We are fairly new to spectroscopic ellipsometry (SE) analysis and we want to be sure we are not utilizing this technique outside its capable range
We have amorphous carbon thin films characterized by both X-ray photoelectron spectroscopy (XPS) and spectroscopic ellipsometry (SE). We know a good deal about the thickness of the films and their composition from XPS analysis. However, we would like to use SE to supplement our data with information about band gap and optical properties. Thus far the results are promising and agree with our XPS characterization and we believe that "qualitatively" they are appropriate, however we want to know if they can be trusted "quantitatively" as well.
Our films are on the nanometer scale, between 3 - 7nm thick. Using a a tauc-lorentz general oscillator to model our films and known thicknesses (obtained vis XPS), we solved for unknown optical and band gap properties. If this is relevant; the Psi and Delta were obtained at an incidence angle of 65.27 degrees and between the wavelengths of 250nm(~5.0eV) to 1000nm (~1.2eV).
My questions:
Is our method of analysis appropriate / can we trust the qualitative results with regard to measured band gaps and optical constants? If we can, to what extent can we trust the quantitative results (i.e. what is the average error when measuring films in this thickness regime)?
While I have found many papers using SE to characterize films in a similar fashion, I cannot seem to find any regarding films between 1-10nm thick. Does anyone know of papers that have done similar SE analysis?
Any information would be greatly appreciated. Thank you!
I've a 25+ years experience in SE, mostly in semiconductor industry.
Repeatability is usually excellent, but for very thin films you have in general a relatively weak solution: TL model is using up to 5 parameters for each oscillator and for such thin films they correlate a lot and you can have excellent fits with very different parameters providing different refractive indexes.
Of course, the film should be very flat and the native oxide of the substrate is not negligible. The refractive index of the substate is very important too.
The fit robustness increases fitting multiple samples with the same refractive index and different thickness.
Enjoy your measurements!
Hello Michael Barclay
I can confirm that we have rigorously used ellipsometry for measurement of samples of upto 50 nanometers, and the repeatability of ellipsometry is also exceptional.
I've a 25+ years experience in SE, mostly in semiconductor industry.
Repeatability is usually excellent, but for very thin films you have in general a relatively weak solution: TL model is using up to 5 parameters for each oscillator and for such thin films they correlate a lot and you can have excellent fits with very different parameters providing different refractive indexes.
Of course, the film should be very flat and the native oxide of the substrate is not negligible. The refractive index of the substate is very important too.
The fit robustness increases fitting multiple samples with the same refractive index and different thickness.
Enjoy your measurements!
- Badges
- Science topic
- Similar topics
- Chemistry
- Nanotechnology
- Nanostructures
- Thin Films