Probing Cosmic Evolution with the Most Distant Quasars Xiaohui Fan University of Arizona Apr 18, 2010 Collaborators: Jiang, Carilli, Kurk, Rix, Strauss, Vestergaard, Walter, Wang Today I would like to present some new Spitzer observations of the hot dust emission from dust tori in the Mosts distant quasars at z~6. The question that we would like to ask is really: Background: 46,420 Quasars from the SDSS Data Release Three

46,420 Quasars from the SDSS Data Release Three 5 Ly forest 3 Ly 2 CIV redshift CIII MgII FeII 1 FeII OIII H wavelength 4000 A 9000 A

Almost 50 Years Ago: First Quasar: 3C 273 by Maarten Schmidt

Quest to the Highest Redshift

Quest to the Highest Redshift 090423 080913 050904 000131 GRBs 970228

30 at z>6 60 at z>5.5 >100 at z>5 Here are the spectra of some 30 quasars at z~6 range, most discovered in the SDSS over the last eight years or so. We notice two features, one is the strong Ly alpha absorption, the presence of complete G-P absorption trough at z>6 Indicating rapid increase in the neutral H density in the IGM as we are approaching the reionization epoch. The other is again the existence of strong metal lines, even at the highest redshift, lines from C, N, O, Si, all the way to Mg, and Fe.

z~6: Crucial Transition in quasar evolution Evolution of quasar density and accretion rate: Formation of the first billion solar mass BHs? Dust-free quasars Connection to the earliest galaxies? Gas and star formation in host galaxies Evolution of M-σ relation? Evolution of IGM neutral fraction End of reionization epoch?

Theorists Tell us Li et al. 2007 These luminous z~6 quasars: The most massive system in early Universe Living in the densest environment BH accreting at Eddington Host galaxies have ULIRG properties with maximum starburst Li et al. 2007

Quasar Evolution at z~6 Strong density evolution Density declines by a factor of ~40 from between z~2.5 and z~6 Black hole mass measurements MBH~109-10 Msun Mhalo ~ 1012-13 Msun rare, 5-6 sigma peaks at z~6 (density of 1 per Gpc3) Quasars accreting at maximum rate Quasar luminosity consistent with Eddington limit Note: 1. They are extremes. Sensitive tests; 2. But also rare object, so don’t generalize too much Low-z z~6 Fan et al. 2006, 2010

Puzzle 1: Are there luminous quasars at z>>7 Black Holes do not grow arbitrarily fast Accretion onto BHs dicitated by Eddington Limit E-folding time of maximum supermassive BH growth: 40 Myr At z=7: age of the universe: 800 Myr = maximum 20 e-folding Billion solar mass BH at z>7 Non-stop, maximum accretion from 100 solar mass BHs at z=15 (collapse of first stars in the Universe) Theoretically difficult for formation of z>7 billion solar mass BHs What if we find them: Direct collapse of “intermediate” mass BHs? More efficient accretion model “super-Eddington”? So where are those earliest quasar? Is it because the universe simply can not make them fast enough? Jury is still out in term of what the constraint on the density of these objects are. But there is an important theoretical consideration that people are taking very seriously now, and that’s the concept of Eddington limit, which in principle is the fastest rate a BH can grow in a steady state, because in a steady state, a particle far away from the BH must at least by balanced by radiation pressure from the BH, otherwise things will be blown apart, that means for a given mass, there is a max luminosity, called Edd luminosity, and since BH grows by converting gravitational potential energy to radiation, luminosity is propotional to the growth rate, M dot, therefore this simply differential equation essentially means that in normal circumstances, a BH can e-fold its mass in 40Myr, at the maximum. Comparing to the age of the universe, 40Myr is a short time, but at high redshift, this is becoming an interesting fraction of the age of the universe, in other words, the BH in early universe might be bound by Edd limit and limited by the number of e-folding available.

