In general, for the Arsenic (As), mercury (Hg) element, you could use the flame AAS to detect the concentrations. However the detection limit is too high: for As, maybe 2-3 ppm; for Hg, maybe 2 ppm, {different instrument has different detection limits}. So the results make no sense = have zero practical meanings. That is the reason why V did not recommend using the Flame AAS.
For Arsenic: one of the main sources of interference in the determination of arsenic is the molecular absorption of flame gases and solution species at the extreme ultraviolet region of spectrum where the most sensitive lines for arsenic occur ( 93.7 nm and 197.2 nm). This non-atomic absorption can be corrected by using the deuterium lamp (D2) as a continuum light source. But do not expect too much on the D2 lamp correction.
For Hg: Hg ( I ) and Hg( II) show different sensitivities in the Air-Acetylene Flame AAS. In my opinion, suppose the Hg (I) is more sensitive due to the disproportionation reactions. Also for Hg at high concentration, there are always memory effects.
So for As, Hg: At low concentrations, these elements can be determined by vapor generation technique. That is why V recommend Hydride.
From my experience, I use following parameters.As: NaBH4 0.6%, NaOH 0.5%, 5 M HCl, KI [ for As( III) is required ]
Hg: SnCl2 25% in 20% HCl, H2O.
For As, Hg testing: should use separate module to analysis for As, the other module for Hg or make 200% sure to clean the glassware to avoid contamination.