I'm doing with niobium oxide on the silicon substrate. I used the Spectroscopic Ellipsometric measurement to analyze the thickness, roughness, and refractive index. After fitting the Si-with-absorbing-films model, MSE is approximate 0.412, and thickness is 235.76 +/- 55.899 nm. However, the deviation of roughness is too high, particularly roughness = 18.18+/- 40.808 nm. I had changed and fit different parameters but the result still looks invalid. Which parameters do I need to consider or fit? Or do we have another way to solve this problem?
Thank you!
Have you verified the roughness by another method, e.g. WLI or AFM? If your film gets too rough, ellipsometry becomes a bad method and there's nothing you can do about it.
Shine an intense white light beam on your sample to see if it is that rough. (as Jürgen Weippert mentioned, an AFM value is even better)
Also:
- Is the back surface of Si rough or polished; do you take that into account when fitting?
- If possible, measure the depolarization spectrum with Psi and Delta (you can do it easily with a JAWoollam spectroscopic ellipsometer); the depolarization spectrum helps seeing any roughness or undesired back reflectance; fitting it at the same time as Psi and Delta is a good way to get the roughness right.
- The confidence limit given after the fit can be a sign of a large correlation between fitting parameters.
Firstly, you have to check parameter uniqueness for roughness value. If it is not sharply falling to a universal minima, your model is wrong. But if it is showing sharp value then there is problem with your deposition. In first case, I suggest you to switch to basic models.
My experience to use a Spectroscopic Ellipsometric measurement is to have another equipment such as a surface profiler to work out the film thickness rather use the model to extract the parameter of thickness. I fully agree with Jürgen Weippert.
When you have a huge error, it means the roughness is correlating with another parameters, so you can try fixing one of them and just try to refine the roughness parameter.
Another important point is that mean roughness analysis in ellipsometry considers a layer of EMA (Effective medium approximation) mixture of 50% material and 50% void. So your top layer could actually be more porous than 50%/50%, and you can try to set the Roughness parameter off, and create an EMA 50%/50% VOID/MATERIAL layer on top to act as the roughness. You can refine the thickness of this layer and the VOID/Material ratio (it should be close to 50/50).
This actually worked for me in the past with the same issue that you are presenting here. I employed the CompleteEASE software.
Regards
- Badges
- Science topic
- Similar topics
- Biological Science
- Physiology
- Physiological Processes
- Stress