Built-in HDR modes on cameras do one of two things:
- Take a single exposure, and compress shadow and highlight detail when creating a JPEG from it.
- Take multiple exposures in quick succession with different (bracketed) exposure parameters, and blend them together into a single JPEG.
I say JPEG in both cases because neither of these make sense for a RAW file.
In case #1, the raw capture just includes all available sensor data, with no particular rendering. It would be possible to provide advisory information saying "hey, the user would like this processed to maximize dynamic range", but the actual data wouldn't be changed.
In case #2, each exposure would be a separate RAW file; it would theoretically be possible to create a multi-exposure RAW container with the data from multiple shots, but to my knowledge no one does that. Just bracket yourself (either manually, or automatically if the camera has this feature) and take your multiple shots into image-processing software yourself.