Thank you for the link, that is what i was looking for. The AD9363 block diagram is not that detailed :( I do not question the non-linear behaviour of the output driver.
@Sigi: Thank you for your explanations. I was asking about your data as I could not see what you consider to be a spurious emission.