ARE (PSEUDO)BULGES IN ISOLATED GALAXIES ACTUALLY PRIMORDIAL RELICS?

Total magnitudes were derived using SExtractor (Bertin & Arnouts 1996) in the gri bands. The magnitudes were corrected applying the Galactic dust extinction computed by SDSS following Schlegel et al. (1998), and the k-correction computed using the code kcorrect (Blanton & Roweis 2007).

Structural modeling of galaxy bulges was performed using the GALFIT package (Peng et al. 2010) applied to the i-band SDSS images. We fit a Sérsic function to the bulges and an exponential function to the disks. We visually identified galaxies hosting a bar using the i-band and color images of SDSS and in these cases we fit an additional Sérsic function for this component. In all cases, the model was convolved with a point-spread function (PSF) generated from the SDSS psField.

Since the fit depends strongly in the input values, we fitted each galaxy with a variety of initial values, inspecting the results and accepting a fit if the median of the residuals in the 5 × 5 innermost pixels was lower than 10%. This was achieved for 189 galaxies, which form our sample hereafter (45% barred). Four galaxies (the three Sdm and one Sd) show no bulge. Eight galaxies show signs of interaction and their disks could not be properly fitted. Four galaxies are affected by a star in the central part and 24 (most of them highly inclined) show dust lines in their centers, which confuse the GALFIT bulge fit. Finally, 50 galaxies present residuals larger than 10% of the flux in the innermost region, mostly caused by unreliable parameters in the fit that could not be solved with any set of initial parameters tested. In the g and r bands, the fits were done using the i-band parameters, leaving free the magnitudes of the components and the disk scale length. We compare this methodology with the free fit in the g and r bands. The parameters obtained for the disk are very stable, with a difference in magnitude lower than 0.01 for 80% of the cases. Differences in bulge parameters are larger although classification based on the Sérsic index (see below) would change for only 5% of the galaxies. Further discussion and details will be presented in a follow-up paper.

Gadotti (2008) showed that the structural properties of bulges can be reliably retrieved when a_e is larger than ∼80% of the PSF half width at half maximum (HWHM). We checked the ratio between our a_e (which is PSF corrected) and the seeing HWHM given by SDSS for each i-band image. Only 6 galaxies (3% of the sample) have a_e/HWHM < 0.8 and 14 galaxies have a_e/HWHM < 1.

GALFIT provides the effective radius along the semimajor axis (a_e). To calculate the circular effective radius, we used R_e = $a_{e}\sqrt{(b/a)_{b}}$, where (b/a)_b is the bulge–axis ratio. We derive the average surface brightness within the effective radius using Equation (1) (Graham & Driver 2005):

$\begin{equation} \langle \mu _e\rangle =m_i+2.5 \log \left(2 \pi R_e^2\right), \end{equation} \tag{ 1 }$

where m_i is the magnitude in the i band given by GALFIT for the bulge component, corrected for Galactic extinction and by k-correction. 〈μ_e〉 was corrected for cosmological dimming, multiplying the total flux by (1 + z)⁴, where z is the redshift of the galaxy.

The bulge parameters obtained from the GALFIT fit are represented in Figure 1. The left panel presents the bulge Sérsic index versus the bulge-to-total (B/T) ratio. We used n_b to separate classical bulges (n_b > 2.5) and pseudobulges (n_b < 2.5). We found that 94% of bulges in the AMIGA sample are classified as pseudobulges. We used different symbols for high-B/T (B/T > 0.1) and low-B/T (B/T < 0.1) pseudobulges in the figures to check if the former represented more evolved pseudobulges. In the right panel of Figure 1 we show the Kormendy relation (Kormendy 1977) between 〈μ_e〉 and R_e for the bulges. For comparison, we show the Kormendy relation for those early-type galaxies (T < 0) in the AMIGA sample that were fitted with a Sérsic function in Fernández Lorenzo et al. (2013). To derive 〈μ_e〉, we apply the same procedure and corrections used for the bulges, but using m_i, the total magnitude of the galaxy in the i band.

Zoom In Zoom Out Reset image size

Classical bulges are expected to form an unbroken sequence with the distribution of early-type galaxies in the 〈μ_e〉–R_e plane (Carollo 1999). Pseudobulges show significantly lower 〈μ_e〉 than classical bulges for a given R_e (Gadotti 2009; Fisher & Drory 2010). The right panel of Figure 1 shows results consistent with these expectations. High-B/T pseudobulges are at the top of the pseudobulge distribution, in a region consistent with the Kormendy relation, which suggests some similarity with classical bulges. Fisher et al. (2009) found a class of non-star-forming pseudobulges which are consistent with the parameter correlation of classical bulges and early-type galaxies (Fisher & Drory 2010). These inactive pseudobulges may be composite systems where both classical and pseudobulge components have similar luminosities. This could be also the case for the high-B/T pseudobulges in our sample. Moreover, some classical and pseudobulges may also be composite systems where one component is more prominent than the other (Peletier et al. 2007; MacArthur et al. 2009).

