Please note, first of all, that the aluminum-oxide layer on the surface of the aluminum is very thin – only about 40-angstroms thick! That’s VERY thin, since an angstrom is only about one ten-billionth of a meter thick.
Even when warmed up a bit, it’s only about 60-angstroms thick. And because it is so thin and because the aluminum base metal expands almost four times as fast as the oxide layer on its surface, the aluminum oxide breaks up on the surface of the aluminum base metal (just as the ice on frozen rivers breaks up in springtime forming individual patches of floating ice). This creates openings between the “floating patches” of oxide (Fig. 2) into which the molten BFM can flow.
But because oxygen is always present in any furnace brazing atmosphere – either as free oxygen (as in our air) or as part of a water molecule in the atmosphere’s moisture content (as measured by its dew point) – these oxygen molecules will very aggressively try to enter those cracks in order to react with the aluminum base metal to form new Al-oxides so as to “heal” those cracks on the metal’s surface.
Aggressive fluxes are used in atmosphere brazing to coat the surface of the joints to be brazed so that the oxygen cannot get through that layer of flux to heal the cracks. In vacuum brazing of aluminum, magnesium is often used to “getter” any oxygen in the vacuum chamber (i.e., to quickly react with and tie up the oxygen) and thus prevent it from reacting with the aluminum base metal.
By tying up the oxygen in that way, using either flux or magnesium, the molten BFM can enter into those “cracks” to alloy with the clean aluminum metal below the oxide layer. Interestingly, the molten BFM can also literally get under those oxides, float them away and make a sound brazed joint. This appears to be a unique feature of aluminum brazing, and I am personally not aware of any other metal that can behave in a similar manner.
Many years ago, a lab test was performed in which a pencil mark was placed on top of the aluminum surface that was to be brazed (thus, the pencil line was sitting on top of the aluminum-oxide layer on the surface of the aluminum part being brazed). The pencil mark was still there after brazing, even though it was now quite visible on top of the well-brazed aluminum joint below it.
All that is really needed when brazing aluminum is to thoroughly clean off any oils, lubricants, etc. from the surface of the parts via degreasing followed by a clean water rinse and then perhaps an alcohol rinse. Trying to remove the adherent aluminum-oxide layer is not necessary since that layer will break apart during brazing, thus allowing the molten BFM to effectively braze the aluminum components together.