A Chandra View of AGN and Supernova Feedback in Galaxy Groups
Jesper Rasmussen, Carnegie Observatories
+ Trevor Ponman, Ria Johnson, Alexis Finoguenov
Black Holes, Big and Small
Impact on Galaxy Formation
Luis C. Ho (何子山)
Kavli Institute for Astronomy and Astrophysics (KIAA) Peking University
Wednesday, February 26, 14
Wednesday, February 26, 14
Wednesday, February 26, 14
Wednesday, February 26, 14
Wednesday, February 26, 14
Wednesday, February 26, 14
Wednesday, February 26, 14
Cosmic backgrounds at different wavelengths
CMB
Optical = stars
X-ray Background = AGN
!"! #
2 keV soft hard Infrared=
Stars
(+AGN?)
Wednesday, February 26, 14
Cosmic backgrounds at different wavelengths
CMB
Optical = stars
X-ray Background = AGN
!"! #
2 keV soft hard Infrared=
Stars
(+AGN?)
Comastri, Gilli & Hasinger (2007)
Wednesday, February 26, 14
Vanden Berk et al. (2001)
SDSS quasars: composite spectrum
Wednesday, February 26, 14
Vestergaard (2004)
Barth et al. (2003)
Wednesday, February 26, 14
☯ M
•
= 3 x 109 M☉☯ Gas chemical enriched
with metals (C, Mg, Si, Fe)
☯ At z = 6.42, age of the Universe only 800 Myr !
Barth et al. (2003)
Richards et al. (2006)
quasar
space density
Wednesday, February 26, 14
Richards et al. (2006)
quasar
space density
dormant
massive BHs?
Wednesday, February 26, 14
27
Figure 9
(top) Schematic views of the almost-edge-on, warped maser disk of NGC 4258 (from Moran 2008) with warp parameters from Herrnstein et al. (2005) and including the inner contours of the radio jet.
The relative positions of the receding, near-systemic, and approaching masers are indicated by red, green, and blue spots, respectively. Differences in line-of-sight projection corrections to the slightly tilted maser velocities account for the departures in the high-|V | masers from exact Keplerian rotation. The near-systemic masers are seen tangent to the bottom of the maser disk bowl along the line of sight. They drift from right to left in ∼ 12 years across the green patch where amplification is sufficient for detection; this patch subtends ±4◦ as seen from the center (Moran 2008).
(bottom) NGC 4258 rotation curve V (r) versus radius in units of pc (bottom axis), Schwarzschild radii (top axis), and milliarcsec (extra axis). The black curve is a Keplerian fit to 4255 velocities of red- and blue-shifted masers (red and blue dots). The small green points and line show 10036 velocities of near-systemic masers and a linear fit to them. The green filled circle is the corresponding mean V (r) point (§ 3.3.2). The maser data are taken from Argon et al. (2007).
Miyoshi et al. (1995) Herrnstein et al. (2005)
Wednesday, February 26, 14
Bower et al. (1998)
speed of gas clouds
d is tanc e fr om t he c ent er ➛ ➛
➙
approaching receding
M84
Barth et al. (2001)
NGC 3245
Wednesday, February 26, 14
The “Nuker” Team
Magorrian et al. (1998)
Wednesday, February 26, 14
b lac k ho le m as s
Gebhardt et al. (2000); Ferrarese & Merritt (2000); Gültekin et al. (2009)
Standard “Paradigm”
☯ All bulges contain BHs
☯ M• ~ Mbulge 〈M• / Mbulge〉 ~ 0.1%− 0.2%
☯ M• ∝ "4
☯ M•−" relation tighter than M• − Mbulge relation
☯ No strong dependence on galaxy mass or type
☯ Mild to strong evolution with redshift
☯ AGN feedback engineers BH-host correlations
1.0
Wednesday, February 26, 14
Wednesday, February 26, 14
Wednesday, February 26, 14
Wednesday, February 26, 14
Crotton et al. (2006)
AGN radio heating
No radio heating
Wednesday, February 26, 14
Recent Developments
Ho (2008, ARA&A): Nuclear Activity in Nearby Galaxies
Kormendy & Ho (2013, ARA&A): Coevolution of Supermassive Black Holes and Galaxies
Wednesday, February 26, 14
Wednesday, February 26, 14
NGC 4889: M
•
= 2 x 1010 M⊙ (McConnell et al. 2011)Kuo et al. (2011)
7 New
Megamasers!
