X-rayObservationsofUltraFastOutflowsinAGN

James Reeves (UMBC/Keele)Main Collaborators:- Emanuele Nardini (Arcetri), Michele Costa,

Andrew Lobban (Keele), Valentina Braito, Gabi Matzeu (Brera), Jane Turner (UMBC), Francesco Tombesi (GSFC/UMD), Stuart Sim

(Belfast), Ken Pounds (Leicester)

Outflows from Active Galactic Nuclei

Introduction – ObservationsofwindsinActiveGalacticnucleiandthediscoveryofultrafastoutflows.

Case Study 1 – Thepowerful,wideangle,diskwindintheluminousquasarPDS456.Whataretheoutflowenergetics?Revealingtheoutflowvariabilityandstructure.Possibledrivingmechanism

Case Study 2 – thelargeXMM-Newtoncampaignonthenearbynarrowlinequasar,PG1211+143.Observationsofarapidlyvariableaccretiondiskwind.ThesoftX-raysignatureofthefastwind.

Discussion – statisticsandoccurrence.Energeticsandroleinfeedback.Whatarethesignaturesofthefastoutflow?Drivingmechanismandacceleration.

DiscoveryofAccretionDiskWindsinAGN• Initial discovery in PG 1211, PDS 456 and APM

08279 (Chartas et al. 2002).• Disk winds simulations of Sim et al. (2010),

Proga & Kallman (2004). Produces blue-shifted Fe K absorption

• Also MHD wind models (Ohsuga & Mineshige2011, Fukumura et al. 2015).

PG1211+143,z=0.0809(Poundsetal.2003)

PDS456,z=0.184,(Reevesetal.03)

King 2010

Tombesi+ 2010

Maiolino+ 2012

Powerful disc winds are naturally expected at high accretion rates:

Tombesi et al. (2015, Nature)

AGN Winds in Charleston (15th October 2011)

Results & Examples16/47 objects found to have significant Fe K absorption (in 20/61 spectra.) => 4 have absorption in more than one obs, w/ evidence for variability (e.g. PDS456)

Lines No.

Fe XXV+Fe XXVI(same Vout)

6

Fe

(RQQSO,z=0.184)

Outflow velocity of 0.25-0.30c, if associated with Fe XXV/XXVI resonance lines (at 6.7-6.97 keV).

Suzaku OutflowSample(Gofford etal.2013,2015)

• 20/51AGNfoundtohavesignificantironKabsorption(18/51at>99%significancefromMonteCarlo).

• Thiscorrespondsto28/73 ofthefittedX-rayspectra(or40%).Theabsorptionlinesystemsbreakdownasfollows:-

9/28systemslikelyassociatedwithFeXXVILyα4/28withFeXXVorlowerionization.12/28associatedwithatleast2linesduetoFeXXVandFeXXVI.3/26mayhavemultiplevelocitysystems.

• Overall12/20oftheAGNhaveoutflowvelocities>10000km/s ofwhich8 havev>0.1c.Only3/19areconsistentwithno-netvelocity-shift.

• Meanionizationparameter,logxi=4,columnlog(NH/cm-2)=23

Suzaku OutflowSample(Gofford etal.2013,2015)

• 20/51AGNfoundtohavesignificantironKabsorption(18/51at>99%significancefromMonteCarlo).

• Thiscorrespondsto28/73 ofthefittedX-rayspectra(or40%).Theabsorptionlinesystemsbreakdownasfollows:-

9/28systemslikelyassociatedwithFeXXVILyα4/28withFeXXVorlowerionization.12/28associatedwithatleast2linesduetoFeXXVandFeXXVI.3/26mayhavemultiplevelocitysystems.

• Overall12/20oftheAGNhaveoutflowvelocities>10000km/s ofwhich8 havev>0.1c.Only3/19areconsistentwithno-netvelocity-shift.

