ABSTRACT
Low-resolution (4.5–5 Å) spectra of 58 blue supergiant stars distributed over the disk of the Magellanic spiral galaxy NGC 55 in the Sculptor group are analyzed by means of non-LTE techniques to determine stellar temperatures, gravities, and metallicities (from iron peak and α-elements). A metallicity gradient of −0.22 ± 0.06 dex/R25 is detected. The central metallicity on a logarithmic scale relative to the Sun is [Z] = −0.37 ± 0.03. A chemical evolution model using the observed distribution of column densities of the stellar and interstellar medium gas mass reproduces the observed metallicity distribution well and reveals a recent history of strong galactic mass accretion and wind outflows with accretion and mass-loss rates of the order of the star formation rate. There is an indication of spatial inhomogeneity in metallicity. In addition, the relatively high central metallicity of the disk confirms that two extraplanar metal-poor H ii regions detected in previous work 1.13 to 2.22 kpc above the galactic plane are ionized by massive stars formed in situ outside the disk. For a subsample of supergiants, for which Hubble Space Telescope photometry is available, the flux-weighted gravity–luminosity relationship is used to determine a distance modulus of 26.85 ± 0.10 mag.
Export citation and abstract BibTeX RIS
1. INTRODUCTION
The galaxies in the Sculptor group seem to form a filament extended along the line of sight (Jerjen et al. 1998; Karachentsev et al. 2003) with three distinct subgroups, one around the starburst galaxy NGC 253 at about 4 Mpc, the other around NGC 7793 at about the same distance, and the third much closer at half of this distance around NGC 300. While NGC 253, as the largest and most massive galaxy, seems to form the dynamical center of the group, the subgroup around NGC 300 appears to be attracted by the gravitational field of the Local Group.
NGC 55 is member of the nearby subgroup. The near-IR Cepheid study of the Araucaria collaboration places NGC 55 and NGC 300 at roughly the same distance of 1.9 Mpc (Gieren et al. 2005a, 2008). These two galaxies have roughly comparable near-IR magnitudes (Ktot = 6.25 and 6.38, respectively, see Jarrett et al. 2003) and mid-IR fluxes (F3.6μ = 2.02 and 1.63 Jy, F4.5μ = 1.39 and 1.20 Jy, see Dale et al. 2009), indicating comparable stellar masses. However, while NGC 300 is a regular spiral galaxy of morphological type Scd with a moderate inclination angle (i = 399), NGC 55 is an almost edge-on (i = 78°) barred spiral of type SB(s)m with the bar apparently oriented along the line of sight (de Vaucouleurs 1961; de Vaucouleurs et al. 1991), which closely resembles the Large Magellanic Cloud (Robinson & van Damme 1964; Westmeier et al. 2013).
NGC 300 has been subject to many very detailed studies of its stellar populations, the interstellar medium (ISM) (atomic and molecular gas distribution, H ii regions, planetary nebulae, supernovae remnants, dust content) and the very extended faint stellar disk (see Bresolin et al. 2009; Vlajić et al. 2009; Westmeier et al. 2011; Stasinska et al. 2013; Kang et al. 2016; Toribio San Cipriano et al. 2016, and references therein). In particular, the metallicity of the ISM and the young stellar population has been investigated by detailed quantitative spectroscopic studies of blue supergiant stars (Kudritzki et al. 2008), red supergiants (Gazak et al. 2015), and H ii regions (Bresolin et al. 2009). While these three investigations used entirely independent methods for metallicity diagnostics, the results with respect to central metallicity and metallicity gradient agreed extremely well. On the other hand, only a handful of H ii regions in the disk of NGC 55 have been studied to date, with an uncertain range in metallicity of [Z] = −0.6 to −0.28 (Webster & Smith 1983; Stasinska et al. 1986; Zaritzky et al. 1994; Tüllmann et al. 2003; Castro et al. 2012; Pilyugin et al. 2014). Tüllmann et al. (2003) also investigated two extraplanar H ii regions located 0.8 and 1.5 kpc above the disk and found a metallicity almost a factor of 10 smaller than solar. Castro et al. (2008) described spectra and spectral morphology of a large sample of hot massive stars, mostly blue supergiants, which were obtained within the Araucaria collaboration (see Gieren et al. 2005b), but so far only a small fraction of these objects—12 supergiants of early-B spectral type—have been subject to a quantitative spectral analysis (Castro et al. 2012). This work indicated an average metallicity very similar to the LMC, but remained inconclusive about the possibility of a spatial trend in metallicity, in particular a radial metallicity gradient.
NGC 55, a galaxy with significant star formation, is very likely subject to mass accretion and gas outflows (Tüllmann et al. 2003; Westmeier et al. 2013) and, thus, accurate information about metallicity and a potential metallicity gradient might allow one to constrain the rates of matter inflow and outflow (see Kudritzki et al. 2015). We have, therefore, resumed the analysis of the spectra obtained by Castro et al. (2008), this time focusing on the supergiants of spectral type B8 to A5, for which the signal-to-noise ratio (S/N) was sufficient for a quantitative spectral analysis. Because of the many metal lines in their spectra and because of their enormous intrinsic brightness, supergiants of these spectral types are ideal for studies of extragalactic metallicity (see Kudritzki et al. 2008, 2012, 2014, and references therein). The selection of targets resulted in a sample of 46 objects distributed over a large range of galactocentric distances. These objects were then analyzed in detail with respect to their effective temperatures, gravities, and metallicities, and the results are presented in this paper. After a brief description in Section 2 of the observations, the analysis method and the geometrical model used for a deprojection of the location of the targets in the galactic disk, we summarize the results in Section 3. Section 4 focuses on the metallicity and the metallicity gradient and applies a chemical evolution model.
