Star formation rate and dynamical mass of 10^8 solar mass black hole host galaxies at redshift 6

0.66 MB·English

by  Chris J. Willott

Draft version January 30, 2015
PreprinttypesetusingLATEXstyleemulateapjv.12/01/06
STAR FORMATION RATE AND DYNAMICAL MASS OF 108 SOLAR MASS BLACK HOLE HOST GALAXIES
AT REDSHIFT 6
Chris J. Willott
NRCHerzberg,5071WestSaanichRd,Victoria,BCV9E2E7,Canada
Jacqueline Bergeron and Alain Omont
UPMCUnivParis06andCNRS,UMR7095,Institutd’AstrophysiquedeParis,F-75014,Paris,France
Draft version January 30, 2015
5
1 ABSTRACT
0 We present ALMA observations of two moderate luminosity quasars at redshift 6. These quasars
2 from the Canada-France High-z Quasar Survey (CFHQS) have black hole masses of ∼108M . Both
(cid:12)
n quasars are detected in the [Cii] line and dust continuum. Combining these data with our previous
a study of two similar CFHQS quasars we investigate the population properties. We show that z > 6
J quasarshaveasignificantlylowerfar-infraredluminositythanbolometric-luminosity-matchedsamples

9 at lower redshift, inferring a lower star formation rate, possibly correlated with the lower black hole
2 massesatz =6. Theratiosof[Cii]tofar-infraredluminositiesintheCFHQSquasarsarecomparable
with those of starbursts of similar star formation rate in the local universe. We determine values of

] velocity dispersion and dynamical mass for the quasar host galaxies based on the [Cii] data. We find
A
that there is no significant offset from the relations defined by nearby galaxies with similar black hole
G masses. There is however a marked increase in the scatter at z = 6, beyond the large observational
. uncertainties.
h
Subjectheadings: cosmology: observations—galaxies: evolution—galaxies: high-redshift—quasars:
p
general
-
o
r
t 1. INTRODUCTION has been considerable success in measuring AGN host
s
galaxy luminosities, morphologies and in some cases ve-
a Improvedastronomicalobservationalfacilitieshaveen-
[ abled the discovery and study of many galaxies at an locitydispersions(Cisternasetal.2011;Parketal.2014).
early phase of the Universe’s history. It is now possi- At higher redshifts (1 < z < 4) the galaxy light is more
1
difficult to separate from the quasar, which, combined
v ble to witness the majority of the stellar and black hole
with greater mass-to-light corrections, lead to larger un-
8 mass growth over cosmic time and identify how physical
certainties(Merlonietal.2010;Targettetal.2012). The
3 conditions at early times differ from now. One of the
results of these studies are mixed with some evidence in
5 majorrelationstobedeterminedasafunctionoftimeis
favour of higher M at a given galaxy mass.
7 thetightcorrelationbetweenblackholemassandgalaxy BH
0 properties observed for nearby galaxies (see Kormendy Atyethigherredshiftsithasprovedimpossibletomea-
. & Ho 2013 for a review). Observations of this relation sure the galaxy light of quasars (Mechtley et al. 2012)
1
before launch of the James Webb Space Telescope and
0 at high-redshift are critical to understanding the cause
instead the main method of determining galaxy mass is
5 because most of the growth occurred at early times.
kinematics of cool gas in star-forming regions (Carilli &
1 Attempts to measure black hole and galaxy masses at
Walter 2013). Facilities such as the IRAM Plateau de
: high-redshiftfaceanumberofproblems. Blackholemass
v Bure Interferometer, the Jansky Very Large Array and
measurementscannotbemadedirectlybyresolvedkine-
i the Atacama Large Millimeter Array (ALMA) have suf-
X matics of gas or stars within the black hole’s sphere of
ficientsensitivityandresolutiontoresolvethegasindis-
influence, nor by reverberation mapping. Instead black
r
tantquasarhostsandprovidedynamicalmasses(Walter
a hole masses, M , of quasars can be measured at any
BH et al. 2004, 2009; Wang et al. 2010, 2013). In partic-
redshiftusingthesingle-epochvirialmassestimatorthat
ular, ALMA has the sensitivity to probe z = 6 quasar
involves measuring a low-ionization broad emission line,
such as Mgii or Hβ, and calibrating the location of the hostswithstarformationrates,SFR,inthetensofsolar
masses per year, rather than only in the extreme star-
emitting gas with low-z reverberation-mapped quasars
bursts previously observable (Willott et al. 2013). The
(Wandel et al. 1999). For AGN with obscured broad
studiesabovefocussedonz ≈6SloanDigitalSkySurvey
lines M can only be estimated from the luminosity
BH (SDSS)andUKIRTInfraredDeepSkySurvey(UKIDSS)
making an assumption about the accretion rate relative
quasars with high UV and far-IR luminosities and found
to the Eddington limit.
that their black holes are on average 10 times greater
Measuring galaxy properties, such as luminosity or
thanthecorrespondingσ forlocalgalaxies, roughlycon-
velocity dispersion, σ, of distant quasars is hampered
sistentwithacontinuationoftheevolutionseeninlower
by surface brightness dimming, the bright glare of the
redshift studies.
quasar and AGN (active galactic nuclei) emission line-
Although observationally there appears to be an in-
contamination of spectral features. Up to z ≈ 1 there
crease in M with redshift at a given galaxy mass or
BH
Electronicaddress: [email protected] σ, it has long been understood that there are selection
2 Willott et al.
biases that affect how closely the observations trace the tors, J0108+0135 and J2232+1143. The amplitude cali-
underlying distribution. In particular, the steepness of bratorwasNeptuneandthebandpasscalibratorsJ2258-
the galaxy and dark matter mass functions combined 2758andJ2148+0657. Totalon-sourceintegrationtimes
with large scatter in their correlations with black hole were 4610s for J0055+0146 and 5490s for J2229+1457.
mass mean that a high black-hole-mass-selected sample The band 6 (1.3mm) receivers were set to cover
ofquasarswillhaveasystematicoffsetinσtowardslower the frequency range of the redshifted [Cii] transition
values. Thiseffect,firstidentifiedbyWillottetal.(2005) (ν =1900.5369 GHz) and sample the dust continuum.
rest
andFineetal.(2006)wasstudiedindetailinLaueretal. Therearefour≈2GHzbasebands,twopairsofadjacent
(2007) and numerous studies thereafter. The magnitude bandswitha11GHzgapinbetween. Thechannelwidth
of the effect depends upon the scatter in the correla- is 15.625MHz (17kms−1).
tion, which has not been conclusively measured at high- The data were initially processed by North American
redshift, but appears to increase with redshift (Schulze ALMA Regional Center staff with the CASA software
& Wisotzki 2014). Willott et al. (2005) and Lauer et al. package1. On inspection of these data it became clear
(2007) showed that the bias is particularly strong for thatthe[Cii]lineofJ0055+0146waslocatedrightatthe
M > 109M quasars such as those in the SDSS at edgeofthebaseband, 1000kms−1 fromthetargetedfre-
BH (cid:12)
z ≈6andthereforethatthefactorof10increaseinM quency defined by the broad, low-ionization Mgii emis-
BH
at a given σ first seen in the quasar SDSSJ1148+5251 sionline(z =5.983; Willottetal. 2010a). TheMgii
MgII
(Walter et al. 2004) could be accounted for by the bias line redshift is usually close to the systemic redshift as
(seealsoSchulze&Wisotzki2014). Incomparison,there measured by narrow optical lines with a dispersion of
would be little bias for a sample of high-z quasars with 270kms−1 (Richards et al. 2002). A large offset for this
black hole masses of M ∼108M (Lauer et al. 2007). quasar was not particularly surprising for two reasons:
BH (cid:12)
An alternative to measuring the evolution of the as- firstly the signal-to-noise (SNR) of the Mgii detection
sembled galaxy and black hole masses is to determine is not very high and the line appears double-peaked due
the rate at which mass growth is occurring. For quasars to noise and/or associated absorption; secondly the Lyα
the bolometric luminosity is a measure of the black hole redshift(z =6.02)isoffsetfromMgiiby1600kms−1
Lyα
massgrowthrate. Forgalaxies,thestarformationrateis (in the same direction as the [Cii] offset) and this would
proportionaltothestellarmassgrowth. Thestarforma- make the size of the Lyα ionized near-zone negative,
tionratecanbedeterminedbytherest-framefar-infrared which is not physically sensible for a quasar with such
dust continuum luminosity. Additionally, the interstel- a high ionizing flux and has not been observed in a sam-
lar [Cii] far-infrared emission line is well-correlated with ple of 27 z ≈6 quasars (Carilli et al. 2010). Due to this
star-formation (De Looze et al. 2014; Sargsyan et al. redshift uncertainty the receiver basebands were set up
2014) so can also be used as a star formation proxy. so that the adjacent band covered the Lyα redshift with
In Willott et al. (2013, hereafter Wi13) we presented zero gap between the two bands.