• Meanionizationparameter,logxi=4,columnlog(NH/cm-2)=230.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35

vout/c

0

2

4

6

8

10

12

Num

ber

ofso

urc

es

⌘=

0.06

⌘=

0.34(b)

1021102210231024102510261027102810291030

log(

Ṁout/

gs�

1)

3C11

13C

390.

34C

+74.

26A

PM

0827

9C

BS

126

ES

O10

3-G

035

MC

G-6

-30-

15M

R22

51-1

78M

rk27

9M

rk76

6N

GC

1365

NG

C32

27N

GC

3516

NG

C37

83N

GC

4051

NG

C41

51N

GC

4395

NG

C55

06P

DS

456

SW

J212

7

10�3

10�2

10�1

100

101

102lo

g(Ṁ

out/

Ṁed

d)

Howpowerfularediskwinds?

Majoruncertainties:launchradius+solidangle

Gofford etal.2015,SampleofironKoutflowswithSuzaku

Howpowerfularediskwinds?

Thedetectionofnarrow,blueshifted X-rayabsorptionlinesdoesnotprovideanysolidconstraintonthetotalenergeticsofawind

★ Solidangle: frequencyofBHwindsignaturesamonglocalAGN

★Columndensity: modelling ofabsorptionbyphoto-ionised gas

★Outflowvelocity:line’senergyshiftfollowingidentification

★ Launchradius:ionisation stateofthegasandescapevelocity

Itisstillunclearwhetherdiskwindshavesufficientmechanicalenergytopowerfeedbackongalacticscales

PDS456:theRosettaStoneofAGNdiskwinds

2013/14campaign:5simultaneousXMM+NuSTAR observations

Rest-frameenergy(keV)

Ratio

Mostluminousradio-quietAGNinthelocalUniverse

Reeves+03

Systematicdetectionofadeeptroughabove7keV rest-frame:evidenceforalargecolumnofhighlyionised matteroutflowingataboutonethirdofthespeedoflight

IdealtargetforstudyingBHwindsintheEddington-limitedregime

MB ∼ −27 Lbol ∼ 1047 erg s−1 MBH ∼ 109 M⊙

TheWideAngleUFOistheQuasarPDS456

Nardini+15,Science

Apersistent,wide-anglewind

Apparentresponsetocontinuumchangesover7-10days

P-Cygni-likeprofileresolvedatanyepoch(aperture>50° fromFWHM)

Nardini+15

Somerelevantnumbers

AlltheinformationcannowbedeterminedfromthedataThesolidangle isobtainedfromtheemitted/absorbedluminosityratio,andthelaunchradius fromthevariabilitytimescale

Thedepositionofafew%ofthetotalradiatedenergyisenoughtopromptsignificantfeedbackonthehostgalaxy (Hopkins&Elvis10).Overalifetimeof107 yr theenergyreleasedthroughtheaccretiondiskwindlikelyexceedsthebindingenergyofthebulge

OneYearintheLifeofPDS456…(Matzeu etal.2016)

(Aug2013)

(Feb2014)

(Feb-March2013)

VariabilityofWindvelocityinPDS456(Matzeu etal.2017)

CentroidenergyofironKabsorptionappearstoincreasewithincreasingluminosityover12XMM/NuSTAR/Suzaku observationsfrom2001-2014.Windvelocityincreaseswithionizingluminosity(asLX1/4)MaybeexplainedintermsofradiativedrivingintheEddingtonlimitedregime.

CentroidsofFeKabsorption Windvelocityvsluminosity

L

Signaturesoffast(0.15c)“BAL”likeprofilesinsoftX-rayswithXMM-Newton/RGS.Velocitywidthsσ≈10000km/s.