Wednesday, February 26, 14
Kuo et al. (2011)
7 New
Megamasers!
b lac k ho le
galaxy bulge
velocity dispersion mass
Kormendy & Ho (2013, ARA&A)
Wednesday, February 26, 14
M • – "
Relation
Wednesday, February 26, 14
54
If we use individual errors in MK,bulge (± 0.2) and log e and add individual errors in log M• to the intrinsic scatter in quadrature and iterate the intrinsic scatter until the reduced 2 = 1, then
log
✓ M• 109 M
◆
= (0.253±0.052) (0.484±0.036)(MK,bulge+24.21); intrinsic scatter = 0.31; (4)
log
✓ M• 109 M
◆
= (0.501 ±0.049)+(4.414±0.295) log
✓
200 km s 1
◆
; intrinsic scatter = 0.28. (5) The di↵erence between the two sets of fits is small. Taking account also of fits that use di↵erent combinations of points, we conclude that the intrinsic log M• scatter in M•–MK,bulge is 0.31±0.02, almost the same as the intrinsic scatter 0.29±0.03 in M•– e. This conclusion has also been reached by other authors who use infrared luminosities (e. g., Marconi & Hunt 2003; Sani et al. 2011).
Rewriting Equations 2 and 3 in physically more transparent forms, M•
109 M =
✓
0.542+0.0690.061
◆ ✓ LK,bulge 1011 LK
◆1.21±0.09
(6) M•
109 M =
✓
0.309+0.0370.033
◆ ✓
200 km s 1
◆4.38±0.29
(7) 6.6.1. The M• – Mbulge correlation and the ratio of BH mass to bulge mass
Galaxy formation work requires the mass equivalent of Equation 6, the M• – Mbulge correlation.
This is tricker to derive than it sounds. It is not just a matter of multiplying the bulge luminosity by a mass-to-light ratio that is provided automatically by the stellar dynamical models that give us M•. Bulge mass is inherently less well defined than bulge luminosity. Mass-to-light ratios of old stellar populations are uncertain; (1) the initial mass function (IMF) of star formation is poorly known; it may vary with radius in an individual galaxy or from galaxy to galaxy; (2) stellar population age and metallicity distributions a↵ect M/L and are famously difficult to disentangle; one consequence is that late stages of stellar evolution – especially asymptotic giant branch stars – a↵ect M/L but but are poorly constrained observationally (e. g., Portinari & Into 2011). Most important of all, (4) dark matter contributes di↵erently at di↵erent radii and probably di↵erently in di↵erent galaxies.
Graves & Faber (2010) provide an up-to-date discussion of these problems. They conclude that all of the above are important, with stellar population e↵ects (age and metallicity) accounting for
⇠ 1/4 of the variations in optical mass-to-light ratios and some combination of IMF and dark matter variations accounting for the rest. However, this field is unsettled; extreme points of view are that even K-band mass-to-light ratios vary by factors of ⇠ 4 from galaxy to galaxy and that all of this range is due to variations in IMF (Conroy & van Dokkum 2012) or contrariwise that IMFs vary little from one place to another (Bastian, Covey, & Meyer 2010).
These problems are background worries that may yet hold unpleasant surprises, but mostly, they are beyond the scope of this paper. The extensive work of the SAURON and ATLAS3D teams (Cappellari et al. 2006, 2013) shows that dynamically determined I- and r-band mass-to-light ratios are very well behaved. For 260 ATLAS3D galaxies, M/Lr / e0.69±0.04 with an intrinsic scatter of only 22 %. Since M/LK almost inevitably varies less from galaxy to galaxy than M/Lr, this suggests that we proceed by finding a way to estimate M/LK. In particular, we want an algorithm that does not involve the use of uncertain e↵ective radii re. Here’s why:
Published studies often derive Mbulge dynamically from re, e, and a virial-theorem-like relation Mbulge = k e2re/G, where k is, e. g., 3 (Marconi & Hunt 2003) or 5 (Cappellari et al. 2006, 2010) or 8 (Wolf et al. 2010). This situation is unsatisfactory; di↵erent assumptions about the density profile are one reason why k is uncertain. Also, re values are less well measured than we think.
