How to Check an IC Is Not Counterfeit Before It Reaches the Line

A tray of microcontrollers can look perfect and still be wrong. Counterfeiters remark lower-grade parts, sand and recoat packages, pull used devices off scrap boards and copy labels well enough to pass a quick glance. The only strong defence is a supply path you can trace back to the manufacturer. When that path is broken, because a part was bought from a broker or an independent distributor, the next best thing is a planned set of checks that looks for the specific ways the lot could be fake.

This guide is about that second situation: a lot of ICs with incomplete pedigree that you are thinking of using. It covers what to decide before any test is ordered, what each inspection method actually finds, how to sample a lot, and what a laboratory report has to contain before quality can sign a release.

Know what testing cannot do

SAE AS6171 is the main test-methods standard for suspect counterfeit electrical, electronic and electromechanical parts (AS6171A was the current revision on SAE's site when we checked in October 2026). Its general requirements apply once a decision has been made to use parts without pedigree back to the original component manufacturer (OCM). It is also blunt about the limit of the whole exercise: no amount of testing can confirm an item as authentic, because that would need a known, unbroken chain of custody to the OCM or an authorised manufacturer.

Hold on to that sentence when a supplier offers a "100% tested" certificate. Testing reduces specific risks. It never turns open-market stock into manufacturer-traceable stock, and Texas Instruments, for one, states that it gives no warranty or support for parts bought outside its authorised sources. If the part is still available through the authorised channel, buying there is cheaper than any test plan. Our guide on how to verify authorised distributor status covers that check, and spot buying electronic components covers when an open-market purchase can be justified at all.

Write the risk statement first

A test plan without a written risk statement tends to become a menu of whatever the laboratory happens to own. Before you send samples anywhere, record:

• the OCM, the complete manufacturer part number (MPN) with every suffix, package and purchased grade;
• quantity, and how many date codes, lot codes, reels or trays the lot contains;
• the seller, the upstream source they have disclosed and any gaps in custody;
• the product the part goes into and what a field failure would cost;
• known counterfeit modes for this device and package (search ERAI's database and your own nonconformance history);
• how many units you can afford to destroy; and
• the decision the result has to support: release, reject, or more testing.

That last line matters more than it looks. A report that says "no anomalies found" is useless if nobody agreed in advance what would count as an anomaly.

Layer 1: paperwork and packaging

Start before the bags are opened. Photograph outer cartons, seals, moisture barrier bags, humidity indicator cards, reels, trays, tubes and every label. Scan the barcodes and compare the encoded data with the printed text; a label printed by a counterfeiter often gets the human-readable part right and the barcode wrong, or the other way round.

Reconcile MPN, quantity, date code, lot code and country of origin across the purchase order, invoice, packing list, certificate of conformance and labels. Where you have known-good stock from an authorised source, compare label layout and packaging with it. Keep the labels and the packaging under the same custody as the parts, because they are evidence if there is a dispute.

A flawless label proves very little. A label that disagrees with its own barcode is a reason to stop.

Layer 2: external visual and dimensional inspection

External visual inspection is the first technical screen and the cheapest. SAE AS6171/2A sets out the methods: general and detailed external visual inspection (including part weight), tests for remarking and resurfacing, dimensional measurement, and surface texture analysis using a scanning electron microscope. The independent distributors' association, IDEA, publishes a separate visual acceptability standard for open-market parts, IDEA-STD-1010, which many brokers inspect against.

Under suitable magnification and lighting, inspectors look for:

• marking fonts, alignment or content that vary within one stated lot;
• sanding marks, a coating over the original surface (blacktopping), or texture that differs between units;
• faint ghost markings, or coating that runs over the package edge;
• mould marks, pin-one indicators or dimensions that differ between devices;
• solder residue, scratched or re-tinned leads, bent pins, or ball damage on BGAs;
• corrosion or contamination suggesting the parts were pulled from boards.

Record every anomaly against a numbered sample. Be careful with rules of thumb such as "the logo always looks like this". Manufacturers change marking lasers, mould compounds and assembly sites through controlled changes, so an unfamiliar look calls for comparison with manufacturer data, not automatic rejection.

Layer 3: marking permanency and resurfacing tests

Where the plan calls for it, the laboratory applies controlled solvent and marking-permanency tests to detect markings or coatings that were added after manufacture. These are specified procedures with defined solvents, exposure and acceptance criteria. An improvised acetone wipe on the receiving bench can damage legitimate parts and destroy evidence, so leave it to the laboratory.

A negative result here only means the surface was not resurfaced in a way this test detects. Plenty of counterfeits are never blacktopped.

Layer 4: X-ray

Radiological inspection is non-destructive, which makes it the usual next step. It shows die size and position, lead-frame shape, bond-wire count and routing, and voids. SAE AS6171/5 describes comparing each part with a known exemplar, or, when no exemplar exists, comparing parts within a homogeneous sample population using whatever technical data is available.

