ORIGIN OF THE UNUSUALLY LOW NITROGEN ABUNDANCES IN YOUNG POPULATIONS OF THE LARGE MAGELLANIC CLOUD

Chemical abundances in gaseous and stellar components of the LMC and the SMC have provided many clues to the formation and evolution of the Clouds (e.g., Da Costa & Hatzidimitriou 1998; Pagel & Tautvaisiene 1998; Venn 1999; Piatti et al. 2001; Hill 2004; Cole et al. 2005; Carrera et al. 2008; Tsujimoto & Bekki 2009). One of the intriguing observational results in terms of chemical abundances of the LMC is that star-forming H ii regions and very young stellar populations in the LMC show apparently low [N/H] that is a factor of 6–7 lower than the solar value (e.g., Korn et al. 2002; Hill 2004; van Loon et al. 2010) while other elements such as O and Ne are underabundant by a factor of ∼2 (Hill 2004). It remains observationally unclear whether only these young components in the LMC have such low [N/H] or whether other older ones also have it. It would be theoretically unlikely that all stellar populations and gas in the LMC have such low [N/H].

In any standard chemical evolution model, the N abundance will increase with time and reach the same level of enrichment as those of other elements in the reasonable scheme that stellar ejecta from asymptotic giant branch (AGB) stars (which are considered to be the major production site for N) in a galaxy is well recycled into and mixed with the ISM and subsequently the mixed gas is efficiently converted into new stars. Therefore, the observed low [N/H] ∼−0.8 in H ii regions and very young stellar populations of the present LMC with [Fe/H] ∼−0.3 appears to be at odds with theoretical predictions of previous chemical evolution models (e.g., Henry et al. 2000; Mollá et al. 2006). One of the possible ways to solve this is to dilute the nitrogen abundance through recent infall of gas with low [N/H] onto the LMC based on the assumption that only young components in the LMC have such low [N/H]. There would be a number of possibilities for the gas resource, such as the Galactic HVCs, the SMC, and some gas-rich dwarfs orbiting the Galaxy.

The purpose of this paper is to discuss a new scenario in which the observed low [N/H] can be closely associated with the recent infall of the Galactic HVCs onto the LMC. Given that some HVCs presumably within the Galactic halo are observed to have low [N/H] ranging from −2.0 to −1.2 (Collins et al. 2007), the proposed scenario appears to be reasonable. In addition, the similarity of the N/O ratio between the H ii region (log(N/O) ∼ −1.5) and the HVCs (log(N/O) ∼ −1.6) implies some connection between them. Using an idealized model, we investigate how much accretion of the HVCs is necessary to dilute the star-forming ISM to the extent that [N/H] of new stars can have the observed low [N/H]. We then estimate the possible number and total mass of the Galactic HVCs for the required accretion rate of the HVCs. Given that previous numerical simulations demonstrated (1) the accretion of HVCs onto the Galaxy (e.g., Cameron & Torra 1994) and (2) the possible presence of a large number of HVCs in the Galactic halo (e.g., Peek et al. 2008), the HVCs can be accreted onto the LMC if they collide with the LMC.

In the present paper, we do not intend to discuss why and how the HVCs have very low [N/H] (e.g., Collins et al. 2007), because this is simply beyond the scope of this paper. The origin of the low [N/H] would be closely associated with tidal stripping of outer H i gas of ancient gas-rich dwarfs, where chemical evolution did not proceed effectively (K. Bekki & T. Tsujimoto 2010, in preparation): our future papers will discuss why the stripped H i gas can have low [N/H] in detail. This paper thus focuses on (1) whether gas infall from the HVCs can explain the origin of the observed low [N/H] in the LMC and (2) what other scenarios can explain it if the HVC infall scenario is not so plausible.

The Appendix in this paper shows the time evolution of [N/O] of stars in the LMC based on the standard one-zone chemical evolution model for the LMC and compares the simulated [N/O] of the present LMC with the observed one. The Appendix thus can help readers to understand (1) that this is a serious inconsistency between the theoretical prediction of [N/O] for the present stellar populations of the LMC and the corresponding observed one and thus (2) that other factors need to be considered to reproduce the observed low nitrogen abundance of stars in the present LMC.

