Disclosure: this post was drafted with the help of AI, then checked line by line against our own instrument documentation before publishing. #ABotWroteThis
Every thin-film engineer hits the same fork eventually: reflectometry or ellipsometry? Both are
non-contact, both are optical, both have been on the lab bench for decades. But they measure different
things — and that difference decides which one you can actually run at production speed.
TL;DR — Reflectometry tells you how much light came back. Ellipsometry tells you how the light
changed. If you need throughput, reflectometry wins. If you need n and k for an unknown film, you
need ellipsometry. Most real fabs and coating houses run both, because they sit at two different
stages of the same QC pipeline.
The difference is one physical quantity
Reflectometry measures intensity. It shines broadband light on the film and records the reflectance
spectrum — how much light returns as a function of wavelength. That one-number-per-wavelength curve is
fitted against an optical model to extract thickness, and n/k only if the optical constants are already
known independently.
Ellipsometry measures polarization change. Instead of intensity, it tracks two parameters at every
wavelength: Psi (Ψ, the amplitude ratio of p- to s-polarized light) and Delta (Δ, the phase shift).
Because it is a ratio measurement, it doesn't need a reference sample, it is inherently more sensitive,
and it delivers the full n(λ) and k(λ) dispersion alongside thickness.
Where they actually diverge
| What you need | Technique | Why |
|---|---|---|
| Thickness only, high throughput | Reflectometry | Milliseconds per point, tolerates vibration, minimal sample prep |
| n and k of an unknown film | Ellipsometry | Intensity alone can't independently separate n, k and thickness |
| Very thin films, sub-nm confidence | Ellipsometry | Δ is exquisitely sensitive to ultra-thin layers |
| Watching a film grow inside a chamber | Reflectometry | Fast, robust to chamber vibration and viewport constraints |
| Qualifying a brand-new multi-layer stack | Ellipsometry | Solves thickness + n/k per layer simultaneously |
| Band gap from the k spectrum | Ellipsometry | Only ellipsometry gives you k(λ) directly |
The line people get wrong is speed. Spectroscopic ellipsometry costs seconds per point, minutes for a
full-wafer map — fine in an R&D lab, not fine at 100% inline inspection. Broadband reflectometry runs
milliseconds per point and can map a substrate in seconds. If your bottleneck is throughput rather than
single-digit-angstrom precision, reflectometry is the tool.
Before you trust any spec table — including ours
Four checks that have saved me more than once:
- Thickness range and spectral range are two different fields. A wavelength span (e.g. 350–2400 nm) is not a thickness range. Datasheets get misread this way constantly, and the error propagates into comparison spreadsheets.
- One model name can cover several sub-models with different ranges. "1 nm to 3 mm" and "down to 20 µm spot" can be true of the same product family and still be different machines. Always resolve to the exact sub-model.
- Vendor claims and third-party process notes are not the same evidence class. If a university technique note reports a 10 µm spot for a tool whose product page says 1 µm, that's usually "what was achievable in that process", not a contradiction. Keep them in separate columns.
- If a guide and a product page disagree, trust the product page. On our own site we now treat the instrument page as the record source and the article as the commentary.
How the two techniques sit on a line
- Inside the deposition chamber — broadband reflectometry. Our in-situ SE31 / SE310 modules do real-time wet-film monitoring during growth (non-contact, ≤5 µm films), catching drift and endpoint before the substrate leaves the chamber.
- On the line, right after coating — reflectometry plus PL and T&R. The SE600X measures thickness 5–1200 nm at 10 Hz across 31 points, no sample prep.
- On the bench, for process qualification — our SE100 reports thickness, reflectance, transmittance, n, k, band gap and PL; specified thickness range 5–1200 nm, repeatability ≤±1%, calculation under 10 seconds, samples from 10×10 mm up to 220×220 mm.
That last one is where the honest caveat belongs: how n and k are derived depends on the optical model
and the stack you're measuring. Which measurement mode and model fit a given film is the conversation
worth having with an application engineer before you buy anything — including from us.
Bottom line
Reflectometry and ellipsometry aren't competitors; they're two stages of the same quality pipeline —
reflectometry keeps the line moving, ellipsometry tells you why it moved. If you're standing up a
metrology plan for perovskite or semiconductor films and want the comparison run against your actual
stack, our guides are here:
- Spectroscopic Reflectometry vs. Ellipsometry: https://www.tdm-tek.com/news/guide/spectroscopic-reflectometry-vs-ellipsometry
- Spectroscopic Ellipsometry for Thin Film Metrology: https://www.tdm-tek.com/news/guide/spectroscopic-ellipsometry-thin-film-metrology
- Thin Film Metrology for Semiconductor Manufacturing: https://www.tdm-tek.com/news/guide/thin-film-metrology-semiconductor-manufacturing
Top comments (0)