Puzzle 2: non-evolution of quasar (black hole) emission z~6 composite Ly a Low-z composite NV Ly a forest OI SiIV XF et al. 2010 Jiang, XF et al. 2008 In some sense, we were quite surprised to see this general lack of evolution in the quasar spectral properties. It is best illustrated here, where we compare the composite spectrum of our sample of z~6 quasars with the composite from all SDSS quasars, at average redshift of 1.5 or so. It is striking that in terms of both emission line strength and continuum shape, there is no evolution whatsoever from the local universe to the highest redshift currently observed, close to reionization epoch, The blue side of Ly alpha emission of course is affected by IGM absorption. People are still debating exactly how significant this is; but the bottom line is that these quasar evnv has reached high metallicity early on with no or little evolution, and the high-z quasars we observe, in term of the hot gas around them, are old, mature. Rapid chemical enrichment in quasar vicinity Quasar env has supersolar metallicity : no metallicity evolution High-z quasars are old, not yet first quasars, and live in metally enriched env similar to centers of massive galaxies

When did the first quasar form? Dust: emitting in infrared To probe this, we have been observing these quasars not only in the optical wavelength we found them, but for photons of all energies, from X-ray to radio, because different wavelength photons come from different part of the region surrounding central BH in quasar, roughly speaking, high energy photons comes from the most inner part, only light hours to light days from BH, and probably can form very quickly, and low energy photons, such as those emitting in the IR, comes from extended dust structure light years away. And these dust emission in IR is what the next discovery came from. radiation from X-ray to radio as a result of black hole accretion and growth

Disappearance of Dust Torus at z~6? typical J0005 3.5m 4.8m 5.6m 8.0m 16m 24m quasars with no hot dust Spitzer SEDs consistent with disk continuum only No similar objects known at low-z no enough time to form hot dust tori? Or formed in metal-free environment? In cycle 4, we found another object, with non-detection in the mid-IR, and SED consistent with disk emission only, as if there is no dust torus at all. As shown in these Spitzer images, a typical quasar has strong mid-IR emission, but these quasars simply dropped out from mid-IR. The crucial point here is that no object with such weak IR emission has ever been detected, at any redshift, before. As shown in this diagram, which plots the host contribution, in term of hot dust flux vs. optical flux, for all Spitzer objects from z=0 to 6, and our IR weak quasars stands out as the only example, and they happen to be some of the most distant object. Is this a coincidence? Or is this showing something fundamental about quasar dust properties? Of course not all quasars at z~6 have weak hot dust emission, so what’s so special about them? Jiang, XF et al. 2010

Epoch of first quasars? Dust-free quasars: Dust/Bolometric Only at the highest redshift With the smallest BH mass First generation supermassive BHs from metal-free environment? How are they related to PopIII? Dust/Bolometric Dust/Bolometric Jiang, XF et al. 2010 BH mass

High-redshift quasars live in the center of star-forming galaxies J1148 (z=6.42) - Spatially resolved CO and [CII] emissions: Size: ~1.5 kpc Star formation rate of: ~1000 Msunyr-1kpc-2 theoretically close to maximum star formation rate ? Gas supply exhaused over a few tdyn Similar SF intensity to the brightest local starburst (Arp 200) but 100 times larger! CO 1kpc J1148 (z=6.42): CO line width ~300 km/s Dynamical mass ~1011Msun? BH formed earlier than completion of galaxy assembly? Walter et al. 2004 Walter et al. 2009

2. IGM transmission: zreion > 6 reionization Two Key Constraints: WMAP 5-yr: zreion=11+/-3 2. IGM transmission: zreion > 6 From Avi Loeb

First detection of Gunn-Peterson Effect

Evolution of Lyman Absorptions at z=5-6 transparent opaque OK, now switch gear and talk about IGM. We all know the strong evolution of lyman absorption at high-z. Now I will show two animations: they are all segments of observed ly alpha or ly beta transmission, with z range of 0.15, or about 60 comoving Mpc at z~6 I will show these independent transmission spectrum from low to high-z. The full scale is 0 - 0.4. They all have same spectral resolution. At z~5 to 5.5: On average the absorption becomes stronger But there are still few completely dark pixels. Most regions have some transmission Although occasional long dark gaps present at z~5.3 A sharp transition happened at z~5.7 From there, forest is dominated by long, dark gaps, complete GP With only occasional transmitting pixels. Show again, two important points The absorption strength evolution accelerated at z~5.7 - 5.8 The appearance changed from most pixels transmitting to most pixels opaque But with very significant line of sight variance And even at z>6, there are still transmitting pixels, I.e., the GP trough is not inifinitely long. Now ly b: the available sight line is shorter due to contamination of high-order foreground abs But the evolution and transition at z~5.7 is sharper. More pronoucd. Due to the smaller o.s. and higher sensitivity of ly b. Now let’s look at the IGM evolution more quantitatively. z = 0.15

