It seems clear that some bulges have central disks (Peletier et al. 2007), often (but not always) with young stars, which is usually linked to disk gas inflow and central star formation caused by internal secular processes related with the presence of a bar (Friedli & Benz 1995, Norman et al. 1996, Noguchi 2000, Immeli et al. 2004). However, this central rotationally younger component does not necessarily form due to internal processes. Major and minor mergers and external accretion of gas may result in the formation of a disky bulge (e.g. Guedes et al. 2013, Querejeta et al. 2015). This idea may be supported by some observational studies. For example, Kannappan et al. (2004) found, in a sample of disky bulges selected to be blue and, therefore, with central young stars, that all of them showed signs of recent interactions.
Hydrodynamical cosmological simulations predict that both secular and external processes contribute to create disky pseudobulges with similar characteristics in both cases, rotationally supported and with a young and metal poor stellar population (Obreja et al. 2013, Guedes et al. 2013). Eliche-Moral et al. (2011) also analyze the effects of minor mergers on the inner part of disk galaxies, finding this process to be efficient in forming rotationally supported stellar inner components, i.e., disks, rings or spiral patterns (see also Domínguez-Tenreiro et al. 1998, Aguerri et al. 2001, Scannapieco et al. 2010).
A way to quantify the importance of secularly formed disky bulges is to compare properties of galaxies with and without bars. While it has been found that Hα emission is enhanced in early-type spirals with bars with respect to those early-type non-barred (e.g. Ho et al. 1997, Huang et al. 1996, Alonso-Herrero & Knapen 2001, Jogee et al. 2005, Ellison et al. 2011), the evidence supporting the bulge building by bars from the ages of its stars has proven to be elusive. Several authors have tried the comparison using samples of face-on galaxies, where it is easy to morphologically identify the bar. The differences, however, have not been firmly established. Moorthy & Holtzman (2006) and Pérez & Sánchez-Blázquez (2011) found hints of lower ages and higher metallicities in barred galaxies compared to their counterparts in unbarred at a given σ 8. They also found higher [α / Fe] abundances in barred galaxies with central velocity dispersion 2.2 > logσ (km/s) > 2.35, but the opposite for 2 > logσ (km/s) > 2.2. At fixed σ and Vmax, barred galaxies appear to have larger central values of [MgFe]′ 9 (which can be used as an indicator of metallicity independent of [α / Fe], see above) than non-barred galaxies (or galaxies with elliptical shape bulges) of the same σ or Vmax. The differences, however, were not very significant in a statistical sense.
On the other hand, Jablonka et al. (2007) found no difference between the stellar population properties of edge-on barred and non-barred galaxies. However, it may be difficult to detect a bar in an edge-on galaxy. de Lorenzo-Cáceres et al. (2012) and de Lorenzo-Cáceres et al. (2013) analyzed the stellar populations in the center of double-barred early-type S0s and spirals finding some signs of gaseous flows and young stellar populations. This population was not very prominent though. Nevertheless, all the above studies were affected by poor number statistics.
Coelho & Gadotti (2011) observed 575 face-on bulges in disk galaxies, of which 251 contain bars. They found that, for bulges with masses between 1010.1M⊙ and 1010.85M⊙, the distribution of ages in barred galaxies is bimodal with peaks at 4.7 and 10.4 Gyr. This bimodality is not seen in non-barred galaxies of a similar bulge mass range. The age distribution of barred and non-barred galaxies is, per contra, similar for bulges of masses lower than 1010.1M⊙ (i.e., the differences are only seen in massive bulges). These authors did not find any difference in the metallicity distribution of barred and non-barred galaxies. These results are summarized in Fig. 5 where the distribution of ages for several mass intervals is shown for samples of barred and non-barred galaxies.
Figure 5. Normalized distributions for bulge ages, for several mass intervals, as indicated. Distributions for barred and non-barred galaxies are shown in red and black lines, respectively. (Figure taken from Coelho & Gadotti 2011).
Therefore, it is still not clear if there are differences between the stellar populations of galaxies with and without bars. What seems clear though is that, if there are differences, they are only visible in massive, early-type galaxies, a result that is supported by other studies analyzing the molecular gas concentration (Sakamoto et al. 1999). This is often attributed to the fact that early-type galaxies host larger and stronger bars than late-type galaxies (e.g. Buta & Combes 1996), although, as pointed out in Laurikainen et al. (2007) and Laurikainen et al. (2004), a longer bar does not implies a stronger bar and, in fact, in early-type galaxies the bar induced tangential forces are weaker because they are diluted by the more massive bulges 10.
