A thin layer changes the timing of returning light
A coating on a lens can be too thin to inspect from the side, yet light encounters two separate boundaries. One part reflects where air meets the coating. Another enters the layer and returns from the substrate underneath. Weaker contributions make further round trips. Their electric fields add before the detector measures the result.
Increasing thickness lengthens the internal journey. Since the field oscillates, that extra path changes whether returning contributions reinforce or oppose one another. Position within an oscillation cycle is called phase. A small thickness change may leave reflected brightness nearly unchanged while shifting its phase appreciably.
Sources: [3]
The electric field can trace an ellipse
Imagine watching the electric-field arrow at one fixed place along a beam. In linearly polarized light it moves back and forth along a straight line. Resolve the motion along two perpendicular directions. If both components cross zero together, their sum stays on a line. Give them different timing and the arrow tip can travel around an ellipse.
An ellipsometer sends a known linear polarization onto the sample at an oblique angle. The beam and surface normal define a plane. The electric-field component in that plane is called p; the perpendicular component is s. Reflection generally changes their amplitudes and phases differently, so a straight-line input can leave as an ellipse.
Two measured angles describe the relative amplitude and phase change. Comparing components lets some common intensity variations cancel, although real optics still have axis offsets, imperfect delays and wavelength-dependent detector response. These effects require calibration.
A rotating analyzer turns polarization into changing brightness
A photodetector integrates light intensity during an exposure. It cannot follow each visible-light field oscillation. An analyzer placed before it transmits one electric-field direction. Rotate the analyzer and the brightness rises and falls. That intensity curve contains the orientation and shape of the polarization ellipse.
Recovering the full phase also requires distinguishing the sense in which the ellipse is traversed. A compensator adds a known relative delay between perpendicular components. Another intensity scan then separates polarization states that a linear analyzer alone can confuse. Known reference states relate optical response to electronic counts; a different reference tests that calibration. Stable reference and sample illumination are part of this measurement.
Several colors can separate films that match at one color
The measured polarization change must still be interpreted. Refractive index describes propagation inside a material, and thickness determines the distance traveled. Both affect the returning phase. An analysis starts with a proposed layer structure and substrate, calculates its reflection, and varies the unknown parameters to match the observations.
For a transparent layer at one wavelength, increasing thickness by a whole internal round-trip cycle repeats the reflection. A thicker film can therefore give the same response as a thinner one. At another color the thickness needed for a whole cycle changes. Candidates that coincide at one wavelength often separate across several wavelengths. Additional incidence angles supply further constraints.
The light must also carry information from the lower boundary. If film and substrate have identical optical properties, that boundary supplies no separate reflection. Strong absorption can prevent a useful field from returning through the film. A close fit can then allow a wide thickness range. Reliable thickness interpretation depends on the optical path and the material model that explains it.