The luminosity-metallicity relation

120 Chapter 4. Star-formation and chemical history in the large-scale structure of the Hercules Supercluster

Figure 4.14: Dierences between O/H estimates given by the four calibrations that we have used: the c-method (CM) of Pilyugin et al. (2012), the N2 calibration of PMC09, the O3N2 calibration of PP04, and the models ofDors et al.(2011) for the MS galaxies in HSC.

derived now are generally in good agreement with the previously derived values.

In Figure 4.15 we represent each one of the three estimates of O/H vs. the mean value. We note rst that the weighted mean value for all MS galaxies is consistent with all three calibrations, within their respective statistical error. The second point is that, despite their dierences, the used calibrations show linear correlations, this implying that the mean value bears only a zero-point dierence.

This is important, given that in the following sections, the relative dierences of the chemical abundances are investigated, between galaxies inside and outside the cluster cores, and a zero-point shift would not introduce any bias. Thus, here we adopt the weighted mean value of 12+log(O/H), considering an error of 0.1 dex.

Accordingly, for the N/O ratio we see that the values derived using the c-method and the PMC09 N2S2 calibration show a linear correlation with a zero-point dierence, and thus we adopt the mean value and an error of 0.1 dex.

In AppendixBwe quote the derived values of O/H and N/O, the adopted values, as well as all the physical properties discussed in this work, for all MS galaxies in the three clusters and DISP (A2151: Table B.1, A2152: Table B.2, A2147: Table B.3, and DISP: TableB.4).

4.4. Chemical history of low-mass galaxies in the Hercules Supercluster 121

Figure 4.15: 12+log(O/H) derived by the c-method (CM), PMC09, and PP04 cal- ibrations, vs. the weighted mean value. The grey points (of dierent tones) cor- respond to the three 12+log(O/H) estimates for each galaxy, and the linear ts to the values derived using the same calibration are ploted with solid lines (blue: CM estimates, red: PMC09, and blue: PP04).

MZR. Optical LZR has been found to show larger scatter than the MZR, attributed to variations in the stellar-mass-to-light ratios, produced by variations in galaxy SFRs (Bell & de Jong 2001).

In Figure 4.16 we plot the 12+log(O/H) vs. the M_B absolute magnitude for the MS galaxies, color-coded to their color g−i. We derive M_B from the SDSS g magnitude, adopting, for the clusters, their distance modulus m −M given in NED (36.0/36.23/35.91 mag for A2151/A2152/A2147 respectively). For DISP, we separate galaxies in three velocity bins of 2000 km s⁻¹ (our total velocity range is 6000 km s⁻¹, see Section 4.2) and we assign to galaxies in each velocity bin the m−M of the cluster to the corresponding velocity range. We also adopt the conversion formula g−B = −0.21 for magellanic-type irregulars (Fukugita et al.

1995), since our MS sample contains low-mass SF galaxies. We add in Figure4.16 the samples of eld dIrr ofLee et al. (2006) andvan Zee & Haynes (2006).

We observe in Figure 4.16 that the MS galaxies show an interesting feature:

in the same bin of magnitude, e.g. −18.5 < M_B < −17.5, MS galaxies span a wide range in metallicity, showing a clear color segregation, where more metal rich galaxies appear to be redder in color g−i. Figure 4.17 shows clear correlations between the chemical abundances O/H and N/O and the g−i color for the MS

122 Chapter 4. Star-formation and chemical history in the large-scale structure of the Hercules Supercluster

Figure 4.16: 12+log(O/H) vs. M_B magnitude for the MS galaxies (points, color coded to the galaxyg−icolor), and the samples of nearby dIrr ofLee et al.(2006) (cyan triangles) and van Zee & Haynes (2006) (gray squares).

galaxies. In Figure4.16, the redder and more metallic MS galaxies produce a large dispersion at −18.5< M_B <−17.5 and appear to be shifted upwards compared to the general LZR dened by eld dIrr.

Before considering a color evolution scenario for the dispersion on the LZR of our MS galaxies, we should investigate possible biases in the g−icolor. For example, while in the wavelength range of SDSS iband (centered at ∼7700Å) there are not bright emission lines, in g band (centered at ∼ 4400Å) are included the Hβ and [Oiii]λλ4959,5007 lines, which could produce a bias for galaxies with high SFRs toward bluer colors. The excess ingmagnitude, due to these nebular emission lines, is equal to their total EW divided by the FWHM of the g lter. We have checked that, for our sample of MS galaxies, an excess in g magnitude due to gas emission should be typically less than 0.02 mag, that is within the error of SDSSgmagnitude.

