1: \documentclass[useAMS,usenatbib]{mn2e}

2: \usepackage{graphicx}

3: \usepackage{natbib}

4: \usepackage{amssymb}

5: \bibpunct{(}{)}{;}{a}{}{,}

6: \newcommand{\aap}{A\&A}

7: \newcommand{\aaps}{A\&AS}

8: \newcommand{\pasp}{PASP}

9: \newcommand{\apj}{ApJ}

10: \newcommand{\apjl}{ApJ}

11: \newcommand{\apjs}{ApJS}

12: \newcommand{\mnras}{MNRAS}

13: \newcommand{\aj}{AJ}

14: \newcommand{\araa}{ARA\&A}

15: \newcommand{\gca}{Geocosmochimica Acta\ }

16: \newcommand{\AuJP}{AuJP\ }

17: \newcommand{\nat}{Nat}

18: \newcommand{\pasj}{PASJ\ }

19: \newcommand{\procspie}{Proc. SPIE\ }

20: \newcommand{\advspres}{Adv. Sp. Res.\ }

21: \newcommand{\iaucirc}{IAU Circ.\ }

22: \newcommand{\qjras}{QJRAS\ }

23: \def\mrk{Mrk~279}

24: \def\ngc{NGC~5548}

25: \def\zw{I\,Zw\,1}

26: \def\ch{{\it Chandra}}

27: \def\hst{HST-STIS}

28: \def\xmm{{\it XMM-Newton}}

29: \def\epi{EPIC-pn}

30: \def\rg{RGS}

31: \def\nl{NLSy1}

32: \def\ho{1H~0707-495}

33: \def\Halpha{\ifmmode {\rm H}\alpha \else H$\alpha$\fi}

34: \def\Hbeta{\ifmmode {\rm H}\beta \else H$\beta$\fi}

35: \def\Hgamma{\ifmmode {\rm H}\gamma \else H$\gamma$\fi}

36: \def\Hdelta{\ifmmode {\rm H}\delta \else H$\delta$\fi}

37: \def\Lya{\ifmmode {\rm Ly}\alpha \else Ly$\alpha$\fi}

38: \def\Lyb{\ifmmode {\rm Ly}\beta \else Ly$\beta$\fi}

39: \def\Lyg{\ifmmode {\rm Ly}\beta \else Ly$\gamma$\fi}

40: \def\fei{Fe\,{\sc i}}

41: \def\feii{Fe\,{\sc ii}}

42: \def\feiii{Fe\,{\sc iii}}

43: \def\feiv{Fe\,{\sc iv}}

44: \def\fexx{Fe\,{\sc xx}}

45: \def\hi{H\,{\sc i}}

46: \def\hii{H\,{\sc ii}}

47: \def\hei{He\,{\sc i}}

48: \def\heii{He\,{\sc ii}}

49: \def\ci{C\,{\sc i}}

50: \def\cii{C\,{\sc ii}}

51: \def\ciii{\ifmmode {\rm C}\,{\sc iii} \else C\,{\sc iii}\fi}

52: \def\civ{\ifmmode {\rm C}\,{\sc iv} \else C\,{\sc iv}\fi}

53: \def\cv{\ifmmode {\rm C}\,{\sc v} \else C\,{\sc v}\fi}

54: \def\cvi{\ifmmode {\rm C}\,{\sc vi} \else C\,{\sc vi}\fi}

55: \def\ni{N\,{\sc i}}

56: \def\nii{N\,{\sc ii}}

57: \def\niii{N\,{\sc iii}}

58: \def\niv{N\,{\sc iv}}

59: \def\nv{N\,{\sc v}}

60: \def\nvi{N\,{\sc vi}}

61: \def\nvii{N\,{\sc vii}}

62: \def\oi{O\,{\sc i}}

63: \def\oii{O\,{\sc ii}}

64: \def\oiii{O\,{\sc iii}}

65: \def\o5007{[O\,{\sc iii}]\,$\lambda5007$}

66: \def\oiv{O\,{\sc iv}}

67: \def\ov{O\,{\sc v}}

68: \def\ovi{O\,{\sc vi}}

69: \def\ovii{O\,{\sc vii}}

70: \def\oviii{O\,{\sc viii}}

71: \def\neiii{Ne\,{\sc iii}}

72: \def\nev{Ne\,{\sc v}}

73: \def\neviii{Ne\,{\sc viii}}

74: \def\neix{Ne\,{\sc ix}}

75: \def\nex{Ne\,{\sc x}}

76: \def\mgi{Mg\,{\sc i}}

77: \def\mnii{Mn\,{\sc ii}}

78: \def\Niii{Ni\,{\sc ii}}

79: \def\mgii{Mg\,{\sc ii}}

80: \def\mgix{Mg\,{\sc ix}}

81: \def\mgx{Mg\,{\sc x}}

82: \def\mgxi{Mg\,{\sc xi}}

83: \def\siiv{Si\,{\sc iv}}

84: \def\siIII{Si\,{\sc iii}}

85: \def\siII{Si\,{\sc ii}}

86: \def\six{Si\,{\sc x}}

87: \def\sixi{Si\,{\sc xi}}

88: \def\sixii{Si\,{\sc xii}}

89: \def\sixiii{Si\,{\sc xiii}}

90: \def\si{S\,{\sc i}}

91: \def\sii{S\,{\sc ii}}

92: \def\siii{S\,{\sc iii}}

93: \def\siv{S\,{\sc iv}}

94: \def\sv{S\,{\sc v}}

95: \def\svi{S\,{\sc vi}}

96: \def\sx{S\,{\sc x}}

97: \def\sxi{S\,{\sc xi}}

98: \def\sxii{S\,{\sc xii}}

99: \def\sxiii{S\,{\sc xiii}}

100: \def\ariii{Ar\,{\sc iii}}

101: \def\ariv{Ar\,{\sc iv}}

102: \def\arv{Ar\,{\sc v}}

103: \def\arvi{Ar\,{\sc vi}}

104: \def\arvii{Ar\,{\sc vii}}

105: \def\arviii{Ar\,{\sc viii}}

106: \def\caii{Ca\,{\sc ii}}

107: \def\fei{Fe\,{\sc i}}

108: \def\feii{Fe\,{\sc ii}}

109: \def\feiii{Fe\,{\sc iii}}

110: \def\fev{Fe\,{\sc v}}

111: \def\feix{Fe\,{\sc ix}}

112: \def\fexiv{Fe\,{\sc xiv}}

113: \def\fexvi{Fe\,{\sc xvi}}

114: \def\fexvii{Fe\,{\sc xvii}}

115: \def\fexviii{Fe\,{\sc xviii}}

116: \def\fexix{Fe\,{\sc xix}}

117: \def\fexxiii{Fe\,{\sc xxiii}}

118: \def\fexxiv{Fe\,{\sc xxiv}}

119: \def\fexxvi{Fe\,{\sc xxvi}}

120: \def\alii{Al\,{\sc ii}}

121: \def\aliii{Al\,{\sc iii}}

122: \def\piv{P\,{\sc iv}}

123: \def\pv{P\,{\sc v}}

124: \def\cliv{Cl\,{\sc iv}}

125: \def\clv{Cl\,{\sc v}}

126: \def\nai{Na\,{\sc i}}

127: %

128: \def\o{\o}

129: \def\kms{km\,s$^{-1}$}

130: 

