Interference Thickness Measurement

Quick Start

Just about any clear coating, wet or dry, on a smooth substrate can be measured via a very simple reflectance spectrometer. The pattern of the reflected light is specific to its thickness. In the app you can see how the pattern changes with thickness nmcoat for a given refractive index of the coating and substrate. So in principle you could slide the thickness slider to match the on-line pattern and read off the thickness. In a real device the pattern is extracted via a Fourier Transform.

We have an optically clear and smooth coating on top of a base substrate. Like all materials they have a refractive index (RI) input here as RICoat and RISubs. The coating has a thickness of nmCoat. You shine a light (usually via a fibre-optic probe) at the coating with a wavelength range of λMin to λMax. The same probe receives the reflected light and sends it to a little USB spectrometer. It measures a spectrum as shown in the graph. In this app we input the thickness and get a spectrum. But the thickness gauge takes the spectrum and calculates the thickness. It does it by (effectively) counting the number of peaks across the wavelength range. The thicker the coating the more peaks you get (try this for yourself).

As you play with the values you quickly find the limits of the method.

  • For very thin coatings there aren't any peaks/troughs to count. You have to use smarter algorithms to extract data from the curvature of the spectrum, or else you use a very wide wavelength range.
  • For very thick coatings there may be too many peaks/troughs.
  • If the RI of the coating is close to the RI of the substrate then the reflected signal is very weak so the thickness is hard to measure.
  • Any colour in the coating will give its own spectrum which might confuse the thickness algorithm
  • The "thickness" is the optical thickness not the physical thickness. You have to divide by the RI to get the physical thickness. Usually it's simple to calibrate the measurement, but a smart device can work out the RI from the height of the reflection peaks - the bigger the RI, the bigger the height.

There is a bonus with this method. If you have a two-layer coating (and if their RI are different from each other) then you get reflections from the top/middle and top/substrate interfaces, giving you two thickness measurements, so both coatings are measured at the same time.

The method works well for wet coatings - so you can get instant read-out straight after the coating head, though calibration of the physical thickness is trickier because the RI is less easy to define.

Summary

  • Accuracy: Dependent on RI differences, thickness, wavelength range - but easily gives you 1μm accuracy for "normal" coatings.
  • Size: Small, light
  • Safety: Safe
  • Scanability: Easy to scan, and to use in spot mode
  • Cost: Low
  • Wet/Dry: Works on both
  • Big limitations: Needs optical clarity and smoothness