Can Sodium Abundances of A-Type Stars Be Reliably Determined from Na I 5890/5896 Lines?

An extensive non-LTE abundance analysis based on Na I 5890/5896 doublet lines was carried out for a large unbiased sample of ~120 A-type main-sequence stars (including 23 Hyades stars) covering a wide v_e sin i range of ~10--300 km/s, with an aim to examine whether the Na abundances in such A dwarfs can be reliably established from these strong Na I D lines. The resulting abundances ([Na/H]_{58}), which were obtained by applying the T_eff-dependent microturbulent velocities of ξ~2--4 km/s with a peak at T_eff ~ 8000 K (typical for A stars), turned out generally negative with a large diversity (from ~-1 to ~0), while showing a sign of v_e sin i-dependence (decreasing toward higher rotation). However, the reality of this apparently subsolar trend is very questionable, since these [Na/H]_{58} are systematically lower by ~0.3--0.6 dex than more reliable [Na/H]_{61} (derived from weak Na I 6154/6161 lines for sharp-line stars). Considering the large ξ-sensitivity of the abundances derived from these saturated Na I D lines, we regard that [Na/H]_{58} must have been erroneously underestimated, suspecting that the conventional ξvalues are improperly too large at least for such strong high-forming Na I 5890/5896 lines, presumably due to the depth-dependence of ξdecreasing with height. The nature of atmospheric turbulent velocity field in mid-to-late A stars would have to be more investigated before we can determine reliable sodium abundances from these strong resonance D lines.