Cycle0ALMAobservationsinthe[Cii]lineand1.2mm The default ALMA Regional Center reduction ex-
continuum for two z = 6.4 quasars from the Canada- cluded 11 channels at each end of the 128 channel band.
France-High-z Quasar Survey (CFHQS, Willott et al. However, only the first 4 channels need to be excluded,
2010b). These quasars have M ∼ 108M , a factor so we re-reduced the data with CASA to include more
BH (cid:12)
of 10–30 lower than most SDSS quasars known at these spectral channels at the baseband edges. We checked
redshifts. Onequasarwasdetectedinlineandcontinuum that the noise does not increase in these extra channels,
andtheotherremainedundetectedinthesesensitiveob- exceptfortheveryfirstandlastchannelstocontaindata
servations placing an upper limit on its star formation so we excluded those. In summary our reduced product
rate of SFR<40M yr−1. contains 118 of the original 128 channels per baseband,
(cid:12)
In this paper we present ALMA observations of two compared with 106 channels in the default reduction.
further CFHQS quasars with similar redshift and black
hole mass with the aim of providing a sample large 3. RESULTS
enough to address the issue of how host galaxy prop- Figure 1 shows the reduced spectrum of J0055+0146
erties such as SFR, σ and dynamical mass depend upon fromthetwoadjacentbasebands. Thefinalgapbetween
blackholeaccretionrateandmassatatimejust1billion the bands is only ≈ 150 kms−1 and crucially the peak
yearsaftertheBigBang. Inparticular,thesequasarsare of the [Cii] line is contained within the higher frequency
notsubjecttothebiasintheMBH−σ relationdiscussed band. The lower frequency band contains only a small
previously because of their moderate black hole masses. amount of the line flux but provides an important con-
Cosmological parameters of H0 = 67.8 km s−1 Mpc−1, straint on the wings and hence the peak and width for
ΩM =0.307andΩΛ =0.693(PlanckCollaboration2014) a symmetric line. A single Gaussian plus flat continuum
are assumed throughout. modelwasfittotheavailabledatausingaMarkov-Chain
MonteCarlo(MCMC).Thisprocessshowsagoodfitfor
2. OBSERVATIONS
a single Gaussian with FWHM=359±27 kms−1. The
CFHQSJ005502+014618 (hereafter J0055+0146) and formal uncertainty in the FWHM is very small consid-
CFHQSJ222901+145709 (hereafter J2229+1457) were ering that there is some missing data. This is because
observed with ALMA on the 28, 29 and 30 November a symmetric line model is used and with the peak and
2013forCycle1project2012.1.00676.S.Between22and wingscoveredbydatathereislittlemarginfordeviation
26 12m diameter antennae were used. The typical long in the missing channels. We add in quadrature an extra
baselines were ∼ 400m providing similar spatial resolu- 10%uncertaintyinboththelinefluxandFWHMdueto
tiontoourCycle0observations. Observationsofthesci-
ence targets were interleaved with nearby phase calibra- 1 http://casa.nrao.edu
Star formation rate and dynamical mass of 108 solar mass black hole host galaxies 3
Velocity offset (km / s)
2.5 2000 1000 0 1000 2000 CFHQS J0055+0146 11.0 CFHQS J2229+1457
Lyα z MgII z 20.0
CFHQS J0055+0146 10.0
2.0 [CII] z=6.0060+/-0.0008
19.0
sity (mJy) 11..05continuum Dec (J2000) 1178..00 Dec (J2000) 0089..00
n 07.0
e
ux-d 0.5 16.0 +14:57:06.0
Fl 0.0 +1:46:15.0
03.10 03.00 02.90 0:55:02.80 01.80 01.70 01.60 22:29:01.50
RA (J2000) RA (J2000)
0.5 Fig. 3.— The color scale shows the integrated [Cii] line maps
for the two quasars. White contours of 1.2mm continuum emis-
269.0 269.5 270.0 270.5 271.0 271.5 272.0 272.5 273.0
Observed Frequency (GHz) sion from the three line-free basebands are over-plotted at levels
of2,4,6σbeam−1. Thequasaropticalpositionsareshownwitha
Fig. 1.—ALMAspectrumofJ0055+0146coveringtwoadjacent
blackplussymbol. Thepositionaloffsetsbetweentheopticaland
basebands. The gap between the bands with no data is shaded
millimeteraremostlikelyduetoastrometricmismatch,ratherthan
in gray. The higher frequency band is centred on the redshift de-
a physical offset. J0055+0146 is well-detected in both continuum
terminedfromtheMgiiemissionlinewhereasthelowerfrequency
andlineemission. J2229+1457hasonlya2σcontinuumdetection
band covers the Lyα redshift. The [Cii] line is found at the edge
that is spatially co-incident with the line emission. The restoring
of the higher frequency band. The blue curve is a Gaussian plus
beamisshowninyellowinthelower-leftcorner.
continuum fit as described in the text. The red circle marks the
continuumlevelindependentlymeasuredinthethreeline-freebase-
bands. Theupperaxisisthevelocityoffsetfromthebest-fit[Cii]
Gaussianpeak. TABLE 1
Millimeter data for the two CFHQS quasars
CFHQSJ0055+0146 CFHQSJ2229+1457
1000 500 Velocity offse0t (km / s) 500 zMgIIa 5.983±0.004 6.152±0.003
z 6.0060±0.0008 6.1517±0.0005
MgII z [CII]
1.5 C[CFIHI]Q zS= 6J2.212591+7+14/-507.0005 FIWHM(J[yCkIIm] s−1) 30.5893±9±450.k1m32s−1 03.5518±2±390.k0m75s−1
[CII]
mJy) 1.0 Lf1[.C2ImI]m(L(µ(cid:12)J)y) 2(81.127±±341.30)×108 5(54.9±62±90.77)×108
nsity ( 0.5continuum LSFFRIR[C(ILI](cid:12)(M) (cid:12)yr−1) 8(43.8±51±30.78)×1011 6(10.2±48±0.67)×1011
Flux-de 0.0 SLF[CRIIF]/IRL(FMIR(cid:12)yr−1) (713.7±01±20.38)×10−3 1(49.8±01±02.67)×10−3
Notes.—
0.5 a DerivedfromMgiiλ2799observations(Willottetal.2010a).
