A jewelry plating thickness number is meaningful only when it names the layer, unit, measurement location and kind of value being reported. A nominal target, the mean of several readings and a minimum local thickness are different statements. None, by itself, predicts a fixed number of months or years of wear.
A report average can conceal a low local reading. The invented example below separates the two, and the worksheet keeps the layer, location and measurement details together. The laboratory still needs an appropriate measurement method.
Read the units correctly
A micrometer, commonly called a micron, is a unit of length written µm. One micrometer equals 0.001 millimeter, and one millimeter equals 1,000 micrometers. Consequently, 0.5 µm equals 0.0005 mm. Confusing these units changes the quantity by a factor of one thousand.
View full-size image ↗Write the unit beside every reading. A cell containing “0.5” without a unit can be misunderstood when copied into a quotation, spreadsheet or product record. Also distinguish micrometers from microinches when comparing documents. They are different units; a repeated number does not imply the same thickness.
Karat and thickness describe different quantities. A gold-alloy fineness designation does not tell you the layer’s depth. “18K gold plated” therefore leaves the thickness question open. Equally, a thickness number does not establish the gold alloy’s composition. Record both fields when both matter.
Identify the layer before comparing the number
A decorative finish can contain more than one layer. A report may concern a named outer layer, an intermediate layer or a defined total stack. The label “coating thickness” is incomplete if it leaves the measured quantity ambiguous.
| Field | Question to answer | Why it changes the interpretation |
|---|---|---|
| Named material | Which layer is being reported? | The total stack is not necessarily the outer gold layer. |
| Value type | Nominal target, local reading, mean or minimum? | These summarize different things. |
| Location | Which component and surface? | A pendant face result does not automatically cover its edge or clasp. |
| Method and model | How was the value obtained and interpreted? | Method capability and assumptions affect what can be concluded. |
| Sample scope | Which items and batch were examined? | One measured sample is not evidence about every supplied item. |
For a multilayer system, ask for the layer order from substrate outward. Keep an outer gold-alloy layer separate from a gold-colored compound or a protective topcoat. A color description is not a chemical identification, and a number for one layer should not be copied into a field for another.
There is also a difference between a production target and an observed result. “Target 0.5 µm” describes an intention or requirement. “Measured 0.5 µm at location A” describes an observation under a method. The target cannot be used as proof that the supplied item reached it.
What XRF actually measures
X-ray fluorescence can be used for coating analysis, but a displayed thickness is the output of an application and its assumptions. ASTM B568’s public scope explains that X-ray spectrometry measures coating mass per unit area, which can be expressed as linear thickness when coating density is known. It also identifies a limit where signal response no longer changes usefully with increasing thickness. ASTM B568-98(2021), public Scope 1.1–1.3.
Extra decimal places cannot compensate for an unsuitable layer model, calibration or measurement scope. Ask the laboratory which layers and substrate the application includes, and whether those assumptions match the construction being supplied.
Fischer describes jewelry instruments for precious-metal and complex-multilayer analysis. That capability depends on the instrument and application. It does not establish that every XRF device can resolve every layer and geometry, or that a test screenshot represents a complete batch. Fischer: jewelry measurement applications.
An elemental-composition result and a coating-thickness result answer different questions. Where the layer identity or construction is uncertain, clarify it with the laboratory before interpreting the displayed value as a thickness result.
Why location and geometry matter
Jewelry’s small, curved and sometimes inaccessible surfaces make the measured spot and orientation important. Hitachi’s guide to XRF results discusses geometry, positioning and calibration. That interpretation guidance supplies no universal jewelry acceptance threshold.
Ask whether a result came from a flat witness coupon, a pendant face, the inside of a ring or an accessible clasp surface. A coupon may be useful process evidence, but it is not the same physical surface as a recessed feature on the finished piece. Keep that distinction visible in the record.
View full-size image ↗Match sampling locations to the concern and the method’s capabilities. If moving contacts also need assessment, check whether they were included in the reported locations. Ask the laboratory how an inaccessible feature could be examined before specifying a measurement there.
List unmeasured locations explicitly. A recess with no reading has neither a zero thickness result nor a pass; it remains outside the measured scope.
Worked example: the mean is not the minimum
Invented teaching data: suppose three named locations produce readings of 0.20, 0.40 and 0.60 µm for the same named layer. Assume the values can be compared for this arithmetic exercise. No instrument uncertainty is modeled, and no JBD item or laboratory report is represented.
The arithmetic mean is (0.20 + 0.40 + 0.60) ÷ 3 = 0.40 µm. The minimum of those three observations is 0.20 µm. Both statements are correct, but they answer different questions.
View full-size image ↗For this exercise, suppose the purchasing criterion requires at least 0.30 µm at every named location. Location A falls below it, although the reported mean is 0.40 µm.
This is not a complete conformity decision. A real agreement also needs an appropriate method, uncertainty treatment, decision rule and sampling plan. The minimum of three sampled points is not necessarily the minimum over the whole object, and three observations do not establish the distribution across a batch.
A mean requirement would ask a different question. Agree the quantity before testing so the acceptance decision uses the specified measure, rather than whichever summary looks most favorable afterward.
