It is obvious: you desorb your silica and there is a lot of surface diffusion on the remaining silicon. You can work out the material volume displaced and so get the diffusivity of the Si atoms. Check whether this agrees with what will be tabulated somewhere for silicon.

Thanks for your input. I'm not very familiar with deoxidizng patterned wafers but only planar ones so have not observed it make such distinct structural changes to a substrate so assumed perhaps something in addition to surface diffusion was occurring.