We here assume that (1) all young stellar populations with ages less than t_sf have unusually low [N/H] in the LMC and (2) the populations are formed exclusively from mixed gas of the original ISM of the LMC and the accreted gas (i.e., the Galactic HVCs). We adopt the first assumption above because previous observations show no clear evidence for the presence of young stars and H ii regions with normal [N/H] (e.g., Russell & Dopita 1990). The second assumption above means that the original ISM of the LMC has a "normal" [N/H] before external accretion of gas and thus needs to be diluted by the accreted gas to have low [N/H]. We discuss how the present results change if we relax these model assumptions in Section 4.

We try to derive the total mass of the accreted HVCs (M_HVC), for t_sf, the observed present star formation rate of the LMC (R_sf), the nitrogen abundance of the original ISM of the LMC (A_g), that of the HVCs (A_HVC), that of the young population observed in the LMC (A_obs), star formation efficiency for the mixed gas (_sf), and the total mass of the original ISM that can be converted into new stars (M_g; therefore this is not the total mass of the entire ISM of the LMC; it is the local gas mass (initially in the LMC) mixing with the accreted HVCs to form new stars with low [N/H]). We assume that t_sf is constant during the formation of young stars (i.e., for ∼10⁷ yr in most models), because there is no/little observational evidence which supports rapid change in star formation within an order of 10⁷ yr: the periodic bursts of star formation are observationally suggested (e.g., Harris et al. 2009), but they are estimated for a time span of 100 Myr to several Gyr.

We use the following two sets of equations to derive M_HVC:

$$\begin{equation} {\epsilon }_{\rm sf}(M_{\rm HVC}+M_{\rm g}) = t_{\rm sf}R_{\rm sf} \end{equation} \tag{ 1 }$$

and

$$\begin{equation} \frac{ A_{\rm HVC}M_{\rm HVC}+A_{\rm g}M_{\rm g} }{ M_{\rm HVC}+M_{\rm g} } = A_{\rm obs}. \end{equation} \tag{ 2 }$$

For Equation (1), we consider that (1) the star formation rate estimated from H ii regions of the LMC (∼0.26 M_☉ yr⁻¹; Kennicutt et al. 1995) is reasonable in the present study and (2) young stellar populations with ages less than t_sf were continuously formed with the star formation rate of R_sf. We choose A_obs corresponding to the observed [N/H] (=−0.8) for H ii regions and B type of stars in the LMC (Korn et al. 2002). It is observationally unclear what the typical value of A_HVC is, though some observations show very low [N/H] ranging from −2.0 to −1.2 (Collins et al. 2007). Therefore, we consider that A_HVC is a parameter with the above observed range. Since the present [Fe/H] of the LMC is observed to be −0.3 (e.g., van den Bergh 2000), we set the gaseous [N/H] to be −0.3 assuming [N/Fe] = 0: we use A_g corresponding to this [N/H] value. We consider that t_sf ∼ 10⁷ yr is reasonable, because the ages of stars (e.g., main-sequence B-type stars) observed for estimation of N abundances (Korn et al. 2002) correspond roughly to the above t_sf. We however investigate models with different t_sf.

Figure 1 shows how M_HVC required to decrease [N/H] of the ISM to ∼−0.8 depends on the nitrogen abundance of HVCs (denoted as [N/H]_HVC) for t_sf = 10⁷ yr in three models with different _sf. The required M_HVC ranges from ∼1.8 × 10⁶ M_☉ to ∼2.2 × 10⁷ M_☉ and is larger for smaller _sf for a given [N/H]_HVC (i.e., A_HVC). The reason for this is as follows: only a smaller fraction of the accreted HVCs can be converted into new stars after mixing with the original ISM in the models with a smaller _sf. Therefore, a larger amount of HVCs needs to be accreted to form the observed total mass of young stars with low [N/H]. The required M_HVC is larger for larger [N/H]_HVC for a given _sf. It should be stressed here that the derived M_HVC is for the ISM that forms young stars: it is not for the entire ISM of the LMC.

