Ufos) Impact of Ufos
Analogies/differences in black-hole driven massive outflows found in AGNs, QSOs, ULIRGs, and galactic BH binaries
Massimo Cappi INAF/IASF-Bologna
Outline
1. Framework/importance A very brief recall on context: AGN feedback
2. From the “classic” X-ray view of winds/outflows to the “new” X-ray view Warm Absorbers (WAs)à Ultra-Fast Outflows (UFOs) Impact of UFOs
3. Understanding analogies and differences comparison with WAs comparison with molecular outflows comparison with binaries/microquasars (on-going work…)
Tombesi F., MC, et al. ’10a+b;’11a;’12a, ‘12b in prep. (and ESA/NASA/INAF press release) Main Collaborators: F. Tombesi, M.Giustini, M. Dadina, V. Braito, J. Kaastra, J. Reeves, G. Chartas, M. Gaspari, C. Vignali, J. Gofford, G. Lanzuisi Framework: Co-evolution of AGN and galaxies ~20 years ago, a somewhat unexpected “revolution” in extragal. astrophysics: not only most (all?) galaxies have SMBHs in their centers, these also correlate with bulge properties
Kormendy & Richstone, 1995, ARA&A
AGN Feedback !? Framework: Three major feedback mechanisms between the SMBH and its environment
• L M c 2 3C 273 1. radiative feedback: acc = η( acc )
Able to quench the star formation and the cooling flow at the center of elliptical galaxies e.g. Ciotti & Ostriker 2001, Sazonov et al. 2005
But it is not enough to reproduce the MBH-σ relation Light € e.g., Ciotti et al. 2009
3C 273 2. mechanical/kinetic feedback: i. mass outflows from collimated, radiatively bright, relativistic radio JETS:
Heat the IGM and the ICM, quench the cooling flow in rich Clusters of Galaxies Jets e.g. Fabian et al. 2009, Sanders et al. 2009
ii. mass outflows from wide angle, radiatively dark, massive WINDS/outflows
e.g., Silk & Rees 1998 e.g., Begelman 2003
Winds The “classic” X-ray view: Warm Absorbers in nearby Seyferts and QSOs
Seyfert galaxies: ASCA… Many details from Chandra/XMM gratings NGC3783 Exp=900 ks
Fabian, et al. ’94 Otani, ’95, PhD Reynolds et al. '97 Georges et al. '97 Kaspi et al. '01; Netzer et al. '02; Georges et al. '03; Krongold et al. ‘03
QSOs: XMM… R. G. Detmers et al.: Multiwavelength campaign on Mrk 509 Mrk 509 RGS: 600 ks
Porquet et al. 2004; Piconcelli et al. 2005
Fig. 1.KaastraFluxed stacked et RGSal. 2011, spectrum Detmers in the 7–38 Ået range. al. 2011 The strongest Fig. 2. Unabsorbed spline continuum model used for the Mrk 509 ob- lines are indicated and the O ISM edge can be clearly seen around servations. ⇒ à Clear now that ~50% of all Seyferts and QSOs present multiple ionization & kinetic 23 Å. components (from Optical, UV and soft X) of outflows/winds with v~100-1000 km/s Table 1. Spline continuum parameters. SPEX 2.03.001 spectral fitting package to fit the spectrum. We Wavelength Flux ⇒ à Typically energetically unimportant for feedback i.e. Blustin et al. 2004, but see Crenshaw & Kraemer, 2012 2 1 1 updated the wavelengths of some important transitions for our (Å) (photons m− s− Å− ) study (see Appendix A). 5.00 0 We constructed the spectral energy distribution (SED) for 5.95 0.7 0.6 Mrk 509, using the EPIC-pn and OM data to obtain the neces- 7.07 13.95± 0.09 ± sary flux points for the XMM-Newton observations and extend- 8.41 14.28 0.06 10.00 15.10 ± 0.04 ing it with other data. This SED is an average of the Mrk 509 ± SED during the time of observations (roughly two months time). 11.89 15.92 0.09 14.14 17.27 ± 0.09 The full procedure on how the SED was derived can be found 16.82 19.36 ± 0.16 in Kaastra et al. (2011b). The ionization balance calculations 20.00 21.31 ± 0.11 needed for our spectral modeling (the xabs components, see 23.78 25.15 ± 0.08 Sect. 3.2)werebasedonthisSEDandperformedusingversion 28.28 27.83 ± 0.11 C08.00 of Cloudy2 (Ferland et al. 1998)withLodders & Palme 33.64 33.62 ± 0.19 (2009)abundances. 40.00 9.80 ±0.18 ±
2.3. Spectral models The ionized outflow is modeled with three different mod- The unprecedented quality of the spectrum requires a rather els, each with multiple (two or three) velocity components to complex spectral model to be described accurately. We describe account for the separate outflow velocities observed. All mod- each component in more detail in separate sections, but we give els take a wide range of ionization states into account. These ashortoverviewofthetotalmodelhere. models are described in more detail in Sect. 3.2.Wealsoin- We modelthe continuumwith a spline (see Fig. 2). The main cluded eleven broad and narrow emission lines, which are mod- reason for doing so is that a spline can accurately describe the eled with a Gaussian line profile. Radiative recombination con- (complex) continuum shape without having to make any physi- tinua (RRCs) are also included using an ad-hoc model that takes cal assumptions about the origin of the shape of the continuum the characteristic shape of these features into account. (powerlaw, blackbody, Comptonization, or reflection, for exam- ple). We use a redshift z = 0.03450, which combines the cos- 3. Spectral analysis mological redshift (Huchra et al. 1993)withtheorbitalveloc- ity of the Earth around the Sun, which is not corrected for in 3.1. Continuum, local absorption, and emission features the standard XMM-Newton analysis (see Kaastra et al. 2011b). 24 2 The continuum is modeled with a spline with a logarithmic spac- Galactic absorption (NH = 4.44 10 m− , Murphy et al. 1996) is also taken into account. We use× three distinct phases for the ing of 0.075 between 5 and 40 Å. We show the spline model in Galactic ISM absorption, a neutral (kT = 0.5eV)phase,awarm Fig. 2 and in Table 1.Thecontinuumissmooth,sothespline (kT = 4.5eV)slightlyionizedphase,andahot(kT = 140 eV) does not mimic any broad line emission features. The softening highly ionized phase (Pinto et al. 2010). Additionally we model of the spectrum at longer wavelengths can be seen clearly. the neutral oxygen and iron edges of the ISM by including a The neutral Galactic absorption is responsible for the narrow dusty component. Details about the Galactic foreground absorp- O (23.5 Å) and N (31.3 Å) absorption lines. To fit the Galactic tion are given by Pinto et al. (in prep.). O absorption line we add a slightly ionized component with a temperature of 4.5 eV and with acolumndensitythatis4%of 1 See http://www.sron.nl/spex the cold, neutral gas (Pinto et al., in prep.). To properly model the 2 http://www.nublado.org/ O edge, we take the effects of depletion into dust into account.
A38, page 3 of 17 The Suzaku view of highly-ionised outflows in AGN. 15
eter values are broadly consistent with those found by T10A using XMM-Newton. The Fe K absorbers detected with Suzaku cover a wide range of column densities, ranging
from 21.5 < log NH 6 24.0, with a peak in the distribution at log NH 22 23. As shown by the dot-dashed (blue) ⇡ and dotted (black) lines, which show the mean log NH value as found with Suzaku and XMM-Newton, respectively, the mean is log NH 23 for both samples. From the middle
⇡ panel the ionisation parameters span from 2.5 < log ⇠ 6 6 with the significant fraction of He↵-Ly↵ pair systems, which persist over only a relatively narrow range in ionisation pa- rameter (see T11 curve of growth analysis), leading to a sharp peak in the distribution at log ⇠ 4. Again, as shown ⇡ by the vertical lines the mean ionisation parameter in both samples is almost identical at log ⇠ 4.5. ⇡ The detection of relatively lowly ionised material in the Fe K band, i.e., log ⇠ 2.5 3.0, is particularly interesting ⇠ and suggests that high velocity absorption could feasibly be detected at softer X-ray energies through weak, moderately ionised, iron lines. Moreover, the detection of a small fraction of absorbers with log ⇠ 5 in both this work and in the The “new” X-ray view: UFOs> (Ultra-Fast Outflows) confirmed and quite common T10A sample raises the possibility that material may be present in some sources which is so highly ionised that even XMM-Newton sampleiron isof not nearby detectable AGNs through (Seyferts spectroscopy.) If this is the case • 36 absorption lines detected in all then the fraction of sources with Fe K absorption ( 40%) ⇠ 104 XMM observations may represent a lower limit on the number of sources with intrinsic nuclear outflows along the line-of-sight. • Identified with FeXXV and The log vout distribution (bottom panel) appears to be relatively continuous over a broad range of velocities; FeXXVI K-shell resonant 1 ranging from as low as vout 0kms up to vout 1 ⇡ ⇡ absorption 100, 000 km s .90%ofthedetectedoutflowshavevout >
1 1000 km s which makes the absorption detected at Fe K • 19/44 objects with absorption almost systematically faster than the traditional soft X-ray lines (≈43%) warm absorber. Only NGC 3227, NGC 4395 and NGC 3783 have Fe K outflows which are consistent with having no out- • 17/44 objects with blue-shifted flow velocity. From Table 5 the broadly binned distribu- -2 tions of outflow velocity appear to be very similar between absorption lines (lower limit ≈39%, Log Nh (cm ) Log ξ (erg s-1 cm) Blueshift (v/c) both the Suzaku and XMM-Newton samples, with the bulk can reach a maximum of ≈60%) 1 Tombesi, MC, et al. 2010,of outflows 2011 in both(A&A, samples 521, having 57; ApJvout, >742,10, 000 44) km s . The Suzaku sample does appear to have slightly more low- • 11/44 objects with outflow intermediate velocity systems but, given the low number velocity >0.1c (≈25%) Suzaku sampleThe Suzaku of statisticsAGNsview of involved,( highly-ionisedSey+RGs+RQQs the di↵erences outflows are) in probably AGN. not signif- 5 icant. Indeed, the similarity between the Suzaku and XMM- Figure• Blue- 8. Histogramshift velocity showing the distribution overall distributions of mean Newton parameter distributions can be quantitatively as- absorber parameters for each source: (a) logarithm of the mean . sessed using the Kolmogorov-Smirnov two-sample test (K-S ~0-0.3c, peak ~0.1c oDf l1 fEs l1 oFf column density; (b) logarithm of the mean ionisation parameter; Test) which uses the maximum di↵erences between the cu- (c) logarithm of the mean outflow velocity. The dot-dashed (blue) • Average outflow velocity - mulative3 fractional distribution of two5 sets of data to de- and dotted (black) vertical lines indicate the mean value of the termine the probability that they are both drawn from the Suzaku0.110±0.004and XMM-Newton c analyses, respectively. same parent. sample. In this case such− a test can be used / to quantify the level at which the column density, ionisa- tion parameter/ and outflow velocity distributionst as mea- Table 5. Outflow velocity comparison g sured with both Suzaku and XMM-Newton are in agree-
1XPEHU ) RI VRXUFHV 1XPEHU ) RI VRXUFHV 1XPEHU ) RI VRXUFHV 1 ment.g For a null hypothesis (H0)thateachoftheg Suzaku Velocity (km s ) Suzaku XMM-Newton and XMM-Newton distributions are drawn from the same No outflow 3/20 2/19 8 parent1 sample we are unable to conclusively1 rule out the íl í841 8 841 l l41 g g41 g6- ξξ6- / /6- - -6- . íg6/ íg íl6/ íl í16/ 1 0
21 −2 24 the measured column densities cover a wide range (i.e., 10 ∼< NH/ cm ∼< 10 ) 22 −2 23 −2 there is a peak at (3 − 10) × 10 cm and a mean value of NH,suzaku ≈ 1 × 10 cm . This is consistent with the analagous values found with XMM-Newton.Thedistribution −1 of ionisation parameter [panel (b)] covers the range range 3.1 ∼< log(ξ/erg cm s ) ∼< 5.5and,withapeakandmeanvalueatlog(ξ/erg cm s−1) ≈ 4.0, is again consistent with the results found by Tombesi et al. (2010). Despite this good agreement in terms of intrinsic absorber properties the current distrubtion ofoutflowvelocitiesappearsto differ somewhat, as is shown in panel (c). Whereas there is a sharp peak at ∼ 0.1c found with XMM-Newton,wefindasmootherandmorecontinuousrangeofvelocities with Suzaku,rangingfrom0.004 ∼< vout/c ∼< 0.5, with very few at the expected peak velocity of 0.1c (e.g.,seeKing2010). However,giventhatanalysisofthe sample is ongoing it is currently too soon to determine whether this is an intrinsic property of the absorbers or a result of, for example, low number statistics or intrinsic detector biases. The full Suzaku outflow sample and a thorough discussion regarding all of the work outlined here will be given in Gofford et al. (2012; in preparation).
References
Bautista, M. A., & Kallman, T. R. 2001, ApJS, 134, 139 Ferrarese, L., Cˆot´e, P., Dalla Bont`a, E., Peng, E. W., Merritt, D., Jord´an, A., Blakeslee, J. P., Has¸egan, M., Mei, S., Piatek, S., Tonry, J. L., & West, M. J. 2006, ApJ, 644, L21. arXiv:astro-ph/0603840 Gofford, J., Reeves, J. N., Turner, T. J., Tombesi, F., Braito, V.,Porquet,D.,Miller,L.,Kraemer, S. B., & Fukazawa, Y. 2011, MNRAS, 414, 3307. 1103.0661 Hopkins, P. F., & Elvis, M. 2010, MNRAS, 401, 7. 0904.0649 King, A. R. 2010, MNRAS, 402, 1516. 0911.1639 Protassov, R., van Dyk, D. A., Connors, A., Kashyap, V. L., & Siemiginowska, A. 2002, ApJ, 571, 545. arXiv:astro-ph/0201547 Reeves, J. N., Gofford, J., Braito, V., & Sambruna, R. 2010, ApJ, 725, 803. 1010.2409 Ross, R. R., & Fabian, A. C. 2005, MNRAS, 358, 211. arXiv:astro-ph/0501116 Tombesi, F., Cappi, M., Reeves, J. N., Palumbo, G. G. C., Braito, V., & Dadina, M. 2011, ApJ, 742, 44. 1109.2882 Tombesi, F., Cappi, M., Reeves, J. N., Palumbo, G. G. C., Yaqoob, T., Braito, V., & Dadina, M. 2010, A&A, 521, A57. 1006.2858
Tombesi et al. 2010, 2011 (A&A, 521, 57; ApJ, 742, 44) The “new” (unifying) X-ray view of UFOs and non-UFOs (WAs)
INAF Press releases in 2010, 2012 and 2013 (also NASA and ESA in 2012) out
log Ė à UFOs kinetic energy >1% of Lbol à Feedback (potentially) effective!
