Example: I have Raman spectra of a silicon wafer obtained by diamond wire sawing, I it is having different phases like Crystalline Si, a-Si, Si-III etc.. as a result there are multiple peaks in the spectra.
Stress can be a highly highly localized phenomenon. Is your Raman setup a confocal microscope? Then you could do spectromicroscopy and resolve stress variations in crystalline and amorphous zones.
However, what bothers me here is that you say the sample stems from sawing a wafer which should be basically single-crystalline. Did you apply a cleaning procedure like RCA/ultrasound/... after the sawing in order to remove "sawdust"?
Jürgen Weippert thank you for your response. Yes it is a confocal Raman microscope, I plan to map surface defects to the residual stress in wafers in as sawn condition after cleaning.
But I am stuck with obtaining residual stress information and I couldn't get the papers related to that.
We here are working on diamond for which I could provide plenty of material, for Si I am not experienced, but have you seen this one here?
https://solinstruments.com/support/learning-center/silicon-stress-measurements-by-raman-spectroscopy/
I reviewed this still there are using single peak shift to quantify residual stress.
Yes, but that is normally the case. Not every peak is suitable to be converted to stress directly, for some vibrational modes the relationships are so complicated they can't suitably be used while for others it's straightforward and these are the single peaks that are used in such guidelines.
Ok, so is there any way by which we could uncover the information about residual stress hidden in the spectrum by any machine learning models?
Because we don't know exact relationship between vibrational modes and stress state.
If the existing methods don't suffice, you could simulate Raman spectra by DFT and compare with and without stress:
https://www.degruyter.com/document/doi/10.1515/zkri-2018-2112/html?lang=en
- Badges
- Science topic
- Similar topics
- Mathematics