I deposited a 50-nm-thick Pt film on a silicon substrate using magnetron sputtering method. TEM results showed the presence of SiO2 layer between Pt and Si-sub. Strangely, between SiO2 and Si-sub, a very thin Pt layer was also found, and furthermore, localized heating revealed that this Pt layer continues to diffuse towards Si-sub. How can I explain this phenomenon?
Thank you!
May be a thin Pt silicide has formed by the impact of Pt atoms. The silcide oxidizes depending of temperature and atmosphere (e.g. residual gas during annealing).
A phase diagram is here: https://www.heraeus.com/media/media/hpm/doc_hpm/pt_labware/Correct-Use-of-Platinum.pdf
Likely the oxide film is very thin. Then Pt atoms could indeed penetrate it during magnetron sputtering as Stefan suggested.
Then Si and Pt is very reactive, especially during heating. The siliciding occurs and then Pt diffuses into Si readily.
Did you remove the native oxide of the silicon just before you started the sputter process? This oxide is usually 2-3 nm thick. Looking at the scale of the picture it might just be this native oxide you see, which is penetrated during sputtering by Pt atoms. Pt then easily diffuses into Si to form Pt silicide. You could check this by removing the native oxide first with a bufffered oxide etch (BOE) and then start the Pt deposition. You'll then probably see a slightly thicker Pt silicide but no oxide layer.
If you did remove the native oxide prior to deposition, I agree with Stefan that it can be post-sputter oxidation of the silicide. This very much depends on the wafer temperature which is reached during sputtering and the oxygen background of the sputtersystem.