AGN Outflows,
the tip of the
Iceberg?
X-ray Spectroscopy
Workshop, Cambridge,
MA, July 2007
James Reeves,
Keele University, UK (Formerly at NASA/GSFC/JHU)
Collaborators:Jane Turner (UMBC/GSFC), Valentina Braito (JHU/GSFC), Alex
Markowitz (JHU/GSFC), Ken Pounds (Leicester)
Outflows in AGN
Outflows (in the form of warm
absorbers) are seen in the majority
of nearby AGN.
Typically velocities ( from a few
100 km/s to a 1000 km/s, which
could carry a few solar masses per
year (out to pc scales).
In some higher luminosity AGN
strong blue-shifted Fe K absorption
features are seen above 7 keV possible high v outflows at v~0.1c
Such outflows can carry significant
Kinetic power - equivalent to the
bolometric output.
(Outflow Schematic; Elvis 2000)
Can provide feedback between
BH/bulge mass in galaxy.
Iron K-shell Absorption in Seyfert 1s.
MCG-6-30-15, Chandra/HETG (500ks)
NGC 1365/XMM, (Risaliti et al. 2005)
Young et al. (2005)
Column densities are high, >1023 cm-2.
Can effect broad Fe K modeling.
McKernan et al. (04, 05) suggested
some such lines could arise locally, e.g.
WHIM, if coincident with AGN
recession vel.
NGC 3783, XMM, Reeves et al. (2004)
Absorption Variability in NGC 3516
Strong changes in source flux
driven by changes in covering
fraction of “heavy absorber”.
XMM 2001 vs
2006
XMM Nov 2006
XMM April 2001
XMM Nov 2001
Suzaku Oct 2005
Increase in column of
highest ionized absorber,
with strong Fe
XXV/XXVI absorption
lines emerging in 2006
obs campaign.
A highly ionized outflow in NGC 3516
(Turner et al. 2007, in prep)
Fe XXVI
NGC 3516 observed for 200ks
with XMM/Newton and
Chandra/HETG in 2006.
Source returned to former
bright state (5e-11 cgs, 2-10 keV)
Neutral
Fe K
Chandra/
HEG
Fe XXV
Observed frame and energies at 6.64, 6.92 (0.02
keV) rules out local (z=0) origin, e.g. WHIM.
Strong (100 eV, EW) absorption
lines near 6.7, 6.97 keV rest
frame, due to Fe XXV, XXVI
1s-2p. NH >5x1023 cm-2 for
vturb=3000 kms-1.
Evidence of P-Cygni profile
from outflow to Fe XXVI.
Velocity shift ~2000 km/s
Neutral Fe K width ~3000
kms-1.
Zones of gas can cover wide range of ionization
NGC 3516, Chandra/HETG, 2006
O/Ne K-shell, Fe L-shell band
Mg/Si K-shell band
Absorbing gas covers a very wide range in ionization from log =0
up to 5. Fe observed from Fe < XVII (UTA) to Fe XXVI.
Highest ionization gas from largest columns and can form an
energetically significant part of outflow.
X-ray evidence for high velocity outflows in AGN
Outflows of ~0.1-0.4c have been claimed from X-ray spectra of several AGN,
mainly via absorption features in the Fe K band. Detection of absorption in the
Fe K band requires a large column density - together with a high velocity that
implies the outflow is both massive and energetic (unless highly collimated)
APM 08279+5255 v~0.2-0.4c (Chartas et al, ApJ, 2002, 579, 169)
PG1211+1431, 3
v~0.08-0.1c (Pounds et al, MNRAS, 2003, 345, 705)
PG1115+080
v~0.1/0.34c (Chartas et al, ApJ, 2003, 595, 85)
PG0844+3492
v~0.2c
(Pounds et al, MNRAS, 2003, 346, 1025)
PDS 456
v~0.17c
(Reeves et al, ApJ, 2003, 593, 65)
IRAS13197-1627 v~0.11c
(Dadina and Cappi, A&A, 2004, 413, 921)
RXJ0136.9-3510 v~0.1/0.14c (Ghosh et al, ApJ, 2004, 607, L111)
Mrk5093
v~0.2c
(Dadina et al., 2005, A&A, 442, 461)
IC 4329a
v~0.09c
(Markowitz et al. 2006, ApJ, 646, 783)
MCG -5-23-16
v~0.1c
(Braito et al, 2007, ApJ, submitted).
NB
1. Disputed by Kaspi et al., who claim the outflow may arise from a lower velocity, depending on the
specific identification of lines in the spectrum.
