Which particular oxide's equilibrium curve should be used when brazing a complex base-metal?
Choose the oxide-curve of the constituent in that alloy that represents the most difficult-to-reduce oxide in that alloy's composition. For example, in a 304 stainless steel, the chrome-oxide curve should be the one to use, since the oxide curves for the other constituents in 304 stainless (iron and nickel) are far to the left of the chromium-oxide curve and, thus, much easier to reduce/dissociate. Always be sure a brazing furnace has the capability to deal with the "worst" of the oxide curves in any base-metal's composition.
Important Note: Only look at the oxide curves for those elements in each base-metal composition that represents about 0.5% or more in the base-metal's composition. When an element's composition is less than this amount, it usually does not have a negative effect on the brazeability of the part.
As another example, suppose you are trying to braze Inconel 738. Looking at the base-metal chemistry of that alloy, it will be noted that it contains small additions of titanium and aluminum in addition to the nickel, chromium, etc. Since the titanium and aluminum metal-oxide curves are the furthest to the right on the chart of all the significant constituents in that Inconel's composition, those are the curves to deal with in evaluating your furnace's capability to braze that base metal. But it appears that getting to the right side of those curves would require furnace operating temperatures well above the operating range of most commercial brazing furnaces and, thus, not easily achievable. In such a situation, it is wise to pre-plate the faying surfaces of the Inconel 738 with a layer of electrolytic nickel prior to brazing in order to prevent any oxygen in the vacuum furnace from reacting with the metal to form tenacious Ti-oxides or Al-oxides that can prevent any effective brazing.
This illustrates why it is critically important that the oxygen content of any gaseous atmosphere be measured right at the brazing furnace. This is easily done by use of a dew-point meter, since the measurement of the dew point of a gas is a good indicator of how much oxygen is in that atmosphere. The chart in Figure 2, taken from the same AWS article (N. Bretz and C. Tennenhouse, AWS Welding Journal, Research Supplement, pp. 189-193, May 1970.), shows this clearly.
Please note, too, that vacuum is an "atmosphere," in that there can be plenty of oxygen present in the partial pressure remaining in the furnace chamber during brazing. A perfect vacuum is only available in deep outer space. In most vacuum furnaces, operated at about 10-4 Torr, there are still plenty of oxygen atoms moving around in the furnace chamber. It's just that the "mean free path" of those oxygen atoms is such that not enough "hits" occur on the metal surface to cause damaging oxidation during a normal brazing run. Another very important part of vacuum-furnace brazing is the "leak-up rate" of the furnace because that item is key to keeping the oxygen level low in a vacuum chamber. Vacuum furnace "leak-up rates" are the equivalent, in many respects, in importance to that of "dew point" in regular gaseous atmospheres used in brazing.