Redshift 6.4 host galaxies of 10^8 solar mass black holes: low star formation rate and dynamical mass
DDraft version October 30, 2018
Preprint typeset using L A TEX style emulateapj v. 12/01/06
REDSHIFT 6.4 HOST GALAXIES OF 10 SOLAR MASS BLACK HOLES:LOW STAR FORMATION RATE AND DYNAMICAL MASS
Chris J. Willott
Herzberg Institute of Astrophysics, National Research Council, 5071 West Saanich Rd, Victoria, BC V9E 2E7, Canada
Alain Omont and Jacqueline Bergeron
UPMC Univ Paris 06 and CNRS, UMR7095, Institut d’Astrophysique de Paris, F-75014, Paris, France
Draft version October 30, 2018
ABSTRACTWe present ALMA observations of rest-frame far-infrared continuum and [C ii ] line emission in two z = 6 . ≈ M (cid:12) . CFHQS J0210-0456 is detected in the continuumwith a 1.2 mm flux of 120 ± µ Jy, whereas CFHQS J2329-0301 is undetected at a similar noise level.J2329-0301 has a star formation rate limit of <
40 M (cid:12) yr − , considerably below the typical value atall redshifts for this bolometric luminosity. By comparison with hydro simulations, we speculate thatthis quasar is observed at a relatively rare phase where quasar feedback has effectively shut down starformation in the host galaxy. [C ii ] emission is also detected only in J0210-0456. The ratio of [C ii ] tofar-infrared luminosity is similar to that of low redshift galaxies of comparable luminosity, suggestingthe previous finding of an offset in the relationships between this ratio and far-infrared luminosityat low- and high-redshift may be partially due to a selection effect due to the limited sensitivity ofprevious continuum data. The [C ii ] line of J0210-0456 is relatively narrow (FWHM = 189 ±
18 km s − ),indicating a dynamical mass substantially lower than expected from the local black hole – velocitydispersion correlation. The [C ii ] line is marginally resolved at 0 . (cid:48)(cid:48) . (cid:48)(cid:48) − across a scale of6 kpc, possibly due to rotation of a galaxy-wide disk. These observations are consistent with the ideathat stellar mass growth lags black hole accretion for quasars at this epoch with respect to more recenttimes. Subject headings: cosmology: observations — galaxies: evolution — galaxies: high-redshift — quasars:general INTRODUCTION
The peak of global star formation occurred about 10billion years ago at a redshift of z ≈ ii ], probe the interstellar medium and the outer partsof star-forming regions. It has been recognized that the Electronic address: [email protected] [C ii ] line will likely become the most useful line for study-ing very high redshift galaxies with the Atacama LargeMillimeter Array (ALMA; Walter & Carilli 2008). In-deed early ALMA observations already show detectionsof [C ii ] in normal star-forming galaxies at z > ii ] has also beendetected in a few z > ∼ a r X i v : . [ a s t r o - ph . C O ] A p r Willott et al.galaxy dynamical masses to measure the ratio of blackhole to galaxy mass in the early universe (Walter et al.2004). Observations in the rest-frame UV or optical arehampered by the overwhelming brightness of the quasarpoint-source (e.g. Mechtley et al. 2012). Wang et al.(2010) showed that CO line widths of the most opticallyluminous z ≈ z ≈ z = 0 to z ≈ z = 6 . M ≥ −
25) and black hole masses of ∼ M (cid:12) .Section 2 details the new observations. The results arepresented in Section 3. Section 4 contains a discus-sion of the results. Cosmological parameters of H =70 km s − Mpc − , Ω M = 0 .
27 and Ω Λ = 0 .
