5. Discussion, conclusion and future plans
A grid of equivalent widths for 23 N ii lines in the region 3950 Å-6500 Å was calculated in both NLTE and LTE approaches. Calculations were performed for the temperature spanning the interval from 16000 K to 31000 K.
Let us briefly discuss the results of these calculations. Note, that similar study of the behaviour with a temperature of the nitrogen lines was carried out by BB (Becker & Butler, 1989). They investigated trends in the equivalent width as a function of , microturbulent velocity and initial nitrogen abundance. Atmosphere models, used for analysis, were those of Gold (1984). Some of the investigated lines showed a strong NLTE effects, while for line 4432.66 Å the difference between LTE and NLTE predictions was negligible.
We decided to undertake the similar study of the line equivalent width behaviour with a temperature, but based on Kurucz (1992) grid of models. To have the possibility to compare our results with those of BB, we used the same initial parameters (, , ) as above mentioned authors did. Note that adopted for this step of the analysis value () differs from that used for Peg, but it can be considered as a more typical for B stars value.
Our results generally agree with those of BB (see Figs. 2a-4a and Table 5 for selected lines. Some dependence of the equivalent width upon the nitrogen abundance for , and are also shown on Figs. 2b-4b). Nevertheless, there are some important supplements, which should be mentioned.
Table 5. Equivalent widths of selected nitrogen lines calculated with , and .
Firstly, at low temperatures the differences between LTE and NLTE equivalent widths are smaller than at higher temperatures. Secondly, the maximal discrepancy between equivalent widths calculated in LTE and NLTE approach slightly greater than derived by BB. And finally, maximum of the calculated equivalent width for all investigated N ii lines is slightly shifted towards the lower temperatures comparably to BB maximum. The difference in temperature achieves 2000-2500 K. To our mind, such a discrepancy is caused by the different grids of the atmosphere models used in the analyses. BB used Gold's (1984) grid, while we employed Kurucz (1992) models.
Similar conclusion was recently made by Cunha & Lambert (1994) in their work on chemical evolution of Orion association. In that extremely interesting study of 18 main sequence B stars authors used for determination of the elemental abundances (of the nitrogen, in particular) LTE calculations based on the use of Gold's (1984) and Kurucz's (1979) grids of models. Simultaneously, they took into account the NLTE corrections for investigated lines using the grid of NLTE equivalent widths calculated by BB. Those calculation, as it was already mentioned, were based on the Gold's (1984) grid too. Then, having inspected LTE equivalent widths calculated with both Gold's and Kurucz's grids, the authors made a conclusion that LTE equivalent width of a given line reaches the maximum at higher effective temperatures for lightly blanketed Gold's models and at 3000 K lower temperatures for heavily blanketed models of Kurucz's grid.
After the statement that more heavily blanketed models would be more realistic, Cunha & Lambert (1994) performed a preliminary assessment of NLTE abundances that would result from use of Kurucz's models, assuming that 1) NLTE effects are identical for two considered grids of models and 2) the difference between LTE abundances derived using these grids of the models should produce the same difference between corresponding NLTE abundances. As a result of such supposition, they indirectly obtained the mean value of NLTE nitrogen abundance for sample of 18 early B stars: =7.65 0.09. The nitrogen abundance determined by us for Peg is in excellent agreement with that inferred by Cunha & Lambert (1994) for their sample of the stars.
Our precise NLTE calculation of carbon (Korotin et al., 1998) and nitrogen (present work) abundances in Peg show that these elements are both deficient in the atmosphere of this star =8.30 and =7.65, when compared with solar values. It implies that CN-cycled material is not present in the superficial layers of program star. Probably the turbulent diffusion, proposed by Maeder (1987), as a mechanism responsible for CN anomalies in the main sequence stars) is not operating in Peg, provided its rotation is slow.
In the next papers we plan to present the results on slowly rotating main sequence B stars and stars with high values. We hope that such detailed study can help to understand whether the turbulent diffusion induced by rapid rotation is operating in some main sequence stars and what is the real magnitude of CN anomalies caused by the turbulent diffusion.
Finally note, that Table 1 provides several nitrogen lines with deviation between LTE and NLTE calculated equivalent widths less than 20%. Such lines can be used in LTE analysis of the nitrogen abundance in the hot main sequence B stars.
© European Southern Observatory (ESO) 1999
Online publication: February 23, 1999