54
If we use individual errors in MK,bulge (± 0.2) and log e and add individual errors in log M• to the intrinsic scatter in quadrature and iterate the intrinsic scatter until the reduced 2 = 1, then
log
✓ M• 109 M
◆
= (0.253±0.052) (0.484±0.036)(MK,bulge+24.21); intrinsic scatter = 0.31; (4)
log
✓ M• 109 M
◆
= (0.501 ±0.049)+(4.414±0.295) log
✓
200 km s 1
◆
; intrinsic scatter = 0.28. (5) The di↵erence between the two sets of fits is small. Taking account also of fits that use di↵erent combinations of points, we conclude that the intrinsic log M• scatter in M•–MK,bulge is 0.31±0.02, almost the same as the intrinsic scatter 0.29±0.03 in M•– e. This conclusion has also been reached by other authors who use infrared luminosities (e. g., Marconi & Hunt 2003; Sani et al. 2011).
Rewriting Equations 2 and 3 in physically more transparent forms, M•
109 M =
✓
0.542+0.0690.061
◆ ✓ LK,bulge 1011 LK
◆1.21±0.09
(6) M•
109 M =
✓
0.309+0.0370.033
◆ ✓
200 km s 1
◆4.38±0.29
(7) 6.6.1. The M• – Mbulge correlation and the ratio of BH mass to bulge mass
Galaxy formation work requires the mass equivalent of Equation 6, the M• – Mbulge correlation.
This is tricker to derive than it sounds. It is not just a matter of multiplying the bulge luminosity by a mass-to-light ratio that is provided automatically by the stellar dynamical models that give us M•. Bulge mass is inherently less well defined than bulge luminosity. Mass-to-light ratios of old stellar populations are uncertain; (1) the initial mass function (IMF) of star formation is poorly known; it may vary with radius in an individual galaxy or from galaxy to galaxy; (2) stellar population age and metallicity distributions a↵ect M/L and are famously difficult to disentangle; one consequence is that late stages of stellar evolution – especially asymptotic giant branch stars – a↵ect M/L but but are poorly constrained observationally (e. g., Portinari & Into 2011). Most important of all, (4) dark matter contributes di↵erently at di↵erent radii and probably di↵erently in di↵erent galaxies.
Graves & Faber (2010) provide an up-to-date discussion of these problems. They conclude that all of the above are important, with stellar population e↵ects (age and metallicity) accounting for
⇠ 1/4 of the variations in optical mass-to-light ratios and some combination of IMF and dark matter variations accounting for the rest. However, this field is unsettled; extreme points of view are that even K-band mass-to-light ratios vary by factors of ⇠ 4 from galaxy to galaxy and that all of this range is due to variations in IMF (Conroy & van Dokkum 2012) or contrariwise that IMFs vary little from one place to another (Bastian, Covey, & Meyer 2010).
These problems are background worries that may yet hold unpleasant surprises, but mostly, they are beyond the scope of this paper. The extensive work of the SAURON and ATLAS3D teams (Cappellari et al. 2006, 2013) shows that dynamically determined I- and r-band mass-to-light ratios are very well behaved. For 260 ATLAS3D galaxies, M/Lr / e0.69±0.04 with an intrinsic scatter of only 22 %. Since M/LK almost inevitably varies less from galaxy to galaxy than M/Lr, this suggests that we proceed by finding a way to estimate M/LK. In particular, we want an algorithm that does not involve the use of uncertain e↵ective radii re. Here’s why:
Published studies often derive Mbulge dynamically from re, e, and a virial-theorem-like relation Mbulge = k e2re/G, where k is, e. g., 3 (Marconi & Hunt 2003) or 5 (Cappellari et al. 2006, 2010)
M • – "
Relation
Wednesday, February 26, 14
M • – M bulge Relation
Wednesday, February 26, 14
M • – M bulge Relation
57
Presumably these galaxies contain larger contributions of dark matter that we choose not to include.
The remaining 22 galaxies satisfy log (M/LK) = 0.287 log σe − 0.637 with an RMS scatter of 0.088.
As expected, the relation is shallower than the one in r band (above). It has essentially the same scatter of ∼ 23 %. Dynamically, M/LK = 1 at σe = 166 km s−1, where the Into & Portinari (2013) calibration gives M/LK # 0.76. Cappellari et al. (2006) argue that the difference may be due to the inclusion of some dark matter in the dynamical models. We use the dynamical zeropoint.
To shift the Into & Portinari log M/LK values to the above, dynamical zeropoint, we first use their Table 3 relation log M/LK = 1.055(B − V )0 − 1.066 to predict an initial, uncorrected M/LK. This correlates tightly with σe: log M/LK = 0.239 log σe−0.649 with an RMS scatter of only 0.030.