73 (Komatsuet al. 2011) are assumed throughout. OBSERVATIONS
CFHQS J021013-045620 (hereafter J0210-0456) andCFHQS J232908-030158 (hereafter J2329-0301) were ob-served with ALMA between June and August 2012 in
Early Science project 2011.0.00243.S. The number of12 m diameter antennae in use ranged from 17 to 24 witha typical longest baseline of 400 m. Observations of thescience targets were interleaved with nearby phase cal-ibrators, J0217+017 and J2323-032. Uranus was usedas the amplitude calibrator and 3C446 as the bandpasscalibrator. Total on-source integration times were 8000 sfor J0210-0456 and 8500 s for J2329-0301.The band 6 (1.3 mm) receivers were set up so thatone of the four basebands (each of width 1.875 GHz) wascentred on the expected location of the redshifted [C ii ]transition ( ν rest =1900.5369 GHz). The redshifts adopted Fig. 1.—
ALMA 1.2 mm continuum images generated from thethree line-free basebands for each of the two quasar fields. Thegreyscale ranges from − σ (black) to +3 σ (white) where σ =35 , µ Jy beam − for J0210-0456 (left) and J2329-0301 (right), re-spectively. Red circles show the expected source positions (circlesdo not indicate the positional uncertainty). There is a 3 . σ detec-tion for J0210-0456 and no detection for J2329-0301. The strongestsource in the field of J2329-0301 is marked with a blue circle andlabelled ’b’. It is co-incident with a blue galaxy in the opticalimages of Willott et al. (2007). were those of the low-ionization broad Mg ii lines of thequasars measured by Willott et al. (2010b). Previousstudies of high-redshift quasars have shown relativelysmall offsets (1 σ dispersion 270 km s − ) between Mg ii and the systemic redshift (Richards et al. 2002). Theremaining three spectral windows were placed nearby tosample the 1.2 mm continuum. Each baseband is sam-pled by 120 channels of width 15.625 MHz (equivalent to ≈
18 km s − ).Data processing was performed by staff at the NorthAmerican ALMA Regional Center using the CASA soft-ware package . The three line-free spectral windowswere combined to generate 1.2 mm continuum images.Both the continuum maps and spectral line datacubeswere spatially sampled with 0 . (cid:48)(cid:48) . (cid:48)(cid:48)
77 by 0 . (cid:48)(cid:48)
52 for J0210-0456 and 0 . (cid:48)(cid:48)
73 by 0 . (cid:48)(cid:48) − for J0210-0456 and0.23 mJy beam − for J2329-0301. RESULTS
Far-infrared luminosity
The 1.2 mm continuum luminosity of z = 6 . µ m radiation, on the Rayleigh-Jeans tail side of the typical star-forming galaxy dustspectral energy distribution (SED; Lagache et al. 2005).This makes it an excellent proxy for the total far-infrared luminosity ( L FIR ; integrated luminosity over42 . − . µ m) which is a reliable tracer of the starformation rate due to dust heated by young stars. Inthe most ultraviolet-luminous quasars (such as those at z ∼ L FIR from dust heated by the AGN (Wang etal. 2008). The CFHQS quasars are an order of magni-tude less UV-luminous than SDSS quasars and thereforeshould have a correspondingly lower contribution fromAGN-heated dust, allowing continuum observations toprobe lower star-formation rates.The ALMA 1.2 mm continuum images generated fromthe three spectral windows that did not include the[C ii ] line were analyzed to determine their flux-densities.These images are shown in Figure 1 where the expected http://casa.nrao.edu edshift 6.4 host galaxies of 10 solar mass black holes 3 Fig. 2.—
Far-infrared luminosity versus AGN bolometric luminosity for z ≈ L Bol for quasars from the H-ATLAS survey andother far-IR/mm data (Serjeant et al. 2010). The dotted line is a fit to high-redshift (2 < z <
7) stack averages (Wang et al. 2011a). Thedashed line converts L Bol to black hole accretion rate and L FIR to star formation rate such that these grow in parallel to match the local M BH /M stellar relationship (Tundo et al. 2007). locations of the quasars are identified by red circles.J0210-0456 is detected at 3 . σ with f . = 120 ± µ Jy. At this significance level it is not possible to de-termine whether the source is spatially resolved. J2329-0301 is undetected with no hint of positive flux at thequasar location. The most significant continuum sourcein the field (7 (cid:48)(cid:48) north of the quasar, labeled ’b’) is identi-fied as a blue galaxy at much lower redshift in the opticalimaging of Willott et al. (2007).This continuum flux-density was converted to a far-infrared luminosity assuming a typical SED for high-redshift star-forming galaxies. To make meaningful com-parison with previous results (in particular Wang et al.2008; 2011a, Omont et al. 2013) we adopt a greybodyspectrum with dust temperature, T d = 47 K and emissiv-ity index, β = 1 .