265.0 265.5 266.0 266.5 UncertaintiesinLFIR,SFR[CII]andSFRFIRonlyincludemeasure-
Observed Frequency (GHz) ment uncertainties, not the uncertainties in extrapolating from a
monochromatictointegratedluminosityorthatoftheluminosity-
Fig. 2.— ALMA spectrum of J2229+1457 covering the single
SFRcalibrations.
baseband containing the [Cii] line. The blue curve is a Gaussian
plus continuum fit and the red circle the independent continuum
level. The upper axis is the velocity offset from the best-fit [Cii] the vicinity, so it is not considered a secure detection,
Gaussianpeak. however the measured flux and uncertainty are included
in Table 1.
the missing channels. Thequasar1.2mmcontinuumflux-densitieswerecon-
The spectrum of J2229+1457 is plotted in Figure 2. verted to far-infrared luminosity, LFIR, assuming a typ-
For this quasar the [Cii] line is centred in the band ical SED for high-redshift star-forming galaxies. As in
with no significant offset from the Mgii redshift. The Wi13 we adopt a greybody spectrum with dust temper-
line is consistent with a single Gaussian with a best ature, Td = 47K and emissivity index, β = 1.6. To
fit FWHM=351 ± 39 kms−1, similar to the value for convert from far-IR luminosity to star formation rate we
J0055+0146. Thecontinuumlevelofthefitisagaincon- use the relation SFR (M(cid:12)yr−1)=1.5×10−10LFIR(L(cid:12))
sistentwiththeindependentcontinuumleveldetermined appropriate for a Chabrier IMF (Carilli & Walter 2013).
from the three line-free basebands. Measurements from We note that this assumes that all the dust contribut-
the spectra are given in Table 1. ing to the 1.2mm continuum is heated by hot stars and
Figure 3 shows maps of the 1.2mm continuum (white not by the quasar. An alternative estimate of the star
contours)plusthe[Cii]line(colorscale)forthequasars. formation rate comes from the [Cii] luminosity. We
For both quasars there is no significant offset between adopt the relation in Sargsyan et al. (2014) of SFR
any of the continuum or line centroids or the optical (M(cid:12)yr−1)=1.0×10−7L[CII](L(cid:12)). Fortheremainderof
quasar position. The < 1(cid:48)(cid:48) mm-optical offset is within thispaper,uncertaintiesonL (andinferredSFR)only
FIR
the relative uncertainty of the optical astrometry. The include the flux measurement uncertainties, not that of
more accurate [Cii] positions for the two quasars are thedusttemperatureandluminositytoSFRconversion.
00:55:02.92 +01.46.17.80 and 22:29:01.66 +14.57.08.30. The synthesized beam sizes are 0(cid:48).(cid:48)63 by 0(cid:48).(cid:48)45 for
For J2229+1457 there is only a marginal 2σ detection J0055+0146and0(cid:48).(cid:48)76by0(cid:48).(cid:48)64forJ2229+1457. Thebet-
of the continuum located coincident with the peak of ter resolution for J0055+0146 is mostly due to higher
the line emission. As seen in Figure 3 there are several elevation of observation. We used the CASA IMFIT task
other continuum peaks of this magnitude or greater in tofit2Dgaussianmodelstothesemaps. ForJ0055both
4 Willott et al.
the continuum and line are resolved with deconvolved
sourcesizesof0(cid:48).(cid:48)51±0(cid:48).(cid:48)13by0(cid:48).(cid:48)35±0(cid:48).(cid:48)26atpositionan- Far-IR luminosity evolution of L 7 1012L quasars
gle87degreesand0(cid:48).(cid:48)50±0(cid:48).(cid:48)14by0(cid:48).(cid:48)18±0(cid:48).(cid:48)27atposition Bol∼ ×Veale etfl al. 2014
angle 62 degrees, respectively. At the distance to this This paper
quasar the spatial extent of 0.5(cid:48)(cid:48) is equal to a linear size Serjeant et al. 2010
Rosario et al. 2012
of2.9kpc. Wenotethatthemissingdataintheredwing Rosario et al. 2013
of the [Cii] line may cause a bias in the size and inclina-
tioniftheemissioncomesfromarotatingdisk,butthere 1012
is no evidence for this based on the similarity of the line )fl
and continuum sizes. For J2229+1457 the continuum is (L
R
too poorly detected to attempt a size measurement and LFI
the line emission is only marginally more extended than
the beam size. In several other z ≈ 6 quasars velocity
gradients across the sources are observed (Wi13; Wang
et al. 2013). Velocity gradients are not seen for either of
these two quasars, although for J0055+0146 the missing
data for 150kms−1 of the red wing hampers our ability
to detect such a gradient. 10110 1 2 3 4 5 6 7
Redshift z
4. DISCUSSION
4.1. Evolution of far-IR luminosity Fig. 4.— Stacked mean far-infrared luminosity for samples of
quasarsatdifferentredshifts. Detailsofthesamplesaredescribed
In Wi13 we reported on the low far-IR luminosities of inthetext,butallsamplesareselectedtoincluderoughlythesame
the two previously observed CFHQS quasars and impli- rangeinbolometricluminositycentredonLBol∼7×1012L(cid:12),the
meanL ofthefourz≈6CFHQSquasarsplottedwiththeblue
cations for the relatively low SFR of these quasar host Bol
square. There is a clear rise in LFIR up to a peak at 2 < z < 3
galaxies relative to the black hole accretion rate. We followedbyadeclinetoz=6. Themagentacurveshowsthemean
now revisit this issue with the sample of four z ≈ 6 LFIR due to star formation for the model of Veale et al. (2014)
CFHQSquasarswithALMAobservations. Wenotethat includingscalingbyafactorof2toaccountforstellarmassloss.
thissampleincludesfourofthesixCFHQSquasarswith
measured black hole masses within the absolute mag- found in the optical quasar phase of co-evolution at a
nitude range −25.5 < M < −24 at a declination time ∼ 1Gyr after the onset of activity for the z = 2
1450
low enough for ALMA observation. The two unobserved model of Lapi et al. (2014). Given that this is the age
quasars have 7×108 < M < 109M , and were not of the universe at z = 6 the evolution must occur more
BH (cid:12)
observed due to limited time available and the desire to rapidly at higher redshift. However, the effect of AGN
study the lowest mass black holes from CFHQS. There- variability may also be important leading to a selection
fore there is a slight bias to low black hole mass in this effect whereby AGN luminosity-selected objects are ob-
sample compared to pure UV-luminosity-selection. served to have lower ratios of L /L than the time-
FIR Bol
Twoofthefourquasarsarewelldetectedinthecontin- averaged values (Hickox et al. 2014; Veale et al. 2014).
uum with fluxes of 211±34 (J0055+0146) and 120±35 Wehavepreviouslyshown(Wi13)thatthetwoz =6.4
(J0210-0456) µJy. J2229 has a marginal 2σ detection of CFHQS quasars observed with ALMA in Cycle 0 have
54±29µJy(Figure3andTable1)andJ2329-0301isun- L lower than quasars of similar AGN luminosity at
FIR
detected with a 1σ rms of 30µJy. We combine the four lower redshift. On the other hand, at fixed AGN lumi-
valuesoffar-infraredluminosityderivedfromthesemea- nosity L is observed to rise from z = 0 to z = 3
FIR
surements assuming that J2329-0301 has a flux equal to (Serjeant et al. 2010; Bonfield et al. 2011; Rosario et al.
its2σ upperlimitof60µJy. Themeanandstandardde- 2012, 2013). We next analyze the evolution of L us-
FIR
viationofthesampleisL =(2.6±1.4)×1011L . We ing our expanded ALMA sample at z = 6 and compa-
FIR (cid:12)
notethisismuchlowerthanthevaluesof1012−1013L rable low-redshift data. At all redshifts we determine
(cid:12)
typically discussed for z ≈6 quasars due to two factors, the mean L for optically-selected quasars and X-ray
FIR
firstly that the CFHQS sample here have lower AGN AGN in a narrow range of L corresponding to the
Bol
luminosity than most known z ≈ 6 quasars and a cor- mean L of the four CFHQS z ≈ 6 quasars in this
Bol
relation between AGN and far-IR luminosities is present paper (L ∼7×1012L ).