We then apply the shift ∆ log M/LK = 0.1258 or a factor of 1.34 that makes the corrected Into &
Portinari mass-to-light ratio agree with the dynamic one, M/LK = 1.124, at σe = 250 km s−1. We then have two ways to predict M/LK that are independent except for the above shift,
log M/LK = 0.2871 log σe − 0.6375; RMS = 0.088; (8) log M/LK = 1.055(B − V )0 − 0.9402; RMS = 0.030, (9) where we use the RMS scatter of the correlation with σe to estimate errors for the latter equation.
We adopt the mean of the mass-to-light ratios given by Equations 8 and 9. For the error estimate, we use 0.5!0.0882 + 0.0302 + (half of the difference between the two log M/LK values)2. We use the resulting M/LK together with MK,bulge to determine bulge masses. For the log Mbulge error estimate, we add the above in quadrature to (0.2/2.5)2. The results are listed in Tables 2 and 3.
Figure 18 shows the correlation of M• with bulge mass Mbulge. A symmetric, least-squares fit to the classical bulges and ellipticals omitting the monsters and (for consistency with M• – σe), the emission-lime M• values for NGC 4459 and NGC 4596 plus NGC 3842 and NGC 4889 gives the mass equivalent of Equation 6,
M•
109 M" =
"
0.49+0.06−0.05
# "
Mbulge 1011 M"
#1.16±0.08
; intrinsic scatter = 0.29 dex. (10) Thus the canonical BH-to-bulge mass ratio is M•/Mbulge = 0.49+0.06−0.05 % at Mbulge = 1011 M".
This BH mass ratio at Mbulge = 1011 M" is 2–4 times larger than previous values, which range from ∼ 0.1 % (Sani et al. 2011), 0.12 % (McLure & Dunlop 2002), and 0.13+0.23−0.08 % (Merritt &
Ferrarese 2001; Kormendy & Gebhardt 2001) to 0.23+0.20−0.11 % (Marconi & Hunt 2003). The reasons are clear: (1) we omit pseudobulges; these do not satisfy the tight correlations in Equations 2 – 7;
(2) we omit galaxies with M• measurements based on ionized gas dynamics that do not take broad emission-line widths into account; (3) we omit mergers in progress. All three of these tend to have smaller BH masses than the objects that define the above correlations. Also, the highest BH masses occur in core ellipticals (more on these below), and these have been revised upward, sometimes by factors of ∼ 2, by the addition of dark matter to dynamical models. Moreover, thanks to papers like Schulze & Gebhardt (2011) and Rusli et al. (2013), we have many such objects.
The exponent in Equation 10 is slightly larger than 1, in reasonable agreement with H¨aring &
Rix (2004), who got M• ∝ Mbulge1.12±0.06 and again a lower normalization, BH mass fraction # 15 %
M • – M bulge Relation
old value
57
Presumably these galaxies contain larger contributions of dark matter that we choose not to include.
The remaining 22 galaxies satisfy log (M/LK) = 0.287 log σe − 0.637 with an RMS scatter of 0.088.
As expected, the relation is shallower than the one in r band (above). It has essentially the same scatter of ∼ 23 %. Dynamically, M/LK = 1 at σe = 166 km s−1, where the Into & Portinari (2013) calibration gives M/LK # 0.76. Cappellari et al. (2006) argue that the difference may be due to the inclusion of some dark matter in the dynamical models. We use the dynamical zeropoint.
To shift the Into & Portinari log M/LK values to the above, dynamical zeropoint, we first use their Table 3 relation log M/LK = 1.055(B − V )0 − 1.066 to predict an initial, uncorrected M/LK. This correlates tightly with σe: log M/LK = 0.239 log σe−0.649 with an RMS scatter of only 0.030.
We then apply the shift ∆ log M/LK = 0.1258 or a factor of 1.34 that makes the corrected Into &
Portinari mass-to-light ratio agree with the dynamic one, M/LK = 1.124, at σe = 250 km s−1. We then have two ways to predict M/LK that are independent except for the above shift,
log M/LK = 0.2871 log σe − 0.6375; RMS = 0.088; (8) log M/LK = 1.055(B − V )0 − 0.9402; RMS = 0.030, (9) where we use the RMS scatter of the correlation with σe to estimate errors for the latter equation.