6. We note that our faint sources havemuch lower mm fluxes than the typical sources used todetermine these parameters. If our sources instead havedust temperature closer to that of nearby luminous in-frared galaxies (LIRGs, 10 − L (cid:12) , T d ≈
33 K, U et al. 2012) then the values of L FIR would be 3 × lower.For the remainder of this paper, uncertainties on L FIR (and inferred SFR) only include the flux measurementuncertainties, not that of the dust temperature.The far-IR luminosity of J0210-0456 is (2 . ± . × L (cid:12) . J2329-0301 is undetected with L FIR < . × L (cid:12) (3 σ limit). We note the incredible sensitivityof these early ALMA observations that reach the lowerend of the LIRG classification in the early universe at z = 6 . L FIR andthe quasar bolometric luminosity L Bol for high-redshiftquasars. L Bol in this case is for the quasar component ofthe galaxy and assumes a typical bolometric correctionfrom the rest-frame UV luminosity at 1450 ˚A of a factorof 4.4 (Richards et al. 2006). L Bol does not include anyexcess FIR luminosity above that of the typical quasar.It is still a matter of debate as to how much of the typicalquasar far-IR emission is due to dust heated by the AGN,compared to dust heated by a starburst (Haas et al. 2003; Willott et al.Hao et al. 2005; Netzer et al. 2007; Lutz et al. 2010).Because most high-redshift quasars have not been de-tected in mm continuum with the sensitivity level of pre-vious studies, the relationship between L FIR and L Bol has been based on stacking of sub-samples with differ-ent bolometric luminosity ranges. Wang et al. (2011a)showed that the stacks based on several samples at2 < z < L FIR ∝ L . .Omont et al. (2013) found that the stacked averagefrom 1.2 mm MAMBO observations of CFHQS z ≈ L FIR at a given L Bol are expected to have asignificant fraction of their L FIR due to quasar-heateddust (Netzer et al. 2007).Figure 2 shows previously published data forindividually-detected z ≈ L FIR and L Bol found by Wang et al. (2011a) for stack averagesof quasars at 2 < z <
7. Note that both the stackedaverages and relationship from Wang et al. 2011a havebeen renormalized according to the bolometric correc-tion adopted here (see Omont et al. 2013 for more de-tails). Gray crosses are a complete sample of low-redshift( z < .
5) optically-selected Palomar-Green (PG) quasars(Hao et al. 2005). L FIR for PG quasars has been es-timated as 2 × the luminosity at 60 µ m (Lawrence etal. 1989). Note that many of the highest luminosity(most distant) quasars in the PG sample are undetectedat 60 µ m and only have upper limits on L FIR . As notedby Wang et al. (2011a), the z ≈ Herschel imaging of quasarsin the H-ATLAS survey plus supplementary publishedIR and mm data to determine the average quasar far-infrared luminosity as a function of both redshift andquasar luminosity. Their data (for all bins containing10 or more quasars) is shown in Figure 2 using an I band bolometric correction of 12.0 (Richards et al. 2006)and L FIR = 1 . × the luminosity at 100 µ m. Thesedata show a similar correlation of the two luminosities asfound for the PG and z ≈ L FIR andredshift up to z ≈ z = 6 data of Wang et al. (2011a) and Omont et al.(2013). A positive correlation between L FIR and redshiftup to z ≈ L FIR to be a factor of ≈ L FIR at least a factor of10 lower than the stacked average from the full CFHQSsample (Omont et al. 2013) and substantially below theH-ATLAS averages.The far-infrared luminosity can be used to derive the
Fig. 3.—
Continuum-subtracted [C ii ] line emission forCFHQS J0210-0456 integrated over 15 channels containing the[C ii ] line. Contours are at [ − , , , × σ where σ =0 .
