I do not understand the question completely, but from my understanding is that you have a 3D structure grown out of the sapphire substrate.  I would recommend using SEM or HRSEM depending on the spatial resolution necessary for your application.  The benefit of SEM is that you do not need to transfer the sample, and you can image it directly on the sapphire. 

Dear professor, Thanks for your answer. Our sample is two-dimensional material and may be it is monolayer. So we want  to characterize it's structure and performance by the TEM techno..So we want remove the sapphire without destroy the area we observed. So i think the method you recommended can't satisfied our need. We want to do more works.

With best wishes.

hello there, to transfer monolayer WSe2 from sapphrie onto cooper grids, you need to firstly spin coating a polymer support (i.e. PMMA) onto your sample. Then, immense your sample into concentrated NaOH solution, and just wait for the WSe2/PMMA membrane detaching from sapphire substrate. 

Dear prof. Qingqing, Thank you for your answer. Cool, we'll discuss the feasibility about this method, Thank you 

Best regards