Why I Still Trust the NETZSCH LFA 447—and What the Theory of Drift Gets Wrong

Some people in thermal analysis think the NETZSCH LFA 447 is old news. I think they're missing the point. The LFA 447 is not the newest laser flash machine on the market. It doesn't have to be. What it does—measuring thermal diffusivity with the classic flash method—is still one of the most reliable things I use in the lab.

Let me be upfront: I've been handling thermal analysis orders for eight years. I've personally made (and documented) four significant mistakes, totaling roughly $18,000 in wasted budget. Now I maintain the checklist that stops our team from repeating those mistakes. This article is not a marketing piece. It's a field opinion.

The opinion, stated plainly

The NETZSCH LFA 447 is not obsolete. There, I said it. It's older than some of the instruments I now use for comparison testing, but it still produces data I trust when the sample prep is right and the calibration history is clean. The growing habit of treating every older instrument as a 'drift risk' is lazy thinking.

The physics hasn't changed

The laser flash method is defined in standards like ASTM E1461 and ISO 22007-4. According to ASTM E1461, you hit the front face of a sample with a short laser pulse and measure the transient temperature rise on the back face. Thermal diffusivity is calculated from that curve. The LFA 447 does exactly that.

Newer instruments do the same thing with different cameras, filters, or data treatment options. Some have nicer software. Some scan faster. But the core physics has not changed. If you already own an LFA 447 that is working and calibrated, you're not measuring with outdated physics. You're measuring with the same physics. Period.

What is the theory of drift?

Let's answer the question directly: what is the theory of drift? There is no single theory. There is a measurement artifact. Drift, in thermal analysis, is a slow systematic change in the measured signal under constant conditions. It can come from electronics aging, detector response, thermocouple contamination, furnace temperature instability, or a dirty sample holder.

The popular version of the drift argument goes like this: older instruments have drifted, so their data cannot be trusted. I disagree. Drift is a symptom, not a lifetime sentence. I remember one case where an LFA 447 was supposedly drifting. It turned out that the sample holder had a small amount of graphite residue on the detector window. We cleaned it, ran the reference again, and the baseline came back. Done.

Oh, and it wasn't a 'detector aging' issue either. It was a piece of graphite. That's the problem with treating drift like a theory instead of a troubleshooting step: you stop looking for the cause.

The woven silica glass mistake

In my first year (2017), I made a classic error. I ran a woven silica glass sample—WSG for short—without thinking about surface coating. The material looked dark, so I assumed it would absorb enough laser energy. The result was a thermal diffusivity value that made no physical sense. I checked it myself, approved it, and sent it to the requester. They caught it in ten minutes. That was embarrassing.

The problem wasn't the instrument. It was my sample prep. The LFA 447 was fine. I was the error. After that, I added a step to my checklist: if the sample is a WSG composite or any material with unknown emissivity, coat it and verify the coating. That lesson has probably saved more value than the original mistake cost. Not ideal, but workable.

This is the kind of thing that makes people blame the machine when the operator was in a hurry. WSG is not a material you run every day. But it's a perfect example of why the checklist matters more than the age of the instrument.

The Ford supplier test

A few years later, I received a request from a Tier 1 Ford supplier. They needed thermal diffusivity measurements for a brake pad formulation: three temperatures, five repeats per temperature, and a tight deadline. The LFA 447 handled the whole run. If I remember correctly, the repeatability was around 1.5% on the standard samples, but don't quote me on that exact number. The important point is that the old machine produced data that got accepted by their customer. The machine wasn't holding us back.

I want to say we finished the run in a day and a half. Again, I might be misremembering the exact timeline. What I remember clearly is that the data had no drama. That is what I want from an instrument.

What NETZSCH UK actually does

One of the reasons the LFA 447 still earns its place in my lab is the support ecosystem. I have worked with technicians from NETZSCH UK, and they have helped me through calibration checks, furnace alignment, and a detector window cleaning that I thought was going to require a full replacement. Without that local support, an older instrument is harder to keep alive. With it, the LFA 447 is not a museum piece. It's a working tool.

I am not saying every older NETZSCH LFA 447 should be kept forever. There are cases where an upgrade makes sense—for example, if you need a wider temperature range or faster automation. My point is that the default assumption should not be 'old means unreliable.' The assumption should be 'check the data, check the calibration history, and then decide.'

The part people hate to hear

Let me address the obvious objection. Newer instruments have better sensors, better software, and automated heating operations. I use them too, and they are good. But when I compare data from a newer system and a well-maintained LFA 447 on the same set of reference materials, the values are close. Not always identical, but close enough for the practical decisions we're making.

You don't need a new instrument to solve every problem. You need a clear method, a clean sample, a valid calibration, and someone who understands what the numbers mean. That last part is the part no software update can fix.

Final word

The industry has changed since the LFA 447 first came out. What was best practice in 2020 may not apply in 2025. In 2020, we were still arguing about whether duplicate measurements were necessary. In 2025, we track drift in the calibration log and treat sample prep as seriously as the instrument itself. That is progress.

But some fundamentals have not changed. Physics has not changed. The value of knowing your sample has not changed. And the LFA 447, if it is looked after, can still be a reliable part of that picture.

So the next time someone tells you to replace an old thermal diffusivity instrument because of the theory of drift, ask them what exactly drifted. Then ask if the sample holder is clean. Then ask if the calibration documentation is up to date. You might be surprised by the answer.