03 Jy km s − beam − . SFR assuming the relation in Kennicutt (1998) with aSalpeter (1955) initial mass function (IMF). For J0210-0456, SFR = 48 M (cid:12) yr − and for J2329-0301, SFR <
40 M (cid:12) yr − (3 σ limit). For both these objects, the as-sumption in deriving SFR is that there is no contribu-tion to L FIR from quasar-heated dust. These quasars lieclose to the lower range of L FIR where it has been sug-gested that the majority of the cool dust is heated by thequasar (Netzer et al. 2007). If this is the case then theSFR would be even lower. The very low SFR impliedfor the host galaxy of J2329-0301 is surprising given thatit has a 2 . × M (cid:12) black hole accreting at the Ed-dington rate (Willott et al. 2010b) and a very luminous,spatially-extended Ly α halo (Goto et al. 2009; Willottet al. 2011). [C ii ] luminosity The datacubes of the spectral windows containing theexpected [C ii ] emission lines for the two quasars wereinspected for line emission. A line was easily detected forJ0210-0456, but not for J2329-0301. Figure 3 shows theimage of [C ii ] emission for J0210-0456 integrated overthe 15 channels (each of width 15.625 MHz) that showline emission. Continuum emission has been subtractedoff using the continuum image of Figure 1. The sourceis elongated east-west, although note this is close to themajor axis of the elongated beam. The spatial structurewill be discussed further in the following section. Noother [C ii ] emitters at the same redshift are seen in thefield.The spectrum of J0210-0456 is plotted in Figure 4. The[C ii ] line is offset from the broad ultraviolet Mg ii emis-sion line by 230 km s − . Note that the rms observationaluncertainty on the Mg ii redshift is 160 km s − , so theredshifts of [C ii ] and Mg ii are consistent. We take theredshift of z [CII] = 6 . ± . solar mass black holes 5 Fig. 4.— [C ii ] spectrum for CFHQS J0210-0456 overlaid withbest fit Gaussian plus continuum model (blue). The red squarewith error bar is the continuum flux measured from the three line-free basebands. The arrow marked Mg ii shows the redshift mea-sured from the quasar broad line region. than Mg ii and is associated with star formation in thehost galaxy rather than gas in the circum-quasar envi-ronment. A simple Gaussian plus constant fit was madeto the observed spectrum. A Gaussian with FWHM= 189 ±
18 km s − provides a good fit to the line pro-file. The best-fit constant is positive showing a non-zerocontinuum level that is consistent with the continuumflux-density of 120 µ Jy measured from the three line-freebasebands in Section 3.1.The improved systemic redshift for J0210-0456 allowsan improvement in the determination of the size of thehighly-ionized near-zone. The size of the region ionizedby the quasar depends upon several factors including theneutral hydrogen fraction of the intergalactic mediumwhen the quasar first turned on (Madau & Rees 2000;Cen & Haiman 2000) and therefore can be used to probecosmic reionization. Willott et al. (2010b) used the Mg ii redshift of this quasar to determine a near-zone size of 1.7proper Mpc, which is lower than any other z ≈ z [CII] = 6 . R ∝ L / makesthe size only slightly lower than the typical size for moreluminous z > . ii ] emission line. There is weak positiveflux at the Mg ii redshift that may correspond to realemission, but it is very uncertain so we assume here anon-detection.The [C ii ] line flux of J0210-0456 was determined byintegrating over the channels containing the line and sub-tracting off the continuum component. This line flux wasthen converted to a line luminosity at the measured red-shift. For J2329-0301 a 3 σ upper limit for the [C ii ] fluxand luminosity was determined by assuming a spatiallyunresolved Gaussian with FWHM = 300 km s − . Thisline width is somewhat broader than that observed forJ0210-0456 but narrower than CO line widths for SDSS z ≈ Fig. 5.— [C ii ] spectrum for CFHQS J2329-0301. There is noconvincing detection of the line for this quasar. The red squarewith downward arrow is the limit on the continuum flux from thethree line-free basebands. TABLE 1Millimeter data for CFHQS z ≈ . quasars CFHQS J0210-0456 CFHQS J2329-0301 M BH (8 . +5 . − . ) × M (cid:12) a (2 . +0 . − . ) × M (cid:12) a z MgII . ± . a . ± . a z [CII] . ± . − FWHM [CII] ±
18 km s − − I [CII] . ± .