Bol (cid:12)
(Wang et al. 2011; Omont et al. 2013), but also that At z < 3 we use three datasets based on Her-
our small sample is selected on quasar rest-frame UV schel imaging of AGN. Serjeant et al. (2010) stacked
luminosity and black hole mass, whereas previous stud- Herschel SPIRE data of optically-selected quasars and
ies have focussed on quasars with pre-ALMA millimeter quoted their results as rest-frame 100µm luminosity.
continuum detections. We adopt L = 1.43νL (100µm) (Chary & Elbaz
FIR ν
The implication is that these quasars have very high 2001) to convert to far-infrared luminosity. The abso-
blackholeaccretionratesasinferredfromtheAGNbolo- lute magnitude bin −26 < I < −24 corresponds well
AB
metric luminosity, L , yet relatively low SFR. Such to L ∼ 7×1012L and we have trimmed the size of
Bol Bol (cid:12)
a scenario is consistent with the well-known evolution- the highest redshift bin from 2 < z < 4 to 2 < z < 2.7
ary model whereby the optical quasar phase comes after because inspection of the luminosity-redshift plane fig-
the main star forming phase (Khandai et al. 2012; Lapi ure in Serjeant et al. (2010) shows all but one of the 52
et al. 2014), possibly due to quasar feedback inhibiting quasars in this bin are at z < 2.7. Rosario et al. (2013)
gas cooling and star formation. The measured ratio of analyzed the Herschel PACS data for optically-selected
L /L = 0.035 for the four CFHQS quasars is only quasars in COSMOS. Due to the shorter wavelength of
FIR Bol
Star formation rate and dynamical mass of 108 solar mass black hole host galaxies 5
PACS than SPIRE they presented results in rest-frame variantofthemodelplottedhereisthe“accretion”model
60µm luminosity. We adopt L = 1.5νL (60µm) where the quasar luminosity is proportional to the black
FIR ν
(Chary & Elbaz 2001) to convert to far-infrared lumi- holegrowthrateandtheEddingtonratiodistributionisa
nosity. From this study we use only the highest red- truncatedpower-lawwithslopeβ =0.6(dashedcurvein
shift, highest luminosity bin as this compares well with Figure 8 of Veale et al. 2014). The model is constrained
L ∼7×1012L . Rosarioetal.(2012)determinedthe bytheobservedevolvingquasarluminosityfunctionand
Bol (cid:12)
mean infrared luminosity with PACS for X-ray-selected the local ratio of black hole to galaxy mass. We have
AGNfromtheCOSMOSsurvey. WenotethattheX-ray scaledthismodelwithafactorof2×increaseinL to
FIR
selected AGN sample contains a mixture of broad-line, account for stellar mass loss.
narrow-line and lineless AGN and these may have dif- As seen in Figure 4 this curve increases from low red-
ferent evolutionary properties, but Rosario et al. (2013) shifttothepeakquasarepochat2<z <3byaboutthe
showedthatthemeanL ofquasarsandX-ray-selected samefactorasthedata,althoughthetotalnormalization
FIR
are similar at a given AGN luminosity and redshift. We of the curve is lower by a factor of 3 to 4. Veale et al.
use the Rosario et al. (2012) data from the AGN lumi- (2014) discuss some of the reasons why the normaliza-
nosity bin 8×1011 < L < 2×1013L . Whilst most tion may be lower than the observations. The decrease
Bol (cid:12)
sources in this bin have L <3×1012L , we consider in L with increasing cosmic time from z =2 to z =0
Bol (cid:12) FIR
the results appropriate to compare to the z ≈6 quasars for fixed luminosity quasars is due to the fact that such
asthecorrelationbetweenL andL isveryshallow quasars are rarer at lower redshift and on the steep end
FIR Bol
at this luminosity in Rosario et al. (2012). of the luminosity function where scatter is more impor-
Figure4plotsdata from thesethreelow-redshiftstud- tant. This behaviour also follows from the general de-
ies with the CFHQS ALMA bin at 6 < z < 6.5. The crease in specific star formation rate with cosmic time.
three low-redshift studies show a rise in L of a factor The high-redshift behaviour of a decline from z = 3 to
FIR
of4fromz =0.3toz =2.4. Thisriseisattributedtothe z = 6 matches our observations, so it is instructive to
generalincreaseinmassivegalaxyspecificstarformation understand why this occurs in the model. It is due to
rate over this redshift range (Hickox et al. 2014). The the assumed (1+z)2 evolution of the ratio of accretion
6 < z < 6.5 bin has a large dispersion due to the range growth to stellar mass growth, but this assumed evolu-
in 1.2mm continuum flux measured for the 4 quasars. tion is also degenerate with evolution in the Eddington
The mean L at z ≈ 6 is a factor of about 2 lower ratio. As discussed previously there is observational evi-
FIR
than the z = 0.3 bin and 6 lower than the z = 2.4 peak dence for positive evolution in the Eddington ratio from
at the so-called quasar epoch. There is clear evidence z = 2 to z = 6, meaning that the ratio of accretion
here for a turnaround that mimics the evolution of the growth to stellar mass growth may change more gradu-
quasar luminosity function (McGreer et al. 2013) and ally than (1+z)2 .
star formation rate density (Bouwens et al. 2014), albeit A possible alternative explanation for the low L at
FIR
with a much less steep high-redshift decline due to the z = 6 is that at these early epochs insufficient dust has
factwe aremeasuringstar formation inspeciallocations been generated so that star formation occurs more often
within the universe where dark matter halos must have within lower dust environments (Ouchi et al. 2013; Tan
collapsed much earlier than typical in order to build up et al. 2013; Fisher et al. 2014; Ota et al. 2014). In this
the observed black hole masses of ∼108M . case there could be a much smaller decline in the typical
(cid:12)
What is the physical reason for this turnaround at SFR of a luminous quasar hosting galaxy. However, two
z > 3? In the evolutionary picture where the opti- lines of evidence point towards this not being the main
cal quasar phase follows the starburst phase one would factor for our quasar sample. First, quasars at z = 6
expect the star formation and black hole accretion to are known to have emission line ratios similar to lower
be more tightly coupled at high-redshift where there is redshift quasars inferring high metallicity at least close
barely enough time for star formation to have decreased to the accreting black hole (Freudling et al. 2003). Sec-
substantially. Acluemaycomefromoneofthefewdiffer- ond, the [Cii] luminosities in three of the four quasars
encesbetweenquasarsatthesetwoepochs. Willottetal. arehigh(seebelow),suggestinghighcarbonabundances
(2010a) showed that the Eddington ratios of matched throughout the host galaxies.
quasarluminositysamplesatz =2andz =6aresignifi-
cantlydifferentwiththez =6quasarshavingafactorof 4.2. The [Cii] – far-IR luminosity relation
3×higherEddingtonratiosandtherefore3×lowerblack ThreeofthefourCFHQSALMAquasarsaredetected
hole masses than at z = 2. Such a difference exists be- in both [Cii] line and 1.2mm continuum emission (two
tween the typical black hole mass of our CFHQS ALMA new detections in this paper and J0210−0456 in Wi13).