This does not mean, necessarily, that bars are not efficient agents in building up bulges. It is still not clear if bars are long-lasting structures or not. If bars are not long-lasting structures but recurrent patterns (Bournaud & Combes 2002) then the fact that we do not find differences between barred and non-barred galaxies would not necessarily imply that bars are not important for secular evolution but, simply, that non-barred galaxies could have been barred in the recent past. However, most numerical simulations show that, once formed, bars are robust structures (Shen & Sellwood 2004, Athanassoula et al. 2005, Debattista et al. 2006, Berentzen et al. 2007, Villa-Vargas et al. 2010, Kraljic et al. 2012, Athanassoula et al. 2013). Furthermore, at least in massive disk galaxies, bars have the same stellar population properties of bulges (old, metal rich, and [α / Fe]-enhanced stellar populations; Sánchez-Blázquez et al. 2011; Pérez et al. 2009) which, in many cases, are very different from that of the disk (see Sect. 3). This result also supports (although it does not prove, see Sect. 3 11) the idea that bars formed long ago. The longevity of bars is also suggested in studies of the bar fraction evolution (e.g. Sheth & et al. 2008), which find a similar bar fraction at z∼0.8 to that seen at the present-day for galaxies with stellar masses M* ≥ 1011 M⊙. In addition, non-axisymmetric structures, such as nuclear spirals, can drive gaseous inflows (e.g. Kormendy & Fisher 2005), which could dilute the differences between barred and non-barred galaxies.
3.1. Stellar population of bars
As the debate of the durability of bars is still open and its influence may be crucial for the formation of bulges, it is also important to study the stellar populations hosted by bars. Very few works, however, have dealt with this problem.
Gadotti & de Souza (2006) obtained the color and color gradients in the bar region of a sample of 18 barred galaxies. They interpreted the color differences as differences in stellar ages and conclude that younger bars were hosted by galaxies of later types (see also Gadotti 2008). However, as we mentioned in Sect. 2, the effects of age-metallicity degeneracy and dust extinction are strongly degenerate in colors and, therefore, conclusions based on only colors remain uncertain. Pérez et al. (2007) and Pérez et al. (2009) performed an analysis of the stellar population of bars in early-type galaxies using line-strength indices. They found that the mean bar values of SSP-equivalent age, metallicity, and [α / Fe], correlate with central σ in a similar way to that of bulges, pointing to an intimate evolution of both components. Galaxies with high central σ (> 170 kms−1) host bars with old stars while galaxies with lower central velocity dispersion show stars with a large dispersion in their ages.
These authors also analyzed the stellar population gradients along the bars and found three different behaviors: (1) bars with negative metallicity gradients. These bars have young/intermediate stellar populations (SSP-equivalent values < 2 Gyr) and have amongst the lowest stellar velocity dispersions of the sample; (2) bars with no metallicity gradients. These galaxies have, however, positive age gradients and (3) bars with a mean old stellar population and positive metallicity gradients (more metal-rich at the bar ends).
The fact that bars are composed of old stellar populations does not mean that they formed long ago, as the bar might have formed recently out of old stars in the disk. One way to disentangle these two options is to compare the stellar populations of the disk and the bar at the same distances. In Sánchez-Blázquez et al. (2011), this comparison is made for two galaxies, finding that stars in the bar are older and more metal rich than those of the disk. Furthermore, the gradient in both parameters is much flatter in the bar. In general, they found that the stellar content of the bar is more similar to that of the bulge than to the disk. However, the sample of this study remains small and biased towards early-type bulges. Clearly, a larger study sampling larger samples of galaxies covering all morphological types is still needed.
8 Although the differences found by Moorthy & Holtzman (2006) in the Hβ-σ relation between barred and non-barred galaxies disappear when Vmax is used instead of σ. Back.
9 This index is defined as [MgFe]′ ≡ √Mgb × (0.72 Fe5270 + 0.28 Fe5335 in Thomas et al. (2003). Back.
10 Although the majority of authors does not distinguish different types of S0, Laurikainen et al. (2007) also show that early-type S0s have shorter bars than later type S0s, i.e., the trend of longer bars for early-type morphologies reverse in this morphological subclass. Back.
11 The stars in the bar can form in the disk long time ago, even if the bar have been recently formed. Back.