A new look at AGN X-ray spectra - the imprint of massive and energetic outflows Ken Pounds University of Leicester Prague August 2006

1H 0419 NGC 509 NGC 5548 Mkn 766 L 2-10 (10 42 ergs/s) most obvious feature is the ‘soft excess’ - a function of luminosity ? also fluorescent Fe K line, sometimes with ‘red wing’ XMM-EPIC spectra for sample of type 1 AGN

but the same spectral features (and changes) are seen for individual AGN example - the luminous Seyfert 1 PG XMM pn data from 2001 (black) and 2004 (red) observations

Appearances can be deceptive - compare the unmodelled data difference lies between ~0.7 and ~2 keV due to a change in photoionised absorption?

Spectral ratio supports change in ionised absorption as cause of spectral difference Gierlinski and Done (2004) and Chevallier et al (2006) have explored this idea as an explanation of strong soft excesses. Problems with relativistically smeared absorber and with ‘fine tuning’ 2001/2004 data

difference spectra showed: dominant spectral change due to steep power law component (Gamma ~ 2.4) Fluxed spectra of the highly variable Seyfert 1H (*) Could another component be moderating the spectral change? Pounds et al, ApJ 2004

PG1211 exhibits a high velocity outflow which, unless highly collimated, has a mechanical energy comparable to the bolometric luminosity (more strictly ~ v/c if continuum driven) * Could this power the steep, variable continuum component ? Pounds et al MN 2003 King and Pounds ApJ 2003 But lower velocity claimed by Kaspi and Behar, ApJ 2006

analysis of higher resolution XMM MOS data has improved constraints on high energy outflow (*) K-shell absorption from H- and He-like ions of Ne, Mg, Si, S and Fe >> v~0.14+/-0.01c Pounds and Page (astro-ph )

Developing a new spectral model components include: power law continuum from disc-corona absorption by high and moderate ionisation gas re-emission from ionised gas components second (power law) continuum powered by outflow

the result - an excellent spectral fit to the EPIC data (chi_sq ~ 840/830 dof) - - with no ‘smearing’

Compare energies and luminosities as further test of model Absorption by warm absorber ~ 8.4 x ergs/s from PL1 and 5.3 x ergs/s from PL2 >>> 9 x ergs/s Absorption by high Xi absorber ~ 1.4 x ergs/s from PL1 and 5 x ergs/s from PL2 >>> 1.9 x ergs/s Luminosity of warm emitter ~ 1.7 x ergs/s >>> CF ~ 0.2 Luminosity of high Xi emitter ~ 1.4 x ergs/s >>> CF ~ 0.1 Luminosity ( ) in second power law ~ 6 x ergs/s while mechanical energy in fast flow and CF of 0.1 ~ 6 x ergs/s >>> efficiency ~ 10% NB warm emission ~ 35 x that of NGC1068 (but only 5-10 x higher L bol )

Critical fit to high resolution RGS data - yes, but requires emission lines to be strongly broadened Steep power law continuum consistent with EPIC model

the broadened emission line structure in the RGS spectral fit N, O, Ne, Mg removed from XSTAR model Fe also removed, to leave continuum only

Finally - how does the same model fit the 2004 data? primarily by increase in weakly absorbed power law

Summary blue-shifted absorption lines in the luminous Seyfert 1 galaxy PG show highly ionized outflow at v ~ 0.14c the mechanical energy in the high velocity outflow is sufficient to power a second X-ray continuum component perhaps via internal shocks in the flow (NB efficiency similar to GRB ?) re-emission of the absorbed flux from primary power law visible in the soft X-ray spectrum (strong line broadening) the need for a relativistic Fe K line is reduced or removed by the spectral curvature caused by continuum absorption similar spectral features in other luminous AGN suggest PG may be characteristic of a high accretion ratio NB. outflow of 1 M Sun / yr for ~ 5x10 7 yr will carry mechanical energy to host galaxy ~ ergs ( cf binding energy ~ erg of galactic bulge with M ~ M Sun and velocity dispersion ~ 300 km/ s; >>> major effect on growth of host galaxy

Warm absorber/emitter Ionisation parameter >>> n.r 2 ~ /35 ~ 2 x column density >>> n.d ~ 8 x assume d ~ 0.1r >>> r ~ 2 x cm and n ~ 5 x 10 5 cm -3 Emission measure ~ cm -3 from observed luminosity spherical shell with above parameters >>> n 2. volume ~ 3 x cm -3 >>> CF ~ 0.3 High Xi absorber/emitter Ionisation parameter >>> n.r 2 ~10 43 /750 ~ 1.3 x column density >>> n.d ~ 6 x assume d ~ 0.1r >>> r ~ 2 x cm and n ~ 4 x 10 6 cm -3 Emission measure ~ 2 x cm -3 from observed luminosity spherical shell with above parameters >>> n 2. volume ~ 2 x cm -3 >>> CF ~ 0.1 Scaling the ionised outflow components