sampleandthatofthehighest luminositybinofRosario With the low L discussed in the previous section,
FIR
et al.(2013). This Eddington ratioevolution is observed these quasar host galaxies probe a new regime in L
FIR
inotherstudies(DeRosaetal.2011;Trakhtenbrotetal. at high-redshift. In Figure 5 we plot the ratio of [Cii]
2011; Shen & Kelly 2012) and predicted by many theo- to far-IR luminosity as a function of L . Also plot-
FIR
retical works due to the increase in gas supply to black ted are several samples from the literature which, due
holes at high-redshift (Sijacki et al. 2014). to ALMA at high-redshift and Herschel at low-redshift,
In Figure 4 we also plot a theoretical curve of mean are rapidly increasing in size and data quality. The low-
L versusredshiftforasimulatedsampleofrest-frame redshiftz <0.4sampleofgalaxiesisfromGraci´a-Carpio
FIR
UV-selected quasars in the same L range as the ob- et al. (2011 and in prep.) and contains a mix of normal
Bol
served quasar samples for the model of Veale et al. galaxies, starbursts and ultra-luminous infrared galaxies
(2014). This model assumes an evolving linear relation- (ULIRGs), some of which contain AGN. The ULIRGs
ship between star formation and black hole growth. The show a [Cii] deficit that has been widely discussed in
6 Willott et al.
tional data, are plotted as dotted lines. These also favor
ashallownegativeslope, notnearlyassteepastheslope
10-2
that would be fit to the previous z > 0.5 data that cov-
ers only a narrow range of L . The L /L ratios
FIR [CII] FIR
of the CFHQS and ULAS z >6 quasars are not greatly
differenttothoseofsimilarL galaxiesatlow-redshift.
FIR
Wangetal.(2013)notethatthelowL /L ratios
[CII] FIR
for SDSS z > 5 quasars may be at least in part due to
R10-3 AGN contamination of the far-IR emission. Future ob-
/ LCII]FI seexravmatinioendsiaffterheingcheesrinsptahteiaslpraetsioalludtiiostnriwbuiltliobneocfritthiceallintoe
L[
and continuum emission (e.g. Cicone et al. 2014). We
z<0.4 Gracia-Carpio et al. 2011 expect to observe that the dust continuum is more com-
0.5<z<5 De Looze et al. 2014 pact than the [Cii] line in the high L z >5 quasars,
FIR
10-4 55.<7<z<z<6.56 .2D eW Laonogz eet e atl .a 2l.0 210314 duetoeithermorecentrally-concentratedstarburstswith
higher dust temperatures, like local ULIRGs, or AGN
z=7.1 Venemans et al. 2012
6.0<z<6.5 This paper dust-heating. In contrast we expect the low LFIR z > 5
quasarshavestarformationspreadmoreevenlythrough-
109 1010 1011 1012 1013 1014
L (L ) out their host galaxies, with similar spatial distribution
FIR
fl of line and continuum emission.
Fig. 5.— The ratio of [Cii] to far-IR luminosity as a function
offar-IRluminosity. High-redshift(z>5)sources,mostlyquasar
4.3. The z ≈6 M −σ and M −M relationships
hostgalaxies,areidentifiedwithlargesymbols. Errorbarsareonly BH BH dyn
plottedfortheCFHQSALMAsourcestoenhancetheclarityofthe Thecombinationofblackholemassestimatesand[Cii]
figure. Thesolidanddottedbluelinesshowthebestfitpower-law
line host galaxy dynamics for these z ≈6 quasars allows
and 1σ uncertainty for the z > 5 sources. The z > 5 relation is
largelyconsistentwiththedistributionofdataatlowerredshift. ustoinvestigatetheblackhole-galaxymasscorrelation
at an early epoch in the universe. The evolution of this
the literature as due to possible factors including AGN
relationship is a critical constraint on the co-evolution
contamination of L (Sargsyan et al. 2012), high gas
FIR (or not) of galaxies and their nuclear black holes. As
fractions (Graci´a-Carpio et al. 2011) or the dustiness,
discussed in the Introduction there are reasons to be-
temperatureand/ordensityofstarformingregions(Far-
lievethatpaststudiesusingonlythemostmassiveblack
rah et al. 2013; Magdis et al. 2014). Previous observa-
holes from SDSS quasars (e.g. Wang et al. 2010) were
tions of high L z > 5 SDSS quasars (Maiolino et al.
FIR prone to a bias where one would expect the black holes
2005; Wang et al. 2013) showed a similar deficit. How-
to be relatively more massive than the galaxies (Willott
ever many 0.5<z <5 ULIRGs do not show this deficit
et al. 2005; Lauer et al. 2007), as observed. With new
and have L /L ratios comparable to low-redshift
[CII] FIR data on M ∼108M quasars we are able to test this
star-forming galaxies (Stacey et al. 2010). This is visible BH (cid:12)
hypothesis and determine any real offset from the local
in Figure 5 for the 0.5<z <5 compilation of De Looze
relationship. Additionally, most previous work in this
et al. (2014).
area has used the molecular CO line to trace the gas dy-
By adding three z > 6 quasars with 1011 < L <
FIR namics. CO is usually more centrally concentrated than
1012L(cid:12) to Figure 5 we have greatly expanded the range [Cii], so [Cii] potentially probes a larger fraction of the
of luminosities at the highest redshift. Large symbols on
total mass (although we note that Wang et al. 2013
Figure 5 identify z > 5 sources. The three z > 5 De found similar dynamical masses using CO and [Cii] for
Looze et al. (2014) sources are HLSJ091828.6+514223
their z ≈6 quasars).
at z = 5.24 (Rawle et al. 2014) , HFLS3 at z = 6.34
In addition to the CFHQS and Wang et al. (2013)
(Riechers et al. 2013) and SDSSJ1148+5251 at z =6.42
quasars we add to our study two other z > 6 quasars
(Maiolino et al. 2005). We note that these sources were observedinthe[Cii]line: SDSSJ1148+5251atz =6.42
mostly selected for followup based on high L . The
FIR and the most distant known quasar, ULASJ1120+0641
quasar ULASJ1120+0641 at z =7.1 has a more moder- atz =7.08. BoththesequasarshaveMgii-derivedblack
ate L and L /L ratio and lies in between the
FIR [CII] FIR hole masses (De Rosa et al. 2011, 2014) with very low
CFHQS and high L objects on the plot.
FIR measurement uncertainties. The black hole masses for
Although we are wary of the selection effects in Fig- all three CFHQS quasars also come from Mgii measure-
ure 5 and the as yet unknown cause for the change in
ments (Willott et al. 2010a). Some of these spectra are
L /L withL athigh-redshift,thenewdatapro-
[CII] FIR FIR ofmoderateSNRandhavesubstantialmeasurementun-
vide the opportunity to make the first measurement of
certainties on the black hole masses. To all the quasars
the slope of this relation at z > 5. We fit the 12 z > 5 with Mgii-derived black hole masses we add a 0.3 dex
sourceswithasinglepower-lawmodelofthedependence
uncertaintytothemeasurementuncertaintiestoaccount
of L on L incorporating the observational (but
FIR [CII] for the dispersion in the reverberation-mapped quasar
not systematic) uncertainties using a MCMC procedure.
calibration (Shen et al. 2008).
The best-fit relation is NoneoftheWangetal.(2013)quasarshaveMgiimea-
log L =0.59+1.27log L . (1) surements so black hole masses are estimated assuming
10 FIR 10 [CII]
that the quasars radiate at the Eddington limit, as ob-
Therefore the L[CII]/LFIR ratio line plotted in Figure 5 served for most z ≈ 6 quasars (Jiang et al. 2007; Kurk
has a logarithmic slope of (1/1.27)−1 = −0.21. The etal.2007;Willottetal.2010a;DeRosaetal.2011). The
MCMC 1σ uncertainties, based solely on the observa- dispersion in the lognormal Eddington ratio distribution
Star formation rate and dynamical mass of 108 solar mass black hole host galaxies 7
atz ≈6is0.3dex(Willottetal.2010a). Weadd0.3dex the quasars are distributed around the local relationship
uncertainty from the observed dispersion in the Edding- rather than all being offset to low σ as is commonly be-
tonratiodistributioninquadraturetothe0.3dexdueto lieved to be the case. As noted by Wang et al. (2010),
the dispersion in the reverberation-mapped quasar cali- using the method of Ho (2007b), rather than calculating
bration fora total uncertainty on the Wang et al. (2013) σ as FWHM/2.35, leads to much higher σ. Note that
quasar black hole masses of 0.45 dex. this is without adopting extreme face-on inclinations for
FirstweconsidertheM −σrelationship. Fornearby most quasars. There are still several quasars, such as
BH
galaxies σ is the velocity dispersion of the galaxy bulge. SDSSJ1148+5251andULASJ1120+0641,thathaveval-
At high-redshift bulges are less common (Cassata et al. ues of σ considerably lower than the local relation.