Read uncertainty and decision rules separately
NIST Technical Note 1297 distinguishes a reported estimate from its associated uncertainty and explains expanded uncertainty. Those reporting concepts help interpret a measurement; the jewelry acceptance rule still has to come from the applicable requirements.
Near an agreed limit, both measurement uncertainty and the decision rule matter. Ask the report issuer: “What uncertainty applies to this result, and which rule did you use to decide whether it meets our agreed limit?” Keep the answer with the report. Do not invent a universal rule that simply adds or subtracts the uncertainty.
Repeatability is another distinct issue. Similar repeat readings at one location do not establish that the layer model is correct or that every location has the same thickness. Keep repeated measurements, different locations and different sampled items in separate fields.
Expanded uncertainty is written U = k × uc, where uc is combined standard uncertainty and k is the coverage factor. The report should say which uncertainty it gives and state k when reporting U. A k of 2 does not, by itself, prove an exact 95% coverage probability; that interpretation depends on the measurement model and distribution assumptions. NIST TN 1297 §6; §7 (1994 guidance).
| Teaching input or rule | Result for this exercise only |
|---|---|
| Local result y = 0.32 µm; invented uc = 0.02 µm and k = 2 | U = 0.04 µm; y ± U spans 0.28–0.36 µm. No coverage probability is assigned here. |
| Invented minimum criterion = 0.30 µm; hypothetical agreement R1 compares y with the limit | 0.32 ≥ 0.30: meets R1’s numerical comparison. |
| Same invented criterion; alternative hypothetical agreement R2 requires y − U ≥ limit | 0.28 < 0.30: does not meet R2’s numerical comparison. |
R1 and R2 are alternative invented contract rules, not recommended laboratory practice or universal rules. The same reading can lead to different decisions because the agreements allocate uncertainty differently. Choose and document the applicable rule before testing with the responsible parties; keep the criterion source, uncertainty type, k and decision notes with the result. The exercise specifies no real method, sampling plan or coverage evaluation, so it makes no real conformity claim.
Build a useful report record
Download the coating specification and report worksheet. Start with the item, named layer, unit, reported value and location. The value_type field accepts the report’s actual category—such as target, local reading, mean or minimum—rather than forcing every number into a minimum-thickness claim. Keep method, model, calibration, uncertainty, report reference and unmeasured scope when you are interpreting a technical report. The worksheet also retains criterion_source, decision_rule, uncertainty_type_and_k, decision and notes. Copy the issuer’s exact uncertainty and rule; the existing three-point teaching rows say not modeled and make no real conformity decision.
Use the report’s exact wording for the measured quantity. Attach the original report and any location map. If information is unavailable, leave an explicit question for the laboratory or supplier rather than filling the field from a product photograph. The sample inspection record helps connect those documents to the received item and its observed condition.
Retain the approved finish variant and document revision for reorders. A change of color or component arrangement needs a scope check before an earlier result is reused. Keep that linkage in the worksheet alongside the laboratory’s original technical record.
Why thickness does not equal wear life
Thickness describes one aspect of a coating system. Adhesion, defects, substrate preparation, layer composition, contact geometry, environment and cleaning also affect the finished object. A thicker layer is not a remedy for every failure mechanism.
A thickness report supports the named layer and measured scope, rather than a fixed number of months of wear. Durability claims need performance evidence under stated conditions, including the relevant exposure; a wearer’s routine cannot be predicted from micrometers alone.
The useful purchasing comparison holds the design and test conditions constant where possible. If one supplier reports only a nominal target and another supplies location-specific results, the offers differ in evidence as well as the stated number. The PVD versus electroplating comparison applies this approach to competing coating systems.
Watch a coating measurement demonstration
Fischer Measurement Technologies’ XRF demonstration shows instrument operation and measurement output. It is useful context for understanding that a result belongs to an application and sample. It is not a JBD test, an independent instrument comparison or a universal recipe for jewelry measurement.
Coating Thickness Measurement with FISCHER XRF spectrometer
Loading this video connects to YouTube. You can also use the original video link below.
Watch the Fischer demonstration on YouTube.
Use the specification while buying
Before reading the number, establish the process and stack using PVD versus electroplating. The measurement should describe the same named layer that the purchasing claim describes.
When browsing the 18K gold-plated jewelry collection, identify candidate items and request the specification relevant to the particular finish. A collection label does not establish a measured layer thickness.
The electroplated jewelry care guide covers handling of electroplated surfaces. Check it against the complete item; care advice does not enlarge the sample or location scope of a measurement report.
Common questions
Is one micron always enough? There is no universal answer from the number alone. Define the layer, construction, intended exposure and required outcome.
Does XRF measure every point on a necklace? A report covers the measured samples and locations. Confirm the actual scope rather than assuming whole-surface coverage.
Can I compare a total stack with an outer gold layer? They are different quantities. Identify the named layer before comparing values.
Does a high average prove a minimum? No. The worked example shows why a mean can conceal a lower observed local reading.
References and limits
References linked above include the public scope of ASTM B568-98(2021), Fischer’s manufacturer application description, Hitachi’s XRF guide and NIST Technical Note 1297. Checked November 7, 2026. The full paid ASTM procedure was not accessed. No real coating report, instrument test, batch acceptance or wear-life validation was performed for this guide.