BroadSoftX-rayAbsorptionProfilesinPDS456(Reevesetal.2016,ApJ)

OBSB

OBSE

SoftX-rayabsorbinggasthelikelysignatureofaninhomogeneouswind,partiallyobscuringtheAGN.Velocitiesofupto0.2c.AbsorptionprimarilyduetohighlyionizedFe(FeXX-XXIV)aswellasNeIX/X

BroadSoftX-rayAbsorptionLines−4

−20

24

OBS CD

ABS2

ABS1

−4−2

02

4

OBS EEMIS

1

ABS1

ABS2

0.8 1 1.2 1.4

−50

Rest Energy (keV)

2001

AB

S1

−4−2

02

4

OBS B

RGS$

0.9

11.

1ra

tio

XMM, OBS B.

ABS1

ABS2

0.9

11.

1ra

tio

XMM, OBS CD.

ABS1

ABS2

0.9

11.

1ra

tio

XMM, OBS E.

ABS1

ABS2

EMIS

1

0.8 1 1.2 1.4 1.6 1.8

0.8

0.9

11.

11.

2

ratio

Rest Energy (keV)

XMM, 2001

ABS1

ABS2

MOS$

TheUVconnection– isPDS456aBALQSO?

1400 1500 1600 1700 1800Observed wavelength (Å)

1

2

3

4

5

Flux

dens

ity(e

rgs�

1cm

�2

�1 )

⇥10�

Figure 2: Comparison between the two HST observations ofPDS 456. The red one was taken in May 2000 with STIS, the blueone in April 2014 with COS. There is a remarkable flux variabilityover this decade. Notably, in the lower state the broad Ly↵ absorp-tion trough looks much shallower. The behavior of the optical/UVflux is therefore fundamental to understand the properties and fate ofthe SMBH X-ray wind at larger scales.

through the UVOT data.At the same time, these observations will allow us toderive a preliminary slope for the power spectral densityof PDS 456. Together with the already available Swiftand XMM–Newton data, the coverage in frequency spacecould be enough to test for a broken power law shape.The break energy is known to correlate with the SMBHmass (McHardy et al. 2006, Nature, 444, 730), nomeasure of which is available for PDS 456 to date.

3. Justification of Requested Observing Time, Feasi-bility, and Visibility

The simultaneous monitoring of optical/UV and X-rayfluxes is fundamental to understand the relation betweenthe accretion disk and the X-ray corona in AGN, andSwift is unrivalled to perform this kind of study. Therecent Suzaku and XMM–Newton/NuSTAR programshave proven that one week is the typical timescale todetect significant spectral changes. On top of that, all thehistorical X-ray light curves of PDS 456 exhibit plenty ofsharp flares, during which the flux increases by factors of⇠3–5 in about a day. These flares have an undoubtedlyintrinsic origin (Matzeu et al. 2015, submitted). Bythe light-crossing time argument, the steep flux riseobserved in a flare poses stringent constraints on thesize of the X-ray corona, which should not exceed a fewtens of gravitational radii. Since the absorption-induced

Figure 3: In spite of the very different range of flux variability ob-served over weeks/months in the optical/UV and X-ray bands (⇠10%against 600%), the emission from the accretion disk and from theX-ray corona in PDS 456 are clearly related. Even the most ex-treme X-ray states are consistent with the optical/UV emission oncecorrected for absorption, returning the correct accretion rate (⇠0.8)expected for this source. The combined data sets are from XMM–Newton/OM (blue, August 2013), Suzaku (black, March 2013) andNuSTAR (red, February 2014) observations, fitted with the energeti-cally self-consistent optxagn model by Done et al. (2012).