131: 

132: 

133: \title[The ionised X-ray absorber of \zw\ observed by \xmm]{A longer \xmm\ look at I\,Zwicky\,1:\\ 

134: physical conditions and variability of the ionised

135: absorbers}

136: 

137: \author[E.~Costantini et al.]{E.~Costantini $^{1,2}$\thanks{E-mail: e.costantini@sron.nl}, L.C.~Gallo$^{3,4}$, 

138: W.N.~Brandt$^{5}$, A.C.~Fabian$^{6}$ and Th.~Boller$^{4}$\\

139: $^{1}$ SRON National Institute for Space Research, Sorbonnelaan 2, 3584 CA Utrecht, The Netherlands\\

140: $^{2}$ Astronomical Institute, Utrecht University, P.O. Box 80000, 3508 TA, Utrecht, The Netherlands\\

141: $^{3}$ SUPA, School of Physics and Astronomy, University of St.~Andrews, North Haugh, St.~Andrews, Fife KY16 9SS\\

142: $^{4}$ Max-Planck Institute f\"uer extraterrestrische Physik, Giessenbachstr 1, 45748, Garching bei Muenchen, Germany \\

143: $^{5}$ Department of Astronomy and Astrophysics, Pennsylvania State University, University Park, PA 16802\\

144: $^{6}$ Institute of Astronomy, University of Cambridge, Madingley Road, Cambridge CB3 0HA}

145: 

146: \begin{document}

147: \date{Received  / Accepted  }

148: \pagerange{\pageref{firstpage}--\pageref{lastpage}} \pubyear{2007}

149: 

150: \maketitle

151: 

152: \begin{abstract}

153: We present a spectral analysis of the narrow-line Seyfert~1 galaxy I~Zwicky\,1, focusing on the characteristics of the ionised absorbers as observed with \xmm\ in 2005. The soft X-ray spectrum shows

154: absorption by two components of ionised gas with a similar column density 

155: ($N_{\rm H}\sim10^{21}$\,cm$^{-2}$) and ionisation parameters 

156: $\log\xi\sim0$ and $2.5$. Comparing this observation with a 2002 \xmm\ data set, we see a clear

157: anti-correlation between the X-ray ionisation parameter $\xi_{\rm X}$ and the $0.1-10$\,keV luminosity. Viable

158: explanations for this effect include transient clouds or filaments crossing the line of sight in a complex geometry or a 

159: gas observed in non-equilibrium. The outflow velocity of

160: the X-ray low-ionisation absorber is consistent with the outflow of the UV absorber detected in a past

161: Hubble Space Telescope observation. In

162: addition, the ionic column densities of \civ\ and \nv\ derived from the X-ray model are consistent with the UV values.

163: This suggests that the low-ionisation outflowing gas may survive for many years, despite large changes in flux, 

164: and that there is a tight connection between the X-ray and UV absorbers that can only be confirmed with a simultaneous UV and X-ray

165: observation.      

166: \end{abstract}

167: 

168: \begin{keywords}

169: Galaxies: individual: I\,Zw\,1 -- Galaxies: Seyfert -- quasars: absorption lines -- Ultraviolet: galaxies --

170:  X-rays: galaxies

171: \end{keywords}

172: 

173: 

174: 

175: 

176: 

177: \section{Introduction}

178: About 50--70\% of active galactic nuclei (AGN) display absorption by photoionised gas in their 

179: X-ray spectra \citep[e.g.][]{reynolds97,george98}. With the advent of high energy resolution instruments, 

180: the study of the brightest sources has allowed us to detail the 

181: characteristics of such gas \citep[e.g.][]{kaastra02,kaspi02,kraemer05,turner05}. 

182: Multiple components are often detected, differing in column density, ionisation and outflow velocity. At least some of the 

183: lower ionisation X-ray components most probably share the same location as the UV components 

184: \citep[e.g.][]{turner05,elisa06} as first

185: suggested by \citet{smita95}. Variability studies point out the complex response of the 

186: ionised absorbers (also called

187: warm absorbers) to central-source flux variations. A linear response of the 

188: ionisation parameter\footnote{The ionisation parameter is defined as $\xi=L/nr^2$, where $L$ is the 1--1000 Ryd ionising 

189: luminosity, $n$ and $r$ are the

190: density and the distance to the ionised gas.} of the gas component to the continuum change provides a constraint on the 

191: distance to the absorber 

192: \citep[e.g.][]{netzer02,turner05,otani96}. A linear response gives a relatively intuitive picture of the geometry of the 

193: absorber as a thin slab of material in thermal equilibrium (either in clouds or filaments) exposed to the ionising flux. 

194: In other variability studies \citep[e.g. NGC~4151,][]{kraemer05} the gas does not obey this simple picture, 

195: suggesting instead

196: that gas outflows often change direction along the line of sight in a unpredictable way.\\    

197: I~Zwicky\,1 (\zw) is a well-studied narrow-line Seyfert~1 galaxy \citep[NLS1; see][and references therein]{vest_wilkes} 

198: located at

199: redshift $z$=0.06112$\pm$0.000015 \citep{condon}.

200: Since the time of ROSAT observations, \zw\ has been known to host intrinsic absorption in the X-ray band \citep{boller96,lawrence97}.

201: From moderate-resolution ASCA data, \citet{leighly99} could distinguish between a neutral component ($N_{\rm

202: H}\sim2.6\times10^{20}$\,cm$^{-2}$) and an ionised absorber, typically highlighted, in lower resolution

203: spectra, by 

204: the \ovii\ and \oviii\ absorption edges. A similar 

205: spectral complexity was reported by \citet{luigi04}, using \xmm\ data. The presence of fast ($v_{\rm

206: out}\sim-1870$\,\kms) outflowing

207: ionised gas was observed in the UV spectrum of \zw\ \citep[using HST-FOS;][]{laor97,crenshaw99}, 

208: through the analysis of the \nv\ and \civ\ absorption troughs. FUSE observed the \ovi\ absorption troughs, 

209: which displayed multiple components \citep[][]{kriss02}.\\      

210: This paper is one in a series devoted to a long \xmm\ observation of \zw. 

211: In the present study we focus

212: on the soft X-ray absorbers, discussing the physical picture returned by the long-term variability 

213: of this gas and its connection with the UV absorbers. Other works deal with the continuum emission and the variable

214: iron K$\alpha$ line (Gallo et al. 2007a, hereinafter Paper~I), and with the spectral variability analysis (Gallo et al. 2007b).

