Your QC Lab Finds the Drift 500 Substrates Too Late
In-situ metrology measures the film while the coater is still running. Here is what that actually buys you — and the four cases where it doesn’t.
Disclosure: I write for TDM Technology, a thin-film metrology manufacturer. This post was drafted with the help of AI and then checked line by line against our own published instrument pages before posting. Links below lead off Medium to our site — they are marked where they appear. #ABotWroteThis
The arithmetic nobody puts on a spec sheet
Take a line running 1,000 substrates a day. Coating happens, then a sample goes to the quality lab, and a result comes back 30 minutes later. If the process started drifting the moment the last calibration ended, up to 500 substrates get coated at the wrong thickness before anyone knows.
That number has nothing to do with how good your measurement is. It is purely a function of where in time the measurement sits relative to the process. And it’s the reason in-situ metrology keeps coming up in perovskite and coating-line conversations even though it doesn’t improve accuracy by a single angstrom.
In-situ does not mean “measured during production”
Three words get used interchangeably and mean three different things:
- Off-line — a sample leaves the line, goes to a separate lab, results return in hours or days.
- In-line — the sample is measured at a station on or beside the line, then returned. Fast, but the process step that made the film is already over.
- In-situ — the measurement happens at the point of processing, on the actual production article, while the process is running.
Only the third one closes a control loop. If you can read film thickness every second during slot-die coating, you can change pump speed, die gap or substrate velocity before the batch is wrong. The other two tell you what you already made.
Why the wet film is the only layer you can still fix
During slot-die coating the wet film is a liquid layer, and its thickness sets the dried thickness after solvent leaves. Measure it optically and you get a signal you can act on:
White light hits the wet film; the reflected spectrum carries interference fringes whose spacing is proportional to film thickness. A spectrometer reads that continuously and converts spacing to thickness using the refractive index of the coating solution.
The conversion most people skip looks like this:
Target dry thickness: 500 nm
Solids in the ink: 20%
=> Wet-film target: 2,500 nm
SE310 published range: 300 nm — 20 µm → 2,500 nm is inside it
VCD published range: 5–1200 nm → 500 nm dry is inside it
If the wet film reads 2,200 nm at substrate centre,
the operator adjusts pump flow now — not after the oven.
(Both ranges are our own published numbers: SE310 from its product page, VCD from the drying-stage module page.)
This is the whole argument. Not “better precision” — an earlier decision point, in units your line actually loses money in.
Then the film shrinks, and that’s a second problem
After coating, vacuum drying pulls the solvent out and a 2,500 nm wet film becomes roughly a 500 nm dry film. Measuring through that transition tells you whether the drying recipe is working, which the coater cannot tell you.
Get Ches’s stories in your inbox
Join Medium for free to get updates from this writer.
At this stage thickness alone is not enough. A film can land on the right thickness and still have poor optoelectronic quality — wrong drying rate, trace contaminants in the atmosphere. That’s why monitoring thickness andphotoluminescence during drying catches failures a thickness-only channel structurally misses.
Four things in-situ does not do — including on our own instruments
- Opaque coatings don’t work. Carbon pastes, metal pastes — no usable interference fringes. Optical wet-film monitoring needs a transparent or semi-transparent film on a reflective or partially reflective substrate.
- Repeatability is not field accuracy. Our SE310 publishes < 1% repeatability, but real accuracy depends on the ink’s refractive index, substrate reflectivity and film uniformity. If the refractive index shifts with temperature or evaporation rate, accuracy degrades with it. Dried-film measurement is generally more stable because solid films have more predictable optical constants.
- The wet film is not the result. Wet-film monitoring tells you what went into the dryer. If you only measure there, you never learn what came out. Lines that need tight control put a sensor at the coater and one at the dryer exit.
- Ask for a feasibility test on your chemistry first. One of our two wet-film models publishes a full spec table; the entry model does not publish numerical ranges at all. Anyone who quotes you a number for that configuration without testing your ink is guessing — including us.
What integration actually asks of you
For the production-grade wet-film module, the published integration picture is:
- the module mounts on the slot-die coating head, so measurement happens while coating runs
- Ethernet and power to a local computing server; local data processing in under 1 second
- SDK exposes measurement data into your control environment
- start / stop signals from the coater synchronise sampling with the coating cycle
- continuous single-spot sampling at 15 Hz, spot 1.5–2 mm, working distance < 5 mm
Those last two are the ones to check against your die before you plan anything — the working distance in particular decides whether the module fits at all.
For a separate wide-band route, our SE31-FM publishes 350–2400 nm transmission/reflection monitoring with dual-mode auto-switching, and serial / TCP-IP / analog / PLC integration for optical coating and semiconductor work — a different job from coating-head wet-film monitoring, and not interchangeable with it.
When you should not buy this
Being blunt: if you coat by hand, or your ink chemistry is still being decided, in-situ monitoring is money spent on a process that hasn’t stabilised yet. Bench measurement answers questions faster at that stage. And if your film is opaque, buy a different technique — no amount of inline optics will fix missing interference fringes.
In-situ earns its cost when three things are true at once: the process is continuous, the material is semi-transparent, and a wrong batch is expensive enough that a 30-minute feedback delay costs more than the module.
If you’re sizing a coating-line metrology plan, our guides are here (external links, lead off Medium):
- In-situ monitoring, two routes and published ranges: https://www.tdm-tek.com/news/guide/in-situ-thin-film-monitoring
- Reflectometry vs. ellipsometry — which technique for which layer: https://www.tdm-tek.com/news/guide/spectroscopic-reflectometry-vs-ellipsometry
- SE310 product page (full spec table): https://www.tdm-tek.com/products/in-situ/se310-slot-die-wet-film-metrology-module