Accelerated Evolution at z>5.7 Optical depth evolution accelerated z<5.7:  ~ (1+z)4.5 z>5.7:  ~ (1+z)>11 > Order of magnitude increase in neutral fraction of the IGM  End of Reionization Dispersion of optical depth also increased Some line of sight have dark troughs as early as z~5.7 But detectable flux in ~50% case at z>6 End of reionization is not uniform, but with large scatter (1+z)11 (1+z)4.5 Now let’s combine those two. We are trying to be careful in converting alpha and beta optical depth. We use both a photoionization model in a clumpy IGM, similar to what we did previously, and use a n empirical calibration, by requiring ly and ly b give consistent average effective transmission at the same redshift, to calculate the conversion. The conversion factor, because of the clumpyness of the IGM, which allows flux to come through in underensed regions, is about 2.5, a factor of 2.5 smaller than the factor of 6.2 for the uniform case. Looking at the green ly beta and blue ly a points, they give consistent measurements when there are detections in both transitions. when there is no ly a transition, ly beta sometimes give clear detections that are consistent with ly a upper limit, some times give a much more stringent upper limit when no flux is detected even in beta transition Combining the two, we see clearly that: 1. The evolution of optical depth accelerated. At z>5.7, the best fit power law has a index (1+z) to 10 or larger, although there are still about 50% case that there are clear detections, I.e., the optical depth is finite. 2. Shaded area is 1-sigma dispersion of tau. The dispersion of optical depths also increased dramatically. At z<5.9, the relative scattering is 0.3 - 0.5, while at z>6, the dispersion is of the order of unity, while some dark regions begins to appear as early as 5.7 or so. XF et al. 2006

Beyond Gunn-Peterson Optical Depth: HII Region Sizes zem Gunn-Peterson test saturates at z>6 Size of HII region Rs ~ (LQ tQ / fHI )1/3 HII region size decreases by ~ 3 from z=5.7 to 6.4 Best estimate: fHI ~ a few percent at z~6 Can be applied to higher z and fHI with lower S/N data Model uncertainties due to radiative transfer Carilli et al. 2010 HII region size The third way to measure IGM ionization is to look at the HII region size that Zoltan talked about In the last talk, which seems to give the largest overall estimate previously. Here we try to look at all objects. Since the absolute values of HII region estimate depends on a number of parameters, such as quasar lifetime, details of radiative transfer, which are not well known, we here concentrate on the the relative evolution of HII region size first, Because if there is some kind of phase transition at z~6 after which the neutral fraction is very high, then we should simply see a dramatic decrease in the HII region size as it is confined by expanding into a largely neutral IGM. We don’t have time to talk about details of defining the HII region size, but basically, if one Uses a consistent definite, one finds that the size of proximity zone around these quasars, esp. after correcting for luminosity difference, deceased by a factor of more than 2 from z=5.8 to 6.4. The measurement error is large, so there is only a 3 sigma result. But it tells us two things: first, there is a measurable difference in the HII region size, indicating that the model of HII region expanding into an increasingly neutral IGM is reasonable, and the most Straightforward explanation is that the neutral fraction increased by, again, order of mag Over this narrow z range. Second, however, we don’t see a huge increase yet, in other words, it doesn’t seem to a jump of several order of mag which is required if we want to IGM to be mostly neutral. z

Probing Reionization History WMAP Finally, what does it mean for the reionzation history? I think what we are seeing is: at z~6, the IGM neutral fraction increased by an order of mag, or more, but it is probably not yet neutral at that time. if this trend continues, then the IGM could be largely neutral by z of 8. in this case, we might really need a double reionization model to explain everything., so we simply need some z~7 to 8 objects, whether quasar or GRB, to answer these questions, if the IGM is neutral, we should detect very long GP troughs in these objects.

Roads ahead Luminous quasars probe the evolution of the most massive systems in the early universe Important changes were happening at z~6-7 Timescale constraints on billion solar mass BH growth Evidence of youngest quasar structure End of reionization epoch with order of magnitude (or more?) increase in IGM neutral fraction Discovery (or lack of ) z~7 quasars might reveal new surprises Requires new generation of large IR sky surveys Quasars and GRBs are complimentary probes to the peak of reionization epoch GRB probes pristine, low mass galaxies and can reach high-z faster So I think two important results in the last few years might be pointing us to the same thing, that is, important changes might be taking place at z=6-7, about 1 billion years after the big bang, because this is the timescale for the growth of the earliest SBHs, and because we are finally detecting changes in the way these BHs emit. What’s more is that this underlies the importance of pushing to higher redshift, when time constraints become stronger, and evolution of quasar propeties might accelerate. This needs more ambitious project to find quasars at higher redshift, z~7 or so, which now requires large IR sky surveys that are sensitive to those higher redshift objects.