2
$\begingroup$

tl;dr I simply can not understand the orientation of the upper 45° optical surface.

In the semiconductor industry optical metrology is widely used to measure a combination of thin film thicknesses via interferometry and 2D/3D nanostructure dimensions not by optically resolving them but by recording their spectral response over a wide range (near UV to near IR) and comparing the spectra to rigorous EM models (based on dispersion models for the materials) as well as recorded spectra from reference samples. This allows the throughput necessary for manufacturing that high resolution scanning electron microscopes struggle with since they require at least a partial vacuum at the sample for direct imaging of nanometer-scale structures.

Scatterometry and Optical Critical Dimension Metrology based on polarized reflectance spectroscopy and/or ellipsometry is described further in (just for example) Optical Critical Dimension Metrology for Semiconductor Manufacturing (Vacuum Technology & Coating, Sept. 2021, pp 33-37)

But the diagram below is driving me crazy because I can not understand how the optical system depicted can work.

On the sides are the two arms of a variable angle spectroscopic ellipsometer - the high NA objectives subtend a significant range of angles and the rotating polarizers are indicated. I'm not asking about this system. One arm has a broadband light source and the other has a spectrophotometer.

There is also a second optical system normal to the sample surface. It has its own light source and spectrophotometer. There are two stand-alone optical elements that are shown with surfaces at 45° to the optical axis. (a third is integral to the spectrophotometer)

The lower one appears to dead-end to the left and that's probably a camera for alignment - the system only works on certain repetitive/periodic structures that can be modeled, sometimes in memory areas of a chip, or on dedicated test structures in the scribe lanes between chips. By necessity these are sometimes small and the system must align the optical spot to them.

Thus I assume this represents a beam splitter or half-slivered mirror of some sort.

The other is associated with the broadband light source, and the thin element between the light source and the 45° element that looks just like the other two linear polarizes may indeed be a polarizer (sans arrow indicating rotation) or just a collimator lens.

But what confuses me is the orientation of this 45° upper optical surface (not the one integral to the spectrophotometer); it seems to be able to send the light directly up into the spectrophotometer, but not towards the sample.

I'm assuming that if the normal incident light never reaches the sample, then it can't be measured.

Question: Am I missing something? Are there broadband beam splitters with surfaces in this orientation that can send the light downward? Could the diagram be essentially "wrong"? How do normal incidence reflectometers work (optically) and why can't I understand the upper beamsplitter in this diagram?

optical metrology system for semiconductor manufacture, from https://www.novami.com/nova-technology/spectral-interferometry/

Source: https://www.novami.com/nova-technology/spectral-interferometry/

Improve this question
$\endgroup$
2
  • $\begingroup$ Ideally this would be asked in Optics and Photonics SE but it doesn't exist yet... $\endgroup$
    – uhoh
    Commented Nov 7, 2022 at 23:44
  • $\begingroup$ @EdV If the lower one is for a camera (there has to be a camera somewhere as explained) then I think it's okay. It's perfectly fine to call it a beamsplitter even when using it as a combiner, something about the direction of time... :-) $\endgroup$
    – uhoh
    Commented Nov 8, 2022 at 1:24

1 Answer 1

Reset to default
1
+200
$\begingroup$

enter image description here

I'm not sure that this was that bothered you though... Anyway, if these 45° optical elements are really beam-splitters, there is really no other way around to make it work (optically speaking). Or maybe there are some hidden optical elements willingly omitted for the sake of simplicity.

Improve this answer
$\endgroup$
4
  • $\begingroup$ I'm happy to know that you've correctly perceived my conundrum, but I don't think we're quite there yet; the light can now pass through and reach the sample, but now there's no way for the returning reflected/scattered light to reach the spectrophotometer. To me, I'm thinking that the fix is to leave all the major components and optical paths in place as-is, but perhaps to simply rotated the beam splitter 180° around a vertical axis? $\endgroup$
    – uhoh
    Commented Dec 22, 2022 at 12:37
  • $\begingroup$ Huh... ok I see. I edited the figure a bit, adding the ray path. It could work like this but with a massive loss of optical power (something like $\frac{1}{2^4}=\frac{1}{16}$ of the input power remaining at the spectrometer). $\endgroup$
    – dolun
    Commented Dec 22, 2022 at 13:30
  • $\begingroup$ To me, rotating the beam splitter by 180° will only turn the reflected part by the same angle. $\endgroup$
    – dolun
    Commented Dec 22, 2022 at 13:31
  • $\begingroup$ Yes that works quite nicely! :-) The alternative way to do the same thing via rotation of the polarizer about the vertical optical axis leaves all the big stuff in their original location: i.sstatic.net/0fjyC.jpg $\endgroup$
    – uhoh
    Commented Dec 22, 2022 at 21:16

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.