037 Jy km s − < .
10 Jy km s − L [CII] (3 . ± . × L (cid:12) < . × L (cid:12) b I CO(2 − < .
014 Jy km s − − L CO(2 − < . × L (cid:12) c − f . ± µ Jy < µ Jy d L FIR (2 . ± . × L (cid:12) < . × L (cid:12) d SFR 48 ±
14 M (cid:12) yr − <
40 M (cid:12) yr − L [CII] /L FIR (1 . ± . × − − L CO(1 − /L FIR < . × − − Notes. — a Derived from Mg ii λ b σ upper limit assuming spatially unresolved and line widthFWHM=300 km s − . c σ upper limit from observations in Wang et al. (2011b) assum-ing spatially unresolved and FWZI=300 km s − . d σ upper limit assuming spatially unresolved. data are quoted in Table 1.The [CII] line is primarily produced in photo-dissociation regions and is strongly dependent on the in-terstellar radiation field (Stacey et al. 1991). The ratio L [CII] /L FIR has an inverse correlation on the radiationfield strength and has been widely studied at lower red-shift. It has been found that the ratio has an inverse cor-relation with L FIR (Luhman et al. 2003). Graci´a-Carpioet al. (2011) found this inverse correlation is even tighterif one normalises the far-IR luminosity by the moleculargas mass M H . They attributed this to L FIR /M H be-ing more closely related to the physical properties of theclouds such as density and temperature.Figure 6 shows L [CII] /L FIR against L FIR for a com-pilation of low ( z < .
4) and high (1 < z <
5) red-shift galaxies (Graci´a-Carpio et al. 2011 and in prep.).The horizontal offset between the low and high redshiftsources was attributed by Graci´a-Carpio et al. (2011) tothe higher molecular gas mass (for a given L FIR ) at highredshift (e.g. Tacconi et al. 2010). Also plotted on Fig- Willott et al.
Fig. 6.—
Ratio of [C ii ] to far-infrared luminosity versus far-infrared luminosity. Open circles show low-redshift ( z < .
4) andfilled circles high-redshift (1 < z <
5) galaxies from the collec-tion by Garci´a-Carpio et al. (2011 and in prep.). Detections ofquasars at z > L FIR sensitivity of ALMA in this
Early Science observation with a mod-est integration time. ure 6 are data for z > L FIR and falls along the sequence of2 < z < L FIR and fallwithin the region occupied by low-redshift galaxies. Inthe interpretation of the offset being due to higher molec-ular gas mass at high-redshift, this would suggest thatnot all high-redshift quasars exist in star-forming hostswith higher gas masses than at low redshift. The hor-izontal offset that is so striking in Figure 6 is at leastpartially due to a selection effect where previous facili-ties did not have the sensitivity to detect more moderate L FIR at high redshifts and only ultraluminous contin-uum sources were followed up with [C ii ] observations. Itis likely there is a large population of hitherto undetectedhigh-redshift galaxies with properties like J0210-0456.Wang et al. (2011b) reported Very Large Array ob-servations aimed at detecting the CO (2 −
1) emissionfrom J0210-0456. The object was not detected and aline flux upper limit assuming a full-width zero-intensityof 800 km s − was reported. We have recalculated theline flux limit for the same width as the observed [C ii ]line (FWZI = 300 km s − ). The 3 σ upper limit on theline flux is then < .
