AGN feedback models: correlations with star formation and observational implications of time evolution

MNRAS 443, 1125–1141 (2014) doi:10.1093/mnras/stu1180 AGN feedback models: correlations with star formation and observational implications of time evolution Robert J. Thacker,1‹ C. MacMackin,1 James Wurster1 † and Alexander Hobbs2 1 Department 2 Institute of Astronomy & Physics, Saint Mary’s University, Halifax, NS B3L 3C3, Canada for Astronomy, ETH Zurich, Zurich 8093, Switzerland Accepted 2014 June 13. Received 2014 June 6; in original form 2014 March 5 We examine the correlation between the star formation rate (SFR) and black hole accretion rate (BHAR) across a suite of different active galactic nuclei (AGN) feedback models, using the time evolution of a merger simulation. By considering three different stages of evolution, and a distinction between the nuclear and outer regions of star formation, we consider 63 different cases. Despite many of the feedback models fitting the M–σ relationship well, there are often distinct differences in the SFR–BHAR correlations, with close to linear trends only being present after the merger. Some of the models also show evolution in the SFR–BHAR parameter space that is at times directly across the long-term averaged SFR–BHAR correlation. This suggests that the observational SFR–BHAR correlation found for ensembles of galaxies is an approximate statistical trend, as suggested by Hickox et al. Decomposing the SFR into nuclear and outer components also highlights notable differences between models and there is only modest agreement with observational studies examining this in Seyfert galaxies. For the fraction of the black hole mass growth from the merger event relative to the final black hole mass, we find as much as a factor of 3 variation among models. This also translates into a similar variation in the post-starburst black hole mass growth. Overall, we find that while qualitative features are often similar amongst models, precise quantitative analysis shows there can be quite distinct differences. Key words: hydrodynamics – quasars: supermassive black holes. 1 I N T RO D U C T I O N A growing body of observational evidence (for a recent review see Alexander & Hickox 2012 and references therein) suggests that the growth of supermassive black holes (SMBH) is intrinsically linked to properties of the host galaxy. Yet these relationships, be they correlations of bulge properties such as mass, luminosity or velocity dispersion relative to the black hole mass (e.g. Magorrian et al. 1998; Gebhardt et al. 2000; Ferrarese & Merritt 2000; Gültekin et al. 2009; Kormendy & Ho 2013) or the similarity of cosmological star formation rate (SFR) and black hole accretion rate (BHAR) histories (e.g. Boyle & Terlevich 1998; Silverman et al. 2008; Aird et al. 2010), are subtle and not easily understood. While at a general level numerous mechanisms are known for fuelling black hole mass growth, such as galaxy mergers (e.g. Sanders et al. 1988; Springel, Di Matteo & Hernquist 2005, hereafter SDH05; Hopkins et al. 2006), determining precise predictions for theoretical models remains a challenge because of the inherent difficulty in understand-  E-mail: † Present address: Monash Centre for Astrophysics, Monash University, Victoria 3800, Australia. ing both accretion down to the SMBH scale (e.g. Shlosman, Frank & Begelman 1989; Hopkins & Quataert 2010) and the accompanying energy release ubiquitously known as active galactic nuclei (AGN) ‘feedback’ (e.g. Silk & Rees 1998; King 2003; Proga & Kallman 2004; Ostriker et al. 2010; SDH05). While, as noted, there appears to be a strong correlation between the cosmological histories of SFRs and BHARs, on an individual object basis the correlation is less clear. Some observations have found positive correlations between SFRs and BHARs (e.g. Lutz et al. 2008; Serjeant et al. 2010; Bonfield et al. 2011), while others have found flat or negative correlations (Harrison et al. 2012; Page et al. 2012). However, AGN have a much shorter variability timescale than global star formation (e.g. Hopkins & Hernquist 2009), meaning that any anticipated correlations may only become clear when averages over populations, which will capture the rapidly accreting objects, are considered. Results presented in Chen et al. (2013) for star-forming galaxies appear to provide support for this assertion. For simulation work it is possible to average over outputs taken at different times thereby averaging over different evolutionary phases. As well as considering global star formation in galaxies, observations have also focused on whether correlations are stronger with nuclear (roughly the sub-kpc scale) or extended star formation  C 2014 The Authors Published by Oxford University Press on behalf of the Royal Astronomical Society ABSTRACT 1126 R. J. Thacker et al. (i) Measure the intrinsic time variation of a single merger event, and thus quantify time variation in the SFR–BHAR parameter space. MNRAS 443, 1125–1141 (2014) While not equivalent to ensemble averaging, it quantifies the evolutionary variation of a single AGN formation event (see Section 2 for a discussion). (ii) Calculate the SFR–BHAR correlations for this merger, considering evolution across all the simulation, and both pre- and postmerger cases. Contrast the different models, including those with explicit accretion time-scales such as the Power et al. (2011) model, to observed correlations to see what can be inferred. (iii) Evaluate the SFR–BHAR correlations for both nuclear and extended star formation regions to see if observational expectations are matched. Because AGN accretion and nuclear star formation are both fed by cold gas in the nuclear region stronger correlations between nuclear star formation and the BHAR are expected. (iv) Extend our model framework to include the Hobbs et al. (2012) model in the merger context. The layout of this paper is as follows. In Section 2 we discuss our handling of evolutionary tracks in the SFR–BHAR parameter space. In Section 3 we review the numerical methodology and simulations, and follow this with a detailed analysis in Section 4. In Section 5 we conclude with a brief review. 2 G A L A X Y E VO L U T I O N I N T H E S F R – B H A R PA R A M E T E R S PAC E As galaxies evolve their instantaneous SFR and BHAR chart an evolutionary track in the SFR–BHAR parameter space. While observationally it is only possible to reconstruct these tracks in an averaged sense (Wild, Heckman & Charlot 2010), for simulations the evolution can be plotted exactly. What can we learn from this analysis? The key issue is generating an underlying qualitative understanding of the SFR–BHAR correlation. If galaxies track along this correlation, in both an exact or time-averaged sense tightly, then the relationship is strongly suggestive of an evolutionary explanation. If, on the other hand, the evolutionary tracks run diametrically opposite the correlation then a large intrinsic scatter is to be expected. To put the simulation results in c (...truncated)