215: 

216: The paper is organised as follows. In Sect.~2 the main features of the data reduction and 

217: the \epi\ analysis 

218: (described in detail in Paper~I) 

219: are summarised. Sect.~3 is devoted to the analysis of the \rg\ data and the modelling of the 

220: spectral features in \zw. In Sect.~4 we discuss our results and in

221: Sect.~5 we present the conclusions. 

222: The cosmological parameters used are $H_0$=70 \kms Mpc$^{-1}$, $\Omega_m$=0.3, and $\Omega_{\Lambda}$=0.7.

223: The quoted errors are at 90\% confidence for one interesting parameter, unless

224: otherwise stated. The assumed Galactic

225: column density is $N_{\rm H}=5.07\times10^{20}$\,cm$^{-2}$ \citep{elvis89}. 

226: 

227: 

228: %--------------------------------------DATA ANALYSIS

229: 

230: 

231: \section{The data analysis}\label{sect:data}

232: \zw\ was observed on July 18-19 2005 (rev. 1027) for 85.5\,ks. 

233: Data from the three EPIC cameras MOS \citep{2001A&A...365L..27T} and pn \citep{2001A&A...365L..18S}, the

234: high resolution spectrometer RGS \citep{2001A&A...365L...7D} and the optical monitor \citep[OM, ][]{mason} 

235: were

236: successfully gathered. After filtering high-background intervals, the net exposure time was $\sim58$\,ks for

237: \epi\ and $\sim84$\,ks for RGS. The event files for each instrument were created with the newest SAS pipeline (ver.~7.0).

238: This version of SAS benefits from improved calibrations of \rg\ time-dependent effective area, absolute \rg\

239: flux and bad-pixel location\footnote{For further details refer to \\

240: http://xmm.vilspa.esa.es/external/xmm\_user\_support/usersgroup\\/20060518/index

241: .shtml.}. The discrepancies between \epi\ and \rg\ are of the

242: order of 10\%, similar as for EPIC-MOS when compared to \epi.\\ 

243: The spectral analysis of the soft X-ray spectrum was carried out using the fitting

244: package 

245: SPEX\footnote{http://www.sron.nl/divisions/hea/spex/version2.0/\\release/index.html} (ver.~2.0). 

246: We rebinned the \rg\ data to reach

247: a signal-to-noise ratio (S/N) of $\sim$9 over the whole energy band. During the observation the source shows two different average flux states, which differ approximately 

248: by 44\% and 36\% in the soft (0.3--2\,keV) and hard (2--10\,keV) energy bands, respectively. 

249: The flux changes after $\sim$27\,ks since the start of the observation and happens on a dynamical time scale of $\sim$5\,ks. 

250: The flux variations in the soft and hard X-ray bands show similar trends (see Fig.~3 of Paper~I). We will study the effect of the 

251: flux change on the RGS spectrum in Sect.~\ref{par:data}.\\ 

252: The data from the EPIC-pn camera were used to constrain the broad-band spectrum. The modelling of the continuum

253: is described in full detail in Paper~I. 

254: For the purpose of studying any narrow absorption/emission features in the \rg\ band, the detailed broad-band continuum shape is not

255: crucial. 

256: For simplicity, here we describe the continuum with a broken power-law, which fits the data well 

257: (Tab.\ref{t:rgs_05}). However in Sect.~\ref{par:conti}, we

258: test our results for different continua. 

259: 

260: 

261: 

262: 

263: %-----------------------------------------MODELLING 05

264: 

265: \section{The soft energy spectrum}\label{par:data}

266: The \rg\ spectrum shows the signatures of absorption, especially around the \ovii\ and \oviii\ absorption edges. 

267: In order to fit a photoionised absorber, we used the SPEX model XABS, which interpolates on a fine grid the $N_{\rm

268: H}$ and

269: $\log\xi$ values. The ionisation balance is constructed 

270: from the spectral energy distribution (SED) specific for the source (solid curve in Fig.~\ref{f:sed}), using Cloudy \citep{fer98}. 

271: Here and in the following analysis, the SED definition takes

272: advantage of the simultaneous \epi\ and OM data. The OM data were collected in fast-mode (see Paper~1) 

273: in the {\it U} (3440\AA), UVW1 (2910\AA), UVW2 (2120\AA) and UVM2 (2310\AA) filters. 

274: We considered the average flux for each filter.  

275: By

276: interpolation, we obtained the flux at 2500\,\AA\ that we used in constructing the SED. 

277: Interpolating the OM points was mandatory in order to 

278: compare consistently the ionisation balance obtained with this SED with the one obtained from a 

279: previous \xmm\ data set for which only one OM filter was available  (Sect.~\ref{par:xmm02}). An adopted SED will 

280: always have a degree

281: of uncertainty, when dealing with limited multiwavelength data. When comparing SEDs of different flux states of the same object it is important to consistently compare the resulting ionisation balances. The SED of 2002 was likely 

282: similar to 2005 in the UV regime, but in 2002 only one OM filter was available. 

283: This would make any ``reconstruction" of the 2002 SED highly arbitrary. Therefore the only way

284: to consistently compare the SEDs at the two epochs is to interpolate the 2005 OM points. Beyond the optical band, we used a standard AGN continuum, characterised by relatively low emission at longer wavelengths, as defined in 

285: Cloudy. The host galaxy of \zw\ is indeed very luminous both in the far-infrared and in the infrared \citep{halpern87}. 

286: In the long-wavelength band it is important to avoid dust contamination of the nuclear emission by the host galaxy. In general, an overestimation of the infrared radiation may lead to an incorrect ionisation balance determination \citep[see e.g.][]{ferl02}.\\ 

287: Tunable parameters of the XABS model, in addition to $N_{\rm H}$ and $\log\xi$, 

288: are the root mean squared width of the absorption

289: lines, the outflow velocity, the line-of-sight covering factor of the gas 

290: and the elemental abundances. We fixed the covering 

291: factor to unity and we assumed 

292: solar abundances \citep{anders} throughout the analysis.\\  

293: In Tab.~\ref{t:rgs_05} we list the best-fit parameters for the absorbers in \zw\ as measured by the RGS and, 

294: for consistency, with the lower resolution \epi\ data. 

295: The model parameters are always consistent between \rg\ and \epi. However, the higher energy resolution makes \rg\ more

296: reliable for the modelling of narrow absorption features, and we will consider the physical 

297: parameters constrained with the \rg\ data for

298: the discussion.\\  

299: A warm absorber (hereinafter labelled 05a), with a column

300: density $N_{\rm H}\sim1.3\times10^{21}$\,cm$^{-2}$ and an ionisation parameter $\log\xi\sim0$ 

301: is highlighted in the RGS data especially by the blend of the \ovii\ K absorption edge and the 

302: iron unresolved transition array (UTA). The gas is found to be outflowing at 

303: $\sim$1700\,\kms. The widths of the lines are loosely constrained ($\sigma=36\pm20$\,\kms), but still consistent with the 

304: more accurate FUSE measurement of 30$\pm$5\,\kms\ (G.~Kriss, private communication). 