Blue supergiants, as the brightest stars in the universe at optical wavelengths, are also excellent distance indicators, because their "flux-weighted gravity" gF ≡ g/ ( in units of 104 K) is tightly correlated with their absolute bolometric magnitude, leading to the "flux-weighted gravity–luminosity relationship (FGLR)" (see Kudritzki et al. 2003, 2008). Since the distance moduli to NGC 55 obtained with different methods appear to be slightly controversial, ranging from 26.4 mag (Cepheids: Gieren et al. 2008) to 26.6 mag (EDD database, http://edd.ifa.hawaii.edu, see Tully et al. 2009) to 26.8 mag (planetary nebulae luminosity function (PNLF): van de Steene et al. 2006), we select a subsample of 13 supergiants, for which Hubble Space Telescope (HST)/Advanced Camera for Surveys (ACS) photometry is available, and determine an independent distance in Section 5 using the most recent calibration of the FGLR method by Urbaneja et al. (2016). Section 6 presents a final discussion.
2. OBSERVATIONAL DATA AND ANALYSIS METHOD
2.1. Target Selection, Spectra, and Photometry
The spectroscopic observations were carried out on 2004 November 6, 7, and 10 using the focal reducer low-dispersion spectrograph FORS2 attached to the ESO VLT-UT2. The Mask eXchange Unit mode was used for the multi-object spectroscopy with slit widths of 1 arcsec in conjunction with the 600B grism providing a nominal resolution of 5 Å in the wavelength range from 3100 to 6210 Å. Four fields (A–D) with a field of view of 6.8 × 6.8 arcmin2 were chosen to cover the whole galaxy, and a total of 200 objects were observed. For the central field C two different mask settings were applied because of the high density of targets in the center of the galaxy. Target selection, field settings, and details of the observations, as well as the data reduction, are described in Castro et al. (2008). Their paper provides comprehensive information for all targets. For our work we use coordinates, spectral type, radial velocity, S/N, and the reduced normalized spectra as displayed in the appendix of Castro et al. We have applied small corrections to some of the radial velocities of Castro et al. based on the comparison with our model atmosphere spectra, and therefore give radial velocity values for all our objects. After careful inspection of the digital spectrum of every target with spectral type B8 to A5 we finally selected 46 objects with spectra suitable for a quantitative spectral analysis. Objects with too strong nebular H ii emission contaminating the stellar Balmer lines and with spectra too noisy for the analysis were dropped from the sample in the selection process. The final list is given in Table 1.
Table 1. Late B and Early A Supergiants in NGC 55
Star | Spectral | F | [Z] | R/R25 | x/R25 | y/R25 | vrad | ||
---|---|---|---|---|---|---|---|---|---|
type | (K) | (cgs) | (cgs) | (dex) | (km s−1) | ||||
(1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) | (9) | (10) |
D 29 | A3 I | 1.15 | −0.290 | 98 | |||||
D 35 | A3 I | 1.00 | −0.237 | 63 | |||||
D 39 | A2 I | 1.23 | −0.193 | 132 | |||||
C1 1 | A0 I | 1.34 | −0.163 | 144 | |||||
C1 10 | A2 I | 1.03 | −0.069 | 100 | |||||
C1 17 | A3 I | 1.15 | +0.001 | 117 | |||||
C2 15 | A0 I | 1.60 | −0.031 | 98 | |||||
C2 13 | A2 I | 1.25 | −0.054 | 90 | |||||
C2 39 | A2 I | 1.20 | +0.107 | 158 | |||||
B 7 | A0 I | 1.15 | +0.237 | 162 | |||||
A 38 | A0 I | 1.33 | +0.893 | 147 | |||||
A 10 | A0 I | 1.41 | +0.650 | 187 | |||||
A 43 | A2 I | 1.15 | +0.960 | 185 | |||||
A 6 | A3 I | 1.18 | +0.608 | 288 | |||||
A 5 | A3 I | 1.45 | +0.597 | 260 | |||||
A 28 | A2 I | 1.67 | +0.785 | 219 | |||||
A 33 | A0 I | 1.86 | +0.840 | 199 | |||||
A 25 | B8 I | 1.78 | +0.757 | 229 | |||||
A 14 | B9 I | 1.37 | +0.700 | 225 | |||||
A 9 | A5 I | 0.85 | +0.643 | 228 | |||||
A 4 | A5 I | 1.08 | +0.585 | 256 | |||||
C1 14 | A5 I | 0.97 | −0.031 | 107 | |||||
C1 52 | A2 I | 0.80 | +0.263 | 149 | |||||
C2 54 | B9 I | 1.65 | +0.263 | 223 | |||||
C1 49 | A5 I | 0.99 | +0.241 | 165 | |||||
C2 51 | B9 I | 1.47 | +0.234 | 213 | |||||
C1 47 | A0 I | 1.25 | +0.229 | 182 | |||||
C1 42 | A2 I | 1.09 | +0.195 | 149 | |||||
C2 44 | A2 I | 1.28 | +0.153 | 142 | |||||
C1 34 | B8 I | 1.60 | +0.111 | 176 | |||||
C2 37 | A0 I | 1.50 | +0.092 | 168 | |||||
C1 32 | B9 I | 1.24 | +0.093 | 171 | |||||
C2 29 | B8 I | 1.55 | +0.033 | 160 | |||||
C2 22 | A2 I | 1.25 | +0.027 | 140 | |||||
C2 19 | A3 I | 1.38 | −0.006 | 91 | |||||
C1 16 | A5 I | 1.03 | −0.013 | 164 | |||||
C1 15 | B8 I | 1.68 | −0.024 | 158 | |||||
C1 11 | A3 I | 1.19 | −0.062 | 123 | |||||
C2 14 | B8 I | 1.65 | −0.049 | 130 | |||||
C1 8 | B8 I | 1.60 | −0.093 | 149 | |||||
C2 8 | A0 I | 1.70 | −0.101 | 95 | |||||
D 45 | A2 I | 1.58 | −0.142 | 129 | |||||
C2 1 | A0 I | 1.58 | −0.162 | 86 | |||||
C2 2 | A2 II | 1.71 | −0.155 | 70 | |||||
D 36 | B9 I | 1.65 | −0.224 | 78 | |||||
D 6 | A3 II | 1.58 | −0.501 | 58 |
Note. Parameters of galaxy NGC 55: R25 = 16.18 arcmin (de Vaucouleurs et al. 1991), position angle PA = 109° (see Section 2.2), inclination i = 78° ± 4° (see Section 2.2), central coordinates = 00h 14m 54009, = −39° 11' 49267 (Hummel et al. 1986; Puche et al. 1991). x is the coordinate along the major axis, y is the deprojected coordinate along the minor axis. / ( in units of 104 K).