EPIC (and RGS) spectra show evidence for outflowing matter with wide range of ionisation parameter. Unless tightly collimated the high velocity implies: (1)a high mass rate with M dot.out ~ 1.0 M Sun /yr, comparable to M dot.Edd ~ 1.2 M Sun /yr for a BH mass of 3 x 10 7 M Sun accreting at an efficiency of 0.1, and (2)an energetic outflow ~ 2 x erg/s (~v/c x L Edd ) with mechanical energy significantly larger than the luminosity of the X-ray PL NB. outflow of 1 M Sun / yr for ~ 5x10 7 yr will carry mechanical energy to host galaxy ~ ergs ( cf binding energy ~ erg of galactic bulge with M ~ M Sun and velocity dispersion ~ 300 km/ s; hence will have a major effect on growth of host galaxy

But is the spectrum compatible with no strong narrow lines in RGS? - model fit (PO2 + EM1) requires smoothing to get acceptable chi-sq - find similar ratio of line emission to continuum and PL slope is again steep (Gamma ~ 3) However, best-fit sigma ~ 18 eV at 0.5 keV Corresponds to v ~ km s -1 FWHM additional broadening from line saturation, turbulence or spread in ionisation parameter?

PG is a bright, narrow emission line quasar at z ~ L bol ~ 6 x ergs/s (H0 = 75) M BH ~ 3 x 10 7 M Sun suggesting accretion rate ~ Eddington rate Pre-XMM observations showed a strong `soft excess’ over a PL of photon index ~ 2, with some evidence for a broad Fe K line XMM made a ~ 60 ks observation of PG in June 2001 and second similar observation in June 2004 Consider the case of PG where an energetic outflow is well established

Lyman-alpha of FeXXVI MgXII SXVI * * * Identification of narrow absorption lines in EPIC pn and MOS with resonance lines of H- like ions of Fe, S, and Mg, consistent with high ionisation, high column density outflow of v ~0.08 – 0.1c see also IC4329A (Markowicz et al astro-ph/ )

in particular the absorption features at ~ 1.5 and 2.7 keV are both resolved into two lines a closer look at the MOS data for PG1211

best fit now if 7 keV feature is identified with 1s-2p resonance absorption line of Fe XXV

Finally - testing the new model with 2004 pn data abundances fixed at 2001 values fit acceptable (chi-sq / dof ~ 1.1) reduced absorption modelled mainly by stronger steep power law component weaker soft X-ray emission as indicated by difference spectrum

The initial analysis used 4 of the 5 most significant absorption features to find an outflow velocity of v~0.09c.

- ignoring absorption near the Si edge and an ‘emission line’

Addition of a trace absorber affecting PO2 improved chi-sq from 877/817 to 832/814 N H ~1.2x10 21 cm -2 log Xi~ 1.95 v_out~ 0.07 – 0.10 c Removing absorber shows absorption due to Fe-L and UTA

Initial ratio of 2004 data to 2001 model New fit with increase in PO2 and weaker soft X-ray line emission

conventional plot with PL fit at 1-10 keV -more spectral structure seen below ~ 3 keV a closer look at the MOS data for PG1211

1H in bright, high flux state and faint, low flux state But, individual AGN show very similar spectral variations Perhaps accretion rate is the key?

Conventional analysis of EPIC spectrum (pn black; MOS red) reported in 2003 paper* Power law fit at 1-10 keV found a photon index ~ 1.85 and a strong SX 1-10 keV spectrum modelled with a power law plus an extreme relativistic line Fe K emission line (EW ~ 1.4 keV) Plus narrow absorption features at ~1.5, ~2.7, ~7 and ~8 keV * (MNRAS, 345, 705, 2003)

MOS has equally good statistics at medium and low energies – and better energy resolution (1 sigma~34 eV at 1 keV) Part of MOS spectrum of NGC 1068 (MNRAS, 368, 707, 2006)

Chi-sq plot shows high significance of the 5 absorption lines and (the same) emission lines (*) as seen in pn data higher resolution MOS spectra confirms the strong, highly ionised outflow in PG , albeit with velocity increased to v~0.12c

The fitted model includes power law continua with photon indices ~ 2.1 and ~3 plus high ionisation absorber A1 and warm absorber A2 affecting both power laws E1,E2 re-emission from absorbing gas N H ~2x10 23 A1 log Xi~ 2.9 v_out~ /-0.005c N H ~3x10 22 A2 logXi~ 1.5 v_out~ 0.05+/-0.02c logXi~ 1.4 E1 v_out~ 3000+/-1000 km/s logXi~ 2.9 E1 v_out~ 3000+/-1000 km/s Chi-sq/d.o.f. ~ 820/820

the main spectral change appears to be an increase in flux between ~ 0.7 – 2 keV narrow absorption lines remain, but weaker Does model describe spectral change in 2004 ? -- the plot shows pn data from 2001 (black) and 2004 (red)

2-10 ~45% 1-2 ~40% ~25% ~32% flare over ~10 4 sec in full keV light curve variation least in soft X-ray band as expected for significant re- processed flux Footnote: Variability of EPIC light curve also supports model