305: Additional ionised absorption is required to fit the region below the 

306: \oviii\ edge ($\lambda=14.22$\AA). 

307: The improvement of the fit after the inclusion of this second absorber is $\Delta\chi^2/\Delta\nu$=27/2, where $\nu$ is the number

308: of degrees of freedom. The second gas component (hereinafter 05b) has a higher ionisation parameter ($\log\xi\sim2.6$,

309: Tab.~\ref{t:rgs_05}) but a column

310: density similar to the 05a component. We fixed the widths of the lines to the value for the lower ionisation

311: component. The absorption lines of the 

312: higher ionisation component lie in a spectral region of lower resolution, and a possible outflow velocity is hard to

313: quantify. The upper limit on the outflow velocity of the 05b component is

314: $v_{\rm out}<800$\,\kms. 

315: Finally, a neutral 

316: component, identified by an \oi\ edge at the redshift of the source and with $N_{\rm H}\sim1.5\times10^{20}$\,cm$^{-2}$, 

317: is required by the data. The addition of this component 

318: improves the fit by $\Delta\chi^2/\Delta\nu$=19/1 (where the column density is the additional parameter). 

319: In Fig.~\ref{f:rgs_tra05}, we plot the transmission of the neutral and warm

320: gas components (upper panel), the observed \rg\ spectrum with the best-fit model (middle panel) and the residuals from the

321: best fit (lower panel).\\ 

322: The above analysis refers to the total \rg\ spectrum. 

323: However (Sect.~\ref{sect:data}), 

324: the soft and hard energy bands undergo variations during the \xmm\ pointing. 

325: Using the \rg\ data we further verified that only continuum changes 

326: are responsible for the 

327: overall variations in the source flux and not opacity variations of the ionised gas. 

328: We divided the \rg\ data into two time segments,

329: corresponding to the higher ($\sim$30\,ks long) and lower state ($\sim$40\,ks). 

330: The higher state best fit does 

331: not vary significantly with respect to the average spectral fit, except for the normalisation and slope of the power law.    

332: We then used the higher flux state best-fit model (see above) as a template for the lower flux state and found that the

333: neutral and warm absorbers do not vary within the uncertainties. 

334: 

335: \begin{figure}

336: \begin{center}

337: 

338: \resizebox{\hsize}{!}{\includegraphics[angle=90]{f1.ps}}

339: \end{center}

340: \caption{\label{f:sed} The adopted SEDs for the 05 observation (solid line) and 02 observation (dashed line). 

341: The X-ray shape of the SED is based on

342: the \epi\ data, while the point at 2500\AA\ is derived by interpolation of

343: the OM data points. The 05 OM points (open diamonds around 10$^{15}$\,Hz) were

344: interpolated to the same wavelength measurement of 02, in order to compare consistently the 05 SED with the 02 one, for

345: which data from only one OM filter was available.} 

346: \end{figure}

347: 

348: 

349: \begin{figure*}

350: \begin{center}

351: 

352: \resizebox{\hsize}{!}{\includegraphics[angle=90]{f2.ps}}

353: \end{center}

354: \caption{\label{f:rgs_tra05} Upper panel: 

355: transmission models (convolved with the data resolution) 

356: of the neutral absorber (dashed line) and the ionised components for the 05 spectrum. 

357: Component 05a is displayed with the 

358: light line and 05b with the dark line. The most relevant transitions are also labelled. 

359: Middle panel: \rg\ data together

360: with the best fit. Lower panel: residuals from the best-fit model.} 

361: \end{figure*}

362: 

363: 

364: %-------------------------------------table 05

365: 

366: \begin{table}

367: \caption{\label{t:rgs_05}2005 best-fit model parameters for RGS and PN. 

368: Fluxes are in units of 10$^{-12}$ 

369: erg\,s$^{-1}$\,cm$^{-2}$, unabsorbed 

370: luminosity in units of 10$^{43}$ erg\,s$^{-1}$. Component 1: broken power law, 

371: 2: neutral absorber, 3,4: warm absorbers. All emission components are modified by Galactic photoelectric absorption.}

372: \begin{center}

373: \begin{tabular}{llll}

374: \hline

375: comp. & param. &  R1+R2 & PN\\

376: 

377: 1& $\Gamma_1$ & 2.5$\pm0.1$ & 2.58$\pm0.05$\\

378:  & $\Gamma_2$ & $\dots$ & 2.21$\pm$0.05\\

379:  & $E_{\rm break}$\,(keV) & $\dots$ & 2.0$\pm$0.1\\

380: 

381: 2 & $N_{\rm H}$ (10$^{20}$ cm$^{-2}$) & 1.7$\pm$1& 1.5$\pm$0.4\\

382: 

383: 

384: 3 & $N_{\rm H}$ (10$^{20}$ cm$^{-2}$) & 13$\pm3$ & 9$\pm1$\\ 

385: & $\log \xi$ & 0.05$\pm0.16$ & 0.16$\pm0.14$\\

386: & $v_{\rm out}$ (\kms) & --1700$\pm400$ & --1700 fix.\\ 

387: 

388: 

389: 4 & $N_{\rm H}$ (10$^{20}$ cm$^{-2}$) & 13$\pm5$ & 8$\pm3$\\ 

390: & $\log \xi$ & 2.6$\pm0.3$ & 2.0$\pm0.4$\\

391: \hline

392: & $\chi^2/\nu$ & 411/358 &530/513\\

393: 

394: &  F$_{2-10\,{\rm keV}}$& $\dots$ &  4.86$\pm0.06$ \\

395: &  F$_{0.5-2\,{\rm keV}}$ &  5.0$\pm$0.3  & 5.13$\pm0.07$ \\

396: &  L$_{2-10\,{\rm keV}}$ &  $\dots$ & 4.50$\pm0.06$\\

397: &  L$_{0.5-2\,{\rm keV}}$&  7.0$\pm0.4$ & 6.10$\pm0.08$\\

398: \hline

399: \end{tabular}

400: \end{center}

401: \end{table}

402: 

403: 

404: 

405: 

406: \subsection{Other continuum models}\label{par:conti}

407: 

408: In the present paper the continuum is represented by a simple broken power law.

409: A discussion of the continuum models and

410: their physical implications is given in Paper~I. Here we test different continuum

411: models only 

412: in order to assess the robustness

413: of the warm-absorber parameters. 

414: We follow the same approach as in Sect.~\ref{par:data}, the broad-band continuum is established 

415: using \epi\ and then applied to the \rg, in order to tune the warm absorbers' parameters.\\

416: First we used a power law plus modified black-body emission and neutral absorption 

417: (both from the host galaxy and from the

418: Milky Way). In the black-body model the effects of Compton

419: scattering are considered \citep{kaastrabarr}. 