Download table as: ASCIITypeset image
Photometry is needed to determine reddening, extinction, and apparent bolometric magnitudes, which can then be used for a determination of distances through the FGLR method. As pointed out by Castro et al. (2012), the original ground-based photometry obtained with the Warsaw 1.3 m telescope at Las Campanas Observatory and provided in the 2008 paper was affected by calibration issues. Recalibrated photometry, which was used for the early B-supergiant targets in Castro et al. (2012), is available for almost all our targets. However, a comparison with a subsample of 13 stars, for which HST/ACS photometry obtained within the ANGST treasury survey of nearby galaxies (Dalcanton et al. 2009) is available, revealed large differences in both the V and I bands of a few tenths of a magnitude. We thus decided to only use the subsample of 13 stars for the distance determination. The photometry for these objects, together with the information needed for the distance determination, is given in Table 4.
Castro et al. (2012) have analyzed 12 early B-supergiants of spectral type B0–B5 and determined effective temperature, gravity, and metallicity. In our investigation of the metallicity and the metallicity gradient we include the results from their paper and add the metallicities to enlarge the sample. The list of these objects together with their stellar parameters, galactic positions, galactocentric distances, and radial velocities is provided in Table 2. For one of these objects, C1 13, HST photometry is available. It is therefore included in the subsample used for the distance determination.
Table 2. Late O, Early B, and Mid B Supergiants in NGC 55
Star | Spectral | F | [Z] | R/R25 | x/R25 | y/R25 | vrad | ||
---|---|---|---|---|---|---|---|---|---|
type | (K) | (cgs) | (cgs) | (dex) | (km s−1) | ||||
(1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) | (9) | (10) |
A 08 | O9.7 I | 27700 ± 1200 | 2.91 | 1.14 ± 0.07 | −0.37 ± 0.15 | +0.631 | 304 | ||
C1 44 | B0 I | 26200 ± 1500 | 2.83 | 1.16 ± 0.07 | −0.53 ± 0.15 | +0.211 | 181 | ||
C1 9 | B1 I | 17400 ± 1500 | 2.37 | 1.41 ± 0.07 | −0.51 ± 0.15 | −0.081 | 118 | ||
C1 13 | B1 I | 24200 ± 1400 | 2.69 | 1.15 ± 0.09 | −0.45 ± 0.15 | −0.042 | 115 | ||
C1 45 | B1 I | 21500 ± 2500 | 2.73 | 1.40 ± 0.12 | −0.44 ± 0.15 | +0.216 | 179 | ||
A 17 | B1 I | 22500 ± 2300 | 2.85 | 1.44 ± 0.08 | −0.45 ± 0.15 | +0.733 | 207 | ||
D 27 | B2 I | 19400 ± 1600 | 2.41 | 1.26 ± 0.10 | −0.42 ± 0.15 | −0.317 | 152 | ||
A 27 | B2 I | 16300 ± 1100 | 2.13 | 1.28 ± 0.09 | −0.57 ± 0.15 | +0.778 | 199 | ||
C1 53 | B2.5 I | 18200 ± 2000 | 2.54 | 1.50 ± 0.11 | −0.21 ± 0.15 | +0.269 | 164 | ||
B 31 | B2.5 I | 16700 ± 900 | 2.09 | 1.20 ± 0.08 | −0.28 ± 0.15 | +0.490 | 190 | ||
A 26 | B2.5 I | 16100 ± 1000 | 2.15 | 1.32 ± 0.09 | −0.15 ± 0.15 | +0.770 | 220 | ||
A 11 | B5 I | 14400 ± 1000 | 1.94 | 1.31 ± 0.08 | −0.37 ± 0.15 | +0.662 | 209 |
Note. Data from Castro et al. (2012). Parameters of galaxy NGC 55: R25 = 16.18 arcmin (de Vaucouleurs et al. 1991), position angle PA = 109° (see Section 2.2), inclination i = 78° ± 4° (see Section 2.2), central coordinates α2000 = 00h 14m 54009, δ2000 = −39° 11' 49267 (Hummel et al. 1986; Puche et al. 1991). x is the coordinate along the major axis, y is the deprojected coordinate along the minor axis. gF = g/ ( in units of 104 K).
Download table as: ASCIITypeset image
2.2. Geometric and Kinematic Model of the Disk of NGC 55
The investigation of the metallicity distribution across the disk of NGC 55 as a function of galactocentric radius requires a deprojection of the targets into the intrinsic plane of the galactic disk in order to calculate galactocentric distances. We use the information given in Table 1 for the deprojection. Our values for the position angle and inclination were guided by the disk model developed by Puche et al. (1991) and are simplified averages of the inner disk. The range of ±4° in the inclination angle is used to estimate the uncertainties in deprojected galactocentric distance for the individual objects. We note that Westmeier et al. (2013) have developed an improved model of the warped disk but the changes relative to Puche et al. are relatively small in the inner disk, where our targets are located, and are mostly important in the outer disk, for which Westmeier et al. present new data. Deprojected coordinates y perpendicular to the major axis and x along the major axis are given in Tables 1 and 2 together with galactocentric distances in units of the isophotal radius R25.
With the deprojected location of our targets we can also compare the radial velocities of our targets. For this purpose we use the kinematic models and rotation curves by Puche et al. (1991) and Westmeier et al. (2013) together with the systemic velocity of Table 1. We note that Westmeier et al. (2013) have found that the systemic velocity given by Puche et al. (1991) is in error and we use their systemic velocity.
2.3. Spectral Analysis Method
The analysis method is described in detail in Kudritzki et al. (2013, 2014) and in Hosek et al. (2014). In short, we compare normalized observed spectra with synthetic spectra drawn from a comprehensive grid of metal line-blanketed model atmospheres with extensive non-LTE line formation calculations using elaborate atomic models (Przybilla et al. 2006). The grid of models is described in Kudritzki et al. (2008, 2012). The fit procedure uses the higher Balmer lines (usually H5,6,8,9,10) in a first step to constrain gravity as a function of effective temperature. This yields a relationship between gravity and effective temperature for each star (see Figure 3 in Kudritzki et al. 2014). Through this relationship, gravity is usually constrained with an accuracy of 0.05 dex.
