Sunyaev-Zeldovich Decrements with No Clusters?

arXiv:astro-ph/9704237v5 2 Jun 1998 o.Nt .Ato.Soc. Astron. R. Not. Mon. coe 2018 October 9 0HA, CB3 Cambridge Road, Madingley Sigurdsson Astronomy, of Steinn Institute & Natarajan Priyamvada clusters? no with decrements Sunyaev-Zeldovich hr r tpeetmn e bevtoso h M in CMB the of observations new well. many potential present local the at of are depends depth There outflow and the density of ambient for the high scale the on physical the for the sites seed clusters, the of to are mation tend subsequently which and regions peaks density col- matter) high-significance the dark from of (cold form lapse CDM quasars models, standard QSO. mean cosmological the the the dominated of of vicinity by variants the in determined in Since gradient are density times the QSO, and late the density of outflow at output the kinematics energy of the the by properties primarily the determined While are mechanical quasars. the by of powered outflows luminosity the electrons by up by swept produced are be that can compa- decrements that of are work temperature cluster this rich rable that in a demonstrate of electrons we well However, by potential galaxies. matter typically dark - the in the electrons confined of hot microwave scattering by Thomson cosmic the photons by the produced are in background decrements Sunyaev-Zeldovich INTRODUCTION 1 000 0–0 00)Pitd9Otbr2018 October 9 Printed (0000) 000–000 , h usr usrotospoue()afeunyindependen outfl frequency gradie the a density of (i) the scale and produce ∆ density physical outflows local the Quasar mean quasar. and elect the the QSO, on hot the both the of luminos depends indu by turn luminosity mechanical quasar-outflow the produced these the on of decrements by magnitude marily of the up potential, case swept cluster the baryons the to the contrast by In produced be can noigqaa uflw,w ru htboth that o optical argue the we in outflows, either detected quasar p are Invoking quasar clusters or virialized assembled, dete quasars produce no around can decrements that Sunyaev-Zeldovich process by physical matic line-o photons alternate the an CMB along propose the encountered we Medium of Intra-Cluster scattering the in Thomson electrons by produced M Cosmic are the (CMB) of temperature the in decrements Sunyaev-Zeldovich ABSTRACT aito eaiet ∆ to relative variation usr general – quasars 0 at 20” ∆ h iigrgo ftefis ope ekat peak c multipole Doppler CMB first the the affect of hence region and power rising the the to contribute will which otx ftenx-eeaino M xeiet A PLANC & MAP poten - words: are experiments Key CMB they of since next-generation interest the of of are context effects These models. defect T/T T/T ∼ ∼ ∼ 10 10 10 − − − 7 4 4 omlg:csi irwv akrud–dffs aito galaxie – radiation diffuse – background microwave cosmic cosmology: i)afeunydpnetflcuto nteitniy∆ intensity the in fluctuation dependent frequency a (ii) ; ee;adteeoe()pthso inl∆ signal of patches (v) therefore and level; nteopst ieo h S;(v ierplrzto inlover signal polarization linear a (iv) QSO; the of side opposite the on T/T - .K. U. a t1 H n 0Gz ii eprtr increment temperature a (iii) GHz; 30 and GHz 15 at say , hmevsaeotie;i eto edsustekine- the discuss we 4 Section manifest to in likely outlined; are ou effects are scales their themselves physical and outflows implied the the which of ener to terms the 3 in Section has constraints in getic outflows; QSO quasar-driven of the role the after vate well observed be ef- can the QSO, gas the faded. the the of exceeds or phase of to luminosity fects comparable high is the time of cooling time-scale the very if e. be interesti i. not are long, particu- time-scales need In cooling/expansion decrement. gas the observable when the an lar, produce that to out order outflows, point in quasar hot to produce objects: instructive may of is that class it objects this For of veloc- decrements. class peculiar such another finite include with and QSOs cross-se ity) around scattering bubbles velo the warm peculiar to and with clas- corrections the clusters relativistic on (clusters, expand ties, we scenarios paper, S–Z this high-redshif In sical of hosts. morphology their radio there and the QSOs and in underway, interest or renewed is planned wave-band millimeter the hsppri raie sflos nScin2w moti- we 2 Section in follows: as organized is paper This l thermal ∼ 0frCMlk oesa elas well as models CDM-like for 60 T/T and tbetemladkine- and thermal ctable ∼ kinematic ilyosral nthe in observable tially ieai decrement kinematic t e ffcsdpn pri- depend effects ced 10 isi ein where regions in airs Background icrowave oscnndwithin confined rons -ih.I hspaper, this In f-sight. -a wave-band. X-ray r ti h iiiyof vicinity the in nt − 4 h ofie hot confined the ndge scales degree on K. J t fquasars. of ity ν w-wihin which - ow luain in alculations /J decrements ν ∼ 20% a ctions ngly ci- s: t t - 2 Priyamvada Natarajan & Steinn Sigurdsson matic and the thermal S–Z decrement induced by outflows of scale R: and their other observational implications; and we present ǫ ǫL 1 Mbh 1 fQ − 1 v/c − 2 3 3 3 3 our conclusions and prospects for detection in the final sec- R ∼ 1( ) ( 11 ) ( ) ( ) Mpc. (1) 0.05 10 M⊙ 1 0.01 tion. For our purposes typical swept up regions are of the order of 0.1–1 Mpc or so; fQ is the local baryon fraction in units of 0.01, and a density contrast of ∼ 103 over the mean 2 QUASAR-DRIVEN OUTFLOWS cosmological density at redshift ∼ 3 is required for the numbers to be consistent, this is a plausible over-density for the The ubiquity and high luminosity of quasars out to high highest significance peaks that will seed the massive clus- redshifts suggests that they might have played an important ters observed at lower redshifts. The total energy budget role in the subsequent formation of structure (Efstathiou & ET ∼ ǫ ǫLMbh; and the restrictions on the available param- Rees 1988; Rees 1993; Silk & Rees 1998). Babul & White eter space in terms of the mass that needs to be swept up (1991) suggest that the ionization produced by UV pho- and the bulk velocity required to produce a kinematic S–Z tons from the quasar might inhibit galaxy formation in their decrement are dictated by the following relations, vicinity due to the ‘proximity’ effect. Thus quasar activity is 2 likely to be one of the important non-gravitational processes ET ∼ Msweepvsweep ; R ∼ tQSO vsweep ; (2) to have influenced structure formation at redshifts z > 2. ∆T vsweep ∼ τ ∼ Msweep vsweep , (3) We propose quasar-driven outflows as a possible phys- T c ical mechanism for concentrating baryons. Such a scenario where tQSO is the quasar lifetime and τ the optical depth. has been explored previously in the context of explosion in- Beam dilution imposes a minimum Msweep assuming the line duced structure formation models (Ostriker & Cowie 1981; of sight length through the gas is comparable to its trans- Daly 1987) as well as in theoretical models of the Inter- verse scale, which is likely for the resulting outflow geometry Stellar Medium (McKee & Ostriker 1977). Mechanically studied here. Scaling to typical Abell cluster parameters, the powered outflows can push baryons around active quasars mean electron thermal speeds are βT = vT /c ∼ 0.1. Since in shells cleaning out regions of Rtoday ∼> 1Mpc. We exam- the kinematic effect is first order in βK = v/c, we require 2 −2 ine the effectiveness of this mechanism in the production of βK ≈ βT ∼ 10 . This is consistent with speeds required massive clumps of baryons that can in turn produce a signal for the mass of swept up gas which would produce the re- in the CMB. The observational evidence for these outflows quired optical depths over angular scales of a few arc min- are seen in a subset of radio-quiet QSOs - the broad ab- utes. It is these mass and velocity requirements which drive sorption line systems (Hazard et. al 1984; Weymann et. al us to postulate ultra–massive central black holes to power 1991) in which speeds v/c ∼ (0.01 – 0.1) are inferred dur- the outflows. Note that τ ∼ neR whereas the mass swept 3 ing a wind-phase. These ‘hyper-active’ quasars can create up Msweep ∼ neR . All the numbers quoted in this work −1 both regions that are baryon-rich as well as voids via their have been computed for H0 = 50 kms Mpc; Ω0 = 1.0 energetic outflows. This powered outflow model is compati- and Λ = 0. As illustrated by eqns. 2 and 3, the best sources ble with both scenarios of quasar population evolution: the for large energy effects are the massive BHs that power the single generation of long-lived quasars as well as the mul- QSOs. We also note here that according to some models of tiple generation