How Deep Does Your Test Actually See?

How Deep Does Your Test Actually See?

November 17, 2026

Every commercial lab knows the call where a client questions a result. A gap pad, a filled adhesive or a thermal interface material came in, your team measured it, the report went out, and a week later the client is on the phone because the number does not match their supplier’s data sheet or the value they have been designing with for years. You rerun the sample under the same conditions and it comes back within a percent of the first result, so you reassure the client that the measurement stands and move on to the next job.

The result is almost certainly repeatable, but there is a question underneath that call that most labs never get around to asking, and it is the one that decides whether the number on your report actually describes the material your client is going to build with: how much of that sample did the test really measure?

Why the length of a test decides what it measures

Most of us were taught that thermal conductivity is a fixed property of a material, in the same way density is, so that any valid instrument should give the same answer no matter the size of the piece or how long the test runs. That holds for a material that is thermally uniform all the way through, where heat moves at the same rate at every depth and in every direction, and it is why reference materials and calibration checks behave so well.

Transient measurements work by putting a small amount of heat into the sample from a sensor and watching how the temperature responds as that heat spreads outward. The important detail is that the heat does not reach the whole sample at once. It travels gradually, so at any moment the instrument is only seeing the part of the sample the heat has reached so far, and the length of the test therefore sets the volume of material that contributes to the result.

The depth the heat reaches does not grow in step with time. It grows roughly with the square root of time, which means a test that runs ten times longer only reaches about three times deeper. A very short test on a typical filled polymer may be measuring little more than the layer closest to the sensor, while a longer test on the same piece reaches well into the body of the material. If the material is the same all the way down, both tests agree and the distinction never shows up, but if it is not, the two tests are measuring two different things and both of them will repeat beautifully.

Same filled gap pad tested for 1 second and for 10 seconds: the longer test reaches deeper into the material. (This avoids the 3x figure, which is still marked Verify first.)

Test time is test depth: 1 second vs 10 seconds on the same filled gap pad

Why electronics materials make this matter

The materials electronics clients send to commercial labs are rarely uniform in the way a reference material is, because they are engineered to be mixtures. A gap pad combines a soft polymer with a heavy load of conductive filler and sometimes a carrier film in the middle; a thermal interface material can form a resin-rich skin at its surface during curing; a potting compound can let its filler settle so the bottom of a cast part differs from the top. Under a microscope a well-made material of this kind often looks evenly dispersed, and that appearance is reassuring, but looking uniform under a microscope is not the same as being uniform to heat, and a thin skin or a buried carrier layer can change how heat travels without ever showing up in a cross-section photo.

When a short test meets a material like this, it reports the thermal behaviour of the region near the sensor, which in many cases is the surface skin rather than the bulk of the part. Nothing in the data warns you, because the instrument is doing exactly what it was asked to do and the surface layer is just as consistent from run to run as the bulk would have been. The result is a clean, tight, repeatable number that answers a question nobody asked.

One materials formulator saw this first-hand when they suspected that a competing supplier’s impressive numbers had more to do with the test method than the material. They measured a single material with four different methods across their own sites and found that every method was highly repeatable, that three of the four agreed with each other within about five percent, and that the fourth, a very short test that mostly sees the surface, read about thirty percent higher than the rest. Nothing in that fourth method’s repeatability data would have revealed the gap, and the formulator went on to publish the comparison because they wanted their material judged on what it was rather than on which test happened to flatter it.

What this means for your lab and your clients

The number on your report does not stay on your report. Your client feeds it into a thermal simulation, uses it to choose between suppliers, or writes it into a specification, and then hardware gets designed around it. When that number describes only the surface of the sample, the consequences show up much later and far from your bench, which is exactly why they are so rarely traced back to the measurement.

A lab report’s thermal conductivity value flows into the thermal model, supplier choice or specification, and hardware. A value that reads low against the bulk leads to an over-built design, a value that reads high leads to parts running hot, and both outcomes erode trust in the lab’s next report.

