Written and maintained by CASRAI Editorial Board
Last updated
An ellipsometer is an optical instrument that measures how the polarization of light changes when it reflects from (or passes through) a sample, and uses that change to work out the thickness and optical constants of thin films. It reports two angles, Ψ and Δ, and the thickness is then fitted from a model rather than read off directly. This page is general equipment information for lab staff, core-facility users and anyone reading an ellipsometry result.
Short answer
An ellipsometer sends polarized light onto a surface at an oblique angle and measures the polarization of the reflected beam. The change is expressed as an amplitude ratio (Ψ) and a phase difference (Δ). Software then fits an optical model of the sample to those values to return film thickness and optical constants (such as refractive index). Because the answer depends on the model, the questions to ask about a reported thickness matter as much as the instrument itself.
Definition and scope
J.A. Woollam, a manufacturer, defines ellipsometry as measuring “a change in polarization as light reflects or transmits from a material structure,” expressed as Ψ and Δ. The measured response depends on the optical properties and thickness of the individual materials, so the technique is used for film thickness and optical constants and, indirectly, for properties tied to the optical response such as composition, crystallinity, roughness and doping concentration. It was developed for nanometer-scale layers in microelectronics and is applied to dielectrics, semiconductors, metals, superconductors, organics, biological coatings and composites (J.A. Woollam, ellipsometry tutorial, read 2026-11-07).
How an ellipsometer is used in practice
The principle: polarization change, not intensity
A reflectometer measures the intensity of reflected or transmitted light. An ellipsometer measures the change in polarization, which is a ratio of values rather than an absolute intensity. According to the manufacturer’s FAQ, this means quantitative results do not need a calibrated intensity standard, and the phase term Δ gives much higher sensitivity to very thin films; the same page says a reflectometer cannot accurately measure intensity to better than 0.1% (J.A. Woollam, ellipsometry FAQ, read 2026-11-07). Measurements are typically made at oblique angles near the Brewster angle, commonly in a 50° to 75° range.
Components
The basic layout described in United States patent 5,877,859 (a broadband spectroscopic rotating-compensator ellipsometer) is a polarizer and an analyzer, one fixed and one rotating. That patent notes that rotating-polarizer or rotating-analyzer designs are “incomplete” polarimeters: they are insensitive to the handedness of circular polarization and perform poorly near fully linear or partly depolarized light. Adding a rotating compensator addresses this. The Woollam FAQ likewise states that its systems include a compensator, and that CCD detection allows a full spectrum to be read in parallel. A patent is a vendor-grade source, so treat it as a description of one design, not of every instrument.
From Ψ and Δ to a thickness
The measured Ψ and Δ are not a thickness. Analysis is model-based (J.A. Woollam, data-analysis tutorial, read 2026-11-07):
- Build a model of the sample layers.
- Calculate the predicted response with the Fresnel equations from the thickness and optical constants.
- Compare it with the measured data and vary the unknowns.
- Quantify the fit with the mean squared error (MSE) and choose the lowest MSE, checking that regression has not stopped in a local minimum.
Direct inversion of Ψ and Δ gives only “pseudo” optical constants and assumes there are no surface layers. Real samples have oxide or roughness, so “regression analysis is required because an exact equation cannot be written.”
Types and related techniques
| Variant | What the sources say |
|---|---|
| Single-wavelength ellipsometry | With two measured values (Ψ and Δ) only two unknown properties can be determined, so even a single film on a known substrate cannot be uniquely determined (Woollam FAQ). |
| Spectroscopic ellipsometry | Data over many wavelengths over-determines the film thickness; the FAQ’s example is 100 wavelengths giving 200 data points for 101 unknowns. Patent 5,877,859 describes serial (change wavelength between measurements) and parallel (many wavelengths at once) approaches. |
| Imaging ellipsometry | A vendor application note (Park Systems, snippet read 2026-11-07) describes setting the polarizer and analyzer for contrast at a chosen wavelength and angle. Its figures of about 1,000 spectra per second and thickness sensitivity below 0.5 nm are the vendor’s own claims. |
| Reflectometry (compared) | Measures intensity rather than polarization change; less sensitive to very thin films according to the Woollam FAQ. |
Where it matters for labs and supply decisions
Before relying on a reported thickness, these checklist questions follow directly from the model-based method described above:
- Is the sample type suitable? For transparent films the only constraint is that light can pass through to the substrate and back. Absorbing films must be thin enough, typically under 50 nm, that the beam is not fully absorbed. The preferred limit for most visible to near-infrared measurements is below 5 microns; near- or mid-infrared can reach films up to 50 microns, but film uniformity becomes much more important (Woollam FAQ).
- Was the film ultra-thin? For very thin layers the FAQ states that thickness and refractive index cannot both be determined with adequate sensitivity, so an approximate index is usually assumed, often from a thicker film. The reported thickness then inherits that assumption.
- Do the unknowns exceed the data? The Woollam tutorial says the number of unknown properties should not exceed the information in the data.
- What MSE was reached, and was a global minimum checked?
- Were surface layers modelled? Pseudo optical constants from direct inversion ignore oxide and roughness.
- How many angles were used? One angle is enough for standard thickness and index measurements; the FAQ recommends 2 or 3 for films over 1 micron, index-graded or anisotropic films and complex multilayers.
For related optical instruments see polarized optical microscopy, instrument checks in spectrophotometer calibration, and sputtering, a common way thin films are deposited before they are measured.
Common misunderstandings
- “It measures thickness directly.” It measures Ψ and Δ; thickness comes from a fitted model.
- “A low MSE proves the answer.” A fit can sit in a local minimum or use more unknowns than the data support.
- “Direct inversion is enough.” It yields pseudo optical constants and ignores surface layers.
- “It only suits semiconductors.” It began in microelectronics but is used on organics, biological coatings, metals and composites.
FAQ
What does an ellipsometer measure?
The change in polarization of light on reflection or transmission, reported as Ψ (amplitude ratio) and Δ (phase difference), from which thickness and optical constants are modelled.
What is the difference between ellipsometry and reflectometry?
Ellipsometry measures polarization change as a ratio and does not require a calibrated intensity standard; reflectometry measures absolute intensity and is less sensitive to very thin films (Woollam FAQ).
What is a spectroscopic ellipsometer?
One that measures across many wavelengths, which over-determines the unknowns and removes the ambiguity of a single-wavelength measurement.
Does the page give ellipsometer prices?
No. Pricing varies by configuration and is quoted by manufacturers.
Sources
- J.A. Woollam Co., What is ellipsometry? (read 2026-11-07)
- J.A. Woollam Co., Ellipsometry data analysis (read 2026-11-07)
- J.A. Woollam Co., Ellipsometry FAQ (read 2026-11-07)
- US Patent 5,877,859, Broadband spectroscopic rotating compensator ellipsometer (USPTO, read 2026-11-07)
- Park Systems, imaging spectroscopic ellipsometry application note (vendor, read 2026-11-07)