Zoom In Zoom Out Reset image size

Figure 1 also shows that the mass ratio of M_HVC to M_g is dependent on [N/H]_HVC in the sense that a larger amount of HVCs is necessary to lower the [N/H] of the ISM to the observed level for larger [N/H]_HVC. The derived large mass ratios M_HVC/M_g (ranging from ∼2.3 to ∼5.8) mean that a significant degree of dilution of original ISM of the LMC by HVC infall is indispensable for explaining the observed low [N/H] in H ii regions and young stars in the LMC. For example, the original gas of the LMC with M_g = 7.8 × 10⁵ M_☉ is converted into new stars for = 1.0 and [N/H]_HVC = −2.0 after being mixed with the HVCs with M_HVC = 1.8 × 10⁶ M_☉. Figure 2 shows that the required M_HVC is quite large (∼6.3 × 10⁷ M_☉) if young stars with ages less than 10⁸ yr uniformly have low [N/H] of −0.8. Figure 2 also shows that the required M_HVC depends weakly on [N/H]_HVC for a given _sf in models with different t_sf.

Zoom In Zoom Out Reset image size

Only a small fraction of the Galactic HVCs can interact with the LMC owing to the small disk size of the LMC. We here estimate (1) a typical timescale for the LMC to collide with one HVC (t_col) for a given number density of the HVCs within the distance of the LMC from the Galactic center and (2) an expected accretion rate of HVCs onto the LMC disk $({\dot{M}}_{\rm HVC})$. Since we can estimate the accretion rate required for explaining the observed [N/H] using the results shown in Figures 1 and 2 (i.e., M_HVC/t_sf), we can compare the expected and the required accretion rates and thereby assess the viability of the present scenario.

The timescale of an LMC–HVC collision event (t_col) can be estimated as follows (e.g., Makino & Hut 1997):

$$\begin{equation} t_{\rm col} = \frac{ 1 }{n_{\rm HVC}\sigma v}, \end{equation} \tag{ 3 }$$

where n_HVC, σ, and v are the mean number density of the HVCs within the Galaxy, the geometrical cross section of the LMC, and a relative velocity between a HVC and the LMC. We here estimate n_HVC for the central 75 kpc of the Galaxy (corresponding roughly to the mean of the pericenter and apocenter distances of the LMC orbit; e.g., Bekki & Chiba 2005) and assume that σ = πR_LMC², where R_LMC is the LMC size and v is velocity dispersion ($= v_{\rm c}/\sqrt{(}2)$, where v_c is the circular velocity thus 220 km s⁻¹) of the Galaxy halo. For convenience, we discuss t_col in terms of the total number of HVCs within 75 kpc from the Galaxy (N_HVC) rather than n_HVC below. Previous observations found more than 600 HVCs (Wakker & van Woerden 1991), yet many initial HVCs have already been destroyed by tides and ram pressure (e.g., see Wakker 2004 for a review). Thus, we consider that it is reasonable to investigate models with N_HVC ranging from a few hundreds to a few thousands.

Figure 3 shows that t_col is shorter for larger N_HVC for a given R_LMC and shorter for larger R_LMC for a given N_HVC. Figure 3 also shows that t_col can be as low as ∼10⁸ yr if the total number of HVCs within the LMC's orbit is as large as ∼1000 for the size of the LMC disk (R_LMC = 5 kpc). It is clear from Figure 3 that if there are only a few hundreds HVCs within the LMC orbit, then the HVCs are highly unlikely to be accreted onto the LMC and consequently dilute the ISM within a timescale of much less than 10⁸ yr (which corresponds to ages of young stellar populations with low [N/H] in the LMC).

Zoom In Zoom Out Reset image size

By assuming a typical mass of the individual HVCs (m_hvc) and using the results shown in Figure 3, we can discuss the possible accretion rate of the HVCs onto the LMC disk for a given set of model parameters. Figure 4 shows that ${\dot{M}}_{\rm HVC}$ is much less than ∼0.1M_☉yr⁻¹ for almost all models with different m_hvc and N_HVC. The minimum value of the required M_HVC shown in Figure 1 is 1.8 × 10⁶ M_☉ for t_sf = 10⁷ yr in different models with different _sf and [N/H]_HVC. Therefore, at least 0.18 M_☉ yr⁻¹ is necessary to dilute the ISM of the LMC to the observed level for t_sf = 10⁷ yr. It should be stressed that the above 0.18 M_☉ yr⁻¹ is for _sf = 1.0 (i.e., 100% star formation efficiency): a realistic value of the required ${\dot{M}}_{\rm HVC}$ is likely to be significantly larger than 0.18 M_☉ yr⁻¹.

Zoom In Zoom Out Reset image size

The results shown in Figure 4 suggest that only models with very large typical masses of HVCs (i.e., m_hvc = 10⁷ M_☉) and large number of HVCs (N_HVC > 2500) can show ${\dot{M}}_{\rm HVC}$ as high as the required rate above (0.18 M_☉ yr⁻¹). Although the required typical mass is similar to the observed mass of Complex C (e.g., Thom et al. 2008), the required total number within the LMC's orbital radius already exceeds the total number of the HVCs (∼2000) observed by the H i Parkes All Sky Survey (e.g., Putman et al. 2002): it should be noted that the observed one is for the HVCs existing in the entire regions around the Galaxy whereas the required one is only for those within ∼75 kpc. These results imply that it is unlikely for the accretion of the Galactic HVCs onto the LMC disk to dilute the ISM. However ${\dot{M}}_{\rm HVC}$ could become large enough in a sporadic way if the LMC can interact with groups of HVCs with locally large number densities.

4.1. On Model Uncertainties

4.2. Infall from the SMC Rather than from the HVCs?

4.3. A Possible Observational Evidence for External Gas Infall onto the LMC

If chemical evolution of the LMC is influenced by accretion of gas from outside the LMC (e.g., from the Galactic halo or other gas-rich galaxies), then the chemical evolution strongly depends on the orbital evolution of the LMC. The LMC may have steadily obtained gas from the accretion events of the Galactic HVCs for at least 3–4 Gyr for the "classical bound orbit" adopted in previous dynamical models for the evolution of the LMC (e.g., Bekki & Chiba 2005). For this classical orbital model, the LMC can show lower [N/H] not only in young stellar populations but also in intermediate-age ones, if the dilution by the continuous gas infall has been overwhelming over chemical enrichment by the AGB stars continuously formed in the LMC for the last 3–4 Gyr. If the LMC has just recently arrived in the Galaxy, as the latest proper motion studies by Hubble Space Telescope (Kallivayalil et al. 2006) has suggested, then the LMC may have started the accretion of the HVCs quite recently (well less than ∼1 Gyr): the low [N/H] can be seen only in young stellar populations.

We have so far considered that all of the young stellar populations in the LMC have unusually low [N/H] and were formed from mixed gas of the LMC's ISM and the accreted gas from outside. However, some local regions where accretion of gas with very low [N/H] does not occur may well form young stellar populations with normal [N/H]. In this case, the young populations in the LMC may well show a large dispersion in [N/H] in the present scenario. The previous observations on the chemical abundances for H ii regions of the LMC indeed show a dispersion in [N/H] (e.g., Russell & Dopita 1990), though the dispersion appears to be smaller (see Figure 5). If future observations confirm that the dispersion in [N/H] is really small in the LMC (i.e., [N/H] is uniformly low for the entire young populations), then it means that young populations in the LMC were formed exclusively from mixed gas of the original ISM of the LMC and the accreted gas for some physical mechanisms.

Zoom In Zoom Out Reset image size

However, it should be stressed that if the entire ISM (i.e., not only cold H i and molecular gas but also H ii regions) of the present LMC has unusually low [N/H] for some physical reasons, no accretion event of gas outside the LMC is required for explaining the origin of the observed young stellar populations with unusually low [N/H]: both old and young stellar populations have unusually low [N/H]. If this is the case, then the next question is why and how the [N/H] of the ISM in the LMC has continued to be very low until the mean metallicity of the LMC has become as high as [Fe/H] ∼−0.3. No previous chemical evolution models appear to have shown very low [N/H] in the present galaxies with lower mean metallicities like the LMC (e.g., Mollá et al. 2006). This implies that it is unlikely that the entire ISM of the LMC has very low [N/H]: it is natural to consider that external gaseous accretion has recently changed [N/H] in the ISM with originally normal [N/H] (see the Appendix on this issue).