We adopt the mean of the mass-to-light ratios given by Equations 8 and 9. For the error estimate, we use 0.5!0.0882 + 0.0302 + (half of the difference between the two log M/LK values)2. We use the resulting M/LK together with MK,bulge to determine bulge masses. For the log Mbulge error estimate, we add the above in quadrature to (0.2/2.5)2. The results are listed in Tables 2 and 3.
Figure 18 shows the correlation of M• with bulge mass Mbulge. A symmetric, least-squares fit to the classical bulges and ellipticals omitting the monsters and (for consistency with M• – σe), the emission-lime M• values for NGC 4459 and NGC 4596 plus NGC 3842 and NGC 4889 gives the mass equivalent of Equation 6,
M•
109 M" =
"
0.49+0.06−0.05
# "
Mbulge 1011 M"
#1.16±0.08
; intrinsic scatter = 0.29 dex. (10) Thus the canonical BH-to-bulge mass ratio is M•/Mbulge = 0.49+0.06−0.05 % at Mbulge = 1011 M".
This BH mass ratio at Mbulge = 1011 M" is 2–4 times larger than previous values, which range from ∼ 0.1 % (Sani et al. 2011), 0.12 % (McLure & Dunlop 2002), and 0.13+0.23−0.08 % (Merritt &
Ferrarese 2001; Kormendy & Gebhardt 2001) to 0.23+0.20−0.11 % (Marconi & Hunt 2003). The reasons are clear: (1) we omit pseudobulges; these do not satisfy the tight correlations in Equations 2 – 7;
(2) we omit galaxies with M• measurements based on ionized gas dynamics that do not take broad emission-line widths into account; (3) we omit mergers in progress. All three of these tend to have smaller BH masses than the objects that define the above correlations. Also, the highest BH masses occur in core ellipticals (more on these below), and these have been revised upward, sometimes by factors of ∼ 2, by the addition of dark matter to dynamical models. Moreover, thanks to papers like Schulze & Gebhardt (2011) and Rusli et al. (2013), we have many such objects.
The exponent in Equation 10 is slightly larger than 1, in reasonable agreement with H¨aring &
Rix (2004), who got M• ∝ Mbulge1.12±0.06 and again a lower normalization, BH mass fraction # 15 % at Mbulge = 1011 M". McConnell & Ma (2013) get a similar range of exponents from 1.05 ± 0.11 to 1.23 ± 0.16 depending on how the bulge mass is calculated (dynamics versus stellar populations).
Wednesday, February 26, 14
? ? ?
Wednesday, February 26, 14
Wednesday, February 26, 14
M33: M • < 1500 M
⊙Wednesday, February 26, 14
M33: M • < 1500 M
⊙Wednesday, February 26, 14
Wednesday, February 26, 14
G1: M • = 2 × 10
4M
⊙Gebhardt, Ho & Rich (2005)
Wednesday, February 26, 14
Fan et al.
J. Wise & T. Abel
Wednesday, February 26, 14
Wednesday, February 26, 14
Wednesday, February 26, 14
Are there mini-quasars in
these ‟simpler” galaxies?
Filippenko & Ho (2003); Barth, Ho et al. (2004)
Wednesday, February 26, 14
Fast-moving gas
Filippenko & Ho (2003); Barth, Ho et al. (2004)
NGC 4395
Sdm
Wednesday, February 26, 14
NGC 4395
Sdm
M • = 10
4−10
5M
☉NGC 4395
Sdm
M • = 10
4−10
5M
☉POX 52
Sph or dE
Wednesday, February 26, 14
NGC 4395
Sdm
M • = 10
4−10
5M
☉POX 52
Sph or dE
M • = 1.6 x 10
5M
☉Wednesday, February 26, 14
BH Mass Function 9
Fig. 7.— Volume-weighted BH mass function in bins of 0.25 dex (# Mpc−3 log MBH−1). The weights used are identical to those for the luminosity function, and as above we show in the inset the mass functions for objects targeted as galaxies (left) and QSOs (right), respectively. Although we are subject to significant incom- pleteness, we will argue below that there is truly a turnover in active galaxy masses at both lower and higher BH masses. We have fit the mass function with both a double power-law (dashed line) and a log-normal function (solid line).
Additionally, as the BH mass increases, the corresponding bulge luminosity is presumably increasing, further decreas- ing the contrast and the S/N in the broad line.
