Research > AGN growth and star formation connection
Some of my results on the growth of AGN and their on star formation
My Related Popular/Press Articles:
ESA Article, Black holes and star formation: a Herschel perspective, also on Herschel website, (2013). [Based on Harrison et al. 2012b]
RAS Press Release, Hidden supermassive black holes brought to life by galaxies on collision course (2023) [Based on Dougherty, CMH et al. 2023]
Obscured AGN enhancement in galaxy pairs at cosmic noon: evidence from a probabilistic treatment of photometric redshifts
Figure from Dougherty et al. 2023 - MNRAS, arXiv:2305.05692, ADS
We find that the most obscured AGN (are enhanced, relative to control galaxies, when located in physically associated close galaxy pairs.
We combined spectroscopic surveys with photometric redshift probability distribution functions for galaxies in the CANDELS and COSMOS surveys, to produce the largest ever sample of galaxy pairs used in an AGN fraction calculation for cosmic noon. Over projected separations 5-100 kpc we find no evidence for enhancement, relative to isolated control galaxies, of X-ray or infrared-selected AGN in major (mass ratios up to 4:1) or minor (4:1 to 10:1) galaxy pairs. However, defining the most obscured AGN as those detected in the infrared but not in X-rays, we observe a trend of increasing obscured AGN enhancement at decreasing separations.
The figure shows the excess (or ‘enhancement’) of obscured active galactic nuclei (growing super massive black hole) found in galaxy pairs, compared to isolated galaxies, as a function of separation to another galaxy. The results show that there is a boost in the number by a factor of ~2 for the closest galaxy pairs.
Cosmological simulations predict that AGN preferentially live in gas-rich, star-forming galaxies despite effective feedback
Figure from Ward et al. 2022 - MNRAS, 514, 2936, arXiv:2204/13712, ADS
We used three major hydrodynamic simulations (EAGLE, IllustrisTNG and SIMBA) to reproduce observational experiments carried out by observers, relating AGN to the molecular gas fractions and star formation rates of their host galaxies. In particular, we were testing if the observation that AGN live in gas rich star-forming galaxies is in contradiction with AGN feedback models (as claimed by some observational studies). We found that AGN typically reside in gas rich, star-forming galaxies in all the simulations, despite effective negative feedback where AGN ultimately shut down star formation in the same simulations. This can be explained by the impact occurring on longer timescales than the visibility of a radiatively luminous AGN. The simulations are in qualitatively good agreement with the observations, but show some quantitative differences due to the different feedback prescriptions.
The figure shows at example presentation of the simulation data used in the work (IllustrisTNG in this case). The data is shown in a molecular gas fraction versus stellar mass plane, with number density contours, and colour-coded by the mean specific star formation rate. The top panel shows rapidly accreting active galactic nuclei (AGN) and the bottom panel non-AGN. AGN are dominated in gas rich, star-forming galaxies. The panels on the right show four example galaxy visualisations from the simulation (coloured by star formation rate).
The impact of AGN feedback on star formation
Figure from Scholtz et al. 2018 - MNRAS, 475, 1288, arXiv:1712.02708, ADS
We used two runs of the EAGLE simulations (with and without AGN feedback) to show that the impact of the AGN feedback is to lower the mode and increase the width of the specific star formation rate distributions in the simulation. The model with AGN feedback agrees well with the trends observed in our data that were obtained from deep ALMA observations of ~100 X-ray detected AGN (also see Stanley et al. 2018). Unlike with stellar mass, no strong trends between star formation distributions and AGN luminosity are observed. We attribute this to the fact that the impact of AGN on the galaxy wide star formation occurs on long timescales and the AGN themselves are rapidly varying. AGN feedback has a more impact on more massive galaxies in the simulation.
The figure shows the mode and width (upper/lower panels, respectively) of the specific star formation rate distributions (i.e., star formation rate / stellar mass) as a function of stellar mass for: (a) the reference EAGLE simulations (with AGN feedback; blue curves); (b) EAGLE without AGN (red curve); (c) out observational data (squares).