model wherein populations fade and flare unified schemes for AGNs, jets and efficient energy release (Haehnelt & Rees 1994). Superwinds of the kind detected occurs only if the QSO is accreting close to or at super- by Heckman et al. (1996) from galaxies can also in principle Eddington rates, possibly via the Blandford-Znajek mech- produce S–Z decrements, but they are typically an order of anism for rotating BHs. This might preferentially happen magnitude weaker than the QSO outflows proposed here. at high redshift, where a higher fraction of high luminosity QSOs are observed, consistent with a picture wherein the most massive BHs form rapidly at high redshift and are fu- eled by the baryons that are present in these high-density 3 ENERGETICS OF THE MECHANISM regions. We do not necessarily need fully assembled, virialized dark halos for this to happen, we only need an efficient The total energy available to power a quasar can be directly way to funnel the baryons already there to the bottom of estimated from the mass accreted by a black hole of mass the deepest potential wells. Mbh (modulo an efficiency factor, ǫ). In general, the total energy available is 10-20% of the rest mass of the central black hole (ǫ ∼ 0.1 − 0.2). We assume that roughly 50% 3.1 Dynamics of the quasar outflow (ǫL ∼ 0.5) of the total energy in turn might be available as mechanical power. By assumption this sets up an outflow The outflow starts out being collimated close to the source that is primarily mechanically powered, that is, a fraction and has some, small, opening angle so that it does not ter- of the energy emitted by the active quasar gets efficiently minate or disrupt the accretion process which is its pow- converted into kinetic energy of the intergalactic medium. erhouse. At a distance of ∼> 10 kpc away from the central Mechanical luminosities up to ∼ 1048 erg s−1 at redshifts engine, the flow opens out. The mass Msweep (of the IGM) swept out by the outflow is: z ∼> 4 for such QSOs are postulated, this is not entirely implausible given inferred X–ray luminosities of comparable Msweep ∼ LQSO; vsweep ∼ (LQSO), (4) magnitude for high z QSOs (Elvis et. al 1994). Clearly such p energy coupled locally to the environment as mechanical lu- where LQSO denotes the mechanical luminosity of the QSO. minosity would set up an outflow that sweeps out a region As was shown by Phinney (1983), jets from QSOs start out Sunyaev-Zeldovich decrements with no clusters? 3 mildly relativistic with a Lorentz factor ∼ a few; it is likely 4 S–Z DECREMENT PRODUCED BY THE that the initial outflow is dominated by a pair plasma. The OUTFLOW total momentum that the outflow can deposit is determined mp The S–Z decrement in the temperature of the CMB is pro- primarily by the ratio of and therefore bulk speeds of the me −2 duced by Thomson scattering of the CMB photons by hot order of v ∼ 10 c are expected, which are indeed observed electrons encountered along the line-of-sight (Sunyaev & Zel- at low redshifts even in low power radio galaxies like NGC dovich 1970; 1972; 1980). The magnitude of the decrement 315 (Bicknell 1994). The mass swept out by a single QSO is independent of the redshift of the hot plasma and is given with an outflow assuming that 50% of its total bolometric by the integral of the pressure along the line-of-sight. The luminosity is expelled mechanically is, effect has two components a thermal and a kinematic one

14 LQSO δt (Sunyaev & Zeldovich 1980). The thermal S–Z effect causes Msweep ∼ 2 × 10 ( )( ) 1048 ergs−1 5 × 108 yr fluctuations in the temperature and the intensity which are vsweep −2 (i) frequency independent, of equal magnitude and the same × ( ) M⊙, (5) 3000 kms−1 sign in the Rayleigh-Jeans limit of the spectrum but (ii) are larger, differ in magnitude and depend on the frequency in where δt is the duration of the outflow phase, vsweep the the Wien limit. The overall sign of the contribution to the ejection velocity and LQSO the mechanical luminosity of the fluctuations (in both the temperature and the intensity) also QSO. The total mass swept up is strictly bounded, differs in the Rayleigh-Jeans and the Wien limit. 