If we could uniquely ascribe a host galaxy luminosity and light profile to each MBH, then we could easily model our incompleteness as a function of MBH and Lbol/LEdd. At high BH mass, this is in fact possible, as there is a re- lation linking MBH and spheroid luminosity, and the fun- damental plane tells us the typical sizes (and thus fiber luminosities) of elliptical galaxies. However, for spiral or dwarf spheroidal host galaxies, there ceases to be a unique mapping between MBH and galaxy luminosity or structure.
The relation between bulge-to-total ratio and galaxy lumi- nosity is poorly quantified and contains significant scatter in any case. Furthermore, at lower masses, as the AGNs become intrinsically fainter, only systems with relatively luminous host galaxies will fall above the magnitude limit of the SDSS. For these reasons, at low MBH the calculated incompleteness is a strong function of the assumed (but unconstrained) host galaxy morphology.
As a matter of practicality, then, we turn the problem around. Rather than attempting to quantify our absolute incompleteness as a function of MBH, we simply quan- tify the range of host galaxy luminosities for which we might hope to detect a BH of a given mass, LHα, and z. Over narrow ranges in all of these parameters, neither the line width nor the galaxy continuum strength changes dramatically and the completeness ought to be constant (provided the BHs are drawn from the same host galaxy population). In such bins, with uniform completeness, we are able to measure true changes in space density. Sim-
ulations allow us to isolate ranges of MBH, LHα, and z with constant sensitivity to galaxy fiber luminosity. Note, however, that in any given interval, we necessarily exclude different members of the population as a function of MBH; at the lowest masses we preferentially exclude those sys- tems in faint hosts, while at the highest MBH we exclude the higher Lbol/LEdd systems. Therefore, we implicitly as- sume that the distribution of MBH is uniform independent of both disk luminosity and Eddington luminosity. With this approach, however, we need not concern ourselves di- rectly with host galaxy structure3 per se, but solely the luminosity. The total host galaxy luminosity must be high enough that the source is spectroscopically targeted, while the fiber luminosity must be low enough to allow detection of the broad line.
We investigate three mass regimes (MBH=105.5 − 106.4, 106 − 107, and 106.5 − 107.5 M!), choosing optimal z and LHα ranges for each. Our procedure is described in most detail for the lowest (and most challenging) mass bin, and then results are presented for all three. In the first bin, we are fundamentally limited by the total number of ob- jects. Therefore, we are forced to use the lowest possible redshift bins: z = 0.05−0.07 and z = 0.07−0.1. These are bins with ≥ 10 objects per bin for the most part; at still lower z a prohibitively large range in distance is needed to populate each bin. In terms of LHα, the highest lu- minosity is set by the Eddington luminosity of the lowest mass bin, in this case 1041 ergs s−1 for a BH with mass 105.5 M!, while the lowest luminosity is set by the paucity of lower-luminosity objects (1040.5 ergs s−1).
Simulations allow us to verify that the selection proba- bility is indeed independent of host galaxy fiber luminos- ity. We make artificial spectra in the appropriate MBH and LHα range, with fiber galaxy luminosities spanning
−14 ≤ MB ≤ −22. The galaxy continuum is modeled as a single stellar absorption-line system, constructed from the eigenspectra of Yip et al. (2004), and the S/N is varied to correspond to typical SDSS spectra over the redshift range of interest. Five realizations are made for each galaxy lu- minosity and S/N, and each spectrum is run through our full detection algorithm. For those with detectable broad Hα, we then investigate whether the galaxy luminosity is sufficient for spectroscopic targeting in the first place. Re- call that this limit depends on total (rather than fiber) luminosity, but there is not a one-to-one conversion from fiber to total luminosity; it depends on galaxy morphology and redshift rather strongly. Therefore, we place an upper limit on the total galaxy luminosity by insisting that the fiber luminosity account for no less than 20% of the to- tal galaxy luminosity (as motivated by the observed range shown in Fig. 9 of Tremonti et al. 2004). Over the entire range of galaxy luminosities we explore, a non-zero detec- tion fraction results only for fiber luminosities in the range
−16 < MB < −18, but the detection fractions at a given host luminosity are very constant across the mass range of interest, as shown in Figure 8.
In Figure 9a we show the resulting mass functions for the two different redshift bins. Visually, it appears that the space density is truly falling at low mass. To quantify
Greene & Ho (2004, 2007); Dong, Ho et al. (2012)
Wednesday, February 26, 14
BH Mass Function 9
Fig. 7.— Volume-weighted BH mass function in bins of 0.25 dex (# Mpc−3 log MBH−1). The weights used are identical to those for the luminosity function, and as above we show in the inset the mass functions for objects targeted as galaxies (left) and QSOs (right), respectively. Although we are subject to significant incom- pleteness, we will argue below that there is truly a turnover in active galaxy masses at both lower and higher BH masses. We have fit the mass function with both a double power-law (dashed line) and a log-normal function (solid line).