Then, in a second step, up to 11 spectral windows in the range from 3990 to 6000 Å dominated by metal lines are selected and a comparison between observed and calculated flux is carried out as a function of metallicity for each point along the gravity–temperature relationship (metallicity [Z] is defined as [Z] = log Z/Z⊙, where Z⊙ is the solar metallicity). In this way, the quality of the fits to the metal lines can be assessed by calculating a χ2 value for each temperature and metallicity. The minimum of χ2 and the isocontours of Δχ2 around the minimum then yield the metallicity and effective temperature of the star with the corresponding fit uncertainties. The final gravity corresponds to the gravity of the gravity–temperature relationship at the temperature determined from the χ2 minimum. The fit uncertainties of the gravities are a combination of the 0.05 dex accuracy at each temperature as mentioned above and the uncertainty in temperature. The next section will show a few examples of the spectral fits obtained. Note that the χ2 fit determines stellar metallicity from the contribution of many elements, including the iron peak and α-elements.
In the fit procedure we take into account the fact that the actual spectral resolution can be better than the nominal resolution of the spectrograph in situations of good atmospheric seeing. We thus measured the effective resolution for the set of spectra in each field. We found that the spectra of fields A–C are well described with a resolution of 4.5 Å, whereas the resolution obtained for field D is 5 Å. The synthetic spectra were degraded to these values of resolution accordingly.
3. RESULTS
The results of the spectral analysis are given in Table 1. Figure 1 shows a fit of the Balmer lines for two stars taken as examples. For almost all stars in the sample, Hβ is affected by stellar winds and H ii region emission. The other Balmer lines, however, allow for an accurate determination of stellar gravities. Figure 2 displays the isocontours of the χ2 minimization in the metallicity–temperature plane. We see that, while there is some degeneracy between metallicity and temperature, both are well constrained. The Δχ2 = 3 isocontour corresponds to a 1σ uncertainty, as we have again verified by detailed Monte Carlo experiments as described by Hosek et al. (2014). This isocontour is used for all stars to determine the errors for temperature and metallicity in Table 1.
Download figure:
Standard image High-resolution imageDownload figure:
Standard image High-resolution imageFigures 3 and 4 give an impression of the fits to the metal lines in different spectral windows. Generally, the fits of the spectra are very good and the stellar parameters are well constrained. Most importantly, the accuracy of the metallicities is about 0.1 dex. We note again that the metallicity [Z] derived here comes from a fit of metal lines over a wide range of elements, mostly iron, titanium, chromium, magnesium, and silicon. In this sense, it represents an average over many elements including the iron group. This is different from metallicities in H ii regions where usually only oxygen is used as a proxy for metallicity. For the comparison of our stellar metallicities with those in H ii regions we convert the latter to [Z] = 12 + log(O/H) − 8.69, where 8.69 corresponds to the solar oxygen abundance (Asplund et al. 2009). Such a comparison assumes that the ratios of α-elements to iron elements are solar.
Download figure:
Standard image High-resolution image3.1. Evolutionary Status
The evolutionary status of the objects of our full sample can be assessed from Figure 5, which shows flux-weighted gravities against effective temperatures compared with calculations of stellar evolution. As discussed in detail by Langer & Kudritzki (2014), this "spectroscopic Hertzsprung–Russell diagram" is a very useful tool to constrain stellar properties in situations where either the distance is unknown or accurate photometry, which can be used to determine luminosity or bolometric magnitudes, is lacking. From Figure 5, which compares with the evolutionary tracks by Eckström et al. (2012), it is apparent that the stars of our sample are in an advanced stage of stellar evolution. From their flux-weighted gravities we conclude that they represent a mass range from 15 to 40 M⊙. Our sample of selected stars is thus comparable to those of our previous studies on other galaxies referred to in the publications already mentioned. The fact that no lower-mass early B-type supergiants are present in our sample is a consequence of the lower V-magnitude limit in the selection of our targets for multi-object spectroscopy. Early B-type supergiants have a much higher bolometric correction than later spectral types. As a consequence, at the same level of V magnitude these objects have higher luminosities and, thus, lower flux-weighted gravities.
Download figure:
Standard image High-resolution image3.2. Metallicity and Metallicity Gradient
In Figure 6 we plot stellar metallicities as a function of galactocentric distance. We see that the two different groups of our sample—early B-type supergiants and late B-type/A-type supergiants—give consistent metallicities. This is reassuring, because different types of model atmospheres and atomic models were used for the calculations of non-LTE line formation and different methodologies were used for the analysis. We conclude that systematic effects caused by these differences are small.
Download figure:
Standard image High-resolution imageThe average metallicity is somewhat lower than in the LMC and between [Z] = −0.4 and −0.5. A linear regression of the form
accounting for errors in both the x-axis and the y-axis reveals a significant metallicity gradient. We obtain [Z]0 = −0.37 ± 0.03 dex and grad[Z] = −0.22 ± 0.06 dex/R25. This is the first detection of a metallicity gradient in the disk of this galaxy.
The local scatter around this relation is about 0.15 dex. This is larger than most of the individual metallicity uncertainties and may be indicative of some chemical inhomogeneity in the disk of this galaxy. A few objects above the regression between 0.5 and 0.8 R/R25 seem to be outliers. We highlight them with different symbols (stars and squares) in Figure 6. In order to test whether these outliers are spatially correlated we also plot the deprojected locations of our objects in Figure 7. We find indeed that these objects cluster on the major axis between 0.5 and 0.8 R/R25.
Download figure:
Standard image High-resolution imageIt is also interesting to check whether some of these chemical abundance outliers are dynamically distinct. For this purpose we use the rotation curve derived by Puche et al. (1991) together with the new corrected systemic velocity of Table 1 to calculate radial velocities of the objects in our sample predicted by the kinematic rotation model. The comparison with the observed radial velocities measured by Castro et al. (2008) is plotted in Figure 8. As is evident from the figure, four of the eight suspected chemical outliers are definite kinematic outliers.