014 Jy km s − giving a CO (2 − L CO(2 − < . × L (cid:12) . Tocompare with other works that usually quote the ground-state CO transition we assume a luminosity ratio of CO(2 −
1) / CO (1 −
0) = 7 . −
0) limitfor J0210-0456 is L CO(1 − < . × L (cid:12) and the ratio L CO(1 − /L FIR < . × − . The limit on this ratio isan order of magnitude higher than the values for typ-ical ultraluminous high-redshift galaxies and AGN (DeBreuck et al. 2011) showing that much deeper observa-tions are required to detect the molecular gas in galaxiessuch as these, highlighting how [C ii ] observations withALMA are the best way to probe the obscured interstel-lar medium in typical high-redshift galaxies. Fig. 7.—
The background shows the continuum map of J0210-0456 and ranges from − σ (purple) to +3 σ (yellow). The blue andred contours near the central dust continuum source show [C ii ]line emission maps for the blue ( − < v < −
45 km s − ) and red(+45 < v < +136 km s − ) wings, respectively. There is an offsetof 0 . (cid:48)(cid:48) ii ] peak velocity map for those pixelswith sufficient flux to enable a Gaussian to be fitted. This mapreveals a smooth velocity gradient across the source. [C ii ] dynamics and spatial extent Wang et al. (2010) showed that CO line widths of z ≈ z ≈ × fasterthan their stellar mass compared to the local ratio. Inthis work we have measured the [C ii ] line width for justone z = 6 . z ≈ ∼ M (cid:12) black hole hostgalaxies have suitable mm interferometry data.CFHQS J0210-0456 has a [C ii ] line FWHM of 189 ±
18 km s − , equivalent to σ = 80 ± − for a Gaus-sian and ignoring any inclination correction. Based onthe local relationship (Gultekin et al. 2009) a galaxywith σ = 80 km s − would be expected to have M BH ≈ × M (cid:12) , a factor of 25 lower than the measured M BH = 8 × M (cid:12) . This difference is comparable tothe factor of 10 found by Wang et al. for more luminous( M BH ∼ M (cid:12) ) quasars. Even after taking accountof possible inclination effects (Carilli & Wang 2006) thisshows the black holes in z ≈ ii ] emissionin SDSS J1148+5251 is concentrated within a radius ofedshift 6.4 host galaxies of 10 solar mass black holes 7only 750 pc of the nucleus. This is consistent with thesmall sizes of luminous z ≈ . ≈
800 pc (Ono et al. 2013).The [C ii ] line image of J0210-0456 shown in Figure3 shows elongation along an E-W direction, roughlyaligned with the direction of the beam. The measuredsize of the source is 879 ±
55 mas × ±
75 mas com-pared to a beam size of 770 mas ×
520 mas. We usedthe
CASA IMFIT task to fit a deconvolved model im-age to the data. This results in a deconvolved source of521 ±
248 mas × ±
297 mas oriented at a position an-gle of 128 degrees east of north. The intrinsic source sizeis not well constrained as it is only marginally resolvedin this image comprised of all fifteen spectral channels ofthe [C ii ] line.A different story emerges when one considers the blueand red sides of the [C ii ] line separately. Maps weremade using only the red and blue wings (5 channels each)and excluding the centre of the line. Figure 7 shows thecontinuum dust emission as the background image. Su-perimposed on this are separate contours for the blue andred sides of the [C ii ] line. It can be seen that there is aspatial offset of 0 . (cid:48)(cid:48) ii ] linefor pixels with sufficient flux to enable a Gaussian emis-sion line to be fitted. There is a clear velocity gradientacross the source along this same axis with magnitude ≈
100 km s − across a size scale of 1 . (cid:48)(cid:48) ii ] velocity gradients over this scale observedby Wang et al. (2013) in some z ∼ DISCUSSION
These observations with ALMA break new ground intheir sensitivity to moderately star-forming galaxies athigh-redshift. Even with this sensitivity, only one ofthe two z ≈ . ± . × L (cid:12) , whereas J2329-0301 remainsundetected with L FIR < . × L (cid:12) , significantly be-low the typical far-IR luminosity for a quasar of thisbolometric luminosity at any redshift. These low far-IRluminosities are surprising and place strong constraintson the star formation rates in the host galaxies of theseEddington-limited quasars.In the simplest black hole/galaxy co-evolution sce-nario, cosmic stellar mass and black hole mass increasein lockstep, ending up at the ratio observed in the localuniverse of M BH /M stellar = 0 .