Additionally, as the BH mass increases, the corresponding bulge luminosity is presumably increasing, further decreas- ing the contrast and the S/N in the broad line.
If we could uniquely ascribe a host galaxy luminosity and light profile to each MBH, then we could easily model our incompleteness as a function of MBH and Lbol/LEdd. At high BH mass, this is in fact possible, as there is a re- lation linking MBH and spheroid luminosity, and the fun- damental plane tells us the typical sizes (and thus fiber luminosities) of elliptical galaxies. However, for spiral or dwarf spheroidal host galaxies, there ceases to be a unique mapping between MBH and galaxy luminosity or structure.
The relation between bulge-to-total ratio and galaxy lumi- nosity is poorly quantified and contains significant scatter in any case. Furthermore, at lower masses, as the AGNs become intrinsically fainter, only systems with relatively luminous host galaxies will fall above the magnitude limit of the SDSS. For these reasons, at low MBH the calculated incompleteness is a strong function of the assumed (but unconstrained) host galaxy morphology.
As a matter of practicality, then, we turn the problem around. Rather than attempting to quantify our absolute incompleteness as a function of MBH, we simply quan- tify the range of host galaxy luminosities for which we might hope to detect a BH of a given mass, LHα, and z. Over narrow ranges in all of these parameters, neither the line width nor the galaxy continuum strength changes dramatically and the completeness ought to be constant (provided the BHs are drawn from the same host galaxy population). In such bins, with uniform completeness, we are able to measure true changes in space density. Sim-
ulations allow us to isolate ranges of MBH, LHα, and z with constant sensitivity to galaxy fiber luminosity. Note, however, that in any given interval, we necessarily exclude different members of the population as a function of MBH; at the lowest masses we preferentially exclude those sys- tems in faint hosts, while at the highest MBH we exclude the higher Lbol/LEdd systems. Therefore, we implicitly as- sume that the distribution of MBH is uniform independent of both disk luminosity and Eddington luminosity. With this approach, however, we need not concern ourselves di- rectly with host galaxy structure3 per se, but solely the luminosity. The total host galaxy luminosity must be high enough that the source is spectroscopically targeted, while the fiber luminosity must be low enough to allow detection of the broad line.
We investigate three mass regimes (MBH=105.5 − 106.4, 106 − 107, and 106.5 − 107.5 M!), choosing optimal z and LHα ranges for each. Our procedure is described in most detail for the lowest (and most challenging) mass bin, and then results are presented for all three. In the first bin, we are fundamentally limited by the total number of ob- jects. Therefore, we are forced to use the lowest possible redshift bins: z = 0.05−0.07 and z = 0.07−0.1. These are bins with ≥ 10 objects per bin for the most part; at still lower z a prohibitively large range in distance is needed to populate each bin. In terms of LHα, the highest lu- minosity is set by the Eddington luminosity of the lowest mass bin, in this case 1041 ergs s−1 for a BH with mass 105.5 M!, while the lowest luminosity is set by the paucity of lower-luminosity objects (1040.5 ergs s−1).
Simulations allow us to verify that the selection proba- bility is indeed independent of host galaxy fiber luminos- ity. We make artificial spectra in the appropriate MBH and LHα range, with fiber galaxy luminosities spanning
−14 ≤ MB ≤ −22. The galaxy continuum is modeled as a single stellar absorption-line system, constructed from the eigenspectra of Yip et al. (2004), and the S/N is varied to correspond to typical SDSS spectra over the redshift range of interest. Five realizations are made for each galaxy lu- minosity and S/N, and each spectrum is run through our full detection algorithm. For those with detectable broad Hα, we then investigate whether the galaxy luminosity is sufficient for spectroscopic targeting in the first place. Re- call that this limit depends on total (rather than fiber) luminosity, but there is not a one-to-one conversion from fiber to total luminosity; it depends on galaxy morphology and redshift rather strongly. Therefore, we place an upper limit on the total galaxy luminosity by insisting that the fiber luminosity account for no less than 20% of the to- tal galaxy luminosity (as motivated by the observed range shown in Fig. 9 of Tremonti et al. 2004). Over the entire range of galaxy luminosities we explore, a non-zero detec- tion fraction results only for fiber luminosities in the range
−16 < MB < −18, but the detection fractions at a given host luminosity are very constant across the mass range of interest, as shown in Figure 8.