Another relevant parameter to take into account is the reddening. In the previous section we derived the c(Hβ) using the Balmer emission lines for our sample of galaxies. Figure 4.18 (left) shows that c(Hβ) correlates with galaxy g −i color, increasing from ∼0.0−1.0 as galaxy g−i color goes from∼0.2−1.2. Adopting the extinction law of Cardelli et al. (1989) with R_V = 3.1, A_V = 2.14c(Hβ); thus the redder galaxies of our sample could suer up to 2 mag higher extinction in the V band. This eect would aect more importantly the higher metallicity galaxies, asc(Hβ) correlates with metallicity too (Figure 4.18, right).

We adopt the simple picture where stellar population and ionized gas suer the

4.4. Chemical history of low-mass galaxies in the Hercules Supercluster 123

Figure 4.17: 12+log(O/H) (left) and log(N/O) (right) vs. g−ifor the MS galaxies.

Figure 4.18: Galaxy g−i color (left) and 12+log(O/H) (right) vs. c(Hβ) for the MS galaxies.

same extinction², and the extinction law of Cardelli et al. (1989) with R_V = 3.1, and we correct for extinction bothM_B magnitude andg−icolor, obtainingM_B,cor the(g−i)_cor.

In Figure4.19we plot the oxygen abundance (upper panel) and N/O ratio (lower panel) vs. M_B,cor magnitude, corrected for extinction. The points are color coded to the galaxy(g−i)_corcolor (also corrected for extinction). We see that galaxies are clearly separated in color in the plane of the LZR and the luminosity vs. N/O. Thus, large part of the dispersion observed in these plots is intrinsically related to galaxy evolution: in the same bin ofM_B, redder galaxies would host a more evolved stellar

2Calzetti et al.(1994) have argued that stars may suer less extinction than the gas, and we have found that this might be true for A2151 galaxies (see Figure3.12). We have checked that considering a factor of 0.8 and 0.5 would not have changed the following considerations.

124 Chapter 4. Star-formation and chemical history in the large-scale structure of the Hercules Supercluster

Figure 4.19: 12+log(O/H) (upper panel) and log(N/O) (lower panel) vs. M_B,cor magnitude, corrected for extinction. The points are color coded to the galaxy (g− i)_cor color.

population, indicating an older mass-weighted stellar age, that would be consistent with the higher metallicity and N/O ratio of these galaxies.

In Figure4.20we present the same relations of Figure4.19, but now we explore the eect of the SFR. We plot in color the EW(Hα)³ and we do not nd any clear trend. It has been claimed that the SFR can explain large part of the scatter of the MZR (Mannucci et al. 2010), and this can be understood by the correlated behavior of gas-phase metallicity and SFR after a gas infall event (we discuss this further in the following chapter, in Section 5.5). The LZR of our MS galaxies, though, does not reveal this kind of eect.

In Figure 4.21 we plot again the oxygen abundance vs. M_B,cor magnitude, cor- rected for extinction and using points color coded to the galaxy(g−i)_corcolor (also

3We remind that Hαemission has been corrected for the underlying absorption.

4.4. Chemical history of low-mass galaxies in the Hercules Supercluster 125

Figure 4.20: 12+log(O/H) (upper panel) and log(N/O) (lower panel) vs. M_B,cor magnitude, corrected for extinction.The points are color coded to the galaxy logEWHα.

corrected for extinction); this is the same as Figure 4.19 upepr panel, but now we add the eld samples of dIrr. We see that the MS galaxies follow a well dened LZR relation, matching the LZR observed for eld dIrr galaxies. The dispersion observed in the LZR of the MS galaxies, after correcting for extinction, is of the same order of that observed in the eld samples ofLee et al.(2006) andvan Zee & Haynes(2006).

Our large sample of galaxies has permitted to clearly identify a second parameter correlation in this LZR with galaxy color, i.e. with the mass-weighted age of the stellar populations.

In the following section we analyze the MZR, which is not expected to be aected by galaxy color evolution, as the mass estimates have been derived using spectral tting based on population synthesis models (see Kaumann et al. 2003); then we explore the eect of the environment on the MZR for our MS galaxies.

126 Chapter 4. Star-formation and chemical history in the large-scale structure of the Hercules Supercluster

Figure 4.21: The same as in Figure4.16usingM_B,cor magnitude and(g−i)_corcolor, corrected for extinction.