420: A black-body temperature of $\sim$0.13\,keV 

421: is required to fit the low energies. Without any warm absorber the resulting reduced $\chi^2$ ($\chi^2/\nu$) is $\sim$1.32. 

422: The addition of one warm absorber (two additional free parameters) improves the fit by $\Delta\chi^2$=93. 

423: A second component improves the fit by another $\Delta\chi^2/\Delta\nu=23/2$. 

424: The final model has a higher black body temperature, $kT\sim$0.2\,keV (see Paper~I for a discussion). 

425: The total black-body

426: luminosity is however very low ($\sim 2.7\times10^{42}$\,erg\,s$^{-1}$) compared to the dominant 

427: power law component ($\sim1.3\times10^{43}$\,erg\,s$^{-1}$) in the 0.3--10\,keV band. 

428: The warm absorbers parameters do not differ within the errors from the ones listed in Tab.~\ref{t:rgs_05}.\\  

429: Next, we tested a relativistically blurred reflection model \citep{ross05}. 

430: We fixed the parameters to the model values determined

431: in Paper~I, i.e. the disc inclination to 51$^{\circ}$, its emissivity

432: $\alpha=2.5$ 

433: and inner radius to 5

434: gravitational radii. We assumed solar abundances for the disc material. 

435: A fit without any ionised absorption results in a $\chi^2/\nu$ of $1.56$. This value decreases to $1.21$, 

436: adding one warm absorber. A second warm absorber component again improves the fit, as in the previous cases ($\Delta\chi^2/\Delta\nu=20/2$). 

437: Note that for this exercise we do not attempt any

438: fit of the Fe K$\alpha$ line. The amount of reflected light with respect to

439: the total luminosity budget is modest (few percent), resulting in a weak soft excess. 

440: We conclude that for the purpose of our study a simple broken-power law fit is acceptable.

441: %---------------------------------------soft xmm 02

442: \begin{figure}

443: \resizebox{\hsize}{!}{\includegraphics[angle=90]{f3.ps}}

444: \caption{\epi\ spectra taken in 02 (light line) and 05 (dark line). The spectrum of 05 is multiplied by 1.7 to best match

445: the hard continuum during 2002 and emphasises the low energy complex variability.}

446: \label{f:ratio0205}

447: \end{figure}

448: \subsection{The recent history of the absorbed X-ray spectrum}\label{par:xmm02}

449: \zw\ was observed by \xmm\ for the first time on June 22, 2002 (rev. 0464) for 20\,ks \citep{luigi04}. At the time the source 2--10\,keV flux was about 1.7 times higher than in the present observation and showed remarkable 

450: spectral complexity at 

451: low energies. In Fig.~\ref{f:ratio0205} we over-plot the two \epi\ data sets, but with the 2005 spectrum shifted up in order to best match the continuum during 2002. A qualitative inspection of the figure shows that the soft energies

452: underwent the most dramatic changes between the two epochs (we refer to the 2005 and 2002 data sets as 

453: 05 and 02, respectively).

454: We reprocessed the 02 \rg\ and \epi\ data using SAS~7.0 and we focused the analysis mostly on the warm absorber. 

455: The \rg\ data were rebinned to reach a S/N ratio $\sim9$ over most of the band. 

456: The best fit shows that the underlying continuum changed between 02 and 05 mostly in flux rather than shape,

457: which is here again described by a simple broken power law (Tab.~\ref{t:rgs_02}). 

458: The shape of the ionising continuum was again constructed using the unabsorbed \epi\ continuum and the fluxes recorded

459: in optical/UV by OM (dashed line in Fig.~\ref{f:sed}). In 02 only the OM-UVW2 filter was used.   

460: Similarly to the 05 data, 

461: in addition to the neutral gas, we find evidence for two

462: ionised absorption components. The corresponding 

463: ionisation parameters are $\log\xi\sim-0.9$ and $\log\xi\sim1.6$. 

464: We identify these two warm absorber components as 02a and 02b, respectively (Fig.~\ref{f:rgs_tra02}). 

465: Interestingly, we find evidence that the 02a component outflows with a velocity similar to the 05a component ($v_{\rm

466: out}\sim1800$\,\kms). The width of the lines is $\sigma=67\pm50$\,\kms.

467: \begin{figure*}

468: \begin{center}

469: \resizebox{\hsize}{!}{\includegraphics[angle=90]{f4.ps}}

470: \end{center}

471: \caption{\label{f:rgs_tra02} Upper panel: 

472: transmission models (convolved with the data resolution)

473: of the neutral absorber (dashed line) and the ionised components for the 02 spectrum. 

474: Component 02a is displayed with the 

475: light line and 02b with the dark line. The most relevant transitions are also labelled. 

476: Middle panel: \rg\ data together with the best fit. Lower panel: residuals from the best fit. 

477: } 

478: \end{figure*}

479: %----------------------------------the ASCA data

480: Going further back into the recent history of the X-ray absorbers of \zw, we reanalysed 

481: the ASCA data obtained in 1995 \citep{leighly99}. The continuum is once again well fitted by a

482: broken power law with parameters comparable to the most recent 05 \xmm\ spectra: $\Gamma_{\rm soft}=2.58\pm0.07$,

483: $\Gamma_{\rm hard}=2.35\pm0.07$ and a break energy at $2.0\pm0.8$\,keV. 

484: We confirm the presence of a neutral gas phase with column density $N_{\rm H}=4\pm1\times10^{20}$\,cm$^{-2}$,

485: consistent with the values found with \xmm.\\

486: For the ASCA data, simultaneous UV observations were unavailable; therefore a true SED could not be constructed. 

487: The ionising SED was then built using the ASCA continuum plus the same $\alpha_{\rm ox}$

488: \footnote{

489: $\alpha_{\rm ox}$ is the spectral index between 2500\,\AA\ and 2\,keV \citep{tan79}.} $=-1.25$ found in 05. We used this SED only to 

490: obtain a best fit that consistently accounted for the absorption features and then only considered 

491: the ionic column densities (and not the ionisation parameter) to compare with our \xmm\ results.\\

492: As the ionisation parameter is dependent on the ionisation balance (i.e. the SED), a good way to compare warm absorbers

493: produced by different (or incomplete) SEDs, is to evaluate the different elemental ionic column densities. 

494: In particular, iron is the most

495: sensitive to changes in the ionisation of the gas \citep[e.g.][]{behar01}. 

496: In Fig.~\ref{f:fe}, we plot the ionic column densities as a function of

497: the ionisation stage of iron for each of the X-ray ionisation components detected in this source. 

498: The column densities are obtained from the XABS best-fit models. 

499: The relative errors are simply the scaled errors from the total hydrogen column densities

500: of the XABS ionisation components.