In Figure 9a we show the resulting mass functions for the two different redshift bins. Visually, it appears that the space density is truly falling at low mass. To quantify
3At a given luminosity, a wide range of galaxy morphologies are permitted. Thus it is still possible to find significant differences in σ∗ (e.g., Greene & Ho 2006b) and potentially host galaxy structure (J. E. Greene, in preparation) as a function of MBH for the SDSS-selected samples of low-mass systems.
200-300 new sources
Greene & Ho (2004, 2007); Dong, Ho et al. (2012)
Wednesday, February 26, 14
HST/ACS Greene, Ho & Barth (2008); Jiang et al. (2011a, 2011b)
Wednesday, February 26, 14
Wednesday, February 26, 14
Ho (2014a, b)
Wednesday, February 26, 14
Recent Updates
☯ Central BHs detected from 104 – 1010 M⊙
☯ All bulges contain BHs, but not all BHs live in bulges
☯ M• ~ Mbulge 〈M• / Mbulge〉 ~ 0.5%
☯ M• ∝ "4.4
☯ M•− " and M• − Mbulge relations have similar scatter
☯ Scaling relations only tight for classical bulges and Es
☯ Scaling relations already in place for high-z QSOs
☯ Mild evolution only for most massive BHs
☯ AGN feedback effective only for classical bulges and Es
1.2
Kavli Institute for Astronomy and Astrophysics (KIAA)
Wednesday, February 26, 14
Featured Science
New Patterns in Planet Distributions
Subo Dong ( 勃) joined the faculty of KIAA in Fall 2013 through the 1000 Talents Program for young researchers (青年千人 ). One of his research interests is to robustly derive the distributions of extrasolar planets in order to find clues on how planet systems form and evolve. In a paper recently published in the Astrophysical Journal (http://arxiv.org/abs/1212.4853), he and Zhaohuan Zhu (Princeton
University) determined the distributions of planets down to Earth size and in orbits closer than Venus.
Wind Braking of AXP/SGRs
Anomalous X-ray pulsars (AXPs) and soft gamma-ray repeaters (SGRs) are believed to be magnetars: peculiar neutron stars powered by their super strong magnetic field. Unfortunately, none of the predictions of traditional magnetar models
successfully explain their properties. In a recent paper, the group of Prof. Renxin Xu (PKU Department of Astronomy, with joint appointment at the KIAA), in
collaboration with Dr. Hao Tong of the Xinjiang Astronomical Observatory, show that a wind braking mechanism in magnetars, where the energy release generates a strong wind, provides a natural understanding of the multiwavelength observational behavior of AXPs and SGRs.
Star Formation and Quasar-host Galaxy Co-evolution in the Most Distant Universe
Dr. Ran Wang and her collaborators have carried out millimeter and radio observations to study the ISM properties and star forming activity in the host galaxies of quasars at the highest redshift. The team detected strong dust continuum and molecular CO line emission in about 30% of the optically bright quasars at z~6. The results indicate huge amounts of 40 to 60 K warm dust and molecular gas in the host galaxies of these millimeter bright quasars.
Latest News
New Science Advisory Committee
Three Astronomy Department Faculty Join KIAA
Experts in Extrasolar Planets, Pulsar Timing, and Distant Quasars Join the Faculty Search @ KIAA-PKU
Search
Upcoming Events
How much cosmological information can be
measured?
Speaker: Yinzhe Ma, University of British Columbia
Time: Mon, 2014-01-06 12:00 to 13:00
Location: DoA, Rm 2907
Finding Transiting Exoplanets and
Characterizing their
Atmospheres: HATSouth and ACCESS
Speaker: Andrés Jordán (PUC) Time: Thu, 2014-01-09 16:00 to 17:00
Location: KIAA-PKU Auditorium
Quick Links
For Visitors
English
Copyright © 2014, The Kavli Institute for Astronomy and Astrophysics at Peking University (KIAA-PKU) Copyright © 2014, The Kavli Institute for Astronomy and Astrophysics at Peking University (KIAA-PKU)
HOME ABOUT KIAA PEOPLE ACTIVITIES NEWS SCIENCE OPPORTUNITIES OUTREACH VISITOR INFO ADMIN
kiaa.pku.edu.cn
Wednesday, February 26, 14