2011) and we do not expect the z ≈ 6 kinematics to The main result of Figure 6 is that whilst there is lit-
matchthatofapressuresupportedbulge. Withthelim- tle mean shift between the z = 0 and z ≈ 6 data, there
ited spatial resolution of current data we cannot be sure is a much larger scatter in the data at z ≈ 6, well be-
the [Cii] gas is distributed in a rotating disk, although yond the size of the error bars. This larger scatter at an
there is evidence of this for some sources (Wang et al. early epoch is expected based on dynamical evolution,
2013). Ho (2007a) discusses the relationship and cali- incoherence in AGN/starburst activity and the tighten-
bration of bulge velocity dispersion and disk circular ve- ing of the relation over time from merging (Peng 2007).
locity and concludes that although there is additional We note that our hypothesis that the bias described in
scatter one can relate molecular or atomic gas in a disk the Introduction would lead to the lower M quasars
BH
to stellar bulges. The major complication is the incli- being located on the local relation with a lower scatter
nation of the disk. For a random sample of inclinations than the high M quasars is not supported by these
BH
this can be modelled, however there is a possibility that observations. The scatter in log σ at M ≈ 108M
10 BH (cid:12)
quasars have disks oriented more often face-on, reducing is about the same as that at M >109M
BH (cid:12)
the line-of-sight velocity dispersion (Ho 2007b). We go one step further from σ to determine dynam-
We determine σ using the method of Ho (2007b), ical masses using the deconvolved [Cii] sizes. For con-
specifically setting the [Cii] line full-width at 20% equal sistency, we follow the method of Wang et al. (2013).
to1.5×theFWHMasexpectedforaGaussiansincemost The dynamical mass within the disk radius is given by
ofthelinesareapproximatelyGaussian. The[Cii]emit- M ≈1.16×105v2 DM where D is the disk diame-
dyn cir (cid:12)
ting gas is assumed to be in an inclined disk where the ter in kpc and calculated as 1.5× the deconvolved Gaus-
inclination angle, i, is determined by the ratio of minor sianspatialFWHM.Theresultingdynamicalmassesare
(a )andmajor(a )axes,i=cos−1(a /a ). The given in Table 2. We note that there is considerable
min maj min maj
circularvelocityisthereforev =0.75FWHM /sini. uncertainty on these values due to the unknown spatial
cir [CII]
For all of the quasars in this study we determine an in- and velocity structure of the gas, the marginal spatial
clination from the [Cii] data or assume an inclination if resolution and limited sensitivity that means we may be
oneorboththemajorandminoraxesareunresolved. All missing more extended gas. Due to the these uncertain-
thequasarsinWangetal.(2013)werespatiallyresolved, ties we do not place formal error bars on the dynamical
although some had quite large uncertainties on a and masses, following Wang et al. (2013). Higher resolution
min
a . We adopt the inclination angles from their paper. data in the future are required to confirm the derived
maj
For the [Cii] emission of J0210−0456, a = 0(cid:48).(cid:48)52± masses.
maj
0(cid:48).(cid:48)25 (2.9kpc) with i = 64◦ (Wi13). For J0055+0146, SDSSJ1148+5251hasbeenextensivelystudiedin[Cii]
a = 0(cid:48).(cid:48)50±0(cid:48).(cid:48)14 (2.9kpc) with i = 69◦ (Section 3, (Maiolino et al. 2005; Walter et al. 2009; Maiolino et al.
maj
assuming no bias from missing red wing data). The 2012; Cicone et al. 2014). The highest resolution ob-
[Cii] emission of J2229+1457 is only marginally spa- servations by Walter et al. (2009) revealed a very com-
tially resolved (0(cid:48).(cid:48)85 versus beam size of 0(cid:48).(cid:48)76) and we pact circumnuclear starburst with radius 0.75kpc and
estimate an intrinsic FWHM of ≈ 0(cid:48).(cid:48)4 (2.4kpc). An in- FWHM =287±28kms−1. For an assumed inclina-
[CII]
clination of i = 55◦ is assumed as this is the median tion of i = 55◦ this gives M = 1.8×1010M . For
dyn (cid:12)
inclination angle for the resolved sources in this paper comparisonWalteretal.(2004)determinedadynamical
and Wang et al. (2013). Neither SDSSJ1148+5251 nor mass from CO emission in this quasar of 5.5×1010M
(cid:12)
ULASJ1120+0641 have published inclination angles, so within a larger radius of 2.5kpc, the larger radius being
we also assume i = 55◦ for both of them. We adopt themaindifferencebetweentheresults. Recentobserva-
FWHM[CII] = 287 ± 28kms−1 for SDSSJ1148+5251 tions have shown more complex [Cii] emission including
(Walter et al. 2009) and FWHM = 235±35kms−1 evidence for gas extended over tens of kpc and at high
[CII]
for ULASJ1120+0641 (Venemans et al. 2012) . Values velocities indicative of outflow (Maiolino et al. 2012; Ci-
of black hole masses and σ for this sample are provided cone et al. 2014). We adopt Mdyn = 1.8×1010M(cid:12) for
in Table 2. SDSSJ1148+5251 noting that the true value could be
Figure 6 shows the M −σ relationship for the z ≈6 several times larger. The most distant known quasar,
BH
quasar sample. Uncertainties on black hole masses in- ULASJ1120+0641 at z = 7.08, has been well detected
clude the scatter in the calibration as described previ- in [Cii], although not yet spatially resolved (Venemans
ously. Uncertainties in σ include FWHM measurement et al. 2012). Based on the published FWHM[CII] =
uncertainty plus a 10% uncertainty for the conversion 235±35kms−1 and assuming a spatial FWHM of 3kpc
fromFWHMtov and15%fortheconversionfromv (similar to the other quasars resolved by ALMA) and
cir cir
toσasseeninthesampleofHo(2007a). Theblacklineis i=55◦ we determine M =2.4×1010M .
dyn (cid:12)
the local correlation of Kormendy & Ho (2013) with the In Figure 7 we plot black hole mass versus galaxy dy-
graybandthe±1σscatter. Thefirstthingtonoteisthat namical mass for the most distant known quasars. The
8 Willott et al.
TABLE 2
Mass determinations for z>5.7 quasars
Name z[CII] MBH (M(cid:12))a σb Mdyn (M(cid:12)) Refs.c
CFHQSJ0055+0146 6.0060±0.0008 (2.4+2.6)×108 207±45 4.2×1010 1,2
−1.4
CFHQSJ0210−0456 6.4323±0.0005 (0.8+1.0)×108 98±20 1.3×1010 1,2,3
−0.6
CFHQSJ2229+1457 6.1517±0.0005 (1.2+1.4)×108 241±51 4.4×1010 1,2
−0.8
SDSSJ0129−0035 5.7787±0.0001 (1.7+3.1)×108 112±21 1.3×1010 4
−1.1
SDSSJ1044−0125 5.7847±0.0007 (1.1+1.9)×1010 291±76 —d 4
−0.7
SDSSJ1148+5251 6.4189±0.0006 (4.9+4.9)×109 186±38 1.8×1010 5,6,7
−2.5
SDSSJ2054−0005 6.0391±0.0001 (0.9+1.6)×109 364±67 7.2×1010 4
−0.6
SDSSJ2310+1855 6.0031±0.0002 (2.8+5.1)×109 325±61 9.6×1010 4
−1.8
ULASJ1120+0641 7.0842±0.0004 (2.4+2.4)×109 144±34 2.4×1010 8,9
−1.2
ULASJ1319+0950 6.1330±0.0007 (2.1+3.8)×109 381±91 12.5×1010 4
−1.4
Notes.—
a DerivedfromMgiiλ2799observationsifpossible,elsefromEddingtonluminosityassumption. Uncertaintiesincludeobservationalerrors
plussystematicsbasedoncalibrations.