Star formation rates of AGN host galaxies
Figure from C.M. Harrison 2017 - Nature Astronomy Review, NatAs, 1E, 165 (arXiv:1703.06889); adapted from original in McAlpine, Bower, Harrison et al. 2017, MNRAS, 478, 3395 (arXiv:1701.01122)
The figure shows mean star formation rate versus instantaneous black hole accretion rate for the EAGLE cosmological model (McAlpine et al. 2017) and versus AGN luminosity (converted from X-ray luminosities) for observations (Stanley et al. 2015). We took a large sample (~2000 sources) of z>0.2 X-ray AGN host galaxies and using Herschel and Spitzer data performed SED fitting to decouple AGN and star formation emission, carefully taking into account both upper limits and detections (Stanley, Harrison et al. 2015). We found an almost flat relationship between mean star formation rate and AGN luminosity; the slight increase in mean star-formation rate with increasing accretion rate is attributed to the increasing average stellar masses. Furthermore, by taking into account detection limits we reproduced these observations using the reference model of the EAGLE simulation (McAlpine et al. 2017). The dotted lines are a linear fit to the running means for the model (solid curves). The logarithm of stellar masses (in stellar mass units) of the first and last bin are labelled. Despite effective star-formation suppression by AGN in the model, this does not result in reduced average star-formation rates for the highest instantaneous black hole accretion rates (i.e., AGN luminosities).
Figure from C.M. Harrison 2017 - Nature Astronomy Review, NatAs, 1E, 165 (arXiv:1703.06889)
The figure shows mean star formation rate versus stellar mass for observed star-forming galaxies (Schreiber et al. 2015) (a) and galaxies in the EAGLE cosmological model run both with and without AGN (Schaye et al. 2015; Crain et al. 2015) (b). More massive galaxies form stars more rapidly; however, the highest-mass galaxies (log[M_stellar]>10 solar masses) are observed to fall below a constant scaling relationship implying a reduction in the efficiency for the available baryons to be converted into stars. In the model, the impact of AGN is to reduce star formation rates of high mass galaxies as well as to reduce the overall number of massive galaxies. It appears that the observational signature of AGN suppressing star formation may be imprinted on the reduced average star formation rates per unit stellar mass for the most massive galaxies (this figure) and not on reduced average star formation rates for the most instantaneously luminous AGN (see above).
ALMA resolves star formation in distant AGN galaxies
Figure from Harrison et al. 2016a, MNRAS, 457L, 122, (arXiv:1510.06740).
Using ALMA we measured the size of the sub-mm emission in 5 X-ray AGN and 1 star forming galaxy. We found that in all cases the emission was extended in our high resolution maps. This can be seen from the fact that the amplitudes drop as a function of uv distance in the ALMA data (see fifth column of figure). Furthermore an extended structure provides a better fit to the images than a point source (see residuals in the third and fourth columns). We obtained sizes of ~1-3 kpc. After demonstrating that the emission traces star formation (using SED analyses), we then used these measurements to infer the star formation rate surface density of these objects. We found that these sizes and star formation rate surface densities were in excellent agreement with non-AGN host galaxies with matched far infrared luminosities and redshifts. Although based on a small sample, this work implies that the star formation rate surface densities are insensitive to the presence of an AGN.
A remarkably flat relationship between the average star formation rate and AGN luminosity for distant X-ray AGN
Figure from Stanley, Harrison et al. 2015: MNRAS, 453, 591, (arXiv:1502.07756)
The figure shows mean infrared luminosity due to star formation (star formation rate) versus X-ray luminosity (AGN luminosity) for ~2000 AGN host galaxies. Using Herschel and Spitzer data, we carefully isolated the infrared emission due to star formation in AGN host galaxies. Taking into account of upper limits, we found that there is almost no dependancy of the (average) star formation rate on the AGN power. This confirmed results from my earlier work, but using a larger sample and more sophisticated methods.
No strong sub-mm observational evidence for the suppression of star formation by luminous AGN
The original work and figure in ApJL, 760:L15 (2012), (arXiv:1209.3016).
The figure shown here, along with a review on the impact of AGN can be found in IAUS 2014, 304, 284, (arXiv:1312.3609).
Popular article on Herschel website, Black holes and star formation: a Herschel perspective and also in an ESA article, (2013).
Many studies have made use of far-infra-red photometric data from Herschel (an excellent indicator of obscured star formation) to investigate AGN host galaxies. However, contradictory results had been found, with all three possibilities being found for the star formation rates (SFRs) of luminous AGN: (1) they are consistent with lower luminosity AGN; (2) they are suppressed compared to lower luminosity AGN; (3) they are enhanced compared to lower luminosity AGN. In this study we stacked 250um data from Herschel for X-ray detected AGN in three fields; the small Chandra Deep Field North (CDF-N) and Chandra Deep Field-S (CDF-S) as well as the much larger COSMOS field. In the figure we show that the previous contradictory results were likely to be due to field-to-field variations: the "enhanced" results were performed in CDF-S; the "suppressed" results were in CDF-N. However, the COSMOS field contains an order of magnitude more luminous sources and demonstrates that there is not apparent variation in the SFR of the most luminous AGN. Furthermore the SFRs appear to be consistent with non-AGN star-forming galaxies at the same redshift (shaded region in the figure)