Chapter 1. Introduction 1.1 Feedback and galaxy formation The role of energy feedback in galaxy evolution is a central problem in extra-galactic as- tronomy today. In the currently-favoured Cold Dark Matter (CDM) cosmology, galaxies form through hierarchical merging of dark matter halos (White & Rees, 1978). Bary- onic matter cools and condenses within these structures to form the gaseous and stellar components of the galaxies. Although the results of CDM cosmological simulations can match observations of large-scale structure in the universe remarkably well, until recently, the distributions of galaxy masses and morphologies were not as closely described (e.g., Cole et al., 2000; Benson et al., 2003; Croton et al., 2006). For example, semi-analytical merger simulations struggled to reproduce the observationally-derived time-scales and ages of formation for early-type galaxies, predicting an excess of the most luminous and massive systems. It was recognised that powerful sources of energy must act in galaxies to reheat the cold gas that fuels star formation. This heating prevents star formation from continuing at late times in early-type galaxies and slows or regulates growth. Fig- ure 1.1 shows an example of how including heat sources in galaxy formation models improves the match between the high end of the predicted and observed luminosity functions (e.g., Benson et al., 2003). Galaxy feedback refers to the processes that return or recycle energy and mass be- tween a galaxy and its environment, and potentially affect the growth of the galaxy itself. Such feedback can provide energy to suppress the cooling of accreting gas or to remove material from the inner regions of the galaxy altogether. This limits the material available for star formation and the buildup of galaxy stellar mass. These effects can provide both an explanation for the deficit of massive galaxies that is observed relative to the predictions of hierarchical merging, and the means to better reproduce the early and rapid formation of the stellar mass in early-type galaxies. Supernovae and starbursts are an important channel for feedback in galaxies (e.g., White & Rees, 1978; White & Frenk, 1991), and are a critical component of galaxy formation models (e.g., see Cole et al., 2000; Benson et al., 2003). They enrich the in- tergalactic medium with metals and produce ultraviolet radiation that heats and ionises the local environment. Starburst winds transport mass and energy, and entrain and uplift material from galaxies that might otherwise form stars (e.g., Chevalier & Clegg, 1985; Efstathiou, 2000). However, the predominant source of feedback and heating that 1 The many lives of AGN 23 V vir new fuel for star formation must come from cooling flows which 0.4). Our morphological resolution limit is marked by the dashed are affected by ‘radio mode’ heating. line at a stellar mass of 4 109 M ; this corresponds approxi- The effect of ‘radio mode’ feedback is clearly substantial. Sup- mately to a halo of 100 particles∼ × in the$ Millennium Run. Recall that pression of condensation becomes increasingly effective with in- the morphology of a galaxy depends both on its past merging history creasing virial temperature and decreasing redshift. The effects are and on the stability of its stellar disc in our model. Both mergers 1 6 large for haloes with V vir 150 km s− (T vir 10 K) at z 3. Con- and disc instabilities contribute stars to the spheroid, as described densation stops almost completely between z 1 and the present in Section 3.7. The build-up of haloes containing fewer than 100 1 =6 in haloes with V vir > 300 km s− (T vir > 3 10 K). Such systems particles is not followed in enough detail to model these processes correspond to the haloes of groups and clusters× which are typically robustly. observed to host massive elliptical or cD galaxies at their centres. A number of important features can be seen in Fig. 9. Of note Our scheme thus produces results which are qualitatively similar is the bimodal distribution in galaxy colours, with a well-defined to the ad hoc suppression of cooling flows assumed in previous red sequence of appropriate slope separated cleanly from a broader models of galaxy formation. For example, Kauffmann et al. (1999) ‘blue cloud’. It is significant that the red sequence is composed 1 switched off gas condensation in all haloes with V vir > 350 km s− , predominantly of early-type galaxies, while the blue cloud is com- while Hatton et al. (2003) stopped condensation when the bulge posed mostly of disc-dominated systems. This aspect of our model mass exceeded a critical threshold. suggests that that the physical processes that determine morphol- ogy (i.e. merging, disc instability) are closely related to those that control star formation history (i.e. gas supply) and thus determine 4.2 Galaxy properties with and without AGN heating