501: The 02 components are

502: well separated: 02a peaks at \fev, while 02b peaks at \fexiv-\fexvi. The 05a component has a relatively lower 

503: total column density (Tab.~\ref{t:rgs_05}) and shares some ions with the 02a components. The 05b component displays the highest ionised iron transitions. The ASCA component

504: (labelled 95) is more in the high ionisation region.\\       

505: A possible second component in the ASCA spectrum would be difficult to detect. We verified that 

506: a higher ionisation component, with a column

507: density of the order of $10^{21}$\,cm$^{-2}$ would produce mostly lines (see Fig.~\ref{f:rgs_tra05}, \ref{f:rgs_tra02}), which

508: would be undetectable in the ASCA spectrum. A much less ionised gas peaking, say at \feiii-\fev, would be easily confused

509: with the neutral absorber. 

510: \begin{figure}

511: \resizebox{\hsize}{!}{\includegraphics[angle=90]{f5.ps}}

512: \caption{Ionic column density as a function of the ionisation level of iron in the warm absorber components

513: detected in recent years by \xmm\ (02a, 02b, 05a and 05b) and ASCA (95).}

514: \label{f:fe}

515: \end{figure}

516: 

517: 

518: \begin{table}

519: \caption{\label{t:rgs_02}2002 best-fit model parameters for 

520: RGS and PN. Fluxes are in units of 10$^{-12}$ erg\,s$^{-1}$\,cm$^{-2}$, unabsorbed 

521: luminosity in

522: units of 10$^{43}$ erg\,s$^{-1}$. Component 1: broken power law, 2:neutral absorber, 3,4: warm absorbers. 

523: All emission components are modified by Galactic photoelectric absorption.}

524: \begin{center}

525: \begin{tabular}{llll}

526: \hline

527: comp. & param. &  R1+R2 & PN\\

528: 

529: 1& $\Gamma_1$ 				& 2.8$\pm$0.1 	& 2.71$\pm$0.07\\

530:  & $\Gamma_2$ 				& $\dots$ 		& 2.32$\pm$0.07\\

531:  & $E_{\rm break}$\,(keV)			& $\dots$ 		& 1.9$\pm$0.2\\

532: 2 & $N_{\rm H}$ 			&3.0$\pm$1.6		& 4.2$\pm$0.6\\

533: 

534: 

535: 3 & $N_{\rm H}$ (10$^{20}$ cm$^{-2}$) 	& 24$\pm$5		& 15$\pm$3\\ 

536: & $\log \xi$ 				& --0.9$\pm$0.2	& --0.8$\pm$0.3\\

537: & $v_{\rm out}$ (\kms) & --1800$\pm400$ & --1800 fix.\\ 

538: 

539: 

540: 4 & $N_{\rm H}$ 			& 13$\pm$4 		& 9$\pm3$\\

541: & $\log \xi$ 				& 1.6$\pm$0.2 		& 1.3$\pm$0.3\\

542: 

543: \hline

544: & $\chi^2/\nu$ & 231/211 & 806/788\\

545: 

546: &  F$_{2-10\,{\rm keV}}$&$\dots$ & 7.9$\pm0.26$\\

547: &  F$_{0.5-2\,{\rm keV}}$ & 8.46$\pm1.34$&  8.1$\pm0.5$  \\

548: &  L$_{2-10\,{\rm keV}}$ &$\dots$ &7.47$\pm0.25$ \\

549: &  L$_{0.5-2\,{\rm keV}}$& 1.49$\pm0.22$& 1.26$\pm0.36$ \\

550: \hline

551: \end{tabular}

552: \end{center}

553: \end{table}

554: 

555: 

556: 

557: 

558: 

559: 

560: 

561: \section{Discussion}

562: 

563: \subsection{The X-ray warm absorber}

564: 

565: The 02 and 05 spectra clearly show absorption by ionised gas. 

566: There were two detected components in 02, with ionisation parameters 

567: $\log\xi$= --0.9 and 1.6. The 05 spectrum was also well fitted with two 

568: warm absorber components, showing absorption with higher ionisation ($\log\xi\sim0\ {\rm and}\ 2.6$). 

569: In Fig.~\ref{f:xit} we plot the gas temperature ($T$) against the ionisation pressure parameter $\Xi$ 

570: for the 02 and 05 SEDs (drawn with the solid and dashed lines, respectively). The ionisation pressure parameter is defined as

571: $\Xi\equiv L/4\pi r^2cp$, where $L$ is the ionising luminosity, $r$ the gas distance from the source, and $p$ the gas

572: pressure. The dark dots mark the position of the 02

573: warm absorbers on the stability curve relative to the SED at that epoch, while the light dots are 

574: the components detected in 05. Our curves show a wide interval of $\log\Xi$ in which 

575: the absorbers exist in a cold regime. This is where the ``cold" components 02a and 05a lie. 

576: The ascending part ($\log\Xi>0.5$) of the curve does not present the typical turnover S-shape 

577: \citep[e.g.][]{crenshaw99}. This is because of the relatively steep X-ray SED of \zw\ \citep[][]{krolik81,guilbert}. 

578: We note that both the two 02 and two 05 components lie far apart from each other 

579: and are not in pressure equilibrium. This behaviour has been observed in other Seyfert~1 galaxies \citep[e.g.,][]{steen05,elisa06}. This

580: suggests that conditions other than pressure equilibrium \citep{kk95} should be present to make the two absorbers co-exist, such as magnetic confinement, as already suggested for the broad line region clouds

581: \citep[e.g.][]{rees87}. The lower ionisation $a$ components show an outflow velocity of roughly 1700-1800\,\kms\ 

582: in 02 and 05. Therefore there are no strong velocity changes apparent over 

583: time, although the constraints on velocity changes are limited.

584: 

585: 

586: \begin{figure}

587: \resizebox{\hsize}{!}{\includegraphics[angle=90]{f6.ps}}

588: \caption{Stability curve for the 2002 (solid line) and 2005 (dashed line) observations. 

589: The dark circles represent the two ionised

590: components detected in 2002, while the light circles represents the 2005 components. 

591: None of the components share the same pressure

592: ionisation parameter.}

593: \label{f:xit}

594: \end{figure}

595: 

596: 

597: 

598: \begin{figure}

599: \resizebox{\hsize}{!}{\includegraphics[angle=90]{f7.ps}}

600: \caption{The log of the X-ray ionisation parameter, $\xi_{\rm X}$, against the 0.1--10\,keV intrinsic luminosity for the 02 (dark dots) and

601: 05 (light dots) components.}

602: \label{f:antic}

603: \end{figure}

604: 

605: \subsection{Long-term changes in the warm absorbers}

606: 

607:   

608: The X-ray ions are 

609: mostly influenced by the 0.1--10\,keV radiation, rather than by the UV photons \citep[e.g.][]{netzer96}. 

610: Therefore in the discussion below we shall use the X-ray ionisation parameter $\xi_{\rm X}$, obtained simply by scaling down the ionising luminosity to the 

611: 0.1--10\,keV band \citep[e.g.][]{george98}. 