2. Disputed on the basis of background subtraction in the EPIC/pn spectrum (Brinkman et al. 2005).
3. Shows red and blue-shifted iron absorption lines.
Relativistic Outflow in PDS 456
(Deep Suzaku Observation, 190ks, Feb 07)
High luminosity QSO z=0.184
LBOL~1047 erg s-1
High v outflow originally claimed
in 2001 XMM observation (Reeves
et al. 2003) and in UV via
HST/STIS (O’Brien et al. 2006).
Pair of blue-shifted absorption
lines observed with Suzaku at
9.08/9.66 keV (rest frame) or
7.68/8.15 keV (observed).
NOT associated with obvious
transition at z=0 frame, ruling out
WHIM or local bubble.
Outflow velocity of 0.26/0.32c, if
associated with Fe XXVI 1s-2p.
In PDS 456, outflow rate is ~20Msolar/yr assuming only 10% covering.
At 1/3c, the KP of outflow is 1047 ergs-1, similar to bolometric output.
Outflow geometry and driving mechanism
Radiation Pressure driven (BAL-like)?
Magneto-rotational?
Kato et al. (2004)
Black holes accreting at Eddington or above can produce optically thick winds,
(King & Pounds 2003). Optically thick within <100Rg. Mass outflow rate similar
to Eddington (Mout ~ Medd).
Alternative is magnetic field driving. Significant energy in magnetic field in
PDS456 from rapid X-ray variability, e.g. factor x2 within 10ks with Eflare=1051
erg (Reeves et al. 2002).
Fe K Absorption line variability in PG 1211+143
XMM (2001) vs Suzaku (2005)
Fe XXVI Ly
Suzaku/2005
XMM-Newton data revealed a highly
ionised outflow (Pounds et al. 2003),
with strong 7.6 keV Fe K absorption
line (7.0 keV observed)
Highest ionization absorber has ~
103.4 and NH ~ 5x1023 cm-2 outflowing
at 0.08c (24000 km/s)
Outflow launched from inner disc at
<130Rs (~1). Mass-loss rate ~ 3 M
yr-1 K.E. ~ 1045 erg s-1 (~LBOL)
XMM/2001
Kaspi et al. (2004) suggested lower
velocity for RGS lines (3000 km/s).
But does not easily account for Fe K.
Iron K absorption variable - intrinsic
to AGN outflow.
A transient blue-shifted Fe K absorption line in Seyfert 2,
MCG -5-23-16 (Braito et al. 2007).
Significant Absn line variabilty
BASELINE MODEL: diskline
+narrow K+K + reflection
EPIC-pn
(100ks)
In the 3rd spectrum a possible
absorption feature is present
at ~7.8 keV (Rest Frame).
|EW|~50 eV (2~35).
A high velocity outflow in IC4329a (Markowitz et al. 2006)
IC4329a, 100ks XMM-Newton obs
Fe K-shell
absorption

Feature is significant at >99.9%
(Monte-Carlo). Not due to bgd (40x
lower)
Well modeled by absorption from Fe
XXVI K (6.97 keV rf) with a
velocity shift of 27000 km/s (note
z=0.016).
Can rule out absorption from Fe XXII
or XXIII K near 7.6-7.7 keV with no
velocity shift - model predicts K
absorption at 6.6 keV (not observed)
Absorption line in IC 4329a spectrum at
7.68 keV (rest-frame). If associated with
Fe XXVI, then outflow velocity is 27000
km s-1
Absorption from low ionization K
(Fe <XVII at 7.1-7.2 keV) with
v=0.07c ruled out - requires too
much bound-free abs below 6 keV.
Do the iron K outflows have to be fast?
(a) High ionization fast outflow. Models the 7.6
keV feature well with an outflow of 0.1c.
(b) Low vel, moderate ionization outflow.
Absorption due to Fe XVII - XXIV K near
7.6 keV. Overpredicts K at 6.6 keV.
(c) Low ionization outflow. Fe K absorption
at 7.2 keV. Also requires fast outflow
(0.06c). Overpredicts abs below 6 keV.
logxi=3.5, NH=1.5x1022cm-2, v=0.1c
logxi=2.5, NH=1.5x1022cm-2, v=0
Fe XVIII-XXIV
K and K
(b) 
logxi=1.5, NH=1.5x1022cm-2, v=0
Fe XXVI K
No Fe K
(a) 
Fe < XVII K
(c) 
Origins of the Blue-shifted Fe K-shell
Absorption
• Local (z=0) absorption, such as the WHIM or local hot
bubble unlikely due to variability. In many cases the
outflow velocity differs from the AGN recessional
velocity
• Therefore likely intrinsic to AGN, via outflow.
• Lower velocity solutions have difficulty in fitting Fe K
band data in physically realistic model.
• High velocity, high ionization outflow seems the most
plausible solution
• Future calorimeter resolution (<6 eV) spectra will
provide a wealth of data at Fe K on high column density
outflows.