galaxy colour. The red and blue sequences both display a strong The suppression of cooling flows in our model has a dramatic effect metallicity gradient from low to high mass (c.f. Fig. 6), and it is this on the bright end of the galaxy luminosity function. In Fig. 8 we which induces a ‘slope’ in the colour–magnitude relations which present K- and bJ-band luminosity functions (left- and right-hand agrees well with observation (e.g. Baldry et al. 2004). panels respectively) with and without ‘radio mode’ feedback (solid By comparing the upper and lower panels in Fig. 9 we can see how and dashed lines respectively). The luminosities of bright galaxies ‘radio mode’ feedback modifies the luminosities, colours and mor- are reduced by up to two magnitudes when the feedback is switched phologies of high-mass galaxies. Not surprisingly, the brightest and on, and this induces a relatively sharp break in the luminosity func- most massive galaxies are also the reddest and are ellipticals when tion which matches the observations well. We demonstrate this by cooling flows are suppressed, whereas they are brighter, more mas- overplotting K-band data from Cole et al. (2001) and Huang et al. sive, much bluer and typically have discs if cooling flows continue (2003) in the left-hand panel, and bJ-band data from Norberg et al. to supply new material for star formation. AGN heating cuts off the (2002) in the right-hand panel. In both bandpasses the model is quite gas supply to the disc from the surrounding hot halo, truncating star close to the data over the full observed range. We comment on some formation and allowing the existing stellar population to redden. of the remaining discrepancies below. However, these massive red galaxies do continue to grow through Our feedback model also has a significant effect on bright galaxy merging. This mechanism allows the dominant cluster galaxies to colours, as we show in Fig. 9. Here we plot the B V colour dis- gain a factor of 2 or 3 in mass without significant star forma- tribution as a function of stellar mass, with and without− the central tion, in apparent agreement with observation (Aragon-Salamanca, Baugh & Kauffmann 1998). This late-stage (i.e. 1) hierarchi- heating source2 (top and bottom panels respectively). In both panels 1. Introductionz we have colour-coded the galaxy population by morphology as es- cal growth moves objects to higher mass without changing their timated from bulge-to-total luminosity ratio (split at L /L colours. bulge total = Figure 8. FigureGalaxy luminosity 1.1 functionsThis figure, in the K (left) from andCrotonbJ (right) photometric et al. (2006 bands,), plotted compares with and without observed ‘radio (blue mode’ feedback points) (solid galaxy and long-dashed lines respectivelyluminosity – see Section functions 3.4). Symbols (the indicate volume observational number results density as listed of in galaxies each panel. vs.As can luminosity) be seen, the inclusion in the of AGNK and heatingbJ produces a good fit tophotometric the data in both colours. bands Without with this those heating predicted source our model by overpredicts galaxy formation the luminosities models of massive that galaxies omit by about (dashed two magnitudes black andfails to reproduce the sharp bright-end cut-offs in the observed luminosity functions. lines) and include (solid black lines) parameterisations of AGN feedback. Without AGN heating these models over-predict the population of massive galaxies by up to two orders of magnitude. C C % 2005 The Authors. Journal compilation % 2005 RAS, MNRAS 365, 11–28 shapes massive galaxies is generally ascribed to active galactic nuclei (AGN), which form when super-massive black holes (SMBHs) in the cores of galaxies actively accrete mass. Several lines of observational evidence indicate that AGN participate in self-regulating cycles of mass and energy feedback in galaxies. For example, it has been found that SMBH masses are tightly correlated with the large-scale properties of the host galaxy, including the host galaxy luminosity (Kormendy & Richstone, 1995), stellar velocity dispersion (Ferrarese & Merritt, 2000; Gebhardt et al., 2000), and bulge mass (H¨aring & Rix, 2004). These relationships suggest that the growth of the SMBH is coupled to that of the host in some, still uncertain, way (e.g., Kauffmann & Haehnelt, 2000). SMBHs grow by accreting matter from their environment and in the process can output power 46 1 exceeding 10 erg s− (Osterbrock & Ferland, 2006). They also generate relativistic bipolar jets that drive expanding lobes of radio-emitting plasma containing energies of up to 1060 1061 erg (Willott et al., 1999). These can extend over several megaparsecs − from the central galaxy (e.g., Machalski et al., 2008; Palma et al., 2000), and certainly across much larger distances than the sphere of influence of the black hole.