Download figure:
Standard image High-resolution imageThe level of chemical enrichment in the disk of NGC 55 obtained from our analysis of blue supergiants is generally in agreement with the oxygen abundances of the few H ii regions studied. We have used the line fluxes published by Webster & Smith (1983) and applied the "direct method" using the auroral forbidden line [O iii]λ4363. We obtained [Z] = −0.39, −0.34, −0.14, −0.36 for their H ii regions 1, 2, 3, 4, respectively. In the same way we find [Z] = −0.52, −0.35, −0.11 for the H ii regions 3, 10, "center" investigated by Stasinska et al. (1986). For the central disk H ii region in Tüllmann et al. (2003) we obtain [Z] = −0.52.
We note that Pilyugin et al. (2014), applying their newly calibrated strong line method on eight H ii regions in NGC 55, four of them in the center, obtain a value of −0.65 below solar and no significant abundance gradient. This is at odds with results obtained by Webster & Smith (1983) and Castro et al. (2012). It is also not supported by our result from the spectroscopy of blue supergiants. Kudritzki et al. (2015) have already pointed to a systematic offset of 0.15–0.2 dex between the H ii strong line calibration of Pilyugin et al. (2014) and their studies of blue supergiants in a sample of galaxies. The difference encountered here goes in the same direction, but is slightly larger.
3.3. Mass–Metallicity Relationship
The relationship between the total stellar mass of a galaxy and its average metallicity, the mass–metallicity relationship ("MZR"), is a Rosetta Stone for understanding the formation and evolution of galaxies. After the pioneering paper by Tremonti et al. (2004), which studied 50,000 galaxies from the Sloan Digital Sky Survey (SDSS), revealed the existence of a tight MZR, great effort has been made to theoretically interpret this relationship either by laborious numerical simulations or by analytical models (see, for instance, Zahid et al. 2014). However, as shown by Kewley & Ellison (2008), the original relationship based on the analysis of strong emission lines from H ii regions is subject to large systematic errors, which are poorly understood. Therefore, Kudritzki et al. (2012) have started to build up an MZR of galaxies in the local universe, which is based solely on the results of stellar spectroscopy. Our analysis of the young stellar population in the disk of NGC 55 now enables us to add an additional data point to this relationship. This is done in Figure 9. Note that, in order to be consistent with Kudritzki et al. (2012) and Kewley & Ellison (2008), we use the metallicity at two disk scale lengths (corresponding to 0.22 R/R25 in the case of NGC 55) for the plot. We have also added the new results by Hosek et al. (2014) and Kudritzki et al. (2014). The stellar mass used for NGC 55 is determined in the next section.
Download figure:
Standard image High-resolution imageAs is evident from Figure 9, the MZR based on stellar spectroscopy does also form a tight relationship and NGC 55 fits nicely into the middle of it. The SDSS MZR by Andrews & Martini (2013), which is based on stacked spectra of galaxies with similar mass so that the auroral lines of H ii regions can be used for a more precise determination of oxygen abundances, shows a very similar shape. However, there is small shift of 0.15 dex in metallicity or 0.3 dex in stellar mass. In view of the different techniques applied for the diagnostics of metallicities and stellar masses we find the agreement compelling. It seems that many of the strong line calibrations discussed by Kewley & Ellison (2008, see their Figure 2) can be ruled out.
3.4. The Nature of the Extraplanar H ii Regions in NGC 55
Tüllmann et al. (2003) have detected two H ii regions in the central region of NGC 55 clearly above the plane of the galactic disk at heights of 0.8 and 1.5 kpc based on their adopted distance of 1.6 Mpc. With our new distance of 2.34 Mpc (see Section 5) the heights above the disk plane increase to 1.13 and 2.22 kpc, respectively. They argue that hydrodynamical considerations rule out an ejection scenario of the ionizing O stars with the surrounding H ii region gas from the central disk. As a consequence, the ionizing massive stars must have been born in situ throughout the disk. In order to support their argument they determine the metallicity of the two extraplanar H ii regions and obtain [Z] = −0.9 ± 0.2, a significantly sub-solar value. They then study one central H ii region in the disk and derive [Z] = −0.64 ± 0.1. The abundances are derived using the auroral line [O iii]λ4363 (or an upper limit for the flux in this line) and the R23 method in the calibration by McGaugh (1991). From the difference in metallicity they conclude that the extraplanar objects are chemically distinct from the disk population, which supports the idea of in situ formation throughout the disk.
However, taking the uncertainties in the metallicties into account, one could argue that at the margin of the errors they almost overlap. The chemical proof of in situ formation all hinges on the metallicity of the one central H ii region studied. With our stellar study using a large sample of objects the central metallicity in the disk of NGC 55 is accurately restricted to [Z]0 = −0.37 ± 0.03 dex. This is a significant difference from the metallicity of the two extraplanar H ii regions and provides very strong support to the scenario proposed by Tüllmann et al. (2003) of in situ formation of massive stars above the disk, facilitated by the presence of large amounts of extraplanar gas due to pronounced mass outflow and inflow in this galaxy (see next section).
3.5. Relationship Between Spectral Type and Effective Temperature
The set of homogeneous spectra of 46 objects of spectral type B8–A5 with well determined stellar parameters offers the opportunity to investigate the relationship between spectral type and effective temperature at a metallicity slightly lower than that of the LMC. For this purpose we use the spectral types assigned by Castro et al. (2008) and the values of our analysis provided in Table 1. Table 3 gives the mean and standard deviation for each subtype as determined by our spectral analysis. (At spectral type A2 two outliers, D 39 and C2 22, were encountered, which are not included in the calculation of the mean.) The comparison with the relationships for the Milky Way (MW) obtained by Kudritzki et al. (2003) and more recently by Firnstein & Przybilla (2012) based on the detailed non-LTE analysis of high-resolution spectra with high S/N shows good agreement. The relationship of Evans & Howarth (2003) for the Milky Way is also in agreement, whereas their relationship for the SMC is significantly cooler at later spectral types. We note that Evans & Howarth (2003) adopted a metallicity [Z] = −0.76 for the SMC, when they calibrated their relationship. This metallicity is significantly lower than for most of our objects studied in NGC 55. In addition, their calibration did not include non-LTE effects for the calculation of the metal lines. LTE is very likely sufficient for their SMC objects, because most of them are of luminosity class II. On the other hand, most of the NGC 55 objects have higher luminosities corresponding to luminosity class I and require non-LTE calculations to fit the spectrum.
