Here's a reference for purity levels for various grades of CO2 (from the same company hvac5646 used). It lists welding/industrial grade CO2 as 99.8% pure, and less than 32ppm H2O.
CO2 is NOT an inert gas as some have stated. At least at steel welding temperatures it is known to react with the weld puddle, causing embrittlement. Not saying that it would be inappropriate as a braze-purge though.
I would like to see a comparison of brazed joints (cut open) using N2, CO2, and no purge gas (not open tubing like I have seen on YouTube, but with ends nearly sealed, just enough to allow pressure to escape but not allowing free flow of air). I am not aware of any published tests thus.