We use (g − i) colors as indicative of the stellar population for bulges and disks in our sample. We used magnitudes from GALFIT to estimate disk colors because the disk fits were very stable (see Section 3.1). In the case of bulges, the magnitudes derived from GALFIT for half of the sample resulted in colors that are more than 3σ redder than the red sequence fitted for early-type galaxies, and hence these colors are unreliable. This happens especially for later-type spirals where the disk contribution is stronger. These anomalous bulge colors might be connected to a change in the disk properties inside the bulge affecting the luminosities obtained from the bulge–disk decomposition. This will be discussed in a subsequent paper. These results motivated us to derive bulge colors using aperture photometry instead. We calculated i-band aperture magnitudes for the galaxies using the IRAF task ELLIPSE, with fits forcing the same center for all the isophotal apertures. We then calculated the aperture magnitudes in the g and r bands, fixing the ellipticity and position angle of each isophotal aperture equal to the i-band values. The bulge is usually defined as the excess inner light over that from the disk. Using our bulge/disk/bar decomposition, we calculate the radius where μ(bulge + disk[+ bar])–μ(disk[+bar]) = –0.1 and adopt the aperture at this radius for calculating the bulge color. This aperture is smaller than 5 pixels for 15% of the galaxies. In these cases, it was set to 5 pixels to ensure a minimum loss of bulge flux because of the seeing (width of the SDSS PSF at 1% level is ∼10 pixels).

In order to compute the rest-frame colors, the magnitudes were corrected for Galactic extinction and by k-correction (see Section 3.1). We derived new a k-correction for the bulge using the aperture magnitudes (the global k-correction results in a difference in the (g − i)_b color lower than 0.02 for 97% of the bulges). Colors were corrected for the dependence of internal extinction with inclination following Masters et al. (2010, Equation (3)), but only for the Sa–Sc galaxies (87%) since no dependence in the color with inclination was found in Fernández Lorenzo et al. (2012) for later types. The same correction has been applied to the bulge and disk. The structural parameters and colors calculated in this work are listed in Table 1.

In Figure 2, we present the (g − i) color obtained for bulges and disks versus the total i-band magnitude. We present as a comparison sample the density diagram obtained from the Nair & Abraham (2010) sample at 0.01 < z < 0.05 (DR8) (Fernández Lorenzo et al. 2012). We find 63% (119/189) of the bulges within 2σ of the red sequence fitted to the E/S0 galaxies of the Nair & Abraham (2010) sample. Splitting by types, we find 64% (113/177) of pseudobulges to be as red as early-type galaxies, with 66% of low-B/T (84/127) and 58% (29/50) of high-B/T pseudobulges located in the red sequence. According to the Kolmogorov–Smirnov (K-S) test, the distributions of colors for low-B/T and high-B/T pseudobulges are statistically equal (probability p(K-S) = 0.88), with a median (g − i) value of 1.06  ± 0.11 and 1.07 ± 0.13, respectively. However, the distributions of colors for their associated disks are statistically different (p(K-S) = 0.003), probably because high-B/T pseudobulges are located in earlier spiral types. The mean difference between bulge and disk colors is larger for low-B/T than for high-B/T pseudobulges, which likely reflects a redder disk (median (g − i)_d = 0.82 ± 0.14 versus 0.75 ± 0.10). We found no difference between the colors of classical bulges and pseudobulges (p(K-S) = 0.90). This result contrasts with Gadotti (2009), who concludes that pseudobulges were 0.2 mag bluer than classical bulges.

Zoom In Zoom Out Reset image size

The colors above are corrected only for differential reddening due to inclination. To check if a full reddening correction could affect the results we used the extinction A_v derived from the starlight fit to the stellar continuous (Cid Fernandes et al. 2005, 2009) for those galaxies with SDSS spectra. This is possible for 108/189 of our galaxies and 8811/8879 of the Nair & Abraham (2010) sample. We used the Calzetti et al. (2000) law and R_v = 3.1 to calculate the reddening in the (g − i) color. In Figure 3, we reproduce Figure 2 with colors corrected for internal reddening. The scatter in the red sequence fitted to the E/S0 galaxies from the Nair & Abraham (2010) sample increases by a factor of two when reddening is taken into account, while the red sequence and the blue cloud are better separated. Sixty percent of our pseudobulges are still located in the red sequence although an important fraction of low-B/T pseudobulges seem to populate the region between the red sequence and the blue cloud, where early-type spirals and lenticulars are located.

Zoom In Zoom Out Reset image size

The bulge–disk color differences found here suggest an early formation for most bulges and a different SF history for both components. Pseudobulges in the red sequence seem to follow the same color–magnitude relation as early-type galaxies, i.e., they are redder for brighter host galaxies. Also, the bluer bulges are located in the faintest spiral galaxies. These results could be indicative of a "downsizing" scenario where the most massive galaxies (and their bulges) are the first to be assembled (Cimatti et al. 2006).