Table 3. Relationships between Spectral Type and of Blue Supergiant Stars
This work | FP12 | K03 | EH03 | EH03 | |
---|---|---|---|---|---|
[Z] | NGC 55 | MW | MW | MW | SMC |
Spectral | |||||
type | (K) | (K) | (K) | (K) | (K) |
(1) | (2) | (3) | (4) | (5) | (6) |
B8 | 11800 ± 490 | 12200 ± 410 | 12000 | 13000 | 12000 |
B9 | 10600 ± 330 | 10920 ± 220 | 10500 | 10750 | 10500 |
A0 | 9800 ± 340 | 9840 ± 290 | 9500 | 9750 | 9500 |
A2 | 9000 ± 330 | 8960 ± 200 | 9000 | 9000 | 8500 |
A3 | 8600 ± 150 | 8430 ± 60 | 8500 | 8500 | 8000 |
A5 | 8130 ± 130 | ⋯ | ⋯ | 8250 | 7750 |
Note. FP12: Firnstein & Przybilla (2012), K03: Kudritzki et al. (2003), EH03: Evans & Howarth (2003).
Download table as: ASCIITypeset image
We conclude that, at the metallicity level of NGC 55 and for objects of luminosity class I, we do not see a systematic difference in the relationship between spectral type and Teffwhen compared with the Milky Way.
4. A CHEMICAL EVOLUTION MODEL FOR THE DISK OF NGC 55
The stratification of metallicity of the young stellar population over the disk of a spiral galaxy compared with the radial profiles of ISM gas mass and stellar mass contains crucial information about the history of chemical evolution in the galactic disk. In the simple closed-box model, which neglects the effects of accretion inflow of material from the cosmic web and of outflow via galactic winds, metallicity at a certain galactocentric radius is uniquley determined by the ratio of the column densities of stellar mass and gas mass at this radius. Metallicity gradients are then a consequence of the variation of this ratio as function of galactocentric radius. However, it is now well known that accretion and galactic winds play an important role in the chemical evolution of galaxies and, thus, the closed-box model can provide only qualitative insight into galaxy evolution. As a simple step further, Kudritzki et al. (2015) have developed a model of chemical evolution in which accretion and outflow are taken into account, but the assumption is made that the rates of mass gain and mass loss in units of the star formation rate have been constant. They provide qualitative arguments in their paper in support of this simplifying assumption. Under this assumption, the equations of chemical evolution can be solved analytically and the observed radial distributions of metallicity and mass can then be used to constrain the rates of mass infall Λ and outflow η in units of the star formation rate ψ:
and the mass accretion factor
In this way, Kudritzki et al. (2015) have determined rates of mass loss and accretion for 20 nearby disk galaxies. With a well constrained determination of central metallicity and the metallity gradient it is an obvious step to apply the same analysis to NGC 55. This is the first application of this approach to a galaxy with a well determined spatial distribution of metallicity of the young stellar population through spectroscopy. Similar to Kudritzki et al. (2015) we obtain profiles of the column density of stellar mass as a function of galactocentric radius from deprojected mid-IR images at 3.4 μm obtained with the WISE (W1 band) and Spitzer (IRAC) space observatories (for details of the conversion from mid-IR surface photometry to column densities of mass we refer to Kudritzki et al.). The column density distribution of atomic gas mass in the ISM can be obtained from deprojected 21 cm radio observations, where the radio fluxes in the 21 cm line are converted into H imasses and then multiplied by a factor of 1.36 to include the mass of helium and metals. We have reanalyzed the original data cube from the Very Large Array obtained by Puche et al. (1991) for that purpose. Westmeier et al. (2013) find that the galaxy's 21 cm flux is higher than that obtained by Puche et al. (1991), which is very likely caused by differences in the flux calibration or missing flux through poor coverage of the uv-plane. We apply a factor of 1.3 to the data of Puche et al. to correct for this effect.
In the inner disks of spiral galaxies the molecular gas can make an important contribution to the total gas mass. An estimate of the molecular gas mass is usually obtained from an observation of CO, which is then converted to the mass of molecular hydrogen, for which the factor 1.36 is then applied again to include helium and metals. Unfortunately, no CO observations are available in the case of NGC 55. We therefore use an indirect way to estimate the column density distribution of the molecular mass. We start from the deprojected observation of the 21 cm continuum radio flux, which is then converted into far-IR flux using the correlation between far-IR and radio flux, thus yielding the radial profile of the star formation rate (see Ho et al. 2010). The star formation rates are then transformed into column densities of molecular hydrogen by application of the molecular Kennicutt–Schmidt law as parameterized by Leroy et al. (2013).
The integrated total stellar mass of the disk of NGC 55 (in units of the solar mass) is log Mstar = 9.29. We note that the mass depends on the distance assumed. We have used a distance of 2.34 Mpc, which we derive in the following section.
Figure 10 displays the column densities of stellar and ISM atomic and molecular masses obtained in this way as a function of galactocentric radius. Obviously, the contribution from molecular gas is important only in the central disk region and has only a weak effect on the constraints for outflow and accretion. The central stellar mass profile does not show a significant indication of a bulge contribution. We thus refrain from an attempt at bulge decomposition as carried out for most of the galaxies in Kudritzki et al.
Download figure:
Standard image High-resolution imageApplying the same fit algorithm of the observed metallicity distribution as Kudritzki et al., we can then constrain the rate of outflow and accretion. We obtain η = 0.72 and Λ = 1.38. This means that the chemical evolution of NGC 55 is characterized by large amounts of gas outflow and infall. Compared with all the other galaxies studied by Kudritzki et al. the infall rate of NGC 55 is by far the largest. Qualitatively, this agrees well with the detailed morphological, kinematic, and dynamic study by Westmeier et al. (2013) and their conclusion that "internal and external processes, such as satellite accretion or gas outflows, have stirred up the gas disc."
