Nitrogen abundances in extragalactic H II regions are almost entirely derived from optical [N II] lines alone, because the other important species, N III, has strong emission lines only in the UV and FIR. Photoionization models generally predict that N+/O+ = N/O under most conditions. Nevertheless, IR measurements of [N III] / [O III] in Galactic H II regions consistently find a steeper N/O gradient than that obtained from optical measurements of [N II] / [O II] (Lester et al. 1987; Martín-Hernández et al. 2002). This suggests that ionization corrections may be important (Garnett 1990). Direct comparison between [N II] and [N III] measurements in H II regions with varying properties is needed to understand the nitrogen ionization balance, so that the variation in N with metallicity can be studied accurately. Comparison with measurements in stars is also helpful, and it should be noted that measurements of abundances in B stars (e.g. Rolleston et al. 2000, Korn et al. 2002) yield O and N abundances in good agreement with the values for H II regions in the Milky Way and the LMC and the Local Group spiral M33 (McCarthy et al. 1995; Monteverde et al. 1997).
With this uncertainty in mind, Figure 18 shows how N/O varies with O/H from optical spectroscopy of H II regions in spiral and irregular galaxies. It has been known for some time that N seems to have two components: one component which follows O in a fixed ratio (log N/O -1.5) for log O/H < -3.7, as inferred from the constant N/O vs. O/H in metal-poor dwarf irregular galaxies (open circles), and a second component that increases faster than O at higher O/H as seen in spiral galaxies (filled circles and squares, crosses). The second component is produced via the classical CNO cycle during hydrogen burning in stars and requires the presence of C and O in the star from birth ("secondary N"), while the first component is postulated to come from the CN cycle on freshly-synthesized C (from He-burning) which has been convectively "dredged-up" into a hot H-burning zone at the base of the convective envelope, and does not require an initial seed of C or O (hence, "primary" N). The latter process is most commonly thought to occur in the asymptotic giant branch (AGB) stage of intermediate mass stars (Iben & Truran 1978), but has been found to occur in models of massive stars with increased convective overshooting or rotationally-induced mixing (e.g. Langer et al. 1997). It has been unclear which primary N source accounts most for the constant N/O in the dwarf galaxies. The massive star primary source does not appear to produce enough N to yield N/O 0.03. For the lower mass stars, the various AGB model calculations give rather discrepant results for the production of N (see Forestini & Charbonnel 1997; van den Hoek & Groenwegen 1997; Marigo 2001). The N production during the third dredge-up is very sensitive to the assumptions that determine the boundary of the convective zone and the overshoot. This is a highly hydrodynamic problem including explosive thermal pulse events and is difficult to model at present (Lattanzio 1998). It is likely that this will continue to be an important topic of study for the near future.
Figure 18. N/O abundance ratios in spiral and irregular galaxies. Open circles: Garnett 1990; open diamonds: Thuan et al. 1995; filled circles: Garnett et al. 1999; filled squares: Garnett and Kennicutt 1994, Torres-Peimbert et al. 1989; plus signs: Díaz et al. 1991; stars: high-redshift absorption line systems from Lu, Sargent, and Barlow 1998. Note the very low N/O in some high-redshift systems.
Some insight may be found by examining more distant objects. Recent studies of high-redshift Lyman-alpha absorption systems (plotted as stars in Figure 18) have found objects with N/S, N/Si, and N/O ratios much lower than in the dwarf galaxies (Pettini, Lipman & Hunstead 1995; Lu, Sargent, and Barlow 1998). A wide range in inferred N/O is seen in the DLAs, but the lowest values are as much as a factor 10 smaller than the average for irregular galaxies. Although S and Si column densities are derived from S II and Si II, which can coexist with both ionized and neutral gas, ionization effects appear to insufficient to account for low N/S and N/Si (Vladilo et al. 2001). The results are consistent with the idea that the DLAs represent lines of sight through very young galaxies, with an age spread of a few hundred Myr, the timescale for enrichment of N from AGB stars. The higher N/O ratios seen in irregular galaxies would then be largely the product of AGB stars.
Some scatter is seen in N/O for the more metal-rich dwarf galaxies (Kobulnicky & Skillman 1998). This may be the result of localized enrichment by Wolf-Rayet stars. The most metal-poor dwarf galaxies seem to show very little scatter in N/O (Thuan et al. 1995). It is possible that this may simply reflect small number statistics (dwarf galaxies with log O/H < -4.5 and bright H II regions are rare). It is also possible to understand these galaxies if they are relatively old systems that experienced an episode of star formation in the past which enriched them to their present composition, and are experiencing a new starburst event after a long quiescent period.