2ǫ ǫL Mbh sweep M ≤ vsweep 2 . (6) ( c ) 4.1 The Kinematic S–Z effect While the initial outflow is likely to be a pair plasma, the number density of electrons in the bulk outflow is dominated Coherent streaming motion with a velocity vsweep with re- by the swept up hydrogen from the IGM and is estimated spect to the frame of reference in which the CMB is isotropic as, introduces a Doppler shift in the scattered radiation producing the kinematic effect which additionally perturbs the −2 Msweep R −3 −2 temperature and intensity across the region. This kinematic ne ∼ 1 × 10 14 ( ) cm , 2 × 10 M⊙ 1 Mpc piece induces a frequency independent temperature gradient but a frequency dependent fluctuation in the intensity; in good agreement with typical values obtained using both of which have the same sign (determined by the direc- the 151-MHz and 1.5-GHz data set of Leahy, Muxlow & tion of the velocity vector; note that if the outflow is spher- Stephens (1989). The optical depth τ to Thomson scatter- ical on large scales the kinematic S–Z effect cancels as the ing is: line of sight passes through regions with opposite sign radial velocity). Sunyaev & Zeldovich (1980); Peebles (1993), show ne R τ ∼ 10−2 ( ) ( ). (7) 10−2 cm−3 1 Mpc that the decrement produced by the kinematic term is given by, With an approximate slab geometry as predicted in this out- ∆T vsweep ∆Jν ∆T xexp(x) flow picture the gas mass required for a given optical depth ≈ −τ ( ) ; = ; (8) can be several times smaller than that for an isothermal T c Jν T exp(x) − 1 sphere whose virialized core provides the most of the optical where x = hν ; the kinematic decrement is first order in kTr depth for the clusters S–Z effect. No dark matter is swept vsweep/c compared to the normal second order thermal effect up, so the total mass involved is considerably less than the produced by hot cluster gas. For typical values of the optical total cluster mass of a virialised cluster producing a com- depth and the velocity of the outflow assumed in the above parable S–Z effect. As is evident from our approach, the ∆T −4 analysis, this yields a T ∼ 3 × 10 . We stress here that micro-physics of the early stages of the jet propagation have the ratio of the predicted thermal to the kinematic not been computed in detail. We assume that the flow is ini- S–Z effects depends entirely on the bulk velocity and tially collimated and breaks out on a scale of a few kpc and mean temperature of the plasma. Below we estimate that approximately 50% of the energy released due to accre- the thermal component of the S–Z effect expected from a tion is efficiently mechanically coupled to the environment. fiducial outflow. The geometry of a typical region through which the outflow The material at the working surface may be shock- propagates is assumed to be a cosmologically overdense re- heated to temperatures as high as ∼ 108 K, but may cool gion with a strong external density gradient. Physically, this down efficiently as we argue below. The bulk of the mass is required so that the initial outflow can remain collimated (which is the material in the IGM; note that only roughly a right out to the advancing surface where the bulk of the thousandth of the mass is expelled directly from the QSO) energy is deposited. The advancing surface breaks out and however is expected to be in a cooler phase. The shock expands once a critical ambient pressure is attained and the heated dense plasma will cool to a temperature Te such that presence of the density gradient serves to anisotropize (i. its cooling time tcool ∼ tdyn, where tdyn is its dynamical e. destroys the spherically symmetry of) the post break-out time (Rees & Ostriker 1977). Thermal brehmstrahlung is expansion. expected to be the primary process for energy loss in the early stages of the evolution of the bubble and X-ray fluxes ≤ 1044ergs−1 are expected, which could be detectable for low-redshift bubbles (see Natarajan, Sigurdsson & Silk 1998 for a more detailed discussion). X–ray cooling is inefficient 4 Priyamvada Natarajan & Steinn Sigurdsson