Ask the laboratory to define views and settings, include a known-good reference where possible, number every sample, keep the original images, and compare across reels, trays and date codes rather than only within one. Two different die outlines in one "lot" is a finding that needs an explanation. X-ray will not tell you the speed grade or whether the markings are original. If you plan to inspect in-house, our pages on AOI, SPI and X-ray machines list equipment suppliers, though counterfeit analysis also needs trained analysts and reference images.

Layer 5: XRF

X-ray fluorescence (XRF) is a different technique from radiography, despite the name. It measures which elements are present in the small area being examined. In counterfeit work it is useful for lead finish and plating, for example when a seller claims a lead-free finish on a part the manufacturer made with tin-lead, or the reverse. SAE AS6171/3 covers XRF methods for piece parts before board assembly or after removal.

The report should state exactly where the beam was aimed, the reference materials, the measured composition and the conditions. XRF says nothing about the die, the firmware, or how many hours a reclaimed part has already run.

Layer 6: decapsulation and die inspection

Decapsulation removes the package so the analyst can read die markings, check die geometry and compare bond arrangements with an exemplar. It is destructive, so it is used on a few deliberately chosen samples. The report should include the preparation method, photographs, die markings found, what they were compared against, and any damage caused by the process itself.

Layer 7: electrical testing

Electrical testing checks that the part meets the purchased specification under defined conditions. SAE AS6171/7 sets out levels of electrical testing a buyer can choose from, minimum capabilities for the test house, and burn-in, temperature-cycling and seal tests where they apply.

A room-temperature function check is weak evidence. A reclaimed or down-marked device can boot happily and still fail leakage, offset, timing, endurance or temperature limits. Build the test list from the datasheet parameters that matter in your application: DC limits such as leakage and supply current, AC timing, memory or interface behaviour, analogue characteristics, and the temperature and voltage corners the product will see. For programmable parts, control the test firmware and check that devices are blank or carry only what you expect.

Sample by population, not by a fixed count

A neat sample size can still miss the problem if it ignores how the lot is built. Before choosing units, separate the lot by reel or tray, date code, label style, surface texture and, once X-ray results are in, internal construction. Sample each population on its own terms.

Suppose 1,000 op-amps arrive under a single seller lot number. Visual inspection shows two surface textures. X-ray then splits the samples into two die-and-bond-wire constructions. XRF finds the same lead finish on both, so it neither confirms nor explains the split. Electrical testing finds a group of high-offset parts outside the purchased grade, all from one construction. No single test settled the question; the combination did. A report that only said "XRF passed" would have been dangerously incomplete.

Pick destructive samples from the suspicious groups on purpose, and keep enough untouched units for a second opinion or a dispute with the seller.

What a usable laboratory report contains

Before quality accepts a report, check that it includes:

• customer, purchase order, supplier and lot identification;
• how samples were selected, how many, and which physical units they came from;
• the standards applied and any deviations from them;
• equipment used and its calibration status;
• methods, conditions and acceptance limits;
• measured data, not only pass or fail;
• labelled images tied to sample numbers;
• the exemplar or documents used for comparison;
• the risks the plan did not cover; and
• the name of the reviewer and the recommended disposition.

A one-page certificate saying "samples passed visual, X-ray and electrical" answers none of these. If a test house is unfamiliar, our listing of testing and certification services is a place to start a shortlist, but check each laboratory's scope, equipment and reference data yourself.

When results conflict

If anything is unexplained, keep the whole lot in quarantine: blocked in the ERP, physically segregated, in its original packaging, and linked to the PO, the supplier and the inspection records. Check whether any of it has already reached work-in-progress, finished goods or service stock. Do not send suspect parts back to the seller informally, since that can put them straight back on the market; follow your contract, your customer's requirements and any reporting obligations. Reporting confirmed cases to a database such as ERAI helps the next buyer.

For how lot and date codes should be recorded so that this containment is possible, see tracking date codes, lot traceability and CoC documents.

Questions buyers ask

Is X-ray the same as XRF?

No. X-ray (radiography) images the internal construction of the part. XRF measures the elements present in a small area, usually the leads or plating. They answer different questions and are often both needed.

Can a counterfeit IC pass electrical testing?

Yes. A limited functional test can pass a part that is the wrong grade, has been used, or fails at temperature. Coverage depends on which parameters were tested and under what conditions.

How many parts should go to the laboratory?

There is no single number. It depends on how many populations the lot contains, the application risk, which tests are destructive and what your customer or contract requires. Write down the reasoning and keep it with the report.

Can a supplier's own test certificate replace our inspection?

Only if you have approved that supplier, the laboratory, the test plan and the reporting format in advance. A certificate that cannot be traced to specific samples and methods should not be accepted on its own.