If the surface reads lower than the bulk, your client’s model under-predicts how well the part will carry heat away, and they may over-engineer the design with thicker pads, larger heat sinks or more expensive material than they needed, paying for that margin on every unit they ship. If the surface reads higher than the bulk, as it did in the formulator’s comparison, the model is too optimistic, and the real part runs hotter than predicted, which tends to be discovered at prototype or in the field as an overheating board or a shortened component life, long after the time and money have been spent.

For your lab, the cost is quieter but just as real. A client who has been burned by a number rarely calls to say the measurement was the problem; they simply trust the next report a little less, ask for more reruns, push back harder when your result disagrees with a data sheet, or take the next job elsewhere. Your reputation rests on data your clients can defend in front of their own management, and a result that is repeatable but unrepresentative undermines that slowly and invisibly.

The good news is that this is one of the more manageable problems in thermal testing, because test time is something your team already controls and can document.

Five steps to make sure your results describe the whole material

1. Ask how the material is built before you choose a test time. When a sample arrives, find out whether it contains fillers, layers, a carrier film, a surface coating or a cured skin, and how it will be used in the client’s assembly. A two-minute conversation at intake tells you whether a short default test is safe or whether the material needs a closer look, and it signals to the client that your lab is thinking about their application rather than just running a sample.

2. Estimate how deep each test reaches, and write it down. Using the material’s approximate thermal diffusivity and the planned test time, work out roughly how far the heat will travel, keeping in mind that depth grows with the square root of time. Compare that depth with the sample’s thickness and with any layers you know about, so you can see in advance whether the test will measure only the skin, reach into the bulk, or run long enough to hit the far boundary of the sample, which would invalidate the result in the other direction.

3. Run a second, longer test on any material you are unsure about, and compare. This is the single most useful habit a lab can adopt, because it turns an invisible assumption into a visible result. If a longer test that reaches deeper gives the same answer as the short one, you have evidence that the material behaves uniformly at that scale and you can report with confidence. If the answer changes as the test reaches deeper, the material is telling you that its surface and its bulk differ, and that difference is exactly what your client needs to know.

4. Report the test time and the approximate depth measured, not just the number. A line on the report stating how long the test ran and roughly how much of the sample it represented costs nothing, protects your lab if the result is later questioned, and gives the client’s thermal engineer the context to use the value correctly. Clients rarely object to a lab that is clear about the limits of a measurement; what they object to is discovering those limits later on their own.

5. Treat disagreement with a data sheet as a reason to investigate, not just to rerun. When your result differs from a supplier’s value, repeating the same test under the same conditions will usually confirm your own number without explaining the gap. Instead, check whether the two results were obtained with different test times or different methods, and consider measuring at a second test time or by a second method. Often the disagreement is not a mistake on either side but a sign that the two numbers describe different parts of the same material.

The question worth asking on every sample

Repeatability tells you that your instrument and your technicians are working well, and that matters, but it cannot tell you whether the test reached far enough into the material to describe what your client is actually going to build with. For uniform materials that question answers itself. For the engineered composites that now make up so much of electronics testing, it is worth asking deliberately, on every sample, because the method has to match the material and the length of the test is one of the simplest ways to make sure it does.

If you have a material that keeps producing results your clients question, or one that reads differently depending on how it is tested, we would be glad to take a look. Send us a single sample and we will test it at no cost and show you not only the result but how far into the sample the measurement reached and whether the result held as the test went deeper. You can reach our applications team at thermtest.com.

FAQ

Is repeatability the same as accuracy?

No. Repeatable means the test gives the same number every run. Accurate means that number describes the material your client builds with. A short test can deliver the first without the second.

What does “thermally heterogeneous” mean?

Heat moves differently at different depths or in different directions within the material. A filled pad can look evenly dispersed under a microscope and still have a skin or carrier layer that changes how heat travels.

Why doesn’t my result match the supplier’s data sheet?

Often neither number is wrong. Different test times or methods can measure different parts of the same material. Rerunning the same test won’t explain the gap; a second test time or method usually will.

Can’t find the right product for your testing?