612: This calculation can be done only for the \xmm\ data sets, for which the complete ionising SEDs were available. 

613: The values for $\xi_{\rm X}$ then become $-1.96\pm0.2$ and $0.43\pm0.22$ for the 02a and 02b 

614: components, respectively. The 05a and 05b values are 

615: $-1.16\pm0.16$ and $1.43\pm0.3$, respectively. In Fig.~\ref{f:antic}, $\xi_{\rm X}$ of both the 02 and 05 components

616: are plotted against the 0.1--10\,keV 

617: luminosity. Note that in general the ionisation parameter value 

618: is SED dependent. For the 02a and 05a components the corresponding 

619: stability curves \citep[which are mostly dependent on the SED shape,][]{bottorff} 

620: are indistinguishable (Fig.~\ref{f:xit}). The stability curves start to differ slightly 

621: at higher values of $\Xi\propto\xi/T$, therefore we compare the 02b and 05b results with this caveat in mind. 

622: We note a significant anti-correlation between the 0.1$-$10\,keV luminosity and the X-ray ionisation parameter. 

623: For a twofold decrease in luminosity between 02 and 05, the low ionisation ($a$ components) $\xi_{\rm X}$ 

624: {\em increases} by a factor of $\sim6.2$. For the 02b and 05b components, the increase factor is $\sim10$. 

625: We may consider the $a$ and $b$ absorbers of 02 and 05 to be low- and high-ionisation components. 

626: If the 

627: observed ionisation comes from gas in equilibrium, we should invoke either a decrease in density 

628: (i.e. the gas thickness should increase, as $N_{\rm H}$ did not change dramatically) 

629: between 02 and 05, or movement of the gas {\it closer} to the central source. 

630: The latter seems an unlikely possibility as blue-shift of the 

631: $a$ components is observed in 02 and 05. If the gas consists of discrete outflowing narrow filaments or clouds, 

632: then we do not expect to see the same structure along our line of sight if the distance $\Delta r$ travelled by this gas is much larger than its characteristic size $d$. 

633: For a rest-frame time difference of $\sim$1057 days between both observations and the 

634: measured outflow velocity of 1700~km\,s$^{-1}$, $\Delta r \sim 1.5\times

635: 10^{16}$~cm.\\

636: For the following calculation we consider the case $\Delta r \gtrsim d$,

637: i.e. the size of the cloud is comparable to the gas travel path. The

638: absolute distance of the centre of the cloud $r$ from the source should be much

639: larger than the sizes considered here, therefore we also assume that the change in

640: $r$ is not a critical term. We then compute\\

641: 

642: \noindent

643: \begin{math}

644: \frac{\xi_{05}}{\xi_{02}}=\frac{L_{05}}{L_{02}}\times\left(\frac{N_{\rm H}}{d}\right)_{02}\ \times\left(\frac{d}{N_{\rm H}}\right)_{05},\\

645: \end{math}

646: 

647: \noindent

648: where $nd=N_{\rm H}$. The observed $\xi_{\rm X}$ ratio between 05 and 02 is 7 for the $a$ components. 

649: The

650: X-ray luminosity ratio is 0.5. With a simple calculation we obtain $d_{05}/d_{02}=7.6$.

651: Keeping in mind the assumption $\Delta r> d_{02}\sim 1.5\times 10^{16}$~cm, for the $a$ components we obtain 

652: $n_{02}<1.6\times10^{5}$\,cm$^{-3}$ and $n_{05}<1.1\times10^{4}$\,cm$^{-3}$. These are very low values 

653: compared to the estimated limits 

654: for the warm-absorber densities $2\times10^{6}-10^{9}$\,cm$^{-3}$ \citep{netzer02,kraemer02}, 

655: therefore it seems unlikely that we are observing the same portion of the gas in both 02 and 05.\\

656: A scenario that could 

657: reproduce the observed anti-correlation sees the gas constantly in a non-equilibrium state \citep[e.g.][]{nicastro99} 

658: because, for example, of a delay in the gas response. 

659: Such a delay may be due to the gas having a very low density \citep[$10^{7-8}$\,cm$^{-3}$, ][]{nicastro99}. 

660: Alternatively, the outflowing gas may be ``transient", in the form of clouds or filaments passing through the line of sight. 

661: A variety of distinct ionisation stages, seemingly unrelated to the ionising flux, could be then 

662: accounted for. This explanation is strengthened if, contrary to what is observed for \zw,   

663: components with very

664: different column densities were observed at different times \citep[e.g.][]{kraemer01,kraemer05,costantini00}.  

665: The outflow velocity may not change in time, if the gas is also moving 

666: transversely \citep{crenshaw_review,kraemer05}. 

667: Finally, the anti-correlation seen in Fig.~\ref{f:antic} may be the result of a shielding of the gas 

668: which is thus experiencing a filtered ionising luminosity. Qualitatively, the

669: (self-)shielding of the gas may happen away 

670: from our line of sight, and is not directly observable in a complex gas environment 

671: as depicted for example in \citet[][]{proga05,progak04} for the UV and X-ray emitting regions. 

672: 

673: \begin{figure}

674: \resizebox{\hsize}{!}{\includegraphics[angle=90]{f8.ps}}

675: \caption{Column density of hydrogen (\hi+\hii), \nv\ and \civ\ predicted by the X-ray absorbers at different epochs. The dashed

676: vertical lines show the range of $N_{\rm H}$ inferred from the HST measurements taken in 1994 \citep[][]{laor97}. }

677: \label{f:uv}

678: \end{figure}

679: 

680: 

681: 

682: \begin{figure}

683: \resizebox{\hsize}{!}{\includegraphics[angle=90]{f9.ps}}

684: \caption{\xmm\ velocity spectra for the 02a (upper panel) and 05a (lower panel) components. The bin size is 200\,\kms. 

685: The transitions chosen to produce the stack of the (unblended) line profiles for each absorber are also labelled. 

686: The vertical dashed line marks the blue-shift as measured by HST-FOS. Note that

687: the redshift used in the HST analysis was 0.0608 which implies a shift in the

688: outflow velocity of only 90\,\kms, which is negligible compared to the errors

689: associated to the X-ray measurements.}

690: \label{f:vel}

691: \end{figure}

692: 

693: \subsection{The connection with the UV absorber}

694: Some of the gas components detected in the X-rays might also absorb in the UV band. 

695: In the HST spectrum of \zw, obtained in

696: 1994 \citep{laor97}, the presence of \nv\ and \civ\ absorption features with an outflow velocity of $-1870$\,\kms, 

697: clearly highlighted the presence of a warm absorber. Unfortunately the column densities of the absorption troughs 

698: were poorly determined in the HST spectrum: $\log N_{\rm CIV}=14.07-18.14$ and 

699: $\log N_{\rm NV}=14.36-18.51$. The derived total hydrogen column density was $\log N_{\rm H}=17-20.77$.