Tombesi, MC et al., 2012b log Lbol The “new” X-ray view: Variable absorption lines Absorbers variability on timescales 1000-10000s NGC1365 Mrk 509 (first long-look, 200ks)
Obs1
Obs2
Obs3
Risaliti et al. 2005 (See also Krongold et al. 2007 on NGC4051; MC et al., 2009 Behar et al. 2010 on PDS456, Braito et al. 2007 on MCG5-23-16; etc.) Dadina et al. ‘05 Variability allows to place limits on location, mass, etc. The “new” X-ray view: Variability in (nearby) PG QSOs Sample: 15 UV *AL QSOs with 32 XMM exposures
Δt=6 months Δt=4 yrs
on time scales of months on time scales of years
Δt=10 ks Δt=3 days
on time scales of hours
on time scales of days Giustini, MC, et al. 2012 Relativistic outflow or absorption edge in the z=2.73 QSO HS 1700+6416? G. Lanzuisi1 ; M. Giustini1; M. Cappi1; M. Dadina1; G. Malaguti1; C. Vignali2 1 INAF/IASF-BO; 2 Universit‡ di Bologna
Abstract We present the detection of broad absorption features in the X-ray spectrum of the quasar HS 1700+6416, indicating either the presence of high velocity out-flowing gas or a huge absorption edge from Fe. HS 1700+6416 is a high-z (z=2.735), high luminosity quasar, classified as a Narrow Absorption Line (NAL) QSO. One broad absorption feature is clearly visible in the 50ks Chandra observation taken in 2000, while two similar features, at different energies, are visible when the 8 contiguous Chandra observations carried out in 2007 are merged together. The XMM-Newton observation taken in 2002, despite strong background flares, shows an hint of such a feature at lower energies.
HS 1700+6416 (z=2.735) is one of the most luminous quasar in the SDSS. It is classified as a NAL-QSO, showing narrow CII, CIV, SiIII and Si IV absorption lines in the SDSS spectrum (Fig. 1, from SDSS_DR3), 1 blueshifted at mildly (~0.1c) relativistic velocities (Misawa et al. 2007).
X-ray coverage The source lies in the same field of 2 clusters (Abell 2246 z=0.225; V1701+6414, z =0.45) and a proto-cluster at z=2.3 (Digby-North et al 2010) and therefore have very good X-ray coverage available. In particular: One 50 ks Chandra observation in 2000; one 30 ks XMM observation in 2002 and 8 x 15-30 ks Chandra observations in 2007.
Short and long term X-ray variability Given the exposure times, and the flux level of the source, the study of short term variability is feasible only for the long, 50 ks Chandra observation. Fig. 2 shows the 0.5-8 keV light-curve of HS 1700+6416 with a bin size of 3 ks (at least 20 net counts per bin). When fitted with a constant, the resulting count rate is 6.8x10-3. The source results to be marginally variable on time scales of few ks (P(2/)=0.25). We also studied the long term variability of the continuum parameters (F(2-10), and NH, fig 3,4,5). A long term variability is clearly detected in the 2-10 keV flux, that varies of a factor 3, from 9x10-14 to 3.5x10-14 erg s-1 cm-2, and in the 22 2 3 4 5 amount of neutral absorption, that is <10 cm-2 in 2000 and 2002 data, and become consistent with 4-8x1022 cm-2 in 2007. UFOs seen also in high-z QSOs For the photon index the error bars are too large to draw any firm conclusion. In the 8 observations of 2007, the source spectral parameters remain almost constant, thus we added together these 646 CHARTAS ET AL. Vol. 706 spectra to increase the statistics. Table 1 Log of Observations of APM 08279+5255
a b c c Observation Date Observatory Observation Time Nsc f0.2 2 f2 1050 ks Chandra spectrum (2000) − − 1 2 1 2 ID (ks) net counts (erg s− cm− )(ergs− Fig.cm− )6 shows the spectrum of HS1700+6416 obtained from the 50 ks Chandra observation. The counts are binned to a minimum significant detection of 3, +0.1 +0.1 2002 Feb 24 (epoch 1) Chandra 2979 88.82 5,627 75 1.8 0.1 4.3 0.1 ± +0− .1 +0−for.1 plotting purpose (we applied the Cash statistic). The fit to a simple absorbed power-law model shows significant residuals around ~3 keV, suggesting the 2002 Apr 28 (epoch 2) XMM-Newton 0092800201 83.46 12,820 139 1.9 0.1 4.1 0.1 ± +0− .1 +0− .1 2007 Oct 06 (epoch 3) XMM-Newton 0502220201 56.38 11,400 114 2.5 0.1 3.9 presence0.1 of a strong absorption feature. When adding an absorption Gaussian line, the C-stat is 20.3 for 3 additional parameters. ± − − 2007 Oct 22 (epoch 4) XMM-Newton 0502220301 60.37 16,698 133 3.5+0.1 5.0+0.1 0.1 The0.1 significance is >4 with F-test, confirmed with extensive Monte-Carlo simulations. The rest frame line energy is E_line=10.26±0.75 keV, the line width ± +0− .2 +0− .2 2008 Jan 14 (epoch 5) Chandra 7684 88.06 6,938 83 1.9 0.2 4.5 0.2 ± − −=1.6±0.5 keV and the equivalent width EW=-0.83 keV (rest frame). Fig 7, 8 show the 68, 90, 99% confidence contours of E_line vs. and E_line vs. Notes. a Time is the effective exposure time remaining after the application of good time-interval (GTI) tables and the removalNormalization, of portions respectively. If the absorption feature is due Fe XXV or Fe XXVI K , the observed E_line translates in an outflowing velocity v_out=0.38±0.10 of the observation that were severely contaminated by background flaring. c. If modeled with the ionized absorber model XSTAR (z=2.73)(Kallman & high-zBautista RQ2001), (NAL) we have QSO to add HS1700+6416 two ionized gas shells, with slightly different v_out, and b Background-subtracted source counts including events with energies within the 0.2–10 keV band. The source counts and effective exposure times for the XMM-Newton observations refer to those obtainedQSO with the space EPIC PN instrument.density See Section 2 forboth details with on highSFR Nh spaceand density source and background extraction regions used for measuring Nsc. 23 -2 c 13 2 1 (Nh>4x10 cm , Log >3.3) and 6 The absorbed fluxes (in units of 10− ergs cm− s− )inthe0.2–2keVand2–10keVobserved-framebandareobtainedusingtheMadau et al. ’96; Wall et al. ‘05 model APL+2AL (model 6; Section 3). The errors are at the 68% confidence level. turbulence velocity v_turb=5000 km/s, to 7 8 Both approaches resulted in values for the fitted parameters reproduce the huge width of the feature. that were consistent within the errors, however, the fits to the Fe XXV Epoch 1 rest-energy If modeled with an absorption edge, the rest higher quality pn data alone provided higher quality fits as 0 indicated by the reduced χ 2 values of these fits. We therefore frame edge energy is E_edge=8.95±0.30 keV consider the results from the fits to the pn data alone more and the absorption depth is =1.85±0.83. reliable especially for characterizing the properties of the X-ray ∆χ absorption features. E_edgeAPM 08279+5255 is consistent with K shell ionization ACIS Chandra For the reduction of the Chandra observations we used -10 -5 thresholds of Fe XVI-FeXXVI with 0 v_out standard CXC threads to screen the data for status, grade, and 2002 Feb 24 time intervals of acceptable aspect solution and background (Hasinger et al. 2002). levels. The pointings placed APM 08279+5255 on the back- Fe XXV For all the Epoch3 models, 5 the C-stat is similar, illuminated S3 chip of ACIS. To improve the spatial resolution, rest-energy 0 we removed a 0"".25 randomization applied to the event i.e. the quality of the data do not allow to positions in the± CXC processing and employed a sub-pixel distinguish between the different scenarios. resolution technique developed by Tsunemi et al. (2001). ∆χ
In both the XMM-Newton and Chandra analyses, we tested the APM 08279+5255 sensitivity of our results to the selected background and source- ACIS Chandra -10 -5 Lanzuisi et al., ’12 extraction regions by varying the locations of the background Merged2008 Jan Chandra14 Spectrum (2007) XMM spectrum (2002) regions and varying the sizes of the source-extraction regions. We did not find any significant change in the background- 12The merged5 spectrum10 has ~1000 countsHS1700: above 1 keV, The and shows 4° high-two featuresz QSO The to 30 ksshow XMM observationvariable of, high-v, 2002 is affected by strong background subtracted spectra. For all models of APM 08279+5255, we Observed-Frame Energyat (keV)~2.2 and ~3.2 keV (fig. 9). The detection for two Gaussian lines, with EW1=- flares. The resulting net exposure is only ~10 ks for pn and MOS cameras included Galactic absorption due to neutral gas with a column Figure 1. ∆χ residuals between the best-fit Galactic absorption and power-law high-Xi absorbers, but the 1° non-lensed 20 2 model and the Chandra ACIS spectra of APM 08279+5255.0.14 and This model EW2=-0.50 is fit keV, and v_out 0.25±0.05c and 0.55±0.08c, has (~300 counts). density of NH 3.9 10 cm− (Stark et al. 1992). All APM 08279+5255 (z=3.91) quoted errors are= at the× 90% confidence level unless mentioned to events with energies lying within the ranges 4.5–10 keV. The arrows indicate the best-fit energies of the absorption lines of thesignificance first and second outflowof ~2 and >3, respectively. Fig 10, 11 show the confidence Despite the bad data 12 otherwise. componentsVout~0.2-0.76 for epoch 1 (top c panel) and epochChartas 5 (lower panel) obtainedet al. in fits 2009 N.B.: Would be nice also to confirm it via longer XMM observations…. that used model 6 of Table 2. contours for the absorption lines parameters. From the XSTAR model results Quality, an hint of the 2.2. Chandra and XMM-Newton Spectral Analysis of 23 -2 APM 08279+5255 column densities Nh=3-5x10 cm , and high ionization parameters (Log>3.2) presence of an absorption We first fitted the Chandra and XMM-Newton spectra of APM We proceed by fitting the spectrain of both APM cases. 08279+5255 In the edge model the two edges have E_edge1=8.14±0.52 and feature around 2 keV 08279+5255 with a simple model consisting of a power law with with the following models: (1) APL; (2)E_edge2=11.20±0.60 APL with a notch keV, the latter consistent with a Fe XXVI edge with can be seen in the neutral intrinsic absorption at z 3.91 (model 1 of Table 2). (APL+No); (3) ionized-APL with a notch (IAPL+No); (4) These fits are not acceptable in a= statistical sense as indicated APL with an absorption edge (APL+Ed);v_out (5) ionized-APL of 0.18c. with residuals (fig. 12). The by the reduced χ 2.Theresidualsbetweenthefittedsimple an absorption edge (IAPL+Ed); (6) APL with two absorption C is 15.7 and the lines (APL+2AL); (7) ionized-APL with two absorption lines absorbed power-law (APL) model and the data show significant 10 absorption for energies in the observed-frame band of <0.6 keV (IAPL+2AL); (8) APL with two intrinsic ionized absorbers 9 confidence level is (APL + 2IA); and 9) APL with two partially covered intrinsic (referred to henceforth as low-energy absorption) and 2–5 keV ~2.5. The E_line=8.05 (referred to henceforth as high-energy absorption). ionized absorbers (APL+PC*(2IA)). The XSPEC notations for To illustrate the presence of these low- and high-energy these models are given in the notes of Tables 2 and 3. 11 ±0.30 KeV, implies absorption features, we fit the spectra from observed-frame The results from fitting these models to the three XMM- v_out~0.14c. 4.5–10 keV with a power-law model (modified by Galactic Newton and two Chandra spectra are presented in Tables 2 absorption) and extrapolated this model to the energy ranges and 3.Forspectralfitsusingmodels3,5,and7,thelow-energy not fit. The residuals of these fits are shown in Figures 1 and 2. absorption is modeled using the photoionization model absori Conclusions: We clearly detected 'at least' 2 strong absorption features, Significant low- and high-energy absorption are evident in all contained in XSPEC (Done et al. 1992). We note that the absori observations. model is just a first approximation to what is likely a more at variable energies, in different X-ray spectra of NAL QSO HS 1700+6416. They can be due to highly ionized (Log >3.2) nearly Compton thick gas with nearly-relativistic outflowing velocities (v_out = ~0.4-0.5c), or to absorption edges at energies consistent with mildly ionized Fe at lower velocities. The source may be one of the few known X-ray BAL QSO with nearly-relativistic v_out. The quality of present data do not allow to draw stronger conclusions, and a long look observation is needed to better 11 constrain the absorption features, and check for variability of their properties on short time scales (few ks).
X-ray Universe, Berlin, 27-30 June 2011 UFOs/outflows/winds in AGNs622 & QSOs:S. A. Sim et al.Possible models
Radiatively driven accretion disc winds
Sim et al., ’08, ’10ab Murray etFigure al. 12.‘95,Sample Proga spectra computed et foral. different ‘00 viewing angle bins (left- to right-hand side) for models with differing mass-loss rates (top to bottom). Except for M˙ , all model parameters are fixed at those of the example model. The plotted spectra are all normalized to the input primary power-law spectrum.
…and/or… for these phenomena in non-spherical outflows require multidimen- These emission-line properties show that highly ionized outflows sional radiative transfer and thus our methods are particularly well may affect the Fe K region beyond imprinting narrow, blueshifted Magnetically driven winds from accretionsuited disk to describing them. absorption lines. This may have consequences for study of the Fe To illustrate the types of line profile, Fig. 12 shows a montage of KfluorescentemissionwhichoriginatesinAGNaccretiondiscs spectra from models with relatively high mass-loss rates. There is a (e.g. Fabian et al. 1989; Nandra et al. 1997; Fabian et al. 2000; wide range of Kα profile types in these models and their emission Miller 2007; Nandra et al. 2007), since it may contaminate the disc EWs, measured relative to a power-law continuum fit, can be in emission and/or lead to an apparently multicomponent emission excess of 200 eV; this is comparable to the typical broad Kα EWs line (see e.g. O’Neill et al. 2007). In this context, the red wings measured in a sample of AGN by Nandra et al. (2007). predicted for the emission lines may be of particular relevance in At low inclination angles, our model Fe Kα profiles have a near- view of the potential for confusion with the effects of gravitational EmmeringP-Cygni character:, Blandford they shows & some blueshifted absorption and redshift – however, more sophisticated 3D models going beyond Shlosmanbroad, redshifted, ’92; emission, the strength of which grows with M˙ . the smooth-flow assumption of our parametrized wind models must KatoSuch et profiles al. ‘ are03 qualitatively similar to those observed in some be examined before such a possibility can be considered in much Seyfert 1 galaxies (Done et al. 2007). greater detail. At intermediate angles, when the line of sight is close to looking down the wind cone, the narrow absorption lines are strongest but are accompanied by moderately strong, emission lines with fairly extended red wings. We note that the narrow absorption lines be- 7.3 Spectral curvature comeFukumura less prominent at very, et high al.M˙ ,2010 a consequence of the increased contributionKazanas of scattered et radiation al. 2012 in the spectra. Absorption by outflowing material has been discussed as a possible For the highest inclinations and M˙ values, the Fe Kα line is explanation for the so-called soft excess in X-ray spectra (Gierlinski´ almost pure emission, peaking at the rest energy of the line and &Done2004,2006;Middleton,Done&Gierlinski´ 2007; Schurch having a long red tail. The redward extension of the line emission is & Done 2007, 2008). In this picture, the decrease in flux above not a consequence of gravitational redshift but arises from electron 1keVisattributedtoabsorptionbylightorintermediate-mass scattering in the outflow. elements.∼ Although somewhat more diverse, our red-skewed emission-line Such absorption does occur in our models, particularly for high profile shapes are qualitatively similar to those obtained via the M˙ values (see Fig. 12) however the scale of the effect is too small in same physical mechanism in spherical outflow models (Laurent & the models presented here: the typical observed soft excess requires Titarchuk 2007; see also Laming & Titarchuk 2004) while they are a drop in flux of nearly a factor of two between about 1 and 2 significantly broader than those computed by Ro´za˙ nska´ & Madej keV (Middleton et al. 2007). Furthermore, Schurch & Done (2008) (2008) for Compton scattering in irradiated accretion discs. The have argued that very large, relativistic velocities are required for the largest Fe Kα emission EWs found by Laming & Titarchuk (2004) absorption model to work since the observed soft-excess spectra are (!4keV)substantiallyexceedthosefoundinanyofourmodels very smooth. Our current models support their conclusions since all but we note that the large EWs in their models mostly arise from cases in which significant spectral curvature arises are accompanied significantly lower ionization stages than are present in our models. by discreet features.
C C " 2008 The Authors. Journal compilation " 2008 RAS, MNRAS 388, 611–624 UFOs potentially significant/dominating AGN feedback
4 Wagner, Umemura, & Bicknell
UFO Feedback 3 that injected by the UFO over a given time. Panel b) in Fig. 4 shows that the mechanical advantage with respect to the clouds, defined here as the ratio of the total ra- dial momentum of the warm-phase at a given time to the total UFO momentum injected up to that time, in both runs B and C is much greater than unity, indicating e - cient momentum transfer.3.Thee ciency is higher than that for AGN jets by almost an order of magnitude (cf. WBU12). Figure 4 c) shows the evolution in time of the fraction of warm-phase kinetic energy to the integrated energy injected by the UFO up to time t, Ekin,w/PUFO t, and 44 1 Fig. 1.— Midplane density slices of the evolution of a 10 erg s UFO in an ISM devoid of clouds (Run A). See the electronic edition the ratio of the warm-phase internal energy to its⇥ kinetic of the Journal for a color version of this figure. energy, Eint,w/Ekin,w. In both runs B and C, the fraction of energy imparted by the jet to the warm phase peaks early at & 50% and subsequently declines slowly. Over- all, however, the energy transfer e ciency, both in terms of heating and accelerating the warm phase is higher in the case of spherically distributed clouds, compared to the case of clouds distributed in a disc. Due to a higher mechanical advantage in the first 100 kyr, the energy transfer rate is somewhat higher for⇠ UFOs than for AGN jets.
4. DISCUSSION AND CONCLUSIONS
Fig. 2.— Same as Fig 1, but for a two-phase ISM with spherically distributed clouds (Run B). The simulations presented in this work confirm that powerful UFOs are capable of generating strong, galaxy- wide feedback. Energy and momentum transfer is achieved by fast, mass-entrained flows through the porous channels of the two-phase ISM, which carry high ram pressure to clouds at all locations of the galaxy, even in the plane of the disc. In the case of spheri- cally distributed clouds, the feedback results in strongly Fig. 4.— Evolution of various quantities that gauge the feedback heated and dispersed clouds, accelerated outward from e ciency for simulations with a bulge-like or a disc-like distribu- the galaxy bulge. In the case of clouds distributed in tion of clouds: a) The density-weighted average radial velocity, the outward (positive) component of the radial velocity, the total ve- a disc, the feedback results in a rapid lift-up of clouds locity dispersion, and the line-of-sight velocity dispersion (at 45 from the plane of the disc followed by compression and Accepted for publication in the Astrophysical Journal Letters on December 20, 2012. inclination); b) The mechanical advantage as measured by the to- net inflow of warm disc material toward the center of the Preprint typeset using LATEXstyleemulateapjv.5/2/11 tal or outward-only radial momenta of clouds. c) The warm-phase galaxy. Because of the strong interaction of the freely Fig. 3.— Same as Fig 1, but for a two-phase ISM with clouds distributed in a quasi-Keplerian dis (Run C). kinetic energy as a fraction of the energy provided by the UFO and the ratio of the warm-phase internal energy to kinetic energy. expanding wind with the first obstructing clouds, the re- sults do not depend on the opening angle of the UFO. galactic plane and central BH by the turbulent, ram- together with the total velocity dispersion, tot, and the Only two ISM distributions have been presented here, pressure dominated back-flow in the bubble. The results (45 ) line-of-sight velocity dispersion, los,45, as a func- whereas negative and positive feedback e ciencies de- ULTRA FAST OUTFLOWS: GALAXY-SCALE ACTIVE GALACTIC NUCLEUSrest FEEDBACK of the simulation as the energy of the UFO is pri- of this run are similar to those of the simulations by SB07 tion of time in Fig. 4 a). We see that for the case of a pend on a range of ISM parameters. For example, for 1 1 2 marily channeled into inflating the bubble beyond. In- and Gaibler et al. (2012) for AGN jets interacting with bulge-likeA. Y. Wagner cloud distributionM. Umemura (run B),G. the V. velocities Bicknell of the jet-mediated AGN feedback the feedback e ciency de- 1 fall brought about by the surrounding overpressure and a dense galactic disc. Accepted forwarm publication phase in reach the Astrophysical several 100 km Journal s ,Letters and keep on increas-December 20, 2012. pends strongly on the mean density and mean size of turbulent backflow dominates the dense gas motions af- Runs B and C demonstrate that the feedback on the ing for the duration of the simulation. At late stages of clouds, but only weakly on cloud volume filling factor warm phase of the ISM depends strongly on the spa- ter 200 kyr resulting in net accretion. The velocity dis- the evolution, the cloudsABSTRACT are predominantly accelerated (WBU2012). Feedback by UFOs and AGN jets operate tial distribution of clouds. In all runs, however, the outward ( v v ), although their radial speed persions also saturate well below 150 km s 1 and do not We show, using global 3D grid-basedr,w hydrodynamicalr,w,out simulations, that Ultra Fast Outflows (UFOs) alike, and there are several reasons to expect this: 1) UFO-blown bubble remains in the energy-driven regime, never quiteh reachesi⇡h the escapei velocity of this system, reach the value predicted by the M– relation for this from Active Galactic Nuclei (AGN) result in1 considerable feedback of energy and momentum into the the injected powers are comparable; 2) the ram pressure despite radiative cooling in the clouds. This is consis- which is 450 km s at 0.5 kpc. The velocity disper- galaxy. interstellar medium (ISM) of the⇠ host galaxy. The AGN wind interacts strongly with the inhomoge- carried by the fast channel flow is comparable because tent with the predictions of recent analytic models by sions, however, reach values beyond those predicted by For a given kinetic power, the ratio of the mass outflow Faucher-Gigu`ere & Quataert (2012),neous, which two-phase also justify ISM consistingthe M– ofrelation, dense clouds which embedded for the simulated in a tenuous galaxy hot using hydrostatic medium. The ˙ ˙ its density is primarily determined by that of the swept our neglect of inverse-Compton cooling.outflow floods through the inter-cloud channels, sweeps up the hot ISM, and ablatesrate of and the disperses UFO to that of the jet is MUFO/Mjet 2( up hot phase; and 3) there is no dependence on open- the relations by Graham (2012) and a black hole mass of 2 ⇠ 7 1 1)/ 1000, where is the jet Lorentz factor and In the following, we use the fourthe quantities dense to clouds. measure The momentum6 10 M ofis the170 UFO km s is. primarily The values transferred of vr,w and to the denseUFO clouds& via the ing angle. Given that the energy and momentum trans- the e ciency of feedback by the UFO:ram pressurethe mean in radial the channelare⇥ comparable flow, and⇠ tothe those wind-blown found in bubbleanalogous evolvesh simulationsi in the energy-driven = v/c. The regime. momentum delivered by the UFO is con- fer mechanisms to the ISM are the same for jet- and velocity, the velocity dispersion, theAny mechanical dependence advan- on UFO opening angle disappears after the first interaction withsequently obstructing larger clouds. by a factor 2( 1)/ jet UFO 50. of AGN jet feedback (c.f. WBU12). ⇠ & UFO-driven feedback, it is reasonable to expect that the tage, and the kinetic energy of theOn clouds. kpc scales, therefore, feedbackIn run C, by the UFOs feedback operates in terms similarly of radially to outward feedback di- by relativisticIn an expanding AGN jets. wind, the momentum transfer leading e ciency dependencies on ISM parameters for the two We define the density-weighted meanNegative radial feedback outflow ve- is significantlyrected cloud stronger acceleration if clouds and are cloud distributed velocity dispersions spherically, ratherto the than acceleration in a disc. of embedded clouds in all directions scenarios are similar. This work shows in addition that locity of the warm phase, v = ⇢ v ˆr / ⇢ h r,wi In thew w latter· casew w the turbulentis noticeably backflow less e ofcient. the wind The radial drives outflow mass inflow velocity towardis the provided central by black the sum of the ram pressure and thermal the spatial distribution of the clouds (e.g., spherical or (Wagner & Bicknell 2011). The evolution of this quantity peaks early (after 50 kyr) as bulk cloud material is pushed pressure integrated over the surface of the clouds. Be- Phole. ConsideringP the common occurrence of UFOs in AGN, they are likely to be important in the in a disc) a↵ects the feedback e ciency substantially. and its outward only (positive) componentcosmological are feedbackplotted cyclesout of of the galaxy galactic formation. disc and then drops throughout the cause the surface area increases over time, this system Subject headings: galaxies: evolution — galaxies: formation — hydrodynamicsexhibits — ISM: a mechanicaljets and advantage greater than unity, and 3 Note that this definition of the mechanical advantage is some- outflows — methods: numerical care is required when assessing momentum budgets: the what di↵erent to that used by other authors, e.g. Faucher-Gigu`ere net (scalar, not vector) momentum may be larger than &Quataert(2012).
42 45 1 1. INTRODUCTION outflow kinetic power is 10 – 10 erg s . These au- Active Galactic Nuclei (AGN) winds have for a long thors also suggested that such outflows ought to have a time been considered an integral part of the feedback cy- strong feedback e↵ect on the evolution of the host galaxy. cle of galaxy formation (Silk & Rees 1998; Fabian 1999). The e↵ectiveness of any mode of AGN feedback de- The kinetic energy fed back by the wind into the inter- pends sensitively on the properties of the ISM, but there stellar medium of the host galaxy is, in principle, suf- are few detailed studies incorporating realistic multi- ficient to heat, disperse, and possibly unbind dense gas phase distributions (Saxton et al. 2005; Sutherland & and therefore inhibit galaxy-wide star formation (Cren- Bicknell 2007, SB07 henceforth; Cooper et al. 2008; shaw & Kraemer 2012). Such a mechanism, whereby the Gaibler et al. 2012; Wagner et al. 2012, WBU12 hence- central supermassive black hole (SMBH) is capable of forth). WBU12 showed with 3D hydrodynamical simu- lations of relativistic AGN jets interacting with a two- controlling the growth of its host galaxy on scales much 43 46 1 greater than its gravitational sphere of influence is an phase ISM that jets with powers 10 – 10 erg s are attractive idea to explain the apparent co-evolution of capable of e cient energy and momentum transfer to SMBH and galaxy as evidenced by the M– relation disperse the dense gas in the bulge of galaxies to veloci- (Ferrarese & Merritt 2000; Gebhardt et al. 2000), and ties commonly observed in (high-redshift) radio galaxies the shape and evolution of galaxy luminosity functions (Morganti et al. 2005; Nesvadba et al. 2008), and beyond those predicted by the M– relation, if the Eddington and BH mass functions over cosmic time (Croton et al. 4 2006; Croom et al. 2009; Merloni & Heinz 2008). ratio of the jets is greater than 10 . The dominant Mass outflows are common in all types of AGN and, force responsible for the e cient energy and momentum those most closely associated with disc winds are ob- transfer was identified as the ram pressure carried by jet served in UV and X-ray absorption lines (Crenshaw et al. streams that percolate the porous two-phase ISM. 2003). An extreme class or component of outflows re- Although UFOs, on average, have kinetic powers an or-
arXiv:1211.5851v3 [astro-ph.CO] 1 Jan 2013 cently detected in highly ionized and highly blueshifted der of magnitude less than AGN jets, their mass outflow Fe K-shell absorption lines in the hard X-ray band are rates are comparable to their accretion rates, and thus, ultra fast outflows (UFOs, Cappi 2006; Tombesi et al. for the same kinetic power, they carry considerably more 2010b). UFOs are thought to be mildly relativistic disc momentum (Tombesi et al. 2012) than jets. UFOs may 3 1 therefore substantially a↵ect the galaxy-scale ISM of the winds, with speeds v 0.01c to 0.1c (several 10 km s 4 1 ⇠ 4 host, in particular the dense, warm and cold phases of to 10 km s ), originating within 100 – 10 gravitational the ISM from which stars could form. The present let- radii of the SMBH. From a sample of 42 local radio-quiet ter tests this proposition with global 3D hydrodynamic AGN, Tombesi et al. (2010a, 2012) determined that the simulations of UFOs interacting on kpc scales with the incidence of UFOs in AGN is greater than 40%, that mass 1 two-phase ISM of the host galaxy. outflow rates are typically 0.01 – 1 M y , and that the 2. EQUATIONS, CODE, AND INITIAL AND BOUNDARY [email protected] CONDITIONS 1 Center for Computational Sciences, University of Tsukuba, 1-1-1 Tennodai, Tsukuba, Ibaraki, 305-8577, Japan Let ⇢, v, p, I, T , ⇤, , =5/3, k, and µ be the 2 Research School of Astronomy and Astrophysics, The Aus- fluid density, the three dimensional velocity vector, the tralian National University, ACT 2611, Australia pressure, the unit tensor, the temperature, the cooling 4 A. R. King
A large-scale shocked wind-ISM seen via (time-integrated, and time-resolved) spectra? 3
2 K.A.Pounds et al.
2003) and probably required for detection of a v∼0.1c flow in a very low redshift AGN with current X-ray missions. The primary aim of the present paper is to examine the complex Fe K emission and absorption profile of NGC 4051 1.5 and attempt to resolve and identify dicrete spectral features carrying information on the highly ionised outflow and re- processing of the primary X-ray continuum. We have sought
to confirm the presence of the highly ionised pre-shock wind ratio discussed in Paper I, for which scaling from the strong ab- − Figure 1. Schematic view of the shock pattern resulting fromsorption the impac seen attofanEddingtonwindontheinterstellargasofthehostgala velocities up to ∼9000 km s 1 predicts a xy. Asupermassiveblackhole(SMBH)accretingatjustabovethewind velocityEddington in the rate range drives∼0.11–0.13c. a fast wind We (velocity also examineu = v the∼ ηc ∼ 0.1c), whose 1 ionization state makes it observable in X–ray absorption liFenes. K The absorption outflow collidesline structure with the at ambient different gas continuum in the host flux galaxy and is slowed in a strong shock. The inverse Compton effect from the quasar’levelssradiationfieldrapidlycoolstheshockedgas,removingits to establish whether the highly ionised outflow compo- thermal energy and strongly compressing and slowing it over a very shortnent radial varies extent. on a This similar gas few-day may be observabletimescale to in that an inverse observed Compton continuum and lower–excitation emission lines associated with lowerinvelocities. RGS spectra The for cooled lower gas ionisation exerts the gas preshock over the ram same pressure ve- on the galaxy’s 5 6 7 8 9 10 interstellar gas and sweeps it up into a thick shell (‘snowpllocityough’). regime This (Pounds shell’s motion and Vaughan drivesa 2011a; milder hereafter outward Paper shock into the ambient − observed energy (keV) interstellar medium. This shock ultimately stalls unless theII). SMBH mass has reached the value M σ satisfying the M σ relation.
Figure 1. Schematic picture of AGN outflows. A wind with vw ∼ 0.1c)impacts the ISM of the host galaxy, producing a shock on either side of the contact disconti-
2 OBSERVATIONS AND DATA ANALYSIS 1.5 nuity. Within ∼ 1kpcofthenucleus(top),theshockscoolrapidlyandradiateaway most of their energy, leading to outflow kinetic energy ∼ (σ/c)LEddXMM-Newton.Inanenergy– 1 2 3 4 RRC driven outflow (bottom), theNGC shocked 4051 regions was expand observed adiabatically, by communicating on 15 orbits be- most of the kinetic energy oftween the wind 2009 to the May outflow, 3 which and isJune then able 15. to Here sweep we the use primarily data galaxy clear of gas. from the EPIC pn camera (Str¨uder et al.2001), which has the best sensitivity of any current instrument in the Fe K band ratio ∼ 3.2. Energy–driven outflow( 6-10 keV). Excluding high background data near the end of each orbit the total exposure available for spectral fitting Alarge–scale(! 1kpc)outflowbecomesenergydriven.Itisessentiallyadiabis ∼600 ks. Further details on the timing,atic, and flux levels and 1 has the wind energy rate, i.e. E˙out " E˙w ∼ 0.05LEdd (from Equation 2). The hot bub- ble’s thermal expansion makesX-ray the driving light into curves the host for ISM each more orbital vigorous revolution than in the are included in momentum–driven case. ObservedVaughan galaxy–wide et al. (2011).molecular Line outflows energies must be are energy– quoted throughout driven, as demonstrated directlythe by paper their kinetic adjusted energy for content the (cf. redshift Equation of 2). NGC The 4051, with the adiabatic expansion of the shockedcalculation wind pushes of line the swept–up shifts andinterstellar velocities medium also in a taking account 56789 ‘snowplow’. In King et al. (2011) we derive the analytic solution for the expansion of observed energy (keV) the shocked wind in a galaxyof bulge a pn with detector an isothermal energy mass off dissettribution. (see below) With AGN by cross-reference Figure 1. pn data summed over the whole 2009 observation plot- luminosity lLEdd,allsuchsolutionstendtoanattractorto the simultaneous data from the MOS camera (Turner et al.2001). Line widths are the intrinsic values, after allowance ted as a ratio to a simple power law between 5 and 10 keV (details in the text). Gaussian fitting in the lower panel models the Fe K for pn1 camera/3 energy resolution. Unless specified otherwise, 2ηlf ˙ = " c 2 " 2/3 1/3 −1 emissionPounds profile andand several Vaughan, possible absorption2012a,b,c lines. Details of R ve uncertaintiesσ c 925σ correspond200(lfc/ fg) tokm 90 s % confidence(4) limits. ! 3 fg " the Gaussian fits and possible identifications of the absorption Figure 2. Impact of a wind from an SMBH accreting at a super–Eddington rate on the interstellar gas of the host galaxy:lines schematic are listed in Table 1. The zero velocity threshold energy of King 2009, 2010 the putative Fe XXVI RRC is also marked. view of the radial dependence of the gas density ρ,velocity2.1u and Spectral temperature structureT .Attheinnershock,thegastemperaturerisesstrongly, in the Fe K region while the wind density and velocity respectively increaseZubovas (decrease) & by King factors 2012∼ 4. Immediately outside this (adiabatic) shock, the strong Compton cooling effect of the quasar radiation severeFigurely reduces1 (top panel) the temperature, shows EPIC and pn spectral slows and data compresses summed the windgasstill ± σ ± further. This cooling region is very narrow compared withover theshockradius(seeFig.1),andmaybeobservablethroughthe the whole 2009 XMM-Newton observation, plotted as a componentand inverse is centred at 6.4 0.1 keV with width =650 70 Compton continuum and lower–excitation emission lines.ratio TheshockedwindsweepsupthehostISMasa‘snowplough’.Thisi to the continuum determined by first fitting the data eV. smuch ∼ ∼ ∼ more extended than the cooling region (cf Fig.1), and itselfbetweendrives an 5 and outer 10 shock keV with into a the power ambient law plus ISM Gaussian of the host. emis- Linestyles: red,Three solid: absorption lines, at 6.8, 7.15 and 8.0 keV ∼ wind gas density ρ;red,dotted:ISMgasdensityρ;blue,solid,windgasvelocitysion line at ∼6.4 keV, andu;blue,dotted,ISMgasvelocity then removing the line. The aimu;green,solid,windand a possible fourth line at 7.75 keV were then modelled gas temperature T .Theverticaldashedlinesdenotethethreediscontinuitieof that simple procedures, inner was shock, to delineate contact discontinuity any emission and and outerwith shock. negative Gaussians. Lines labelled 1,3 and 4 in figure 1 absorption fine structure without introducing potential arte- were found to be narrow and were therefore fixed at a width facts from modelling the characteristic curvature in the con- corresponding to the pn resolution (σ∼60 eV). Component 2 is clearly resolved, and the Gaussian fit yielded an intrinsic liu tinuum with reflectionthis cooling or ionised shock absorption. in NGC 4051 (Pounds et al., in prep). =constant (15) 1σ width of 113±17 eV. Measured line energies (adjusted ξ The ratioPounds plot of pn et dataal (2004) to this had simple already power noted law con- a correlation of out- to the rest frame of NGC 4051) are listed in Table 1 where tinuum showsflow a narrow velocity Fe K with emission ionization line with in anthis apparent source. in this region. We thus expect a correlation betweenred veloc- wing, together with several absorption lines to higher alternative identifications are used to infer relevant outflow ity and excitation. The rapid cooling in this region impliesenergy. Gaussian fitting to the individual spectral features velocities. arapidtransitionbetweentheimmediatepostshockregimefinds the Fe K emission can be modelled with a narrow and a Component 1 is rather unambiguously identified with ( v/4, keV excitation) and the much slower andbroad cooler component. The narrow line energy (adjusted for the the resonance line of He-like Fe XXV (although a more blue- ∼ 1 compressed state. There is direct observational evidenceredshift for of NGC 4051) is 6.437±0.006 keV ( ). The broad shifted inner shell UTA is theoretically possible, Kallman et al. 2004). The Fe XXV identification yields a projected out- − flow velocity of ∼5500 km s 1,similartothestrongabsorp- 1 The summed MOS spectrum finds a narrow line energy of 6.409±0.004 keV, consistent with fluorescence from stationary and cold matter. The MOS energy calibration is believed to be used here to calculate outflow velocities are adjusted by a corre- more stable than that of the pn and all absorption line blue-shifts sponding factor of 0.996
!c 2010 RAS, MNRAS 000,1–?? How WAs/UFOs compare/relateA&A 549, A51 (2013) to (low-z) colder molecular/gas outflows?
NGC6240 A&A 549, A51 (2013) The Astrophysical Journal,742:23(18pp),2011November20NGC4151 Wang et al.
of the nucleus is fully consistent with this value of vs ,e.g.,cloud 28, which has v 190 32 km s 1 (Kaiser et al. 2000). = ± − 3.5. Mass Outflow and Kinematic Power For two of the brighter, spatially resolved [O iii]-emitting clouds seen in the HST images (clouds 1 and 2 in Figure 14;see also Figure 4 in Kaiser et al. 2000), we were able to constrain the number density for the X-ray-emitting gas with Cloudy modeling of their X-ray spectra when an inner radius r 100 pc The Astrophysical Journal,742:23(18pp),2011November20 to the nucleus and an ionizing photonWang rate etQ al. 1053 are= set. The 3 = best-fit density is log nH 2.9 0.5cm− .Weusethisnumber to resolvedensity and characterize to estimate the X-ray the mass-loss emission= ± in therate NLR. and The the kinematic power findingsof are the summarized hot phase as outflow, follows. which are key quantities that gauge the impact of AGN outflow on the host galaxy ISM. 1. The softFollowing X-ray emission Barbosa line images et al. of NGC (2009 4151)andStorchi-Bergmann (O vii, O viiiet,andNe al. (2010ix)areclearlyextendedandshowremarkable), we calculate the mass-loss rate using Mx ˙ = morphologicalnHmHvr C coincidencegA,where withvr theis biconical the radial NLR velocity mapped (taken as the ap- proximate outflow velocity), mH the proton mass, Cg the gas (a) byfilling the [O iii factor,]emission,whichsupportsacommonemission and A the area in the cross section of the out- mechanismflow. For for athe radial hot and distance cool phases of ofr the NLR4 gas.1020 cm (130 pc) and 2. Extendedthe hollow emission cone in the geometry X-ray image of Das is detected= et al.× ( along2005), we obtain A 2 2 2 40 2 = 10" πr (sin θout sin θin) 1.2 10 cm .AdoptingCg 0.11 the NW–SE sectors,− which is the= direction× of a putative 1 = (Storchi-Bergmann et al. 2010)andvr 750 km s− (Crenshaw torus.et al. This2000 may;Storchi-Bergmannetal. explain the faint rogue clouds identified2010∼ ), we obtain a mass 1 inoutflow previous rateHST Mstudiesx 2 that.1 M requireyr− ionizationand a kinetic in this luminosity for the ˙ = % 2 41 1 outflow Loutflow 1/2Mx v 1.7 10 erg s− ,or 0.3% of direction, indicating leakage= of˙ nuclear∼ ionization× instead ∼43 1 the bolometric luminosity of the AGN (Lbol 7.3 10 erg s− ; of fullKaspi blocking et al. by2005 a continuous). obscuring torus. = × 3. SpectralThe modelsM derived involving from smooth the hot continuaphase outflow (a is comparable ˙ x bremsstrahlungto the value plus derived a power law) from with near-IR emission studieslines pro- of ionized gas in 1 1 C. FeruglioFig. et al.: 5. Hα Extendedmap of NGC CO 6240in (color NGC image). 6240 CO(1–0) emission at different velocities: 350 km s− (green contours), 100 km s− (magenta NGC 4151 (Storchi-Bergmann et al. 2010), which estimated a Fig. 5. Hα map of NGC 6240 (color image). CO(1–0) emission at different velocities: 350 km s 1 (green contours), 100 km s 1 (magenta vide good descriptions of the spectra. The emission1 lines contours), with respect to the system velocity. Contours are calculated by merging D and− A configuration− data. Chandra−1.6–2 keV emission is − mass outflow rate at 1.2 M yr− along each cone. The Mx contours), with respect to the system velocity. Contours are calculated by merging D and− A configuration data. Chandra−1.6–2 keV emission is ≈ % ˙ shown by white contours. cannotvalue be uniquelyis over 10 identified times with higher the present than the spectral outflow res- rate derived from shown by white contours. NGC6240 extended X−ray emission 1 Thermal equilibrium plus shock model olution,UV andbut are optical consistent spectra with the (0.16 brighterM linesyr− seen;Crenshaw&Kraemer in % the2007Chandra), butHETGS the latter and XMM-Newton is measuredRGS much spectra closer be- to the nucleus (at low0.1 2 keV. pc). The The absorption-corrected optical-emitting X-raygas likely luminosity has a of much smaller filling in this streamer.in this The streamer. physical The physical size of size the of southwestern the southwestern tidal tidal tail tail outflowing from from the southernthe southern nucleus nucleus and colliding and with colliding the with the (b) factor than the X-ray-emitting gas, which explains the difference thein extended the measurements. emission between Ifr our130 study pc and measuredr 2kpc an outer part of the is however atis least however 15"" at,i.e.7kpc,fourtoseventimeslargerthan least 15"",i.e.7kpc,fourtoseventimeslargerthan− 4 surroundingsurrounding molecular molecular gas. Intriguingly, gas. Intriguingly, Max et al. Max (2005 et)find al. (2005)find 40 = 1 = Figure 17. (a) Superposition of a continuum-subtracted Hα image (contours; is Lsame0.3 2 keV outflow,(1.1 it0.2) could10 haveerg s− accelerated. to a higher velocity the streamersthe in streamers M 82 (Walter in M 82 (Walter et2 × 10 al. 2002 et al. ).2002 The). The CO CO extended extended thatthat excited excited H2 Hemission2 emission closely closely follows follows the Hα filament the Hα ex-filament ex- from Knapen et al. 2004) on the central 1 1 of the smoothed 0.3–1 keV – ! × ! or have= loaded± ISM× between 0.1 pc and 100 pc, and hence a emission toemission the north to the (see north Fig. (see2,leftpanel)coincideswitha Fig. 2,leftpanel)coincideswitha tendingtending eastward. eastward. The The position position of one of of the one peaks of the of the peaks east- of the east- ACIS image. The box region is enlarged in panel b. (b) A composite image 4. Photoionizationhigher M . models successfully reproduce the soft showing the relative distribution of H i emission in red and Hα emission in ˙ x
dust lane seen in HST images (Gerssen− 4 et al. 2004)andmight ern blueshifted CO emission coincides with the position where ) X-rayHow emission, does supportingM compare the dominant to the accretionrole of nuclear rate of the supermas- dust lane seen in HST− 1 images (Gerssen et al. 2004)andmight ern blueshifted CO emission coincides with the position where green. Overlaid white contours are the soft X-ray emission, and blue contours x be associated with another streamer.10 The detection of molecu- Bland-Hawthorne et al. (1991)andGerssenetal.(2004)founda ˙ be associated with another streamer. The detection of molecu- Bland-Hawthorne et al. (1991)andGerssenetal.(2004)founda the 12CO emission. photoionization.sive black holeThere in are NGC considerable 4151? variations The bolometric in ion- luminosity of lar streamer(s) in NGC 6240 confirms keV that the molecular gas is large velocity gradient of the ionized gas (the velocity decreas- 43 1 − 1 (A color version of this figure is available in the online journal.) NGC 4151, Lbol 7.3 10 erg s− (Kaspi et al. 2005), is lar streamer(s) in NGC s 6240 confirms that the molecular gas is large velocity gradient of the ionized gas (the velocity decreas- ization states across the circumnuclear region. A high- − 2 − 5 = × severely affected by galaxy interaction and that the redistribution ing, roughly, north to south). No significant radio continuum about 1.2% of its Eddington luminosity, which is LEdd 6 of molecularseverely gas is averyffected likely by galaxy the trigger interaction for and the that strong the redistribution starburst ing,emission roughly, is north detected to south). in this No region significant inthe radio VLA continuum maps of Colbert ionization45 (log U1 1) component is present in most re- +0.57 ∼7 × 5 × 10 The scenario where part of the CO gas lane is photoionized by 10 erg s− for the black hole mass MBH 4.57 0.47 10 M activity in theof centralmolecular region gas is very of NGC likely the6240. trigger for the strong starburst emissionet al. (1994 is detected). The in blueshifted this region in molecular the VLA maps gas of in Colbert this region might ∼ = − × %2 the AGN and produces the Hα arc seems plausible with a mod- gions.of NGC A low-ionization 4151 (Bentz (log U et al. 20060.25)). component Adopting is Lbol ηMaccrc activity in the central region of NGC 6240. etbe al. a ( tidal1994). tail, The blueshiftedleft behind molecular during gas the in merger. this region However, might its asso- ∼− = ˙ The blueshifted eastern CO emitting region is not associated est covering factor of fc 0.1%, implying that the arc has a presentand η along0 the.1fortheefficiencyoftheaccretiondisk(Shakura& bicone direction (NE–SW), whereas a ∼ = with the dust lanesThe mentionedabove,butblueshifted eastern− 5 CO emitting follows region an is H notα filament, associated beciation a tidal with tail, left strongly behind duringshocked the gas merger. suggests However, that its a asso- shock is propa- keV (Photons cm filamentary structure with a thickness of a few parsecs. Sunyaev 1973), this implies that the central engine of NGC lower ionization (log U 1) component is found in the 1 and PAH emission observed at 8 µm(Bushetal.2008). The ciationgating with eastward strongly and shocked is also gas comp suggestsressing that a shock the molecular is propa- gas, while On the other hand, we consider the possibility that the arc 4151 is accreting at Maccr 0.013 M yr− .Sincethemea- with the dust lanes mentionedabove,but2 × 10 follows an Hα filament, ∼− NW–SE direction, which is consistent˙ = with filtered nuclear% emission-lineand nebula PAH emission seen in observed Hα images at 8 (Gerssenµm(Bushetal. et al.20082004).)is The gatingcrossing eastward it. and is also compressing the molecular gas, while could be a bow shock feature from interaction between the bi- sured outflow rate is 160 times the accretion rate necessary interpreted as evidence of a superwind that is shock heating am- The integrated luminosity of CO in this region is L"(CO) = conical outflow and dense gas in the host disk piled in the CO emissionto feed by warm the active absorbers nucleus, instead of there a continuous must absorb- be either significant en- emission-line nebula seen− 5 in Hα images (Gerssen et al. 2004)is crossing it. Fig. 6. The Chandra X-ray map of NGC 6240 in the energy range 9 1 2
10 gas lanes by the large-scale stellar bar. Assuming that the X-ray trainment of the host galaxy’s ISM while the nuclear outflow bient ISM. Theinterpreted Hα emitting as evidence filaments of a superwind are aligned that is east-westward, shock heating am- 1.3The10 integratedKkms luminosity− pc ,correspondingtoagasmass of CO in this region is L (CO) =M(H2) = 7 ing torus. The lowest ionization component (log U 2) 0.3–4 keV. The circles indicate the regions where we extracted the X-ray 8× " × emission is due to shock heating as the outflow encounters the is expanding, or part of the gas fueling∼− the nucleus is ejected perpendicularbient to ISM. the The dust Hα lanesemitting and filaments to the are line aligned connecting east-westward, the 110.3 10M9 Kkms(assuming1 pc2,correspondingtoagasmass again α = 0.5). HypothesizingM(H ) = 7 that this out- appears to be associated with the dense gas in the host spectra. The region for the background extractiontwo is outside nuclei. theThe limits X-ray emission is associated with12 the Hα fila- flow originates$ − in the southern, 5 luminous AGN2 (or from both dense CO lane, and that the shock is strong and adiabatic, we es- before radiating, or a high fraction (over 99%) of the X-ray gas Energy (keV)8× × perpendicular to the dust lanes and to the line connecting the 10 M (assuming again α = 0.5). Hypothesizing that this out- timate the shock velocity vs ,adoptingapostshocktemperature plane.does A thermalnot take component part in the may outflow still be present and is at illuminated!12% by the AGN of this map, at 2–5 arcmin from the nuclei. ments (Lira et al. 2002). In particular, strong soft X-ray emis- the AGN),$ we derive here a mass loss rate. We assume that the 2 two nuclei. The X-ray emission is associated with the Hα fila- flow originates in the southern, luminous AGN (or from both Tps (3/16)(µv /kB )(Lehnertetal.1999), where µ is the mass of the“in total situ.” soft However, emission, perhaps the latter related seems to hot unlikely ISM in the as the grating spec- sion is coincident with theFig. northeastern 8. The background-corrected filament (N–E region Chandramolecularspectrum outflow is (also distributed corrected in a spherical volume of radius = s ments (Lira et al. 2002). In particular, strong soft X-ray emis- the AGN), we derive here a mass loss rate. We assume that the per particle (CO and HI) and kB the Boltzmann constant. For tra strongly indicate that the emission lines have outflow veloc- NGC6240 extended X−ray emission in Fig. 6). We re-analyzed the Chandra X-ray data and found Rof = 7kpc(i.e.thedistanceoftheeasternblobfromtheAGN), galactic disk or shocks associated1 with the outflow. for the response matrix of the instrument) extracted at the position the kT Figure0.3keVX-raytemperaturemeasuredatthearc,the 18. Schematic drawing of the complex circumnuclear environment of ities of 200–300 km s− (e.g., Armentrout et al. 2007). Signifi- Two temperature model strong evidencesion is of coincident shocked with gas the at northeastern the position filamentFeruglio of the (N–E Hα regionfila- et molecularcenteredal. 2013 outflow on the is AGN. distributed If the in a gas spherical is uniformly volume of distributed radius in this ∼ 1 5. The measured ratios of [O iii]/soft X-ray flux are consistent of the Hα filaments (see Fig. 6). Symbols are as in Fig. 7.Reddot- requiredNGCvs 4151.is (a)150 The km inner sfew− hundred,whichismuchhigherthanthelocal parsec radius region; (b) The 3 kpc across cant mass loading of the outflow by the host ISM may be most ments. The presencein Fig. 6). ofWe shocked re-analyzed gas the inChandra the NGCX-ray 6240 data system and found is Rvolume,of = 7kpc(i.e.thedistanceoftheeasternblobfromtheAGN), the volume-averaged density of molecular gas is given ∼ 1 Wang4 et al. 2012a,b,c 0.1 ted histogram: thermal equilibrium model component; blue3 dashed his- sound speedregion. Thecs features10 are km not drawn s− toin exact the spatial10 scale.KH The cyanα-emitting wiggly lines gas. withplausible a constant and ratio common, of 15 for the since 1.5 kpc previous radius spanned estimated NLR outflow confirmed bystrong the evidencedetection of ofshocked strong gas H at2(1–0) the position S(1) of emission the Hα fila- at centeredby ρof on= the3 M AGN.of/Ω IfR the,where gas is uniformlyΩ is the distributed solid angle in this subtended by = ∼ ∼ 1 % & of Althoughrepresent there possible is no leakage direct/scattering measurement of ionizing photons, at the whereas exact the position orange of byrates these inmeasurements. Seyfert galaxies This suggests ( 0.1–10 a similarM relativeyr− ) generally exceed 2.12 µm. NGC 6240 has thetogram: strongest shock H2 line model emission component. found in the outflow, and Mof is the mass of gas in the outflow (Feruglio ∼ % ments. The presence of shocked gas in the NGC 6240 system is volume, the volume-averaged density of molecular gas is given the Hα arc, the measured outflow velocity in [O iii]atr 8!! NE the accretion rate by tenfold (Veilleux et al. 2005). 3 wedges denote the lower ionization sectors perpendicular to the bicone (yellow= contribution from the low- and high-ionization gas phases 0.01 any galaxy (Goldaderconfirmed by et the al. detection1995). Shocks of strong are H (1–0) usually S(1) identified emission at byet al.ρ 2010= 3 M;Maiolinoetal./Ω R ,whereΩ 2012is the solid). This angle assumption subtended by is an approx- − 1 2 of of of wedges). as the excitation mechanism for this line. Ohyama et al. (2000, imation,% & and evidently cannot represent the complexity of the 15at different radii. If the [O iii] and X-ray emission arise
keV 2.12 µm. NGC 6240 has the strongest H2 line emission found in the outflow, and Mof is the mass of gas in the outflow (Feruglio
− 1 (A color version of this figure is available in the online journal.) 2003)suggestthatsuchshocksareproducedbyasuperwindany galaxy (Goldader et al. 1995). Shocks are usually identified etsystem, al. 2010;Maiolinoetal. but can provide2012 a). rough This assumption estimate is of an the approx- mass loss rate. from a single photoionized medium, this further implies an 10−3 2 as the excitationprevious mechanism for estimates this line. Ohyama if α etis al. significantly (2000, imation, higherand evidently than cannot 0.5. represent To date, the complexity of the outflow with a wind-like density profile (ne r− ). A51, page 6 of 8 Nevertheless, even with such a high M in NGC 4151, the ∝ 2003)suggestthatsuchshocksareproducedbyasuperwindthis is the lowest conversion factorsystem, measured but can provide in ana rough extragalac- estimate of the mass loss rate. ˙ x 6. The estimated mass outflow rate in NGC 4151 is kinematic power Loutflow suggests that only 0.3% of the available 1 10−4 tic object. This said, it is unlikely that the mass loss rate is less 2 M yr− at 130 pc and the kinematic power normalized counts s A51, page 6 of 8 1 accretion power is extracted to drive the outflow, similar to the ∼ % 41 1 than several tens M yr− ,anditislikelyasbigasafewhun- of the ionized outflow is 1.7 10 erg s− ,0.3%Lbol.This 1 % findings of Storchi-Bergmann et al. (2010)andHoltetal.(2006, × dred M yr− .Thekineticpoweroftheoutflowinggasisgiven value is significantly lower than the expected efficiency in % 2 42 1 2011). For a sample of Seyfert NLR outflows (NGC 4151 not 1.4 by the relation Pk = 0.5 v M˙ = 6 10 erg s− .Theageofthe the majority of quasar feedback models, but comparable to 7 × included) studied in the [S iii] λ9069 emission, Barbosa et al. 1.2 outflow is >2 10 years, since it is observed at about 7 kpc dis- 5 4 the two-stage model described in Hopkins & Elvis (2010).
ratio × (2009)reportedalowerL /L 10 to 10 .As 1 tance from the southern nucleus. The star formation rate at the outflow bol − − pointed out in Mathur et al. (2009), this= poses a problem for Placing all our findings in the context of previous studies, we 0.8 position of the eastern filament can be evaluated through both obtain a comprehensive view of the NGC 4151 circumnuclear 12 5the Hα luminosity and the X-ray luminosity (e.g. Kennicut 1998; the majority of quasar feedback models which require a higher Energy (keV) region at various spatial scales, from the innermost r 50 pc fraction of the accretion power of the black hole ( 5% of Lbol to Ranalli et al. 2003). The 0.5–2 keV X-ray luminosity at the posi- ∼ (as illustrated in the schematic drawing in Figure 18(a))∼ to as far 41 1 unity) to be thermally coupled to the host ISM. Our Loutflow/Lbol Fig. 7. Top panel:thetopcurverepresentsthebackgroundcorrected tion of the eastern filament is 0.5 1 10 erg s− ,whichaccord- − × is in rough agreement with a two-stage feedback model recently as r 2kpc(Figure18(b)). The key points are recapped here. Chandra spectrum (uncorrected for the response of the instrument; ing to Ranalli et al. (2003)wouldimplyastarformationrateof Photoionization∼ by the nucleus is important at all scales. In crosses denote spectral widths and amplitude uncertainties) extracted 1 proposed by Hopkins & Elvis (2010), which requires only 0.5% 10–20 M yr− .Thisestimateisderivedbyassumingthatallthe the nuclear region, except for a few clouds that show interaction at the position of Hα filaments (see Fig. 6), and fitted with a two- % of the radiated energy to drive the initial hot outflow to efficiently X-ray luminosity is due to star formation.In Sect. 4.2 we showed with the radio jet (Wang et al. 2009, 2011a), most of the X-ray temperature thermal equilibrium model (solid line). The lower curve that at least a fraction of the X-luminosity is due to a shock, shut down star formation in the host. represents the background spectrum. Data, represented by crosses, have NLR clouds are consistent with being part of a photoionized therefore the SFR derived from the X-ray is an upper limit. biconical outflow, with an n r 2 density profile expected for been binned to give a signal-to-noise ratio >5ineachbin(forplotting 4. CONCLUSIONS ∝ − purposes only). Bottom panel:thedata-to-modelratio(thegreensolid This would suggest that the outflow is not pushed by SN winds. anuclearwind. line representing a ratio of unity), showing the residuals at about 1 keV, Indeed, the power transferred to the ISM by a star formation- In this paper we present spectral analysis and emission line In the ENLR of NGC 4151, photoionization is still the dom- 41 42 1 1.4 keV, 1.8 keV, 2.2 keV, and 5 keV. driven wind is given by PSF = η 7 10 SFR= 10 erg s− , images from deep Chandra observation of NGC 4151, aiming inant ionization mechanism for the observed X-ray emission where η 0.1isthestandardmass-energyconversion(seee.g.× × × ∼ Lapi et al. 2005). We conclude that it is unlikely that the molec- 16 Based on this geometry, we can derive a mass loss rate by using ular flow is powered by star formation. Instead, star formation in the relation: this region is likely to be in the process of being quenched by the outflow. However, we cannot exclude that star formation in this ˙ 2 Mof M(H2) v Ω R ρof = 3 v area is induced by the compression caused by the propagating ∼ of " # R of shock. 1 where v is the terminal velocity of the outflow ( 400 km s− ). We detected a redshifted component with velocity 400 to ∼ 1 This relation yields a mass loss rate of M˙ 120 M /yr. The Hα, 800 km s− with respect to the systemic velocity (Fig. 2), cen- ∼ % H2,andCO(1–0)mapssuggestthattheoutflowismostlikely tered on the two AGN nuclei. Interestingly this emitting region conical. In this geometry, since the mass loss rate is independent is elongated in the same east-west direction as the blueshifted of Ω,themassoutflowratewouldbeequaltothesphericalcase emission discussed above, although on smaller scales ( 1.7 kpc if there are no significant losses through the lateral sides of the in diameter). We derive a CO luminosity for this component∼ of 8 1 2 cone. As it is observed in other local massive outflows (Cicone L'(CO)= 3.0 10 Kkms− pc ,whichconvertsintoagasmass × 8 et al. 2012;Aaltoetal.2012), the outflowing gas is likely char- of M(H2) = 1.5 10 M ,underthesameassumptionsasgiven × % acterized by a wide range of densities, ranging from low density above for the CO-to-H2 conversion factor. The high velocity of gas to dense clumps, which would increase the mass flow rate. In this component suggests that the AGN might contribute to the addition, the mass flow rate would obviously be larger than our dynamics of this gas (Sturm et al. 2011).
A51, page 7 of 8 2 Maiolino et al. for a tiny fraction of the total gas in the host galaxy. So far, no ob- a) servational evidence was found for the massive, quasar driven out- flows at z>6requiredbyfeedbackmodelstoexplainthepopulation of old massive galaxies at z∼2. Here we focus on one of the most distant quasars known, SDSSJ114816.64+525150.3 (hereafter J1148+5251), at z=6.4189 Submm and CO detection (Fan et al. 2003). CO observations have revealed a large reser- Evidence of strong quasar feedback in the early Universe 3 10 voir of molecular gas, MH ∼ 210 M",inthequasar in the highest-redshift quasar: 2 −1 a) host galaxy (Walter et al. 2003; Bertoldi et al. 2003b). The strong reveals broad [CII] wings extending to about ±1300 km s .These 8 9 far-IR thermal emission inferred from (sub-)millimeter observa- are indicative of a powerful outflow, in analogy with the broad • Dust mass: 10 – 10 M wings that have been observed in the molecular and fine struc- sun tions reveals vigorous star formation in the host galaxy, with ture lines of local quasar host galaxies. The spectrum is fit1 with 10 −1 SFR ∼ 3000 M" yr (Bertoldi et al. 2003a; Beelen et al. 2006). anarrow(FWHM=345 km s−1)andabroad(FWHM=2030 km • H2 mass: 10 Msun + −1 2 = J1148 5251 is also the first high redshift galaxy in which the the s )Gaussian,asshowninFig.1,resultingintoχred 1.11. By 3 [CII]158µmlinewasdiscovered(Maiolinoetal.2005).Highres- removing the broad component the χ2 increases from 195 to 229 • Star formation rate: 10 /yr olution mapping of the same line with the IRAM PdBI revealed (with 175 degrees of freedom), implying that the broad component that most of the emission is confined within ∼ 1.5kpc,indicating is required at a confidence level higher than 99.9%. co-formation of SBH and Theoretical models predict that starburst-driven winds can- that most of the star formation is occurring within a very compact not reach velocities higher than 600 km s−1 (Martin 2005; young galaxies region (Walter et al. 2009). How WAs/UFOs compare/relateThacker etto al. 2006), (high-z) therefore the high velocitiescolder observedinthe molecular/gas outflows? Previous [CII] observations of J1148+5251 did not have a [CII] wings of J1148+5251 strongly favor radiation pressure from bandwidth large enough to allow the investigation of broad wings b) the quasar nucleus as the main driving mechanism of the outflow. tracing outflows, as in local quasars. In this letter we present new Quasar radiation pressure is favored, relative to SN-drivenwinds, also based on energetics arguments, as discussed later on. IRAM PdBI observations of J1148+5251 that, thanks to the wide ULIRG SDSSJ1 14816.64+525150.3 (z=6.42) – IRAM PdBI bandwidth offered by the new correlator, have allowed us to dis- Evidence of strong quasar feedback in the early Universe 3 cover broad [CII] wings tracing a very massive and energetic out- flow in the host galaxy of this early quasar. We show that the prop- reveals broad [CII] wings extending to about ±1300 km s−1.These a) b) erties of this outflow are consistent with the expectations ofquasar are indicative of a powerful outflow, in analogy with the broad 3.2 Outflowing gas mass feedback models. wings that have been observed in the molecular and fine struc- The luminosity of the broad [CII] component allows us to infera 1 We assume the concordance Λ-cosmology with H0 = ture lines of local quasar host galaxies. The spectrum is fit with lower limit of the outflowing atomic gas mass, using, −1 . −1 −1 Ω = . Ω = . anarrow(FWHM=345 km s )andabroad(FWHM=2030 km 70 3kms Mpc , Λ 0 73 and m 0 27 (Komatsu et al. −1 2 −4 −91K/T s )Gaussian,asshowninFig.1,resultingintoχ = 1.11. By Moutfl(atomic) 0.7L[CII] 1.410 1 + 2e + ncrit/n red = . 2011). 2 0 77 −91K/T (1) removing the broad component the χ increases from 195 to 229 M" ! L" "! XC+ " 2e (with 175 degrees of freedom), implying that the broad component + is required at a confidence level higher than 99.9%. (Hailey-Dunsheath et al. 2010), where XC+ is the C abundance per hydrogen atom, T is the gas temperature, n is the gas den- 2OBSERVATIONSANDDATAREDUCTION Theoretical models predict that starburst-driven winds can- not reach velocities higher than 600 km s−1 (Martin 2005; sity and ncrit is the critical density of the [CII]158µmtransition 3 −3 Observations with the IRAM PdBI were obtained mostly in April Thacker et al. 2006), therefore the high velocities observedinthe (∼ 3 × 10 cm ). By assuming a density much higher than the critical density, Eq. 1 gives a lower limit on the mass of atomic 2011 in D configuration (mostly with 1.5 < PWV < 3.5mm), [CII] wings of J1148+5251 strongly favor radiation pressure from gas. The quasar-driven outflows observed locally are characterized while a few hours where also obtained in January 2011 in C+D the quasar nucleus as the main driving mechanism of the outflow. Quasar radiation pressure is favored, relative to SN-drivenwinds, by a wide range of densities, including both dense clumps and dif- configuration (PWV < 1.5mm).Theresultingsynthesizedbeam also based on energetics arguments, as discussed later on. fuse, low density gas (Aalto et al. 2012; Fischer et al. 2010, Cicone is 2.2$$ × 1.8$$.Thereceiversweretunedto256.172GHz,which et al. submit.), therefore our assumption on the gas density gives is the rest frame frequency of [CII] at the redshift of the quasar, aconservativelowerlimitontheoutflowinggasmass.Weassume Figure 1. IRAM PdBI continuum-subtracted spectrum of the [CII]158µm z=6.4189 (Maiolino et al. 2005). The following flux calibrators atemperatureof200K,howeverthedependenceonthetempera- line, redshifted to 256.172 GHz, in the host galaxy of the quasar tureb) is weak (going from 100 K to 1000 K the implied gas mass were used: 3C454.3, MWC349, 0923+392, 1150+497, 3C273, [CII] 158 µm broad wings (FWHM~2000$$ km/s) + extension à J1148+3.25152 Outflowing extracted from gas mass an aperture with a diameter of 4 ,top,and Figure 2. Map of the continuum-subtractedMaiolino [CII] line narrowet al. componen 2012t is within 20% of the value obtained• at 200 K). Furthermore, we − + −1 − 3C345, 1144+542, J1208+546, J1041+525, 1055+018. Uncer- $$ −1 + −4 (a), 300 < v < 400 km s ,andofthe[CII]linewings(b), 1300 < 6 ,bottom.Thespectrumhasbeenresampledtoabinsizeof85kms . assume a C abundance typical of PDRs, i.e. X + = 1.4 × 10 −1 −1 The luminosity of the broad [CII] component allows us to infera ≥ 3500C v < −300 km s and +500 < v < +1300 km s .Seetextfordetails tainties on the absolute flux calibration are 20%. The total on- The red lines show a double Gaussian fit (FWHM=345 km s−1 and (Savage & Sembach 1996), whichM is also conservative, since the on the continuum subtraction of the narrow component. The dashed circles lower limit of the−1 outflowing atomic gas mass, using, source integration time was 17.5 hours, resulting in a sensitivity FWHM=2030 km s )tothelineprofile,whilethebluelineshowsthegas-1 in the outflow is certainly, on average, at a higher state of indicate the extraction apertures of the two spectra shown inFig.1.Levels − − − of 0.08 Jy km s 1 beam 1 in a channel with width 100 km s 1. sum of theM two Gaussian components.> 3500−4 M−91K/T yrionization.; Theand luminosity ofQuasar the [CII] broad component isdriven in- are in steps of 0.64outflow Jy km s−1 beam−1 (i.e. 3σ)inthenarrowcomponent (not SB) Moutfl(atomic) 0.7L[CII] 1.410 1 + 2e + ncrit/n out= . ¤ −1 −1 0 77 −91K/T (1) ferred from the flux of this component in the spectrum extracted map (a) and in steps of 0.36 Jy km s beam (i.e. 1σ)inthebroadwings The data were reduced by using the CLIC and MAPPING M" ! L" "! X + " 2e C %% −1 map (b). The beam of the observation is shown in the bottom-left corner. At from anH apertureig ofh 6 -(Fr[CII]e(broad)so=lu6.8Jykmstion), yieldingCO packages, within the GILDAS-IRAM software. Cleaning of the re- + %% @>F&8G+B + L (broad) = 7.3 ± 0.9L .Weobtainalowerlimitontheout- the redshift of the source 1 corresponds to 5.5 kpc. sulting maps was run by selecting the clean components on an area (Hailey-Dunsheath et al. 2010), where XC is the C abundance [CII] " fully consistentper hydrogen with atom, the T value is the expected gas temperature, (4 mJy) n from is the the gas bol den-omet- flowing atomic gas mass of of ∼ 3” around the peak of the emission. For each map the resulting ULIRG Mrk231 - CO (resolved) mapO bservation of z=6.42 Quasar ric observationssity and ncrit (Bertoldiis the critical et al. density 2003a), of the once [CII]158 the frequencµmtransitionyrange 9 3 −3 M (atomic)-1> 7 × 10 M" (2) 3.3 Extension residuals are below the 1σ error, ensuring that sidelobesA&A are 518, prop- L155of (2010)FWHM~700 the(∼ latter3 × and10 thecm steep). km/s, By shape assuming M ofthe a density thermal ~ much 250-2200 spectrum higher than are the taken Moutfl yr erly cleaned away. Anyhow, as discussed in the following, thesize critical density, Eq. 1 gives a lowerout limit on the mass of atomic ¤ into account. We emphasize that this is a conservative lower limit on the total Fig. 2(a) shows the map of the [CII] narrow component integrated gas. The quasar-driven outflows observed locally are charac• terizedSpatial Distribution 2 −1 determination has been investigated directly on the uv data, there- Fig. 1a shows the continuum-subtracted spectrum, extracted mass of outflowing gas, not only because of the assumptions on the within −300 < v < +400 km s .Notethatinthiscasewe by a wide range of densities, including both dense clumps and dif- Morganti et al. (2010)fore independently reported evidence of the cleaning. for AGN-induced mas- from an aperture of 4$$ (corresponding to a physical size of 11 kpc). physical properties of the outflowing atomic gas, but also because a have subtracted a pseudo-continuum defined by the flux observed fuse, low density gas (Aalto et al. 2012; Fischer et0.03 al. 2010, Cicone– Radius ~ 2 kpc −1 0.3 $$ significant fraction of the outflowing gas is likely in the molecular at 400 < |v| < 800 km s (resulting in a level of 5.6 mJy), to mini- sive and fast outflows of neutral H in powerful radio galaxies, Fig. 1bet al.shows submit.), the spectrum therefore our extracted assumption from on a the larger gas density aperturegives of6 form. mize the contribution of the broad component. Fig. 2(b) showsthe possibly driven by the AGN jets. that, althoughaconservativelowerlimitontheoutflowinggasmass.Weassu noisier than the former spectrum, recovers residualme– Two peaks separated by 1.7 kpc map of the [CII] broad wings, where we have combined the blue 3RESULTS flux associatedatemperatureof200K,howeverthedependenceonthetempera with the beam wings and with any0.02 extended com-- Z=6.42 quasar(−1300 < v < −300 km s−1)andred(+500 < v < +1300 km s−1) The bulk of the gas in QSO hosts, i.e. the molecular phase, 0.2 ponent.ture is weak (going from 100 K to 1000 K the implied gas mass wings to improve the signal-to-noise. 3.1 Detection of broad wings • VelocFigureity D 2. Mapistr ofi thebu continuum-subtractedtion [CII] line narrowCO componen(resolved)t map is within 20% of the value obtained at 200 K). Furthermore, we − + −1 − appears little affected by the presence of the AGN. Indeed, most The spectrum+ shows a clear narrow [CII]158µmemis-−4 (a), 300 < v < 400 km s ,andofthe[CII]linewings(b), 1300 < assume a C abundance typical of PDRs, i.e. XC+ = 1.4 × 10 −1 −1 The continuum was subtracted from the uv data by estimating sion line, which was already detected by previous0.01 observations 1 vNote< − that,300 although km s and Fig.+ 1500 shows< v the< spectrum+1300 km resampled s .SeetextfordetailsV~250 in channels ofkm/s studies of the molecular gas in the host galaxies of QSOs and 0.1(Savage & Sembach 1996), which is also conservative, since t–he CO l−ine width of 280 km/s − −1 85on km the s continuum1 for sake of subtraction clarity, the of fit the was narrow performed component. on the The unbinned dashed spec circles- 2 The line core integration limits are asymmetric because the narrow com- its level from the channels at v < 1300 km s and at v > (Maiolinogas in et the al. outflow 2005; Walter is certainly, et al. on 2009). average, However, at a higher thanks state t oothef indicate the extraction apertures of the two spectra shown inFig.1.Levels Seyfert galaxies have found− narrow CO lines (with a width trum to avoid any fitting artifact associated with the binning. ponent is slightly skewed towards positive velocities. + 1 −1 −1 10 13001 km s .Theinferredcontinuumfluxis3.7mJy,whichis muchionization. wider bandwidth The luminosity relative of to the previous [CII] broad data, component our new spec is i–n-trumDyaren ina stepsmi ofc 0a.64l Jym kma s sbeam wit(i.e.hin 3σ )inthenarrowcomponentcentral 2 kpc: ~ 10 of a few 100 km s− ), generally tracing regular rotation pat- ferred from the flux of this component in the spectrum0 extracted map (a) and in steps of 0.36 Jy km s−1 beam−1 (i.e. 1σ)inthebroadwingsBertoldi et al. 2003 0 %% −1 from an aperture of 6 (F[CII](broad) = 6.8Jykms ), yielding Mmap_s (b).un The beam of the observation is shown in the bottom-left corner. At terns, with no clear evidence forprominentmolecularout- %% Walter et al. 2004 L (broad) = 7.3 ± 0.9L .Weobtainalowerlimitontheout- the redshift of the source 1 corresponds to 5.5 kpc. [CII]-1000 -500 0" 500 1000 -1000 -500 0 500 1000 flows (Downes & Solomon 1998;Wilsonetal.2008;Scoville flowing atomic gas mass of 11 Velocity [Km/s] –VelocityTo t[Km/s]al bulge mass ~ 10 M_sun et al. 2003), even in the most powerful quasars at high redshift 9 Moutfl(atomic) > 7 × 10 M" (2) 3.3 Extension ;3+))- (Solomon & Vanden Bout 2005; Omont 2007). Yet, most of the < M-sigma prediction Fig. 1.WeContinuum-subtracted emphasize that this is a conservative spectrum lower limit of on the total CO(1 Fig.0) 2(a) transition shows the map in of the [CII] narrow component integrated Feruglio et al. 2010 − 2 −1 past CO observations were obtained with relatively narrow band- Mrk 231.mass Theof outflowing spectrum gas, not was only because extracted of the assumptions from a region on the twicewithin the−300 beam< v size< +400 km s .Notethatinthiscasewe physical properties of the outflowing atomic gas, but also because a have subtracted a pseudo-continuum defined by the flux observed &%B widths, which may have prevented the detection of broad wings (full width at half maximum, FWHM), and the level ofat 400 the< | underlyingv| < 800 km s−1 (resulting in a level of 5.6 mJy), to mini- significant fraction of the outflowing gas is likely in the molecular 1 of the CO lines possibly associated with molecular outflows. continuumform. emission was estimated from the region withmizev> the800 contribution km s of− the broad component. Fig. 2(b) showsthe 1 map of the [CII] broad wings, where we have combined the blue and v< 800 km s− . Left panel:fullfluxscale.Right panel:expanded Even worse, many CO surveys were performed with single dish, − • BH fo(−1300rm
Blum et al. 2010; Neilsen et al. 2011b). Thus, the ‘static absorber’ interpretation may be unlikely. Early works on accretion disc theory (Shakura & Sunyaev 1973) already predicted the formation of winds from the outer disc. Compton-heated winds (see Fig. 3) can be launched if the inner disc is geometrically thin and thus the central source can illumi- nate and heat the outer disc, creating a hot outflowing disc atmo- ∞ 7–8 sphere with temperature T TIC 0 hνLν dν/4kL 10 K (Krolik et al. 1981; London∼ et al.= 1981; Begelman et al.∼ 1983a,b; ! Woods et al. 1996; Shields et al. 1986) whose ionization parameter is expected to be primarily linked to the spectral energy distribution (SED; characterized by TIC) of the illuminating source. Assuming that the Fe XXV and Fe XXVI absorption lines are unsat- urated and are on the linear part of the curve of growth, we can esti- mate the Fe XXV and Fe XXVI ion abundances (see formula 1 of Lee et al. 2002). Fig. 4 shows the Fe XXV-to-Fe XXVI ion abundance ratio as a function of TIC for all the observations in which the two lines Figure 4. The Fe XXV-to-Fe XXVI ion abundance ratio as a function of TIC. are detected. For all sources, we systematically observe the lower For each source, the Fe XXV/Fe XXVI ion abundance ratio is decreasing with XXV XXV XXVI ionization states (Fe ) at low TIC, with the ratio Fe /Fe TIC (which is a tracer of the spectral hardness). This suggests that the wind decreasing for harder spectral shapes (higher TIC), as expected if the ionization increases with spectral hardness, thus suggesting that ionization ionization parameter (ξ) increases linearly with TIC (see solid lines effects play an important role here. The solid lines show the expected relation in Fig. 4). This result suggests that, during the soft states, ionization between ion ratios and TIC, assuming a linear relation between ξ and TIC 4 effects might play an important role in determining the properties (ξ $ TIC/1.92 10 ,where$ Fion/nkTc, Fion is the ionizing flux = × 3 × = of the wind. However, this conclusion is based on the ionization between 1 and 10 Rydberg; Krolik et al. 1981) and the ion fraction versus ξ balance for a low-density gas illuminated by a # 2powerlaw; as computed by Kallman & Bautista (2001; see their fig. 8) for an optically = thin low-density photoionized gas (# 2). in order to verify the disappearance of the wind in hard states as = due to overionization, a detailed study of the actual ionizing spectra thin at the centre, have an optically thick region with H/R 1or and wind densities would be required. Although important, this is have a significant optical depth as seen from the outer disc (see∼ also beyond the scope of this Letter. Alternatively, the wind disappear- Neilsen et al. 2011b). Alternatively, if the disc ionization instability ance in the hard state might arise from the fact that illumination is at work in these transient sources, the Compton radius of the wind of the outer disc is critical for the production of Compton-heated might lie in a low-temperature, and thus unflared, part of the outer winds or from some other phenomenon (e.g. organization of mag- disc (Dubus et al. 2001). netic field) which is related to the accretion states. Irradiation only occurs when the outer disc subtends a larger solid angle than the How WAs/UFOs compare/relate to binariesinner flow. Thus,winds the formation and of jets? thermal winds might be prevented 5DISCUSSION if harder states are associated with thick discs that, even if optically How important are these winds for the accretion phenomenon? We estimate the wind mass outflow rate using the equation % L % Ubiquitous equatorial accretion disc winds L13 M˙ 4πR2nm v 4πm v X , (1) wind = p out 4π = p out ξ 4π
where mp is the proton mass, vout the wind outflow velocity and % is the solid angle subtended by the wind. Chandra observations pro- vide reliable measurements of the outflow velocities; the detection of the wind in each soft state spectrum suggests a high filling factor; moreover, we measured a wind opening angle of 30◦. Thus, once the ionization parameter is estimated, we can measure∼ the mass outflow rate and compare it to the mass inflow rate (assuming an efficiency η 0.1). We estimate the ionization parameter ξ from = the Fe XXV/Fe XXVI ion ratio and assume the ion versus ξ distribution computed by Kallman & Bautista (2001) and obtain values between log(ξ) 3.5 and 4.2. However, we caution the reader that these val- ues might∼ change significantly once (instead of assuming the ion Figure 2. Left-hand panel: HLD of the high-inclination (dipping) LMXBs studied and of all the low-inclination (non-dipping) LMXBs. Right-hand panel: high-inclination (dipping) sources show Fe XXVI absorption every time they are in the soft state and upper limits in the hard states. In low-inclination (non- abundances of Kallman & Bautista 2001) the ion abundances ver- dipping) LMXBs, the Fe XXVI absorption line is never detected. We interpret these as due to a ubiquitous equatorial disc wind associated with soft statesFigure only. 3. Several physical mechanisms can explain the properties of the sus ξ are computed using the properly tailored ionization balances We note that high-inclination sources tend to show a more triangular HLD, while the low-inclination sources exhibit a boxy one. Ponti et al., 2011 H1743-3222 disk-wind detected in soft, disc-dominatedJ. M. Miller et al. stateobserved equatorial disc winds. Here the thermal winds scenario is sketched. from the self-consistent SED.4 Fig. 5 shows that the mass outflow population of sources which are close to edge-on. Fig. 2 (left-hand presence of the wind through the detection of weak ionized emissionIn soft states, associated with geometrically thin discs, the central source panel) shows the HLD of all the high-inclination LMXBs and lines. 3. ANALYSIS AND RESULTS 1.2 does probably illuminate the outer disc and thus it might heat it, increasing reports the measured Fe XXVI absorption-line EW. These sources Is it theoretically plausible for the disc winds to have a strong 4 show clear evidence for a high-ionization disc wind (vout angular dependence? IndirectFeXXV evidenceThe for central an and angular questions dependenceFeXXVIthe thermal in thisare pressure paper variable, that require then estimates drives away of a the wind, which is flattened above For example, under various assumptions about the gas density, different 2.5 3.5 1 1 ∼ 10 − km s− ) during all 30 observations in the soft state. of the wind in GBH was alreadyionizing inferred from flux the lack and of emissionthe column disc. Thus, density only in each in high-inclination observation of sources H our line of sight to the authors studying the same data set found ionization parameters that vary by On the other hand, whenever these sources are observed in the lines associated with the X-rayand1743 absorption have322. lines This (LeeVout et can al.~300-670 2002; be done through km/s photoionization model- hard X-ray state, they show only upper limits. We, in fact, observe Miller et al. 2006b). This suggests that− the wind subtendscentral a small source crosses the wind, allowing us to detect it. In hard states, a 2 orders of magnitude (Miller et al. 2006a, 2008; Netzer 2006; Kallman stringent upper limits for 16 out of 17 observations and just one fraction of 4π sr. Moreover, discing, wind but models a simpler and magnetohydro-geometrically and more direct thick approach and optically is to thin measure corona the and the jet are present, while no ∼
1 et al. 2009). However, our estimated ionization parameters here are within ∝ detection of a weak wind quasi-contemporaneous with a weak jet dynamic simulations predict aequivalent strong angular dependence width ofwind of absorption the is observed. lines [Correction since EW addedNwhenthe after online publication 2012 April 4: the typical range of measured values, and we believe they are useful for our (Lee et al. 2002; Neilsen & Lee 2009). This demonstrates that for wind (Begelman et al. 1983a,b;absorbing Melia et al. 1991; gas 1992;is on Woods theSee linear Maria part of the Diaz- curve ofTrigo’s growth. talk this set of sources, the presence of the disc wind is deeply linked to et al. 1996; Proga et al. 2002; Luketic et al. 2010). In fact,figure if the colours fixed] purposes here. the source state. In particular, the wind is present during spectrally disc wind is produced by X-ray irradiation (i.e. Compton heating, ratio soft states, when the jet emission is strongly quenched. line driving), it is expected to be stronger3.1. inThe edge-on Spectral sources simply Continuum in the 2010 Low/Hard State C 2012 The Authors, MNRAS 422, L11–L15 The right-hand panel of Fig. 2 shows the HLD for the non- because once the material is liftedWe fromIonization, the fit disc, the it will combined experi- Nh, first-ordervariabilityChandra similar/HEG to and UFOs % C dipping LMXBs,0.8 GX 339-4, XTE J1817 330, 4U 1957 115, XTE ence an asymmetric push from the radiation field of the central Monthly Notices of the Royal Astronomical Society % 2012 RAS J1650 500 and GRS 1758 258, which− have accretion+ discs which source. Flattened disc winds haveRXTE/PCA also beenLarge assumed spectra to velocities explain of the H 1743 −(322wrt jointly. mass) The too HEG spec- are inclined− more face-on to− the observer. None of these source has winds of broad absorption-line QSOtrum and was other AGNfit was outflows limited (e.g. to the 1.2–9.0 keV band, owing to a detection of a highly ionized wind in any state. Several spectra Emmering et al. 1992; Murray etcalibration al. 1995; Elvis 2000).uncertainties and poor signal on either side of this have a signal-to-noise ratio good enough to measure upper limits as small as a few eV, even during the soft state. For this reason, we range. The PCA spectrum was fit in the 3.0–30.0 keV band, confidently state that these sources do not present the signatures of 4IONIZATIONEFFECTSMilleragain owinget al., to calibration 2006, uncertainties 2012 on either side. In the 0.6 highly ionized5.5 Fe K winds.2 6 6.5 7The 7.5 strong connection 8 betweenjoint winds fits,and source a simple states requires constant was allowed to float between the This difference in behaviour can be easily understoodEnergy (keV) if both the an explanation. Ueda et al. (2010),spectra during oscillatingto account X-ray for states differences in the flux calibrations. high- and low-inclination sources have the same wind present in FIG.1.—Thefigureaboveshowsthelinespectrafromtwoof GRS 1915 105,Chan- observe the ionizationAfitwithasimpleabsorbedpower-lawmodelwith parameter of the wind Γ = soft states and absent in the hard states, but the wind is concentrated to vary with the+ source luminosity, suggesting the importance of dra/HETGS observation of H 1743−322. The data have been divided by a ± 2 in the planesimple ofcontinuum the disc; thus, and our binned line of sight for visual intercepts clarity. the wind The observaionization.tion in Can black an wasoverionization1. effect77 explain0.01 the does disappearance not give a formally acceptable fit (χ /ν = only inobtained high-inclination in a disk–dominated sources. If this idea phase; is correct, it is listed we expect as “Observaof thetion wind 1” during in Miller hard states?1 If.56), the absorber but it is does in the characterize form of the flux well. It is likely that the that deeperet al. observations (2006a). ofA low-inclination disk wind was sources detected may revealthrough the the blue-shi‘static’ cloudsfted Fe with XXV approximatelypoor constant fit densityis drivenn and distanceby uncertainties in the cross-calibration of and Fe XXVI absorption lines in that specturm. The deepR low/hfrom theard ionizing state ob- source and assuming that the spectral shape changes have a minor impact onthe the instruments. ionization state of When the wind, each spectrum is permitted to derive its servation considered in this paper is shown in red. No significant absorption Γ ± 1 One observationlines are of evident, GRS 1915 and105 restrictive with lower luminosity upper limits and Compton are obtainedthroughdirectfits.then the absorber ionization parameterown power-lawξ will be directly index related toin a joint fit, a value of =1.93 0.01 + 3 2 XXVI XXV the source luminosity: ξ L/nR .Theleft-handpanelofFig.2 Γ ± temperature does not show any Fe but only Fe absorption, thus = is found for the HEG spectrum while =1.71 0.01 is found suggestingabsence the importance of the of ionization wind. effects. shows that at the same luminosityfor, the the winds RXTE/PCA are present in the spectrum, soft and a much better fit is derived 2 but not in hard states (see also Lee et al. 2002; Miller et al. 2006b; The majority of the low-energy absorption lines detected in these LMXBs 2 (Miller et al. 2004) are consistent with being produced by the interstellar (χ /ν =1.13). If the steeper index is assumed to be the right medium (Juett et al. 2004;2. 2006;OBSERVATIONS Nowak et al. 2008). Most AND of the DATA remaining REDUCTION time-averaged value for the lengthy Chandra observation, it structures are consistent with being at rest, thus unlikely associated with the 3 Where L has been computed as the integral of the disc emission (in the Fe K wind (JuettDuring et al. 2006). its 2003 outburst, H 1743−322 was0.001–100 observed keV band)simul- plus the powerleads law (1–100 to keV) a lowone. value for the 8.8–30 keV ionizing flux (see taneously using the Chandra/HETGS and RXTE on four occa- below). To be conservative, then, we simply adopt the power- C 2012 The Authors, MNRAS 422, L11–L15 # Γ sions (Miller et al. 2006a). AllC four of those observations cap- law index derived in the joint fit ( =1.77) as an approximate Monthly Notices of the Royal Astronomical Society # 2012 RAS tured relatively soft flux states. Three could be roughly classi- value, and derive the unabsorbed 8.8–30 keV flux based on fied as “high/soft” states. The remaining observation (second that model. in the sequence) likely captured a harder (but still luminous) “very high” state. 3.2. Limits on Absorption Lines in the 2010 Hard State H1743−322 was again observed with the Chandra/HETGS To test for the presence of Fe XXV and Fe XXVI absorp- and RXTE in 2010. The Chandra observation started on 2010 tion lines in the low/hard state spectrum of H 1743−322, we August 08 at 23:03:48 UT, and was 60.5 ksec in duration. added Gaussians at 1.850 Å and 1.780 Å (6.700 keV and RXTE observed H 1743−322 simultaneously with Chandra, 6.970 keV,respectively; Verner, Verner, & Ferland 1996). The starting on 2010 August 09 at 05:35:51 UT for a total du- range of line widths and velocity shifts measured in the line ration of 6.1 ksec. Rapid analysis of the RXTE observation detections reported in Miller et al. (2006a) were sampled in confirmed that H 1743−322 was in a “low/hard” state at the order to ensure consistency and conservative limits. We mea- time of these observations (Belloni et al. 2010). sure 90% confidence upper limits of EW ≤ 0.58 mÅ and The Chandra High Energy Transmission Gratings (HETG) ≤ were used to disperse the incident flux onto the Advanced EW 0.42 mÅ for Fe XXV and Fe XXVI, respectively. CCD Imaging Spectrometer “spectroscopic array” (ACIS-S). These limits are a factor of several lower than the line de- To prevent photon pile-up, the ACIS-S array was operated tections reported in Miller et al. (2006a). Figure 1 shows in continuous clocking or “GRADED_CC” mode, which re- data/model ratio spectra from a prior Chandra observation duced the frame time to 2.85 msec. For a discussion of and the low/hard state considered here. this mode, please see Miller et al. (2006a). All Chandra Larger potential velocity shifts were also examined, since data reduction was accomplished using CIAO version 4.4. one means of increasing the ionization of a gas is to accelerate Time-averaged first-order HEG and MEG spectra were ex- it. A potential feature is apparent at 1.834 Å (6.760 keV). tracted from the Level-2 event file. Redistribution matrix However, this feature is merely noise: first, the feature is not files (rmfs) were generated using the tool “mkgrmf”; ancil- significant at even the 2σ level; second, it is unlikely that Fe lary response files (arfs) were generated using “mkgarf”. The XXV would be observed in the absence of Fe XXVI (see, e.g., first-order HEG spectra and responses were combined using Kallman & McCray 1982), especially if the gas is potentially “add_grating_orders”. The spectra were grouped to require a more highly ionized than when both were detected. There is minimum of 10 counts per bin. no evidence for an Fe XXVI line at a velocity shift required if The standard RXTE pipeline spectral files and responses for the feature at 6.760 keV line is identified with Fe XXV. the PCA and HEXTE were downloaded from the HEASARC Limits on the line equivalent widths obtained with Chandra archive, and employed for spectral fitting. A systematic er- (again, 0.58 mÅ and 0.42 mÅ, for Fe XXV and Fe XXVI, re- ror of 0.6% was added to the PCA spectrum in quadrature. spectively), are tighter than those obtained in a prior Suzaku All spectral analyses were conducted using XSPEC version observation in the low/hard state (0.97 mÅ and 0.64 mÅ , 12.6.0. All errors quoted in this paper are 1σ errors. Blum et al. 2010). Importantly, whereas the ionizing flux de- How WAs/UFOs compare/relate to microquasars?
GRS1915 The disappearance of the inner accretion disk, coincides with the beginning of the ejection of a relativistic plasma cloud. As the ejected cloud expands it becomes transparent to radio waves, with a peak radio-wave flux that is delayed by 15 min relative to the infrared peak.
Mirabel and Rodriguez, 1998, Nature The Astrophysical Journal,734:43(16pp),2011June10 Chatterjee et al. 2011 Chatterjee et al. 7 2010 x+0.15
6 2009 ) -1
s 5 2008 -2
d2 4 2007 erg cm d8 d4 -11 d6 3 d3 d9 2006 Knots ejections
Flux (10 d5 2 2005 d7
d1 Times of Corresponding Ejection Dips K1 K2 K3 K4 K5 K6 K7 K8 K9 2004 1 2004 2005 2006 2007 2008 2009 2010 2004 2005 2006 2007 2008 2009 2010 Start Times of X-ray Dips Year Figure 15. Times of ejection of new VLBA knots vs. times of X-ray dips from Figure 14. X-ray light curve of 3C 111 as in Figure 4,withdataindicated Table 10. TheJet dashed and line disc is outflowsthe best-fit in straight 3C 111 line through757 the points. It fits by small circles. The curve corresponds to the same data, smoothed with a the data extremely well with small scatter. This indicates that there is a clear Gaussian function with a 10 day FWHM smoothing time. The arrows indicate association between X-ray dips and superluminal ejections. The y-intercept of the times of superluminal ejections and the line segments perpendicular to the this line minus 2004.0 indicates the value of mean time delay between the times arrows represent the uncertainties in the times. X-ray dips and VLBA knots of dips and ejections. listed3C111 in Table 10 are marked with the respective ID numbers. Over the six years of observation, significant dips in the X-ray light curve are followed by ejections of bright superluminal knots in the VLBA images. This shows a clear connection between the radiative state near the black hole, where the X-rays are of VLBA images of sufficient duration to define the trajectory produced, and events in the jet. Dip d3 and the two significant dips just before and speed of a knot, as is evident from Figures 6 and 7. No new and after that are considered to be one dip with multiple branches and is related bright, moving knots after K9 appeared in subsequent imaging to K3 which was also extended along the jet axis. with the VLBA during the first eight months of 2010 (images not shown here). all of the X-ray variability observed, if we assume that most of The time delay between the minimum ofTombesi the X-ray et dipsal. 2011 the X-ray emission we observe is not relativistically beamed. and the time of ejection of the corresponding superluminal knot is distributed between 0.03 and 0.34 yr (Table 10)with 5. X-RAY/RADIO CORRELATION ameanof0.15 0.08 yr. We plot the times of ejection of new knots along± with the corresponding times of X-ray dips in UFO We can see in Figure 4 that,UFO from the middle of 2009 to the Figure 15. A straight line fits the data extremely well with small present, the X-ray flux undergoes only minor fluctuations about scatter, which indicates that there is a clear association between 11 2 1 ameanvalueof5 10− erg cm− s− . During this interval, X-ray dips and superluminal ejections. The best-fit line through the 230 GHz flux is× at a very low level, while the 15 GHz flux the points has a slope of 1 and y-intercept of 0.15, which is is decreasing. Therefore, the high-state X-ray plateau coincides identical to the mean delay of 0.15 0.08 yr given above. with a low state of the radio jet. We can extend this X-ray/ We have performed a numerical simulation± to calculate the radio connection to the observations of earlier years as well. significance of the dip-ejection correlation. If we keep the times Figure 1. Long-term 2.4–10 keV flux RXTE light curve of 3C 111 between 2008 and mid-2011. The vertical solid/dotted lines refer to the detection/non- The times of “ejection”detection of UFOs of in new the Suzaku radioand XMM–Newton knots in thespectra. jet The (shown detections of UFOsof are dipsmarked withfixed ‘u’. The and dates choose relative to one the X-ray million dips and the sets appearance of nine ejections, by the arrows) areof new related jet knots toin the the VLBA low images X-ray are marked states. with We ‘d’ and convolve ‘k’, respectively. times of which are drawn from a uniform random distribution the X-ray lightTable curve 2. withTimesof a GaussianX-ray dips, observations smoothing of UFOs function and appearance with of betweenasuddenspikeinfluxandweadoptaconservativeapproachnot 2004.3 and 2010.3 (the duration of our observations), a10dayFWHMsmoothingtimetoidentifythemajorlong-radio knots. theconsidering probability it in the thatfollowing there discussion. is at least one ejection event following term trends. Figure 14 shows the smoothed X-ray light curve allFrom nine Fig. dips 1, we by derive 0.03–0 that. overall34 yr the is UFOs 0.007. seem In to a be similar pref- simulation, Dip TXmin UFO Tufo Knot Tknot βapp and denotes the times of superluminal ejections with arrows. weerentially found detected that the during probability intervals of increasing that all nine flux. In dips order will to have exactly d7 2008.51 u7 2008.65 k7 2008.83 0.07 4.54 0.38 estimate the statistical confidence of the possible relation between ± ± 5 Based on inspectiond8 2008.98 of this – light curve, – k8 we 2009.07 consider0.08 an 4.07 X-ray0.43 onethe UFOs following and the periods ejection of rising is flux, 3 we10 tested− .Thisshowsthatthedip- the null hypothe- ± ± d9 2009.26 – – k9 2009.29 0.04 4.33 0.66 sis that UFOs are not detected during phases× of ascending flux but fluctuation to be a significant dip if the smoothed± X-ray flux± ejection association in 3C 111 is highly significant. To further d10 2010.57 u10 2010.69 k10 2010.85 0.02 5.66 0.09 only in steady or decreasing intervals. This hypothesis is satisfied ± ± falls by more thand11 30% 2010.78 and – remains – at k11 a low 2011.01 level0.07 for 5.22longer0.35 investigatein none of the four the cases relationship described before, between yielding the a probability characteristics of the ± ± than one month. We calculate the central time of each dip by X-rayof <1/4. dips Therefore, and conservatively, VLBA knots, we can we say thatcalculate the statistical the average flux Note. βapp is the apparent speed of the radio knots in units of c. determining the local minimum of the X-ray flux. We note that overprobability all ofepochs the claim of that the UFOs latter. are preferentially We do this observed via dur- modeling of the ing phases of rising flux is P 1 (1/4) ! 75 per cent. Given the minimum amplitudethe RXTE light measured curve and also from allows usunsmoothed to oversample by data a factor of of brightnessthe limited number distribution of observations at= each available,− epoch we stress with that multiple the sta- components 2 the typical variability time-scale of the UFO in 3C 111 of about every dip is below∼ the average of the X-ray flux over the length characterizedtistical significance by of this circular relation Gaussian is only marginal brightness and it should distributions, as 7 days (Tombesi11 et al. 2010b, 2011b).2 1 If the difference is positive of the light curve∼ (4 10− erg cm− s− ). It can be seen describedbe regarded only in Section as an indication.2.4.Thisshowsthattheaveragefluxofthe However, we note that a similar it indicates× a rising flux, instead if null or negative it indicates a behaviour was observed also in other sources showing UFOs (e.g. that each of thesteady/decreasing ejections is preceded flux. We find by that a the significant first and fourth dip observations in the VLBABraito et al. knots 2007; corresponding Giustini et al. 2011). to the shallower dips (d2, d4, and X-ray flux. Thein prolonged Table 1, the ones low with level detected of UFOs, X-ray happened flux during in 2005 periods is of d8 in Figure 14)issignificantlysmaller(0.3 0.1Jy)thanthat increasing flux.4 Instead, the non-detections in the second and fifth ± associated withobservations knot K3, occurredwhich duringwas extended intervals of decreasing along the flux. jet Following axis, for the larger dips (d1, d3, d5, and d7), 0.9 0.4Jy.Dipd6is 3RADIOOBSERVATIONSOFTHEJETON± as is apparent inthis the criterion, images the starting non-detection in late in the 2006 third observation (Figure 6 occurred). This in also shallow, but this may be due to the absence of data during SUBPARSEC SCALES suggests that aan prolonged interval of steady/decreasing disturbance near flux too. the However, base of we the note jet that the Sun-avoidance gap, while d9 is of intermediate depth. Hence this latter case is less stringent because it happened very close to 3C 111 is actively monitored with the VLBA at 43 GHz at roughly resulted in an elongated superluminal knot. Hence, d3 and the K6monthly and intervals K9 were by the not blazar included group at the in Boston the University. average Here calculation. That two significant dips just before and after that are considered to thewe present brighter a temporal knots extension correspond of the VLBA to more analysis pronounced of Ch11 (see dips supports be one dip with multiple4 We note that branches the observation and of isthe related UFO u7 occurred to K3. at Detection the beginning of thetheir conclusion fig. 6) from 2008 that up the to mid-2011. X-ray minima The sequence and of the VLBA ejection of knots of the ejection ofa period a new of rising knot flux, requires just after the the major analysis X-ray dip of d7. aIf thesequence rising period inimages the shown jet are in physically Fig. 2 provides related. a dynamic view of the inner jet is related with the acceleration of the outflow, this might explain why the between 2010 November and 2011 September at an angular resolu- velocity of u7 is much lower than u10, which instead was detected close to tion 0.1 mas, corresponding to 0.094 pc. The VLBA data have ∼ ∼ amaximuminflux. 12 been processed in the same manner as described in Ch11. We can
C # 2012 The Authors, MNRAS 424, 754–761 C Monthly Notices of the Royal Astronomical Society # 2012 RAS UFOs compared/relate to binaries winds and jets? 21
Fig. 5.— The above plot shows the power emitted either from the jet power (AGN:red and BHB:orange) or wind power (AGN:black and BHB:blue) as a compared to the mass accretion rate, which is approximated by the bolometric luminosity on the x axis. A clear turnover −2 at M˙ acc ≈ 10 M˙ Edd indicates whereKing the power A. emittedet al. is becoming2012, less Churazov efficient. Interesting et isal. the dichotomy2005 between where the jets lie at lower mass accretion rates and where the winds lie at higheraccretionrates.Thethickblacklinedenotestheoutputpower by outflows, where as the thin line is the power generated by radiation as described by Churazov et al. (2005) Summary
Ø General framework/importance ⇒ Recognized importance of UFOs (to AGN feedback AND wind/outflows/ jets physics)
Ø Critical/remaining open Issues for UFOs/ winds ⇒ Acceleration mechanism? ⇒ Covering & filling factor in high-z QSOs ? ⇒ How/where energy released in ISM?
Ø How they relate/compare to ⇒ WAs? ⇒ Cold molecular outflows in QSOs, ULIRGs, high-z QSOs? ⇒ Accretion/state/jet formation/wind quenching?