and intrinsic variations occur with clearly differenttimescales, the best approach to disentangle the twoeffects is to combine short- and long-term monitoringinto a single campaign.During cycle 12, PDS 456 is not visible for Sun con-straints from October 28, 2016 to January 28, 2017.This leaves a continuous visibility window of morethan 6 months from April 2016. We then request toperform the proposed monitoring during this period, fora total span of 26 weeks. We request a 2-ks snapshotevery week with (reasonably) exact spacing, plus a moreintense sampling over two consecutive weeks with twoadditional pointings (separated by 2 or 3 days) in orderto better explore the frequent short-term flares. As theflares are almost ubiquitous, there is a large chance offinding at least one of them with 7 visits over any periodof 14 days, which can be chosen freely according tothe mission schedule. In summary, the campaign willconsist of 30 observations (26 plus 4) of 2-ks exposureeach. Moon constraints are fully compatible with thesuggested monitoring strategy.Considering the average flux state of PDS 456, weexpected at least 400 counts per observation in the X-rayspectrum, which are enough to determine with goodaccuracy the spectral slope and reveal the presence

3

C IV!

Lyα / N V Lyα absn!

×10-13!

HST/STIS2000HST/COS2014

VerybroadUVemissionprofiles,e.g.Lyα/CIV12000-15000km/s.Highlyblueshifted,CIV=5000km/s.Broadabsorptiontroughbluewards ofLyalpha orhighlyblueshifted CIV?

TheXMM-NewtonLargeProgramme onPG1211+143(Lobban etal.2015,2016;Poundsetal.2016a,b)

PG1211+143,luminous(Lbol~1046 erg/s),nearby,narrow-linedtypeIquasar(z=0.0809),theinitialproto-typeexampleofanultrafastoutflow(Poundsetal.2003).

7SequenceswithXMM-NewtoninJune/July2014.Totalexposure~630ks.Notelowestfluxduringrev2659withXMM-Newton.

UltraFastFeKOutflowinTheQuasarPG1211+143

600ks observation of PG 1211 from XMM/2014 campaign.

Spectrum shows a wealth of high ionization absorption lines from Fe XXV and Fe XXVI.

The line profiles are fitted with a disk wind model (Sim et al. 2008, 2010), with two streamlines at different radii/velocities

Wind launch radius of Rmin=64RGand 250RG

Resultant wind terminal velocities of v=0.07c and 0.14c (consistent with analysis in Pounds et al. 2016).

10−3

1.5×10−3

2×10−3

Flux

(keV

cm

−2 s

−1 ) | | | | | |

PG 1211+143, XMM 2014

4 5 6 7 8 90.80.9

11.11.2

ratio

Observed Energy (keV)

Overallwindparametersin2014:-MassoutflowrateMout/MEdd =0.72±0.13IonizationLX/LEdd =3%,inclination=57°.

Disk wind model

10 15 20 25

0.8

11.

21.

4ra

tio

Rest Wavelength (Å)

Ne IX−XFe XIX−XXI

Fe UTA O VIII O VII N VII

| | | | | | |

IsthereevidenceforthefastoutflowinthesoftX-rayspectrumofPG1211?MeanRGSspectrum(June2014,600ksexposure)

Blue-shiftedabsorptionprofilesrevealedindeepRGSexposureofPG1211+143,e.g.fromN,O,Ne,Fe.Systematicvelocityshiftof0.06-0.07c.NotepositionoflowionizationUTAcomponentandstrongemission(OVII/OVIII).

8 10 12 14 16 18

0.8

0.9

11.

1

ratio

Rest Wavelength (Å)

PG 1211+143 (z=0.0809)

ComparisonbetweenXMM-Newton(RGS)spectraoftheQSOsPG1211+143(z=0.0809)vsMR2251-178(z=0.064)

FastWind(0.062c)

NeX NeIX FeUTA

SlowWind(

18 20 22 24 26

0.8

11.

2

ratio

Rest Wavelength (Å)

PG 1211+143 (z=0.081)

Softbandcomparison– notealsothepresenceofbroadsoftX-rayemissionlines(e.g.OVIIandOVIII)inadditiontotheblueshifted absorption.

OVIIOVIII NVII

18 20 22 24 26

0.8

11.

21.