With the large value of Λ detected it is interesting to investigate in what way NGC 55 differs from the other galaxies analyzed with the same approach to chemical evolution. We compare with the sample of galaxies with well determined η and Λ in Kudritzki et al. (see their Figure 11) and find the striking anticorrelation of Λ with total galaxy stellar mass shown in Figure 11. Of all the spiral galaxies studied so far with our method NGC 55 has by far the lowest mass, and it seems that Λ increases significantly with decreasing mass. A power law with α = 0.8 provides a reasonable fit to the data, except at the high-mass end, where very small values of Λ are observed. We note that on the "main sequence of galaxies" star formation increases with stellar mass following a power law with β between 0.7 and 1.0 (see Lee et al. 2015 and references therein). Consequently, for the sample displayed in Figure 11 the mass accretion rate can be expressed as . This means that the mass accretion rates depend only very weakly on galaxy mass, because α and β almost cancel each other in the exponent. (At high galaxy masses accretion is probably reduced by shocks formed by the accretion process, which heat the infalling material and slow down accretion onto the disks.)
Download figure:
Standard image High-resolution imageTo confirm this scenario is beyond the scope of this paper. We will need more studies, particularly at the low-mass end of the observed relation. This will require more detailed, spatially resolved observations of the atomic and molecular gas in addition to multi-object stellar or H ii-region spectroscopy and a careful investigations of star formation rates. The corresponding work is currently under way.
Figure 6 shows the metallicity distribution calculated with our best-fit chemical evolution model compared to the observed metallicities. The model provides a good fit to the observed data. We take this as a confirmation that the concept of retrieving important information on galaxy evolution from the spatial distribution of the metallicity of the young stellar population or the ISM is a promising approach with great potential.
5. DISTANCE
The subsample of objects in Table 4 with HST/ACS photometry can be used to determine a distance using the FGLR technique briefly described in Section 1. With temperature, gravity, and metallicity determined by spectroscopy we can calculate the intrinsic spectral energy distribution of our targets and their V − I colors. Comparison with the observed colors then yields the reddening (we apply the relation and, by assuming , the value of extinction AV. Adding the bolometric correction calculated from the intrinsic energy distribution to the dereddened V magnitude, we then obtain dereddened apparent bolometric magnitudes for each object. The values obtained for the reddening, the bolometric corrections, and the bolometric magnitudes are also given in Table 4, together with the logarithm of the flux-weighted gravities.
Table 4. Supergiants in NGC 55 with HST Photometry Used for FGLR
Star | F | mbol | V | I | E(B – V) | BC | (V – I)0 |
---|---|---|---|---|---|---|---|
(cgs) | (mag) | (mag) | (mag) | (mag) | (mag) | (mag) | |
(1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) |
C1 17 | 18.905 ± 0.038 | 19.295 ± 0.002 | 19.072 ± 0.003 | 0.109 ± 0.010 | −0.041 ± 0.020 | +0.083 ± 0.012 | |
C2 15 | 19.024 ± 0.048 | 19.709 ± 0.003 | 19.546 ± 0.003 | 0.116 ± 0.006 | −0.314 ± 0.044 | +0.014 ± 0.006 | |
C2 13 | 18.768 ± 0.059 | 19.085 ± 0.002 | 18.936 ± 0.002 | 0.061 ± 0.011 | −0.123 ± 0.046 | +0.071 ± 0.014 | |
C1 14 | 18.551 ± 0.068 | 18.790 ± 0.002 | 18.584 ± 0.002 | 0.077 ± 0.018 | +0.008 ± 0.037 | +0.107 ± 0.008 | |
C1 34 | 18.155 ± 0.065 | 19.484 ± 0.003 | 19.204 ± 0.003 | 0.229 ± 0.007 | −0.598 ± 0.062 | −0.013 ± 0.008 | |
C2 37 | 18.864 ± 0.075 | 19.534 ± 0.003 | 19.338 ± 0.003 | 0.129 ± 0.009 | −0.256 ± 0.069 | +0.030 ± 0.010 | |
C2 29 | 17.417 ± 0.100 | 19.239 ± 0.002 | 18.863 ± 0.003 | 0.314 ± 0.012 | −0.819 ± 0.092 | −0.026 ± 0.015 | |
C2 22a | 17.394 ± 0.107 | 18.737 ± 0.002 | 18.336 ± 0.002 | 0.277 ± 0.015 | −0.456 ± 0.096 | +0.046 ± 0.019 | |
C2 19 | 19.532 ± 0.061 | 19.532 ± 0.003 | 19.729 ± 0.003 | 0.099 ± 0.014 | −0.064 ± 0.042 | +0.056 ± 0.017 | |
C1 16 | 18.872 ± 0.051 | 19.281 ± 0.002 | 19.025 ± 0.002 | 0.131 ± 0.011 | +0.010 ± 0.037 | +0.088 ± 0.014 | |
C1 11a | 19.418 ± 0.058 | 19.848 ± 0.003 | 19.621 ± 0.003 | 0.116 ± 0.012 | −0.058 ± 0.042 | +0.078 ± 0.015 | |
C2 14 | 18.554 ± 0.047 | 19.328 ± 0.002 | 19.321 ± 0.003 | 0.027 ± 0.005 | −0.689 ± 0.044 | −0.027 ± 0.005 | |
C1 13 | 16.328 ± 0.133 | 19.378 ± 0.003 | 19.301 ± 0.003 | 0.211 ± 0.008 | −2.375 ± 0.130 | −0.303 ± 0.014 |
Note. Bolometric correction (BC) and intrinsic color 0 calculated from model atmospheres (see Kudritzki et al. 2008 for details).
aNot included in the distance determination because of variability (see text).Download table as: ASCIITypeset image
The NGC 55 Cepheid search carried out by Pietrzyński et al. (2006) in V and I provided us with repeated photometry of the targets in Table 4 over a period of 900 days, allowing us to assess their photometric variability. According to Bresolin et al. (2004), blue supergiants usually show only a small amount of photometric variability, which does not significantly affect determinations of distance through the FGLR method. But in a few cases larger photometric variations are encountered. Of the objects in Table 4, star C2 22 showed clear signs of variability, increasing in brightness in both V and I by 0.2 mag over the observed period. It was therefore not included in the final sample used for the distance determination. In addition, C1 11 showed brightness variations up to 0.1 mag and was also dropped from the sample. For all other objects the level of variability inferred from correlated changes in V and I is smaller than or equal to 0.05 mag. For those objects we add an additional uncertainty of 0.05 mag to the uncertainties in bolometric magnitude in Table 4 to account for variability of blue supergiants in the distance determination described below.