002 (Tundo et al. 2007).Detailed simulations show that in individual galaxies thephases of star formation and black hole accretion are notsynchronized (Li et al. 2007) and this accounts for thesignificant scatter of points in Figure 2. It is trivial to cal-culate the relationship between star formation rate (lin-early related to L FIR ) and black hole accretion rate (lin-early related to L Bol and assuming an accretion efficiencyof 10%) necessary to achieve M BH /M stellar = 0 . z = 0. At z ≈
2, mm-selected galaxies mostly lieto the upper left of the line showing that they are grow-ing their stellar mass more rapidly than their black holes(Alexander et al. 2005). Lutz et al. (2010) also showedthat low-luminosity AGN from deep X-ray surveys arefound on the left side of such a line.J2329-0301 is found to be growing its black hole ata rate of > × faster than its stellar mass in orderto reach the local ratio. It has a black hole accretionrate of ˙ M BH ≈ (cid:12) yr − and SFR <
40 M (cid:12) yr − (3 σ limit, assuming T d <
47 K, no AGN-heated cool dust andSalpeter IMF). Khandai et al. (2012) report the results ofhydrodynamic simulations of z ∼ (cid:12) yr − asthey evolve during the main black hole accretion phase.The simulations are designed to match the properties ofthe most luminous quasars from the SDSS. J2329-0301has a black hole accretion rate ∼ × lower than typi-cal SDSS quasars and therefore scaling down the lowestsimulated SFR by this amount would result in approxi-mately the SFR limit observed for J2329-0301. This sug-gests that J2329-0301 is observed at a rare phase whereit has very low SFR compared to its black hole accre-tion rate. In the Khandai et al. (2012) simulations, theSFR usually drops at the epoch of peak quasar accretiondue to feedback heating the host galaxy gas. J2329-0301appears to have very effectively shut off star formation.Although the far-IR data of high luminosity PG quasarshas many non-detections and the H-ATLAS data are juststacked averages (Figure 2), it would appear that few lowredshift quasars have such a low ratio of SFR to blackhole accretion as J2329-0301. This quasar is known tobe surrounded by a luminous Ly α halo at least 15 kpcacross (Goto et al. 2009; Willott et al. 2011), which sig-nifies a huge reservoir of diffuse gas likely photo-ionizedby the quasar. When this gas cools and accretes on to thegalaxy, a further bout of star formation is likely. Hayeset al. (2013) noted that the Ly α emission from galaxieswith low dust content tends to be more extended thanthat in dusty galaxies, but do not provide a simple ex-planation for why this happens. J2329-0301 certainlyfits this pattern with a low dust content as measured bythermal dust emission and a very extended Ly α halo.The [C ii ] line detection in J0210-0456 is narrow(FWHM = 189 ±
18 km s − ) and shows only a small veloc-ity gradient ( ≈
100 km s − ) across a scale of 6 kpc. Theinclination angle of the [C ii ] emission is unconstrained,but the narrow line suggests the dynamical mass of thesystem is much lower than would be found in the localuniverse for a galaxy hosting a 10 M (cid:12) black hole. Thisis in agreement with the results for more massive blackholes at z ≈ z ≈ z ≈ ∼
100 pc will reveal the dynamical state of the starforming gas and enable more accurate determination ofdynamical masses.Thanks to Javier Graci´a-Carpio for providing unpub-lished far-IR data on the comparison sample of galax-ies. Thanks to staff at the North America ALMA Re-gional Center for processing the ALMA data. Thanksto the anonymous referee for suggestions that improved the manuscript. This paper makes use of the fol-lowing ALMA data: ADS/JAO.ALMA
Facility: