THE EXTENDED He II λ4686-EMITTING REGION IN IZw 18 UNVEILED: CLUES FOR PECULIAR IONIZING SOURCES

The emission-line fluxes were measured using the IRAF⁶ task SPLOT. The flux of each emission line was derived by integrating between two points given by the position of a local continuum placed by eye. For each line, this procedure was repeated several times by varying the local continuum position. The final flux of each line and its associated uncertainty were assumed to be the average and standard deviation of the independent, repeated measurements (e.g., Kehrig et al. 2006).

We used our own IDL scripts to create the emission-line maps presented in Figure 2. The spaxels where we measure He iiλ4686 are indicated with pluses.

Zoom In Zoom Out Reset image size

An extended He iiλ4686-emitting region with a diameter of ≈ 5" (≈440 pc) is revealed from our IFS data (see Figure 2). The narrow line profile for the He iiλ4686 emission and its spatial extent are evidence of its nebular nature. The spectra of certain Of stars may exhibit somewhat narrow He iiλ4686 lines (e.g., Massey et al. 2004), so He ii emission observed in starburst galaxies could be thought to arise in the atmospheres of such stars (e.g., Bergeron 1977). However, if these Of stars were present in appreciable numbers in IZw 18, then broad Hα emission would be expected (Massey et al. 2004). Thus, the non-detection of this feature in our spectra supports the nebular origin for the He iiλ4686 line in IZw 18.

By adding the emission from the spaxels showing He iiλ4686 (see Figure 2), we created the one-dimensional spectrum for the He iiλ4686 region (see Figure 3). Using this spectrum, we obtained a very small logarithmic extinction coefficient c(Hβ) = 0.08 ± 0.02 from the observed ratio Hα/Hβ = 2.92 ± 0.04, assuming an intrinsic case B recombination Hα/Hβ = 2.75 (Osterbrock & Ferland 2006, OF06). Using PYNEB (Luridiana et al. 2015), we derived the electron temperature, T_e=(2.18± 0.05)×10⁴ K, and density, n_e ≤ 100 cm⁻³, from the [O iii]λ4363/[O iii]λλ4959,5007 and [S ii]λ6717/[S ii]λ6731 ratios, respectively. The integrated flux of the He iiλ4686 line, corrected for reddening, is (2.84 ± 0.18) × 10⁻¹⁵ erg s⁻¹ cm⁻² which translates to an He iiλ4686 luminosity of L $_{{\rm HeII}\lambda 4686}$= (1.12 ± 0.07) × 10³⁸ erg s⁻¹. The corresponding He ii ionizing photon flux, Q(He ii)_obs = (1.33 ± 0.08) × 10⁵⁰ photon s⁻¹, was derived from the measured L $_{{\rm HeII}\lambda 4686}$ using the relation Q(He ii) = L $_{{\rm HeII}\lambda 4686}$/[j(λ4686)/${{\alpha }_{{\rm B}}}$(He ii)] (assuming case B recombination, and T_e = 2 × 10⁴ K; OF06). We checked that different CLOUDY models (Ferland et al. 2013) computed for the typical n_e and T_e in the He iiλ4686 region, and considering several effective temperatures and geometries with no dust, provide a ratio L(He iiλ4686)/Q(He ii) which agrees with the assumed ratio (OF06) within 10%. The total Q(He ii)_obs, a quantity not reported before for IZw 18, will allow us to constrain possible ionizing sources of He ii in IZw 18.

Zoom In Zoom Out Reset image size

One widely favored mechanism for He ii ionization in H ii galaxies involves hot WRs (e.g., Schaerer 1996). Nevertheless, it has been demonstrated that nebular He iiλ4686 does not appear to be always associated with WRs, as is the case of the He ii nebulae LMC N44C, LMC N159F, and M33 BCLMP651, among others (Kehrig et al. 2011 and references therein). This indicates that WRs cannot explain He ii ionization in all cases, particularly at low Z (e.g., Guseva et al. 2000; Kehrig et al. 2008, 2013; Schaerer 2003; Shirazi & Brinchmann 2012). Besides, it has been shown that for a sample of He iiλ4686-emitting star-forming galaxies, current models of massive stars predict He iiλ4686 emission and He iiλ4686/Hβ ratios only for Z > 0.20 Z $_{\odot }$ instantaneous bursts (Shirazi & Brinchmann 2012). In the particular case of IZw 18, previous work claimed that the ratio He iiλ4686/Hβ could be reproduced using highly density-bounded photoionization models while underpredicting the electron temperature measurement (Stasińska & Schaerer 1999); these models have been challenged by Péquignot (2008).

Faint broad emission signatures, attributable to WRs, are observed in the spectrum of IZw 18 despite its low Z (e.g., Izotov et al. 1997; Legrand et al. 1997; Brown et al. 2002). A comprehensive study of WRs in IZw 18, using UV STIS spectroscopy, revealed signatures of carbon-type WRs (WC) in two clusters: one in the NW star-forming region and a second one on the outskirts of this region (Brown et al. 2002). Here we deal with the NW one, since it is in the He iiλ4686-emitting region (see Figure 2). The C ivλ1550 flux measured from the NW WR cluster (Brown et al. 2002) provides a C ivλ1550 luminosity of L₁₅₅₀ = 4.67 × 10³⁷ erg s⁻¹. Taking the L₁₅₅₀ luminosity of the metal-poor, early-type WC (WCE) model by Crowther and Hadfield (2006, CH06), which mimics a single WCE star in IZw 18, this implies ∼9 IZw 18-like WCE stars present in the NW cluster. From these nine WCE stars, a total flux of Q(He ii) = 2.8 ×10⁴⁸ photon s⁻¹ is expected (assuming Q(He ii) = 10$^{47.5}$ photon s⁻¹ for one IZw 18-like WCE; CH06), i.e., about 48 times lower than the Q(He ii)_obs = (1.33 ± 0.08)×10⁵⁰ photon s⁻¹ derived from our data (see Section 3.2).

Based on the He ii-ionizing flux expected from these IZw 18-like WRs, a very large WR population is required to explain the He ii-ionization budget measured; for instance, taking the Q(He ii) = 10$^{47.5}$ photon s⁻¹ for one IZw 18-like WCE (CH06), the number of these WCEs needed to explain our derived Q(He ii)_obs = (1.33 ± 0.08)×10⁵⁰ photon s⁻¹ would be >400. In principle, the presence of hundreds of WRs in IZw 18 should not be discarded on the basis of empirical arguments for reduced WR line luminosity at low Z (CH06). However, assuming a Salpeter initial mass function (IMF; Salpeter 1955; M_up = 150 M $_{\odot }$) and the initial mass needed for a star to certainly become a WC (Meynet & Maeder 2005), a cluster with >8 times the total stellar mass of the NW region (M $_{*,{\rm NW}}$= 2.9 × 10⁵ M $_{\odot }$ from Stasińska & Schaerer 1999 scaled to 18.2 Mpc distance) is required to provide >400 WC in IZw 18. Also, such a high number of WRs is not supported by state-of-the-art stellar evolutionary models for single (rotating and non-rotating) massive stars in metal-poor environments (Leitherer et al. 2014). Furthermore, given the decrease in the ratio of WR/O stars with decreasing metallicity, shown by observations and theoretical models (Maeder & Meynet 2012 and references therein), such a large number of WRs appears clearly unreasonable considering the extremely low Z and the O star content of IZw 18 (CH06). All this suggests that WRs are not solely responsible for the He iiλ4686 emission in IZw 18.

The binary channel in massive star evolution is suggested to increase the WR population (e.g., Eldridge et al. 2008), but the WR population in Local Group galaxies does not show an increased binary rate at lower Z (e.g., Foellmi et al. 2003; Neugent & Massey 2014); thus, the binary channel does not seem to favor the formation of WRs at lower-Z, in contrast to what we need. Nevertheless, we should bear in mind the possible uncertainties still unsolved in the models (Maeder & Meynet 2012). Further investigation awaits the calculation of evolutionary models for binary stars at very low Z.

As mentioned before, nebular He iiλ4686 emission observed in the spectra of Hii galaxies and extragalactic Hii regions has also often been attributed to shocks and X-ray sources (e.g., Pakull & Mirioni 2002; Garnett et al. 1991; Thuan & Izotov 2005). In the following, we discuss these two candidate sources for He ii ionization in IZw 18.

The X-ray emission from IZw 18 is dominated by a single X-ray binary apparently located in the field of the NW knot (Thuan et al. 2004). We have computed a CLOUDY photoionization model (Ferland et al. 2013) using as input a power-law spectral energy distribution (SED) with the same X-ray luminosity, column density, and slope that have been reported for IZw 18 (Thuan et al. 2004). This CLOUDY model provides an He iiλ4686 luminosity of L $_{{\rm HeII}\lambda 4686}$ = 10$^{35.7}$ erg s⁻¹, which is ∼100 times lower than the L$_{{\rm HeII}\lambda 4686}$ measured. This result rules out the X-ray binary as the main source of He ii ionizing photons in IZw 18. We note here that the emission from X-ray ionized nebulae has been successfully reproduced by CLOUDY models before (e.g., Pakull & Mirioni 2002).

Guided by the existence of He iiλ4686 emission associated with supernova remnants (e.g., Kehrig et al. 2011) we explored the conjecture that the He iiλ4686 region represents such a shock-ionized nebula. The [O i]λ6300 line, often strong in remnants, has frequently been used as a sensitive shock-emission test (e.g., Skillman 1985). We checked that, in fact, most of the [O i]λ6300 emission in IZw 18 is concentrated on the SE knot (see the [O i]λ6300 map in Figure 2) with only 12% of the He iiλ4686-emitting spectra showing [O i]λ6300 flux above the 3σ detection limit. Additionally, we find no evidence for [S ii] enhancement (a usual sign of shock excitation; e.g., Dopita & Sutherland 1996) associated with the He iiλ4686 region (see the [S ii] map in Figure 2). Therefore, the He iiλ4686-emitting zone in IZw 18 is unlikely to be produced by shocks.

Our new observations have allowed us to empirically demonstrate why conventional He ii-ionizing sources (e.g., WRs, shocks, X-ray binaries) cannot account for the total He ii-ionization budget in IZw 18. What could the nebular He iiλ4686 emission in IZw 18 originate from?

We have also explored the possibilty of very massive, metal-poor O stars to account for our observations of IZw 18. Using current wind models of very massive O stars at low Z, we can derive their He ii-ionizing fluxes (Kudritzki 2002). According to the hottest models (T_e = 60,000K), between 10 and 20 super-massive stars with 300 M $_{\odot }$ (with Q(He ii) ≈(0.70–1.4) × 10⁴⁹ photon s⁻¹ each) would be sufficient to explain the derived Q(He ii)_obs budget. Very massive stars of up to 300 M $_{\odot }$ were claimed to exist in the LMC R136 cluster (Crowther et al. 2010); however, the existence of such super-massive 300 M $_{\odot }$ stars remains heavily debated (Vink 2014). Additionally, assuming a Salpeter IMF, 10–20 stars with 290 ≤ M_*/M $_{\odot }$ ≤ 310 would imply a cluster mass of ∼10–20 × M $_{*,{\rm NW}}$. We should bear in mind that an extrapolation of the IMF predicting 300 M $_{\odot }$ stars remains unchecked up to now. If we instead consider the 150 M $_{\odot }$ star hottest models (with Q(He ii) ≤1.9 × 10⁴⁷ photons⁻¹ each, for Z ≤1/32 Z $_{\odot }$; K02), the number of these stars required to explain the Q(He ii)_obs would be >650. For a Salpeter IMF, 650 stars with 145 ≤ M_*/M $_{\odot }$ ≤ 155 would require a cluster mass ∼200 × M $_{*,{\rm NW}}$. Besides the Q(He ii)_obs budget, in the He iiλ4686 region we have measured He iiλ4686/Hβ ratios as high as 0.08. These values appear too big to be explained even by the models for the hottest, most metal-poor super-massive 300 M $_{\odot }$ stars (K02) under ionization-bounded conditions. Further constraints to the observations should await the calculation of new evolutionary tracks and SEDs for single O stars at the metallicity of IZw 18 including rotation.

Searches for very metal-poor starbursts and PopIII-hosting galaxies have been carried out in the distant universe using He ii lines (e.g., Schaerer 2008; Cassata et al. 2013). This search is based on the high effective temperature for PopIII-stars which will emit a large number of photons with energy above 54eV, and also on the expected increase of the He ii recombination lines with decreasing Z (e.g., Guseva et al. 2000; Schaerer 2003, 2008). Predictions for burst models of different metallicities show how their corresponding Q(He ii) can increase by up to ∼10³ when going from Z = 10⁻⁵ to 0 (Schaerer 2003). These models cannot explain the Q(He ii)_obs when Z ≥ 10⁻⁵ (for a Salpeter IMF, M_up = 100 M $_{\odot }$; see table 3 in Schaerer 2003) even assuming that the total M $_{*,{\rm NW}}$ would come from He ii-ionizing stars. So another more speculative possibility to explain the derived Q(He ii)_obs could be based on nearly metal-free ionizing stars. These stars should ionize He ii via their strong UV radiation expected at nearly zero metallicity (e.g., Tumlinson & Shull 2000; Schaerer 2003).

As an approximation of nearly metal-free single stars, we have compared our observations with the He ii-ionizing radiation expected from state-of-the-art models for rotating Z = 0 stars (Yoon et al. 2012). According to these models, we found that a handful of such stars could explain our derived Q(He ii)_obs (e.g., ∼8–10 stars with mass M_ini = 150 M $_{\odot }$ or ∼13–15 stars with M_ini = 100 M $_{\odot }$; with Q(He ii) ≈1.4 × 10⁴⁹ (0.9 × 10⁴⁹) photon s⁻¹ for each star with M_ini = 150 M $_{\odot }$(100 M $_{\odot }$)). Additionally, we note that the ionizing spectra produced by these star models are harder than the ones expected from the hottest models of super-massive 300 M $_{\odot }$ stars (K02), so they would also explain the highest He iiλ4686/Hβ values observed, providing that ionization-bounded conditions are met. While gas in IZw 18 is very metal-deficient but not primordial, Lebouteiller et al. (2013) have pointed out that the Hi envelope of IZw 18 near the NW knot contains essentially metal-free gas pockets. These gas pockets could provide the raw material for making such nearly metal-free stars. Clearly, in this hypothetical scenario, these extremely metal-poor stars cannot belong to the NW cluster, which hosts more chemically evolved stars.