The $15,000 Acid Bath

Some hardware questions can only be answered by opening a component. That is an awkward proposition when the component is scarce, expensive, or the only sample associated with a failure. Destructive analysis can reveal information and remove the opportunity to run another test on the intact device.

The title's $15,000 is a framing device, not a published laboratory quote or the price of a test I commissioned. Actual scope and cost depend on the package, preparation, imaging, reference material, and question. The useful issue is when destroying a sample is justified by what you expect to learn.

Start with the question, then the instrument

A suspicious package marking, an unexplained electrical failure, and a suspected silicon modification are different problems. They should not all lead straight to chemical decapsulation.

Suppose, hypothetically, a critical replacement part has markings that differ from an approved lot. I would first preserve its identity and handling history, compare the packaging and supplier documentation, and ask the manufacturer what variations are legitimate. A different mark can be a manufacturing change. It is not automatically a counterfeit, much less a malicious implant.

The next tests should distinguish plausible explanations. Package inspection and X-ray imaging can help assess assembly features. Electrical characterization may reveal a mismatch with the claimed part. Their usefulness depends on the device and an appropriate reference, rather than a generic promise to detect every modification.

Opening the package changes the work

Decapsulation removes enough package material to expose the die or internal assembly for inspection. Texas Instruments describes a failure-analysis sequence that uses nondestructive techniques before partially destroying a package to expose the die. Its workflow also includes electrical analysis and methods for locating and characterizing a failure site.

After preparation, a laboratory may use optical or electron microscopy and further sectioning to examine the relevant structure. An exposed top surface does not reveal every buried layer, firmware behavior, or condition-dependent circuit effect. The analyst still needs a hypothesis and a reference against which an observation means something.

This is specialized laboratory work. Chemical preparation adds handling and waste-management requirements, while mechanical or other preparation techniques have their own limitations. The selection belongs to the qualified lab working with the actual package. An article about security value should not pretend to be an etching procedure.

Comparison is the difficult part

An unexpected bond, die marking, or layout feature deserves explanation. Calling it malicious requires more than a surprising image. It might reflect a legitimate revision, a different manufacturing source, repair, damage, or an incorrect reference sample.

For that reason, I would want the report to separate observations from interpretations. Identify the inspected surfaces and regions, explain which reference was used, and state what was not accessible. If the conclusion is that the die marking is inconsistent with the supplied part description, say that. Don't inflate it into a universal hardware-integrity determination.

Negative findings need the same care. Seeing no unexplained feature in the inspected region does not establish the absence of all implants. Detection coverage depends on resolution, preparation, design knowledge, sampling, and the classes of change being considered.

Preserve options before sacrificing the sample

Before authorizing destruction, I would settle several practical details with the lab:

• Which question requires opening the package, and what result would change the decision?
• Which intact-device measurements must happen first?
• Is a second sample available, and how representative is it of the relevant lot?
• What reference material is available from the manufacturer or a trusted source?
• Which photographs, measurements, remnants, and preparation notes will be retained?
• What conclusion remains possible if preparation damages the area of interest?

These questions help avoid paying for an impressive image that cannot resolve the original concern. They also force the decision-maker to acknowledge sampling limits. A single inspected component describes that sample; extending the result to a shipment requires a sampling rationale.

Destructive inspection makes sense when the consequence of uncertainty warrants the work and the available methods can discriminate between the explanations. The next step is a scoped laboratory plan with explicit limitations, not an acid bath selected because the suspected failure sounds sophisticated.