not be measured. Decide whether another method or specimen stage is needed when that location is important to the acceptance requirement.

Automated feature detection can help repeat location rules across several printed steps, but the recipe must be checked against narrow, irregular or merged features. Save the actual windows it selected. A successful measurement count does not establish that the software used the intended geometry on every feature, especially after artwork or print dimensions change.

Do not turn a spatial map into a population minimum

Report the smallest observed value with the locations and footprint that support it. It is the minimum of the inspected measurements, not automatically the minimum anywhere on the part or across production. A full surface method also has finite resolution and invalid regions, which remain relevant when a narrow local condition could control performance.

If the map is intended to support a production control, define how specimens are selected across panels and lots separately from the within-part location grid. Repeatedly mapping the same convenient specimen cannot establish between-part coverage. Keep the spatial map, sample selection and measurement repeatability as three related but distinct parts of the evidence.

Deliver a reproducible map and an honest coverage statement

The final package should include the drawing-linked planned locations, actual acquired footprints, valid and missing statuses, raw thickness values and the chosen spatial contrasts. Show edge and center results without an interpolation palette that obscures their separation. State any smoothing or surface-fitting method used for visualization.

Revise the location plan when feature width, terminal geometry, print direction or critical region changes. Reusing an old coordinate list can leave a newly important edge unmeasured. A professional map makes this coverage boundary visible and gives engineering a specific next observation when the existing sample does not resolve the question.