XRF vs Microscratch for Coating Thickness

Why coating thickness measurement causes disputes

Coating thickness measurement is a common source of disagreements between suppliers and buyers. A part ships, the buyer pulls a sample, and the reported thickness does not match the purchase order. The gap often comes down to method.

Two methods dominate shop-floor and lab comparisons: XRF and microscratch. Both are widely used for metallic and some organic coatings. Neither is perfect. The choice depends on what you are measuring, how fast you need the answer, and whether the part can be touched.

How XRF works for thickness

X-ray fluorescence instruments measure the amount of X-rays reflected or emitted from a coated surface. The instrument compares the signal against a calibration curve built from reference standards of known thickness.

For flat, smooth, conductive metal, XRF is quick. A technician places the probe against the part, waits for a reading, and records it. Multiple spots are usually taken to average out local variation. This makes XRF suitable for incoming inspection, in-process checks, and final audit sampling.

The method is non-destructive. The part leaves the lab as it arrived. That matters when the buyer wants to retain evidence or re-inspect later.

XRF works best when the substrate and coating materials are known. If the base metal changes from steel to stainless, or the coating alloy shifts, the calibration may drift. Operators who skip recalibration after a material change often see false readings.

How microscratch works for thickness

Microscratch, also called cross-section or optical sectioning, uses a fine wire or diamond scribe to cut a small groove through the coating into the substrate. The cut surface is viewed under a microscope, and the coating thickness is measured optically.

This method is destructive. The part is permanently marked. That limits its use to sacrificial samples or areas where the scratch will not affect function or appearance.

Microscratch handles rough, textured, or uneven surfaces better than XRF. It also works when the coating is very thin, where XRF signal can be weak or ambiguous. Because the measurement is a direct optical view of the layers, it is often easier to explain to a customer or auditor.

The trade-off is speed. A single scratch and image takes longer than an XRF spot. A batch of ten parts will take noticeably longer. In a high-volume production line, that difference adds up.

XRF vs microscratch comparison

Option	Best for	Limitations
XRF	Flat, smooth metal; high-speed incoming and in-process checks; non-destructive sampling	Calibration drift with unknown substrates; less reliable on rough or curved surfaces; can struggle with very thin coatings
Microscratch	Thin coatings; rough or textured surfaces; dispute resolution; visual documentation	Destructive; slower per part; requires trained operator and microscope access

When to pick XRF

Pick XRF when speed and non-destructiveness matter.

Typical use cases include:

1. Incoming inspection of plated or coated sheet, bars, or components from a supplier.
2. In-process checks during production where the part must remain usable.
3. Final audit on flat panels or housings where a small number of spots are acceptable.
4. Repeat checks on the same part type over time, where a stable calibration exists.

XRF also fits well when the buyer wants a quick pass/fail decision at the dock. A probe and a reference chart can settle the question in minutes. If the part is a flat steel plate with a known zinc or nickel coating, XRF is rarely the wrong first choice.

Avoid XRF when the surface is heavily roughened, bead-blasted, or coated with a material that scatters X-rays unpredictably. The reading may be inconsistent from spot to spot, and the average may not represent the true minimum thickness.

When to pick microscratch

Pick microscratch when XRF gives inconsistent results or when the measurement must be visually verifiable.

Common scenarios:

1. Coatings below the reliable lower limit of the XRF instrument.
2. Rough, matte, or textured finishes where a flat probe cannot make consistent contact.
3. Disputes where both parties need an unambiguous, image-based record.
4. Parts with complex geometry where a single optical cut can capture a representative cross-section.

Microscratch is also the method of choice when the coating is an organic or mixed material that does not respond cleanly to XRF. The optical measurement bypasses the need for a fluorescence model.

The cost of destruction is real. If the scratch lands in a visible area or near a critical feature, the part may be rejected. Plan the scratch location carefully. Many labs use a standard, agreed-upon spot or a sacrificial coupon.

Calibration and acceptance criteria

Both methods are only as good as their calibration and the acceptance criteria behind them.

For XRF, the calibration curve should be built with standards that match the substrate, coating material, and thickness range of the parts in use. If the curve was built for steel with zinc and the line switches to aluminum with a different nickel alloy, the readings will not be valid without rework.

For microscratch, the operator must control the cutting depth, angle, and microscope magnification. A shallow cut may miss the full thickness. A deep cut may distort the layers. Standard operating procedures should define the scribe type, the number of cuts per sample, and the number of measurements averaged.

Acceptance criteria must state the method explicitly. A purchase order that says “thickness 100 microns, +/- 10 microns” without naming the measurement method leaves room for argument. If XRF is allowed, the supplier will use it. If microscratch is required, the cost and sample handling change.

A practical rule: specify the method, the number of measurement points, the calibration standard, and the acceptance tolerance. If the method is not stated, the buyer is likely to get the cheapest method the supplier can justify.

Common mistakes in thickness testing

The most frequent errors are not exotic. They are procedural.

Using an XRF calibration built for a different substrate is the top one. The instrument may still display a number, but the number is not the coating thickness.

Taking only one spot per part is the second. Coating thickness varies across a part. One spot may pass while an edge or corner fails. Most quality plans require multiple points, often three or more, and then the lowest value determines acceptance.

Allowing rough surfaces to be measured with XRF without checking variation is the third. If the surface texture changes the X-ray path, the reading is a proxy, not a true thickness.

For microscratch, a common mistake is using the scratch area for visual appearance checks. The scratch is for measurement. If the buyer expects the scratch to be invisible, the method will fail.

How to specify thickness testing on a purchase order

Write the requirement in a way that removes ambiguity.

State the method. Example: “Thickness measured by XRF, non-destructive, three spots per part, minimum value must meet tolerance.” Or: “Thickness measured by microscratch, destructive, one cut per sample, average of three optical readings.”

State the tolerance. Include the unit and the sign convention. “100 microns, +/- 10 microns” is clear. “About 100 microns” is not.

State the calibration expectation. Require the supplier to use a calibration standard that matches the part material and coating. If the supplier does not maintain such a standard, they should state so.

State what happens if the method is not available. For example, if XRF is specified but the part is rough, does the buyer accept microscratch as an alternate? If not, the part is nonconforming by method even if the true thickness is within tolerance.

A clear specification protects both sides. The supplier knows what to deliver. The buyer knows what to check.

Final method selection

There is no universal winner between XRF and microscratch for coating thickness measurement. The right tool fits the part.

For flat, smooth metal with a known coating and a need for speed, XRF is the practical default. For thin, rough, or disputed coatings, microscratch gives a clearer picture. Many quality systems use both. XRF for routine checks, microscratch for verification or when XRF results are questionable.

The goal is not to find the most advanced method. The goal is to find a method that produces a defensible number under real production conditions, and to write that method into the specification before the first part is measured.