Formal proof that if a grammar is LR then it isn’t ambiguous

Title says it all. I’m looking for a formal proof that there can’t be two parse trees for a same sentential form if the grammar is LR.

Can you help me?