bDerivedfromGaussianFWHMfitto[Cii]spectrumusingmethodofHo(2007b)includinganinclinationcorrection(seetextforindividual
inclinationsassumed). Uncertaintiesincludeobservationalerrorsplussystematicsbasedoncalibrations.
c References: (1) This paper, (2) Willott et al. (2010a), (3) Willott et al. (2013), (4) Wang et al. (2013), (5) Maiolino et al. (2005), (6)
Walteretal.(2009),(7)DeRosaetal.(2011),(8)Venemansetal.(2012),(9)DeRosaetal.(2014).
d ThisquasardoesnothaveadynamicalmasscalculationinWangetal.(2013)duetothedifferenceinthe[Cii]andCOlineprofiles.
1011 1011
z=0 Kormendy & Ho 2013 z=0 Kormendy & Ho 2013
5.7<z<6.2 Wang et al. 2013 5.7<z<6.2 Wang et al. 2013
6.0<z<6.5 This paper 6.0<z<6.5 This paper
z=7.1 Venemans et al. 2012 z=7.1 Venemans et al. 2012
z=6.4 Walter et al. 2009 z=6.4 Walter et al. 2009
1010 1010
)fl )fl
M M
( 109 ( 109
H H
B B
M M
108 108
107 107
100 200 400 800 1010 1011 1012
σ(km/s) M (M )
dyn
fl
Fig. 6.— Black hole mass versus velocity dispersion calculated Fig. 7.— Black hole mass versus host galaxy dynamical mass
from the [Cii] line using the method of Ho (2007b) for z ≈ 6 for z ≈ 6 quasars. Symbols as for Figure 6. The black line with
quasars. QuasarsfromtheCFHQSareshownasbluesquaresand grayshadingisthelocalcorrelation±1σ scatterfromtheworkof
theothersymbolsshowquasarsfromtheSDSSandULASsurveys. Kormendy & Ho (2013) equating M to M . The CFHQS
dyn bulge
Theblacklinewithgrayshadingisthelocalcorrelation±1σscatter quasarslieonthelocalrelationshipanddonotshowthelargeoffset
ofblackholemassandbulgevelocitydispersion(Kormendy&Ho displayedbythemostmassiveblackholes. UncertaintiesinM
dyn
2013). The z ≈ 6 quasars are distributed around the the local havenotbeencalculatedduetothereasonsgiveninthetext.
relationship,butwithamuchlargerscatterandsomequasarswith
significantlylowerσ fortheirMBH. pseudobulges, galaxies with uncertain M and ongoing
BH
black line and gray shading represent the local correla- mergers.
tion of M with bulge mass M (Kormendy & Ho The position of the high-z data with respect to low
BH bulge
2013). In the absence of gas accretion and mergers the redshift is fairly similar to Figure 6, not surprising be-
present stellar bulge mass represents the sum of the gas cause v derived from the [Cii] velocity FWHM is a
cir
and stellar mass at high-redshift, so it is a good com- majorfactorinbothσ andM . Thepointsareshifted
dyn
parison for the dynamical mass within the central few somewhatfurtherfromthelocalbulgemassthanforthe
kpc. Kormendy & Ho (2013) note that their correlation local velocity dispersion. This shift is due to the smaller
(their equation 10) gives a black hole to bulge mass ra- size of galaxies at high-redshift, as the size is the only
tio of 0.5% at M = 1011M that is 2 to 4 times term in the derivation of dynamical mass not in σ. We
bulge (cid:12)
higher than previous estimates due to the omission of note the much greater dynamic range in black hole mass
Star formation rate and dynamical mass of 108 solar mass black hole host galaxies 9
(2 dex) than in dynamical mass (1 dex) in our sample. tion with similar slope and normalization to locally, but
Thisislikelyduemoretoourselectionoverawiderange with much greater scatter. Similar results are obtained
of quasar luminosity than to a non-linear relationship fortheM −M relationwithasomewhathighernor-
BH dyn
between these quantities at z =6. malization at z = 6 and a higher scatter at high M .
BH
AllthreeoftheCFHQSquasarsliewithinthelocal1σ As discussed in
scatterandtheoneM ∼108M quasarinWangetal. Combining our results on the relatively low L for
BH (cid:12) FIR
(2013) is only a factor of 4 greater than the local rela- M ∼ 108M z ≈ 6 quasars with their location on
BH (cid:12)
tionship. IncontrasttheM >109M quasarstendto the M −σ relation leads to something of a paradox.
BH (cid:12) BH
showalargerscatterandlargeroffsetabovethelocalre- The fact these quasars lie on the local M −σ relation
BH
lationship as previously found (Walter et al. 2004; Wang suggests that their host galaxies have undergone consid-
etal.2010;Venemansetal.2012;Wangetal.2013). We erable evolution to acquire such a high dynamical mass.
cautionthatthereareconsiderableuncertaintiesinsome So why is it that this mass accumulation is not leading
of these measurements as already discussed, but in dy- to a high star formation rate? As discussed in Wi13,
namical mass the results look more like we would expect simulations such as those of Khandai et al. (2012) and
based on the quasar selection bias effect. Lapi et al. (2014) predict that such low ratios of SFR to
black hole accretion occur after episodes of strong feed-
5. CONCLUSIONS backthatinhibitsstarformationthroughoutquasarhost
During ALMA Early Science cycles 0 and 1 we have galaxies. Another possibility mentioned in Section 4.1 is
observedacompletesampleoffourz >6moderatelumi- that LFIR fails to trace star formation so effectively in
nosityCFHQSquasarswithblackholemasses∼108M . these high-redshift galaxies, due to lower dust content
(cid:12)
Three of the four are detected in both far-IR continuum (e.g. Ouchi et al. 2013). Note that using L as a star
[CII]
and the [Cii] emission line. The far-IR luminosity is formation rate tracer instead of LFIR, would give higher
foundtobesubstantiallylowerthanthatofsimilarlumi- SFR by a factor of three for one of the CFHQS quasars.
nosity quasars at 1<z <3. Assuming that far-IR lumi- Higher resolution follow-up [Cii] observations of these
nosity traces star formation equally effectively at these quasars are critical to measure more accurately the dis-
redshifts this implies that at z ≈ 6 quasars are grow- tribution and kinematics of the gas used as a dynamical
ingtheirblackholesmorerapidlythantheirstellarmass tracer in order to reliably determine the location and
comparedtoatthepeakofthequasarepoch(1<z <3). scatter of the correlations between black holes and their
The ratios of [CII] to far-IR luminosities for the host galaxies at high-redshift.
