Studying AGN with highresolution X-ray spectroscopy
Jelle Kaastra
SRON
Nahum Arav, Ehud Behar, Stefano Bianchi, Josh
Bloom, Alex Blustin, Graziella Branduardi-Raymont,
Massimo Cappi, Elisa Costantini, Mauro Dadina, Rob
Detmers, Jacobo Ebrero, Peter Jonker, Chris Klein,
Jerry Kriss, Piotr Lubinski, Julien Malzac, Missagh
Mehdipour, Stéphane Paltani, Pierre-Olivier Petrucci,
Ciro Pinto, Gabriele Ponti, Eva Ratti, Katrien
Steenbrugge, Cor de Vries
Introduction:
The influence of AGN outflows
•
•
•
•
Dispersal heavy elements into IGM & ICM
Ionisation structure IGM
evolution host galaxy
How created? Structure? Mass & energy?
Confinement?
• Crucial to understanding central engine:
– Accretion process
– Energy budget
2
AGN outflows in a nutshell
• Photo-ionised gas
• v = -100 to -1000
km/s
• Seen through line
and continuum
absorption
• Spectrum  ionic
column densities 
ionization
parameter ξ=L/nr²
Kaastra et al. 2000
3
Photoionisation modelling
• Radiation impacts a volume (layer) of gas
• Different interactions of photons with atoms
cause ionisation, recombination, heating &
cooling
• In equilibrium, ionisation state of the plasma
determined by:
– spectral energy distribution incoming radiation
– chemical abundances
– ionisation parameter ξ=L/nr2 with L ionising
luminosity, n density and r distance from ionising
source; ξ essentially ratio photon density / gas density
4
Question 1:
Structure outflow
• Is it more like
rather uniform
density clouds in
pressure
equilibrium?
• Or is it more like
coronal streamers,
with lateral density
stratification?
5
Emission measure
Column density
Absorption measure distribution
Discrete components
Continuous
distribution
Ionisation parameter ξ
Temperature
6
Separate components in pressure
equilibrium?
• Not all components in
press. eq. (same Ξ)
• Division into ξ comps
often poorly defined
•  Continuous NH(ξ)
distribution?
• Others fit discrete
components
• What's going on?
Steenbrugge et al. 2005
7
Question 2:
Where is the gas?
• Photo-ionization
modeling  ξ=L/nr²
• L obtained from
spectrum
•  only the product
nr² known, not r or n
• Is gas accelerating,
decelerating?
8
Density estimates: line ratios
• C III has absorption lines near 1175 Å from
metastable level
• Combined with absorption line from ground (977
Å) this yields n
•  n = 3x104 cm-3 in NGC 3783 (Gabel et al.
2004)  r~1 pc
• Only applies for some sources, low ξ gas
• X-rays have similar lines, but sensitive to higher
n (e.g. O V, Kaastra et al. 2004); no convincing
case yet
9
Density estimates: reverberation
• If L increases for gas at fixed n and r, then
ξ=L/nr² increases
•  change in ionisation balance
•  column density changes
•  transmission changes
• Gas has finite ionisation/recombination
time tr (density dependent as ~1/n)
•  measuring delayed response yields
trnr
10
Reverberation: NGC 3783
• RGS data (Behar et al. 2003): no change in
Warm absorber  n<300 cm-3, r>10 pc.
• EPIC data (Reeves et al. 2003): change in
Warm absorber (larger columns)  n>108 cm-3,
r<0.02 pc.
11
Observation campaign Mrk 509
• Core: 10 x 60 ks XMM, spaced 4 days (RGS, EPIC & OM all used!)
• Simultaneous Integral 10 x 120 ks
• Followed by 180 ks Chandra LETGS, simultaneous with 10 orbits
COS (HST)
• Preceded with Swift (UVOT, XRT) monitoring
• Supplemented with ground-based (WHT, Pairitel) photometry &
grism
• Period: 4 Sept – 13 Dec 2009 (100 days)
• 7 papers submitted/accepted, 8 in progress
12
General data analysis issues
• Excellent quality  many new steps developed
• RGS: full usage multi-pointing mode, refinements
combining spectra with variable hot pixels, λ-scale,
effective area, reducing response 2Gb8 Mb,
rebinning…) See Kaastra et al. 2011; see auxilary
programs in SPEX distribution www.sron.nl/spex
• PN: Triggered by our campaign improved gain cal
• OM: extended wavelength range optical grism
• HST/COS: extensive efforts to improve data analysis for
this high-quality spectrum
• SPEX: allows simultaneous fitting high-res UV & X-ray
spectra
13
Broad-band spectrum & variability
(Mehdipour et al., Petrucci et al., talks