4

ratio

Rest Wavelength (Å)

MR 2251−178 (z=0.064)

OVIII OVII NVII

RapidVariabilityofUltraFastOutflowinPG1211+143

10.5 2 5

10−3

2×10

−3

νFν F

lux

(keV

cm

−2 s

−1)

Energy (keV)

rev 2659rev 2661rev 2663rev 2664

Lowfluxspectruminrev2659.Sourcethenprogressivelybrightens.

SoftX-rayabsorptionisrapidlyvariable(within2orbits),decreasingwithincreasingflux.

PG1211undergoesabsorptioneventduringrev2659,increasinginlowerionizationabsorption.

SoftX-rayabsorbernotaconventionalwarmabsorber.Fast(0.07c)andrapidlyvariableontimescalesof<1week.Impliesabsorberiscompact,onscalesof10sRs,likelypartofclumpydiskwind.

10 15 20 25

0.5

11.

5ra

tio

Rest Wavelength (Å)

REV 2652

Ne IX/X Fe UTA O VIIO VIII N VII

VariabilityofPG1211windseeninRGSspectrum

VariabilityofPG1211windseeninRGSspectrum

10 15 20 25

0.5

11.

5ra

tio

Rest Wavelength (Å)

REV 2659

Ne IX/X Fe UTA O VIIO VIII N VII

VariabilityofPG1211windseeninRGSspectrum

10 15 20 25

0.5

11.

5ra

tio

Rest Wavelength (Å)

REV 2661

Ne IX/X Fe UTA O VIIO VIII N VII

VariabilityofPG1211windseeninRGSspectrum

10 15 20 25

0.5

11.

5ra

tio

Rest Wavelength (Å)

REV 2663

Ne IX/X Fe UTA O VIIO VIII N VII

VariabilityofPG1211windseeninRGSspectrum

10 15 20 25

0.5

11.

5ra

tio

Rest Wavelength (Å)

REV 2664

Ne IX/X Fe UTA O VIIO VIII N VII

VariabilityofPG1211windseeninRGSspectrum

10 15 20 25

0.5

11.

5ra

tio

Rest Wavelength (Å)

REV 2666

Ne IX/X Fe UTA O VIIO VIII N VII

VariabilityofPG1211windseeninRGSspectrum

10 15 20 25

0.5

11.

5ra

tio

Rest Wavelength (Å)

REV 2670

Ne IX/X Fe UTA O VIIO VIII N VII

VariabilityofSoftX-rayWindZonesinPG1211vsXMMorbit.

2652

2659

26612663/64 2666

2670

AbsorberrapidlyrespondstothelevelofthesoftX-rayexcessbelow2keV ontimescalesof~2XMMorbits(about300ks).Opacityisinverselyproportionaltothecontinuum.Windisclumpyandmulti-phase,lowestionizationphaseismostcompact(△R~1014 cm)onscalesofR~1017cm.

TheWindStructure

101610171018 102 103 104

R(cm) R(Rg)

Innermosthighlyionizedwind launchedfromwithin100Rg (1016 cm)ofblackhole– ultrafastironKabsorption(0.3c).InhomogeneoussoftX-rayabsorberR≈1017-1018 cm,ne≈107-108 cm-3,withthicknessΔR≈1015 cm.Fillingfactorf≈10-3.UVBLRemission (absorption)profilesR≈1018 cm– highvelocityCIVinUV.

Discussion

Whatisthestatisticaloccurrenceoftheoutflows.Istheirstillanydoubtabouttheirdetectionandorigins?

Howcanwemeasureaccuratelytheenergeticsofthewindsandaretheylikelytobesignificantforfeedback?

WhatistheevidencefortheultrafastoutflowsasidefromatironK?

Whatisthedrivingmechanismforthediskwind?Radiation,Magneticdrivingorboth?Whatistheroleoftheclumpygas,isitneededtoacceleratethegas?

Future- Whatistheoccurrenceofthewindsamongstthehigherluminosity(orhigherredshift)QSOs.