Is there, then, any possible observational evidence for recent accretion events of gas in the LMC? In order to answer this question, we have investigated the dependences of log (N/O) on 12+log (O/H) for the H ii regions of the LMC using already existing data sets (Peimbert & Torres-Peimbert 1974; Dufour 1975; Pagel et al. 1978; Russell & Dopita 1990; Simpson et al. 1995). Figure 5 shows that (1) there is no clear correlation between log (N/O) and 12+log (O/H), and (2) H ii regions with larger 12+log (O/H) do not show larger log (N/O). These findings are inconsistent with previous chemical evolution models (e.g., Henry et al. 2000) demonstrating that stars and gas with larger 12+log (O/H) have larger log (N/O): it should be noted here that infall of gas with low [N/H] is not included in the models. We thus suggest that the observed lack of H ii regions with larger 12+log (O/H) and larger log (N/O) are due to accretion of gas with low [N/H]: original H ii regions with larger log (N/O) and larger 12+log (O/H) can disappear owing to the dilution of [N/H] caused by the gaseous accretion.

4.4. Observational Implications: Effects on Other Elements

4.5. Other Possible Scenarios

Given that the observed unusually low [N/H] in young populations of the LMC cannot be simply reproduced by canonical chemical evolution models for the LMC, we have investigated a new scenario in which the observed low [N/H] can be closely associated with recent collisions and subsequent accretion of HVCs that surround the Galaxy and have low nitrogen abundances. We have shown that even if the observed low [N/H] is limited to very young stars with ages less than ∼10⁷ yr, the collision/accretion rate of the HVCs onto the LMC needs to be ∼0.2 M_☉ yr⁻¹ (corresponding to the total HVC mass of 10⁶–10⁷ M_☉) to dilute the original ISM before star formation.

We have demonstrated that if the typical mass of HVCs accreted onto the LMC is ∼10⁷ M_☉, the Galaxy needs to have ∼2500 massive HVCs within the LMC's orbital radius with respect to the Galactic center in order to explain the required accretion rate of the HVCs. Although we have adopted a number of model assumptions in deriving the required number of the HVCs, the required number of massive HVCs suggests that the HVC infall scenario is possible yet unlikely to efficiently dilute the ISM of the LMC and consequently lower the [N/H]. We thus have discussed the alternative SMC-transfer scenario in which the transfer of gas with low [N/H] from the SMC to the LMC can explain the observed low [N/H].

The SMC-transfer scenario has some advantages in explaining the observations (e.g., a higher probability of an enough amount of gas to be accreted onto the LMC) over the HVC infall one. However it also appears to have some problems in explaining self-consistently the observed chemical properties of the LMC. For example, it is not so clear in the SMC-transfer scenario why the gas accretion can continue to occur until quite recently. We thus conclude that although the observed low [N/H] of young populations in the LMC has an external origin (i.e., gas infall from outside the LMC), the host objects that the external gas originates from are yet to be determined.

We are grateful to the anonymous referee for valuable comments, which helped improve the present paper. K.B. acknowledges the financial support of the Australian Research Council throughout the course of this work. T.T. is assisted in part by Grant-in-Aid for Scientific Research (21540246) of the Japanese Ministry of Education, Culture, Sports, Science and Technology. This work was supported in part by the Graduate University for Advanced Studies.

We consider that it is useful to show how the nitrogen abundance in the LMC evolves with time in a reasonable one-zone chemical evolution model of the LMC, because no theoretical works have yet clearly demonstrated that canonical chemical evolution models cannot reproduce the observed low [N/H] in the LMC. We here discuss whether the observed [N/O] in the LMC can be reproduced by a canonical one-zone chemical evolution model which is consistent with the observed age–metallicity relation of stellar population in the LMC. Here, we assume the standard Salpeter IMF. Details of model description are presented in Tsujimoto & Bekki (2010), including the chemical yields from stars with different masses. We will discuss the time evolution of [N/H] in different galaxies with different physical properties using these models in our future papers.

Figure 6 shows both the time evolution of [Fe/H] and [N/O] for the last ∼13.5 Gyr and the observed values so that we can clearly see how serious the observed low nitrogen abundance is in the standard chemical evolution models with no external infall of gas. As shown in Figure 6, [N/O] can be low in the early stage of the LMC evolution when log (N/O)+12 was lower than 8 (when the LMC was much younger). However it is clear that log (N/O) (∼−1.1) in the present LMC (at log (O/H)+12 ∼8.4) for the standard chemical evolution model is much larger than the observed one (∼−1.5). This result demonstrates that some additional factors such as unique IMFs and external gaseous infall from outside of the LMC need to be considered in order to reproduce the observed low [N/O] in the LMC. We suggest that HVCs and gaseous components of the SMC can be the promising candidates for the external infall of gas.

Zoom In Zoom Out Reset image size