this afternoon)
14
Line Intensity
Fe-K line & variability
(Ponti et al.; talk Petrucci et al.)
• Broad and neutral Fe K emission
well correlated with continuum
emission on few days timescales.
• No relativistic profile
• Origin outer disc or inner BLR
3-10 keV flux
.
OM Grism spectrum
16
UV-optical variability
(OM, Swift)
• Source was in
outburst (0.1 dex in
UV) right in the
middle of ourt
campaign!
17
ISM absorption
(talk Pinto et al. on Monday)
18
Abundances
(Steenbrugge et al., poster G45)
19
COS & FUSE UV Absorption lines in Mrk 509
(Kriss et al., talk this afternoon)
• O VI,Lyβ, & Lyγ from FUSE
(Kriss et al. 2000)
• 14 velocity components seen
in COS UV spectrum
• C IV doublet split by only 500
km/s, so grey regions can’t be
used for optical-depth
• Red lines: velocities X-ray
absorbers XMM-Newton/RGS
• Blue lines: velocities X-ray
absorbers Chandra/LETGS
20
LETGS + COS data
(Ebrero et al., poster G14)
• X-rays: outflow with 3 distinct
ionization phases in form of
multi-velocity wind. UV
spectra: absorption system
with 13 kinematic components
• Analysis kinematic properties
& column densities absorbers
 UV-absorbing gas colocated with & embedded in
lower density high-ionization
X-ray absorbing gas
21
Stacked RGS spectrum
• Galactic O I edge
• Several narrow absorption
lines
• Detection 31 individual ions
• Tight upper limits column
densities 18 other ions
• Two main velocity
components (+40 and -300
km/s)
• Highly ionised ions in
general higher velocity
22
No O I from host galaxy
O I host galaxy
(not detected,
NH<5x1018 cm-2)
23
X-ray analysis in more detail
• Fit spectra using a power law + modified
blackbody (or even a spline) continuum
• Where needed, add emission lines: BLR or NLR
X-ray lines (no relativistic lines needed)
• Fit warm absorber using a model  ionic or total
column densities
• Using photo-ionisation model, derive absorption
measure distribution NH(ξ)
• Spectral fits done with SPEX, global fits
24
Sample high-resolution spectra
25
Example of broad emission
lines
O VIII Lyα
O VII triplet
26
Photoionisation modelling RGS spectrum
(Detmers et al. 2011)
•
•
•
•
Use time-averaged SED
Test dependence on SED
Proto-solar abundances
Multiple ionisation, 3
velocity components
• Same ion may be present
in more than one
velocity/ionisation
component!
Mehdipour
et al. 2011
27
Discrete versus continuous
absorption measure distribution?
• Column densities well
approximated by sum
of 5 components (see
table)
• Higher column for
higher ionisation
parameter
• But what about a
continuous model?
E
D
A
B
C
28
Continuous model
• Fitted columns with
continuous (spline) model
• Surprise: comps C & D
pop-up as discrete
components!
• Upper limits FWHM 35 &
80 %
• Component B (& A) too
poor statistics to prove if
continuous
• Component E also poorer
determined: correlation ξ
and NH.
D
E
C
B
29
Where is the gas?
• Needs t-dependent model
• For each of the 5 components: if L increases, ξ
must increase (as ξ=L/nr2)
• From 10 continuum model fits  predicted ξ
change for each of 10 observations, compared
to average spectrum
• If gas responds immediately, transmission
outflow changes immediately
30
Expected response absorbers for
0.1 dex luminosity increase
31
Time-dependent photoionisation
• Time evolution ion
concentrations ni:
• dni/dt = Aij(t) nj
• Aij(t) contains tdependent
ionisation &
recombination
rates
32
Location photoionised gas
(work in progress)
• Component A: Rotating, low ionisation
disk + high velocity outflow, 3 kpc (direct
imaging [O III] 5007, Philips 1986)
• Component C: >10 pc (lack of response)
• Component E: > 1 pc? (lack of response?)
• See also Kriss (talk this afternoon) and
poster Ebrero
33
Mass loss through the wind

M loss   m p nr v

2

M loss  M acc
( / v)

2
m p c
nr  v  ( L /  )  v
2

L   M acc c 2
v (km/s)
-100
-1000
ξ=1
0.001
0.0001
ξ=1000
1.0
0.1
34
Conclusions
• Deep, multi-wavelength monitoring
campaigns (AGN) are rewarding:
• High quality spectra, not limited by
statistics
• “Continuous” light curves, allowing to
monitor the variations
35