the thermal effect, for warm (T ∼ 106 K) plasma with significant bulk motion, as postulated here. This mechanism might be relevant to a recent detection of a decrement in the region of the quasar pair PC1643+4631 where no cluster is detected. The reported observation was obtained by Jones et. al (1997) using the Ryle telescope at 15-GHz in the direction of the quasar pair PC1643+4631 A& B. The pair separation on the sky is ∼ 200” and their r4 magnitudes are 20.3 and 20.6 respectively. The redshifts of the quasars are (A) z = 3.79 and (B) z = 3.83 (Schneider, Schmidt & Gunn 1994). The central value of the measured decrement is 560 µK. Interpreting the source as a cluster at ∼ 5 keV, Jones et. al estimate the implied gas mass to be ∼> 14 2 × 10 M⊙ and conclude that the result demonstrates the existence of a massive high redshift (z ∼> 1) system. How- ever, serendipitous and pointed ROSAT PSPC observations (Kneissl et. al 1997) as well as ground-based optical follow- ups (Saunders et. al 1997) have failed to reveal a cluster of galaxies as the source of the hot plasma to produce this ‘cold’ spot in the CMBR. The hypothesis of lensing by an intervening massive cluster is therefore hard to justify. An- other S-Z decrement has been measured at the VLA (3.6 cm) by Richards et. al 1996, also around a pair of radio- quiet QSOs at a projected separation of 1 Mpc. In this case Figure 1. The ratio of the predicted thermal decrement (denoted the pair have the same redshift (z = 2.561) and although by the subscript [T]) to the kinematic component (denoted by the subscript [K]) of the decrement in the temperature and intensity their spectral features are not identical their velocity sepa- in the Rayleigh-Jeans limit [left side of the graph] and the Wien ration being small, they are likely to be a lensed pair split limit [right side] plotted as a function of the dimensionless pa- by a high-redshift cluster with a large time delay of the or- hν rameter x, defined as x = (where Tr is temperature of the der of 500 years. The physical scales involved in both these kTr radiation i. e. 2.7 K). The ratios plotted above have been com- observations are consistent with the role of quasar outflows 6 puted for the plasma at Te = 10 K and an optical depth τ = and therefore our proposed model might provide a possible 0.01. The arrow on the x-axis marks the value corresponding to a physical explanation. frequency of 15 GHz. The x range varies from roughly λ = 5 cm at x = 0.1 to λ = 1 mm at x = 5.0. 4.2 The Thermal S–Z effect 4.2.1 Emission from the shock-heated bubble for these high power outflows. The plasma can cool to lower temperatures rapidly via turbulent mixing which will tend In a recent analysis Aghanim et. al (1996), consider another to homogenize the temperature of the material on the dy- scenario where the kinematic S–Z effect is also much higher namical time-scale. Synchrotron cooling is however, unlikely than the thermal S–Z effect. They consider the bulk motion to be important, since for the typical estimated value of the of highly ionized baryons in the proximity of QSOs at speeds −8.5 −1 5 IGM magnetic field at high-redshifts |BIGM|∼ 10 G (see of ∼ 300 kms and at temperatures of ∼ 10 K, essentially Kronberg 1994) the cooling time exceeds both the Hubble stationary bubbles around the QSO moving with the QSO time and the lifetime of the QSO, therefore, unless there at the local peculiar velocity. Although the physical picture are significantly stronger B fields on very small scales en- that we present above is entirely distinct from that proposed ergy losses via synchrotron processes will be negligible. The by Aghanim et. al, we find the same scaling of the decrement thermal S–Z decrement produced by this warm plasma is: with the relevant parameters:

∆T 2 kTe − Te ∆T − τ vsweep LQSO 1 1+ z 6 4 3 2 = ( 2 τ) = 3.45 × 10 ( 6 ), (9) ∼ 3 × 10 ( −2 )( −1 )( 48 −1 ) ( ) . T me c 10 K T 10 3000 kms 10 erg s 1 + 3 much lower than that expected for virialized gas in a core The magnitude of the kinematic S–Z effect from outflows of a cluster of galaxies. For a range of frequencies, in Fig. scales primarily with the mechanical luminosity of the QSO 1, we plot the ratio of the expected kinematic and ther- and the redshift, therefore, we expect stronger decrements mal components for the plasma in the outflow. As shown by preferentially from high luminosity and high redshift QSOs. Birkinshaw (1996), the ratio of the thermal to the kinematic Bicknell (1994) finds for a local FRI radio galaxy NGC 315, effect is independent of the optical depth and can be written L ∼ 1043 erg s−1; β ∼ 10−2 and an associated jet that as, extends to scales of the order of R ∼ 0.15 Mpc; our model considers QSOs that are essentially the high-redshift, scaled- ∆T K vsweep Te −1 = 40( −1 ) ( 6 ) , (10) up (scaled up in luminosity and hence the physical scale on ∆T T 3000 kms 10 K which the outflow occurs) versions of objects like NGC 315. therefore, the contribution from the kinematic effect could We stress here that spherically symmetric outflows do easily be one or even two orders of magnitude higher than not generate a kinematic S–Z signal, an anisotropy induced Sunyaev-Zeldovich decrements with no clusters? 5

may be detectable through narrow band imaging or long slit spectroscopy, it is interesting to note that somewhat smaller scale diffuse emission has been seen around both a HS low redshift, low luminosity radio–loud QSO 4C 03.24 (van Ojikpine et al. 1997) as well as in two high redshift radio– loud QSOs (Lehnert & Becker 1998). The diffuse emission predicted for the QSO pair (PC1643+4631) is of comparable power to that seen around the quasar 4C 03.24, but spread over a larger angular scale due to the greater spatial extent x required of the outflow. Any fortuitous alignment of two QSO QSOs such that their outflows contribute additively to the S–Z decrement would make the effect unusually strong and hence detectable. We would expect this effect to be rare in linear & circular general, even in the proximity of bright, high redshift, radio- quiet QSOs with the optimal opening angle (θ ≈ 45◦). It polarization also strongly depends on the efficiency of the conversion of the bolometric QSO luminosity to mechanical outflow with time and environment. The luminosity of radio quiet QSOs is seen to increase strongly with z (Barvainis, Lonsdale & Antonucci 1996; Taylor et. al 1996). For instance, in the scheme of Falcke et. al (1996) the QSOs in PC1643+4631 at their brightest would have been classified as FRIIs with lu- S-Z decrement minosities in excess of 1048 erg s−1 (see Falcke et. al’s Figure Figure 2. Schematic of the configuration showing the relative 1); the radio jets have now faded and the accretion power positions on the sky of the QSO; the observed S–Z decrement decreased by many orders of magnitude, possibly due to de- CS; the predicted temperature increment HS. pletion of the cooling flow in the inner kpc (see eg. Fabian and Crawford 1990) or decreased efficiency at lower mass by the bubble expansion into a density gradient is required. accretion rates. Since the outflow must not be spherical at More importantly with every observed kinematic S–Z decre- large radii for a kinematic S–Z effect to be observable, sub- ment a corresponding increment is produced due to the op- stantial amounts of gas ought to remain in the vicinity of the posite sign bulk velocity on the other side of the QSO. QSO in the regions outside the out-flowing cone, and would Several immediate predictions follow from our proposed probably condense into the lyman absorbers or even a popu- scenario: while the temperature decrement is frequency in- lation of baryon-rich dwarf galaxies (Natarajan, Sigurdsson dependent, the intensity change is not (shown explicitly in & Silk 1998) observed in the vicinity of QSOs. eqn. 8 above), so we expect a ∼ 20% variation in ∆J/J The presence of bright radio sources in the immediate relative to ∆T/T at say 15 GHz and 30 GHz. We also ex- vicinity could make this effect difficult to detect. In the con- pect a temperature increment of the same amplitude on text of unified AGN models, the viewing angle implied for the opposite side from the decrement (see the marked ‘hot’ radio loud QSOs means looking straight down the jet, there- spot HS and the cold spot as CS in Fig. 2), as we expect an fore detection of such extended bulk flows from them is im- outflow with the opposite sign radial velocity on the other possible. Hence one would only see such effects after the peak side of the QSO. emission had diminished by several orders of magnitude and Since the optical depth τ and v/c are of comparable the extended radio lobes faded. order, both polarization effects (see Sunyaev & Zeldovich In order to predict how many such objects are expected 1980) arising due to (i) the finite angle subtended by the to be detected, we need to estimate the number density of line-of-sight with respect to the velocity vector (linear po- QSOs which had strong outflows. If we believe the jet power v 2 scalings with luminosity and redshift predicted by models, larization - ∼ 0.1 τ( c ) ) and (ii) the finite optical depth (circular polarization ∼ ±0.1 τ 2 v ) are also roughly equal. the relevant number density is then of QSOs with lumi- c 47 −1 Therefore, we expect polarization at the 10−7 level on 20” nosities ∼> 10 erg s within the optimum redshift range scales from flow transverse to the line of sight, along the line z ≥ 1. In order to not violate (the admittedly weakly con- from the QSO to the peak decrement. Given the density and strained) local BH mass functions, we expect the ultramas- the temperature of the outflow material, the correspond- sive BHs needed for this scenario to have a mean space den- 44 −1 −7 −3 ing X-ray luminosity LX is expected to be ∼< 10 erg s , sity of 10 per Mpc or less. below the flux limit of current X-ray observations, and sub- stantially redshifted to lower energies. We also note that the −2 presence of extended cool gas with ne ∼ 10 as predicted by this outflow picture could be detected via absorption fea- 5 CONCLUSIONS tures seen blueward of the emission peak (offset by ∼ 3000 km s−1) in the QSO spectrum. Given the available energy S–Z decrements measured in the vicinity of QSOs where budget, diffuse Lyman-α emission from recombination in the no clusters are detected either in the optical or the X-ray warm gas is expected with approximate surface brightness may be due to a concentration of baryons aggregated by the within a narrow band filter (width ∼ 60 Angstrom) that is power of mechanical outflows from these quasars. The above of the order of ∼ 25 mag arcsec−2. Such diffuse emission phenomenon leads to the following predictions (note that 6 Priyamvada Natarajan & Steinn Sigurdsson the numbers below are for the kinematic S–Z decrement would produce a hot spot, therefore producing a one-sided caused by the outflow from a single QSO): thermal SZ effect. Such a scenario is therefore easily distin- guished from the pure kinematic effect discussed above, as • (i) it produces a frequency independent decrement the thermal decrement would show as a strong increment ∆T/T ∼ 10−4; at shorter wavelengths and the predicted long wavelength • (ii) a frequency dependent fluctuation in the intensity increment on the opposite side of the QSO would be absent. ∆Jν /Jν ∼ a 20% variation relative to ∆T/T at say 15 GHz Although we have presented a somewhat speculative and 30 GHz; model, it makes several falsifiable predictions. Our model • (iii) a temperature increment ∆T/T ∼ 10−4 on the does require ultra-massive black holes, but that is not for- opposite side of the QSO; bidden, in fact observations imply precisely that some of • (iv) a linear polarization signal over 20” at ∼ 10−7 level the high redshift QSOs have total luminosities in excess of (pushing at the limits of the detection by the proposed new 48 −1 10 erg s , implying central black hole masses of the order instruments on MAP and PLANCK); 10 of 10 M⊙. We note that the estimated BH mass in the case • (v) patches of signal ∆T/T ∼ 10−4 on degree scales of the nearby BL Lac OJ287 (Lehto & Valtonen 1996) at z which will contribute to the power and hence affect the CMB 10 = 0.306 is ∼ 2 × 10 M⊙, and this is unlikely to be the multipole calculations in the rising region of the first Doppler most massive black hole in the universe. How ultramassive peak at l ∼ 60 for CDM-like models as well as defect mod- black holes could form at high redshift to produce such high els; luminosity sources; and whether high luminosity QSOs are • (vi) we expect this to be a rare phenomenon, occurring particularly efficient at producing mechanical outflows are only in 1 – 3 % of all high-redshift QSOs; and finally, separate issues to be contemplated if further observations • (vii) the thermal effect is one or two orders of magnitude are consistent with the model proposed here. smaller than the kinematic one. In principle, the effect ought to be observable for QSOs in cases with the appropriate opening angle and alignment ACKNOWLEDGEMENTS of outflow to the line of sight, and preferentially for high mechanical luminosity, high redshift, radio-quiet QSOs. We We thank Alastair Edge, Andy Fabian, Mike Jones, Richard note with interest the very large scale diffuse radio emission McMahon, Martin Rees and Richard Saunders for valuable seen on two sides of A3667 (Röttgering et al. 1997). If the discussions. SS acknowledges the support of the European source there is due to outflow from a now quiescent black Union, through a TMR Cat. 30 Marie Curie Personal Fel- hole in the central galaxy of the primary cluster, which is lowship. We thank the referee Sterl Phinney for invaluable one of the possible explanations, then future observations feedback that led to significant reorganizing of the paper. should reveal a supermassive black hole in the cD, primarily through a sharply rising central dispersion in the inner regions of the galaxy. From the scale of the outflow, REFERENCES if it was powered by an AGN outflow, a black hole mass 10 Aghanim, N., Desert, J., Puget, J., & Gispert, R., 1996, A&A, of ∼ 10 M⊙ accreting at near Eddington luminosity for 311,1 at least a Salpeter time is required, and the activity would Babul, A., & White S. D. M., 1991, MNRAS, 253, 31 > 9 have peaked ∼ 10 years ago. Barvainis, R., Lonsdale, C., & Antonucci, R., 1996, AJ, 111, 1431 Since the physical model that we have presented here is Bicknell, G., 1994, ApJ, 422, 542 speculative, we briefly summarize below the problems pre- Birkinshaw, M., 1996, ‘The Sunyaev-Zeldovich Effect’, Phys. Re- sented by the observations of S–Z decrements in QSO fields ports, in press with no detected clusters: Blandford, R., Phinney, E.S., & Narayan, R., 1987, ApJ, 313, 28 (i) A wind-driven model composed of a pair plasma is Daly, R.A., 1987, ApJ, 322, 20 somewhat analogous to the state of the bubble during the Elvis, M., Fiore, F., Wilkes, B., McDowell, J., 1994, ApJ, 422, 60 early phases of our proposition here but such material is Fabian, A.C., Crawford, C.S., 1990, MNRAS, 247, 439 too hot and would have a short lifetime, and, more impor- Falcke, H., Gopal-Krishna, & Biermann, P., 1996, A&A, 298, 395 Goodrich, R., 1997, ApJ, 474, 606 tantly, the maximal scale out to which such plasma could be Haehnelt, M.G., Rees, M.J., 1993, MNRAS, 263, 168 swept out is only of the order of ∼ 100 kpc, insufficient to Hazard, C.,Morton, D., Terlevich, R., & McMahon R., 1984, ApJ, explain decrements on large scales ∼ 1 Mpc. Furthermore, 282, 33 such plasma should produce easily observable diffuse radio Jones, M., et. al, 1997, ApJ Lett., 479, 1 emission as it collides with the IGM. The observational im- Kronberg, P., 1994, Reports on Prog. in Physics, 57, 325 plications of this variant have been discussed in Natarajan, Kneissl, R., et. al, 1997, preprint Sigurdsson & Silk (1998). Leahy, J., Muxlow, T., & Stephens, P., 1989, MNRAS, 239, 401 (ii) It is possible that decrements can be caused by Lehnert, M. D., & Becker, R. H., 1998, A&A, 332, 514 warm gas in the potential well of proto-clusters. The gas Lehto, H. J., & Valtonen, M. J., 1996, ApJ Lett., 460, 207 would not be in virial equilibrium, such a scenario might McKee, C. & Ostriker, J., 1977, ApJ, 218, 148 Natarajan, P., Sigurdsson, S., & Silk, J., 1998, MNRAS, in press fit with the essential outflow model we present here, but van Ojik, R., Röttgering, H.J.A., Carilli, C.L., Miley, G.K., Bre- in the limit of a more modest spherical outflow. Any de- mer, M.N., Macchetto, F., 1997, A& A, 321, 389 tected decrement would then be the signature of the hot gas Ostriker, J., & Cowie, L., 1981, ApJ Lett., 243, L127 formed by an asymmetric collision of the unconfined out- Peebles, P., 1995, Principles of Physical Cosmology, Princeton flowing gas with cold infalling gas. If the infall is strongly Univ. Press, Princeton, pg. 587. non–spherical, as is likely in unvirialized proto–clusters, this Phinney, E.S., 1983, PhD Thesis, Univ. of Cambridge Sunyaev-Zeldovich decrements with no clusters? 7

Phinney, E.S., Blandford, R.D., 1986, Nature, 321, 569 Rees, M. J., 1993, Proc. Natl. Acad. Sci. USA, Vol. 90, pp. 4840 Rees, M. J., & Ostriker, J., 1977, MNRAS, 179, 541 Richards, E., 1997, AJ, 113, 1475 R¨ottgering, H.J.A., Wieringa, M.H., Hunstead, R.W., Ekers, R.D., 1997, MNRAS, 290, 577 Saunders, R., et. al, 1997, ApJ Lett., 479, 5 Schneider, D., Schmidt, M., & Gunn, J., 1994, Astron. J., 107(4), 1245 Sunyaev, R., & Zeldovich, Y., 1970, Astrophys. Space Sci., 7, 3 Sunyaev, R., & Zeldovich, Y., 1972, Comments Astrophys. Space Phys., 4, 173 Sunyaev, R., & Zeldovich, Y., 1980, MNRAS, 190, 413 Taylor, G., Dunlop, J., Hughes, D., Robson, E., 1996, MNRAS, 283, 930 Weymann, R., Morris, S., Foltz, C. & Hewett, P., 1991, ApJ, 373, 23