Figure 12 demonstrates that the bolometric magnitudes and flux-weighted gravities of our objects are tightly correlated. This is the FGLR introduced in Section 1. As in previous work (Urbaneja et al. 2008; U et al. 2009; Kudritzki et al. 2012, 2014; Hosek et al. 2014) it can be used to determine extragalactic distances. Most recently, Urbaneja et al. (2016) have introduced a new calibration of the FGLR through a spectroscopic study of 90 blue supergiants in the LMC. Based on the analysis of the (Mbol, log gF) diagram of their 90 LMC supergiants they find that the FGLR changes its slope at log gF = 1.29. They fit a two-component FGLR with two slopes to their data and determine a new zero point. The new FGLR in absolute bolometric magnitude Mbol is given by
and
where
with log = 1.29, a = 3.46 ± 0.12, b = −7.92 ± 0.03 mag, and alow = 7.93 ± 0.25.
Download figure:
Standard image High-resolution imageTo determine the distance to NGC 55 we use the observed flux-weighted gravities to calculate absolute magnitudes by means of the calibration FGLR. We then calculate individual distance moduli for each object and obtain the final distance from a weighted mean that accounts for the observational errors in bolometric magnitude and the logarithm of flux-weighted gravity. We obtain a distance modulus of . The systematic error comes from uncertainties in the FGLR parameters. Figure 12 also shows the calibration FGLR shifted to the distance determined.
We realize that our FGLR-based distance modulus is 0.42 mag larger than the m − M = 26.43 ± 0.09 obtained by Gieren et al. (2008) in our Araucaria collaboration. Gieren et al. use ground-based J- and K-band photometry of Cepheids in conjunction with the V and I photometry by Pietrzyński et al. (2006). This is a very reliable technique, particularly because of the use of the K-band, which minimizes the effects of reddening for the Cepheids (we note that the average reddening of the targets in Table 4 is mag, very similar to the reddening found in the Cepheid study). Usually the agreement between Cepheid and FGLR distances is satisfactory. For instance, in the case of NGC 300, a neighboring galaxy to the Sculptor group, our new FGLR calibration yields a distance modulus of , while Gieren et al. (2005a) obtain 26.37, again from ground-based photometry combining the V, I, J, and K bands. As mentioned in Section 1, the physical parameters of the two galaxies are very similar. However, there is one fundamental difference between NGC 300 and NGC 55, namely the fact that the latter is an almost edge-on galaxy. As a result, NGC 55 has a significantly larger surface brightness, for instance in the K-band, than NGC 300 (see Figure 13). This increases the potential for blending with nearby fainter objects, which would increase the measured brightness of the Cepheids and lead to a shorter distance. We recall that Bresolin et al. (2005) have tested the effects of crowding for the case of NGC 300 by comparing V- and I-band HST photometry with the ground-based observations and found only a small systematic effect of 0.04 mag. However, in the case of NGC 55 with a surface brightness a magnitude higher the systematic effect might be significantly larger. On the other hand, blue supergiants are several magnitudes brighter than Cepheids, and are thus much less affected by blending. However, even for the supergiants, we detect large differences between HST and ground-based photometry in the case of NGC 55. We note that the TRGB (tip of the red giant branch) distance to NGC 55 is 26.62 ± 0.1 mag (EDD database, http://edd.ifa.hawaii.edu, see Tully et al. 2009). The PNLF distance modulus found by van de Steene et al. (2006) is 26.8 mag, albeit with a relatively large error of ±0.3 mag. Should our FGLR distances be correct, then NGC 55 would be at a slightly larger distance than NGC 300, 2.34 Mpc versus 1.85 Mpc. With an angular separation of 8° (Westmeier et al. 2013) the physical separation would increase from 270 kpc (with the two galaxies at the same distance) to 570 kpc.
Download figure:
Standard image High-resolution imageAt this point, the reasons for the discrepancy between the FGLR and Cepheid distances remain unresolved. A larger sample of blue supergiants with HST photometry and a measurement of HST magnitudes for the Cepheids will be crucial for investigating this issue further.
6. CONCLUSIONS
The comprehensive spectroscopic study of blue supergiant stars distributed over the disk of NGC 55 has led to the first detection of a metallicity gradient in this almost edge-on late-type spiral galaxy. The application of a chemical evolution model indicates the effects of intensive infall and outflow, in agreement with recent radio observations of the galaxy, which conclude that the disk is very likely stirred up and disturbed by infalling and outflowing gas. We also find indications of chemical inhomogeneities, which support this picture. The significant difference between the central metallicity of the disk and the metallicity of two extraplanar H ii regions above the central disk provides strong evidence for in situ star formation outside the galactic plane. A distance determination using the FGLR method leads to a distance that is larger than the distance to NGC 300, a neighboring galaxy to the Sculptor group. Further HST photometry will be needed to settle the issue of distance determination for this galaxy.
The authors express their thanks to the referee whose critical and constructive remarks helped to improve the paper. R.P.K. and F.B. acknowledge support by the National Science Foundation under grants AST-1008789 and AST-1108906. R.P.K., M.A.U., and W.G. were supported by the Munich Institute for Astro- and Particle Physics (MIAPP) of the DFG cluster of excellence "Origin and Structure of the Universe." W.G. and G.P. gratefully acknowledge financial support for this work received from the BASAL Centro de Astrofísica y Tecnologías Afines (CATA), PFB-06/2007. W.G. also acknowledges support from the Millennium Institute of Astrophysics (MAS) of the Iniciativa Científica Milenio del Ministerio de Economía, Fomento y Turismo de Chile, grant IC120009. In addition, support from the Ideas Plus grant of the Polish Ministry of Science and Higher Education and TEAM subsidies of the Foundation for Polish Science (FNP) is acknowledged by G.P.
Facilities: VLT:FORS2 - , HST:ACS - .
Footnotes
- 8
We transform the nebular oxygen abundances (O/H) to metallicity relative to solar [Z], adopting 12 + log(O/H)⊙ = 8.69 from Asplund et al. (2009).