CFHQS quasars lie in the range 0.001 to 0.01, similar
to that of low-redshift galaxies at the same far-IR lumi-
nosity. This suggests a similar mode of star-formation Thanks to staff at the North America ALMA
spread throughout the host galaxy (rather than in dense Regional Center for processing the ALMA data.
circumnuclear starburst regions that have lower values Thanks to Melanie Veale for useful discussion and
for this ratio in local ULIRGs). Combining with pre- providing her models in electronic form. This
vious z > 5.7 quasar data at higher L we find that paper makes use of the following ALMA data:
FIR
the far-IR luminosity dependence of the [Cii]/FIR ra- ADS/JAO.ALMA#2012.1.00676.S. ALMA is a partner-
tio has a shallow negative slope, possibly due in part to ship of ESO (representing its member states), NSF
an increase in L due to quasar-heated dust in some (USA) and NINS (Japan), together with NRC (Canada)
FIR
optically-luminous high-z quasars. and NSC and ASIAA (Taiwan), in cooperation with the
The three CFHQS quasars well-detected in the [Cii] RepublicofChile. TheJointALMAObservatoryisoper-
emissionlineallowthisatomicgastobeusedasatracer atedbyESO,AUI/NRAOandNAOJ.TheNationalRa-
of the host galaxy dynamics. Combining with published dio Astronomy Observatory is a facility of the National
data on higher black hole mass quasars we have investi- Science Foundation operated under cooperative agree-
gated the M −σ and M −M relations at z ≈6. ment by Associated Universities, Inc.
BH BH dyn
We show that the z =6 quasars display a M −σ rela- Facility: ALMA.
BH
REFERENCES
Bonfield, D. G., Jarvis, M. J., Hardcastle, M. J., et al. 2011, Fine,S.,Croom,S.M.,Miller,L.,etal.2006,MNRAS,373,613
MNRAS,416,13 Fisher, D. B., Bolatto, A. D., Herrera-Camus, R., et al. 2014,
Bouwens,R.J.,Illingworth,G.D.,Oesch,P.A.,etal.2014,ArXiv Nature,505,186
e-prints,arXiv:1403.4295 Freudling, W., Corbin, M. R., & Korista, K. T. 2003, ApJ, 587,
Carilli,C.L.,&Walter,F.2013,ARA&A,51,105 L67
Carilli,C.L.,Wang,R.,Fan,X.,etal.2010,ApJ,714,834 Graci´a-Carpio, J., Sturm, E., Hailey-Dunsheath, S., et al. 2011,
Cassata,P.,Giavalisco,M.,Guo,Y.,etal.2011,ApJ,743,96 ApJ,728,L7
Chary,R.,&Elbaz,D.2001,ApJ,556,562 Hickox,R.C.,Mullaney,J.R.,Alexander,D.M.,etal.2014,ApJ,
Cicone,C.,Maiolino,R.,Gallerani,S.,etal.2014,ArXive-prints, 782,9
arXiv:1409.4418 Ho,L.C.2007a,ApJ,668,94
Cisternas, M., Jahnke, K., Bongiorno, A., et al. 2011, ApJ, 741, —.2007b,ApJ,669,821
L11 Jiang,L.,Fan,X.,Vestergaard,M.,etal.2007,AJ,134,1150
DeLooze,I.,Cormier,D.,Lebouteiller,V.,etal.2014,A&A,568, Khandai,N.,Feng,Y.,DeGraf,C.,DiMatteo,T.,&Croft,R.A.C.
A62 2012,MNRAS,423,2397
DeRosa,G.,Decarli,R.,Walter,F.,etal.2011,ApJ,739,56 Kormendy,J.,&Ho,L.C.2013,ARA&A,51,511
De Rosa, G., Venemans, B. P., Decarli, R., et al. 2014, ApJ, 790, Kurk,J.D.,Walter,F.,Fan,X.,etal.2007,ApJ,669,32
145 Lapi,A.,Raimundo,S.,Aversa,R.,etal.2014,ApJ,782,69
Farrah, D., Lebouteiller, V., Spoon, H. W. W., et al. 2013, ApJ, Lauer, T. R., Tremaine, S., Richstone, D., & Faber, S. M. 2007,
776,38 ApJ,670,249
10 Willott et al.
Magdis,G.E.,Rigopoulou,D.,Hopwood,R.,etal.2014,ApJ,796, Serjeant,S.,Bertoldi,F.,Blain,A.W.,etal.2010,A&A,518,L7
63 Shen, Y., Greene, J. E., Strauss, M. A., Richards, G. T., &
Maiolino,R.,Cox,P.,Caselli,P.,etal.2005,A&A,440,L51 Schneider,D.P.2008,ApJ,680,169
Maiolino,R.,Gallerani,S.,Neri,R.,etal.2012,MNRAS,425,L66 Shen,Y.,&Kelly,B.C.2012,ApJ,746,169
McGreer,I.D.,Jiang,L.,Fan,X.,etal.2013,ApJ,768,105 Sijacki,D.,Vogelsberger,M.,Genel,S.,etal.2014,ArXive-prints,
Mechtley, M., Windhorst, R. A., Ryan, R. E., et al. 2012, ApJ, arXiv:1408.6842
756,L38 Stacey, G. J., Hailey-Dunsheath, S., Ferkinhoff, C., et al. 2010,
Merloni, A., Bongiorno, A., Bolzonella, M., et al. 2010, ApJ, 708, ApJ,724,957
137 Tan,Q.,Daddi,E.,Sargent,M.,etal.2013,ApJ,776,L24
Omont,A.,Willott,C.J.,Beelen,A.,etal.2013,A&A,552,A43 Targett,T.A.,Dunlop,J.S.,&McLure,R.J.2012,MNRAS,420,
Ota,K.,Walter,F.,Ohta,K.,etal.2014,ApJ,792,34 3621
Ouchi,M.,Ellis,R.,Ono,Y.,etal.2013,ApJ,778,102 Trakhtenbrot,B.,Netzer,H.,Lira,P.,&Shemmer,O.2011,ApJ,
Park, D., Woo, J.-H., Bennert, V. N., et al. 2014, ArXiv e-prints, 730,7
arXiv:1411.6334 Veale,M.,White,M.,&Conroy,C.2014,MNRAS,445,1144
Peng,C.Y.2007,ApJ,671,1098 Venemans, B. P., McMahon, R. G., Walter, F., et al. 2012, ApJ,
PlanckCollaboration.2014,A&A,571,A16 751,L25
Rawle, T. D., Egami, E., Bussmann, R. S., et al. 2014, ApJ, 783, Walter,F.,Carilli,C.,Bertoldi,F.,etal.2004,ApJ,615,L17
59 Walter,F.,Riechers,D.,Cox,P.,etal.2009,Nature,457,699
Richards,G.T.,VandenBerk,D.E.,Reichard,T.A.,etal.2002, Wandel,A.,Peterson,B.M.,&Malkan,M.A.1999,ApJ,526,579
AJ,124,1 Wang,R.,Carilli,C.L.,Neri,R.,etal.2010,ApJ,714,699
Riechers, D. A., Bradford, C. M., Clements, D. L., et al. 2013, Wang,R.,Wagg,J.,Carilli,C.L.,etal.2011,AJ,142,101
Nature,496,329 —.2013,ApJ,773,44
Rosario,D.J.,Santini,P.,Lutz,D.,etal.2012,A&A,545,A45 Willott,C.J.,Omont,A.,&Bergeron,J.2013,ApJ,770,13
Rosario,D.J.,Trakhtenbrot,B.,Lutz,D.,etal.2013,A&A,560, Willott,C.J.,Percival,W.J.,McLure,R.J.,etal.2005,ApJ,626,
A72 657
Sargsyan, L., Samsonyan, A., Lebouteiller, V., et al. 2014, ApJ, Willott,C.J.,Albert,L.,Arzoumanian,D.,etal.2010a,AJ,140,
790,15 546
Sargsyan,L.,Lebouteiller,V.,Weedman,D.,etal.2012,ApJ,755, Willott,C.J.,Delorme,P.,Reyl´e,C.,etal.2010b,AJ,139,906
171
Schulze,A.,&Wisotzki,L.2014,MNRAS,438,3422