700: The UV measurements (displayed with the dashed vertical lines) are compared in Fig.~\ref{f:uv} 

701: with the values obtained in the

702: X rays at different epochs.

703: The ionic column densities of \nv\ and \civ, which are visible only in the UV band because of their low ionisation, 

704: were derived from the best-fit models of the X-ray absorbers. 

705: The $a$ X-ray components have values of $N_{\rm NV}$ and $N_{\rm CIV}$ which are compatible with the 

706: UV column densities. Also the ASCA column density measurements are consistent with the UV values, while 

707: the $b$ components clearly do not produce enough of these low-ionisation ions.

708: Interestingly, the UV

709: flux in 1994 was about four times higher than in 02 and 05. This at least suggests

710: that a low ionisation outflowing component may survive years, despite large changes in the UV flux (and presumably of 

711: the overall level of the SED). 

712: The UV total hydrogen column density is slightly lower than for the X-ray

713: absorbers (Fig.~\ref{f:uv}). 

714: The outflow velocity measured from the X-ray spectra is consistent with the value found in the HST-FOS spectrum. This is shown in Fig.~\ref{f:vel} where 

715: the velocity spectra for the 02a and 05a components are compared \citep[see e.g.][for details on the procedure]{kaspi02}. The vertical dashed line marks the blue-shift as measured from the UV \nv\ line \citep[--1870\,\kms;][]{laor97}.  

716: The outflow velocity of the UV-X-ray gas did not change in time, within uncertainties. 

717: This suggests that we are observing a gas not dramatically influenced by changes of the central source (e.g. changes in the acceleration of the flow, or strong variability of the source). 

718: Therefore the observed gas may be in the terminal-velocity portion of the flow, at a considerable 

719: distance from the central source, for example at a parsec scale, produced either by 

720: evaporation in a molecular torus \citep{kk01}, or by instabilities of 

721: the outer rim of the accretion disc, which in principle can extend to large distances.

722: As a note of caution, 

723: the comparison between the X-ray and the UV absorbers discussed here is based on 

724: non-simultaneous data, which may be

725: misleading \citep{cren03}. In addition,

726: the UV information given by HST gives only partial information on the SED, which would be important for an

727: unambiguous comparison between the UV and X-ray bands. 

728: 

729:  

730: 

731: 

732: \section{Conclusions}

733: We have presented the \xmm\ low-energy X-ray data of \zw, collected in 2005. In the absorbed spectrum  

734: we find evidence of a neutral phase, stable at least

735: over the last ten years and probably associated with the host galaxy or with the interaction stream to the companion galaxy of

736: \zw\ (see also Paper~I). In addition, two ionised absorption 

737: components with similar column densities of $\sim few\times10^{21}$\,cm$^{-2}$ and ionisation parameters

738: $\log\xi\sim0\

739: {\rm and}\ 2.5$ are detected. Comparing these results with a previous \xmm\ observation and archival

740: ASCA and HST data we conclude:

741: \begin{itemize}

742: \item The two ionised gas components detected in both 05 and 02 cannot be in pressure equilibrium with each 

743: other. For the two

744: components to coexist, a different mechanism, like magnetic confinement and/or gas shielding has to play a role.

745: 

746: \item The X-ray ionisation parameter $\xi_{\rm X}$ is clearly anti-correlated with the X-ray luminosity. 

747: A viable

748: explanation for this considers a gas 

749: organised in filaments crossing the line of sight. However the similarity

750: of the column densities of the $a$, $b$ components and also of the gas detected in the ASCA spectrum, make this

751: possibility not as clear as for other cases \citep[e.g. NGC~4151,][]{kraemer05}. Other possibilities, like

752: a non-equilibrium state of the absorber, would need continuous monitoring to be rigorously tested. Finally, we may think that 

753: the observed gas might be shielded from the continuum source in the context of a complex and turbulent 

754: gas environment and experience a filtered ionising flux.

755: 

756: \item The gas producing the X-ray lower ionisation absorbers detected in 02 and 05 can also produce the

757: low-ionisation species (\nv\ and \civ) seen in the HST spectrum \citep{laor97}, as the column densities 

758: and outflow velocities are consistent. Interestingly however, the UV absorber was observed when the UV flux was

759: about 4 times higher than the ones recorded by \xmm-OM. This shows that  

760: a low ionisation component can survive for at least ten years, despite significant variations in the UV flux. 

761: The constant outflow velocity of the UV and the X-ray absorbers in time suggests also that we are observing 

762: the terminal velocity portion of the gas, probably located at a large distance from the source. The caveat is that 

763: the X-ray-UV comparison is performed with non-simultaneous

764: observations, which might be misleading.

765: \end{itemize} 

766: 

767: 

768: 

769: 

770: \section{Acknowledgements}

771: We thank Jelle Kaastra for useful discussion. 

772: The Space Research Organisation of the Netherlands is supported

773: financially by NWO, the Netherlands Organisation for Scientific

774: Research. WNB acknowledges support from NASA LTSA grant NAG5-13035 and NASA grant NNG05GR05G. 

775: This research has made use of the Tartarus (Version 3.1) database, 

776: created by Paul O'Neill and Kirpal Nandra at Imperial College London, 

777: and Jane Turner at NASA/GSFC. Tartarus is supported by funding from PPARC and NASA 

778: grants NAG5-7385 and NAG5-7067.

779: 

780: 

781: \begin{thebibliography}{}

782: \bibitem[\protect\citeauthoryear{Anders Grevesse}{1989}]{anders} Anders E., 

783: Grevesse N.\ 1989, \gca, 53, 197 

784: \bibitem[\protect\citeauthoryear{Behar et al.}{2001}]{behar01} Behar E., Sako M., 

785: Kahn S.~M.\ 2001, \apj, 563, 497 

786: \bibitem[\protect\citeauthoryear{Boller et al.}{1996}]{boller96} Boller T., Brandt W.~N., Fink H.\ 1996, \aap, 305, 53

787: \bibitem[\protect\citeauthoryear{Bottorff et al.}{2000}]{bottorff} Bottorff M.~C., 

788: Korista K.~T., Shlosman I.\ 2000, \apj, 537, 134 

789: \bibitem[\protect\citeauthoryear{Condon et al.}{1985}]{condon} Condon J.~J., 

790: Hutchings J.~B., Gower A.~C.\ 1985, \aj, 90, 1642 

791: \bibitem[\protect\citeauthoryear{Costantini et al.}{2000}]{costantini00} Costantini E., et 

792: al.\ 2000, \apj, 544, 283 

793: \bibitem[\protect\citeauthoryear{Costantini et al.}{2007}]{elisa06} Costantini E., et al.\ 2007, \aap, 461, 121

794: \bibitem[\protect\citeauthoryear{Crenshaw et al.}{1999}]{crenshaw99} Crenshaw D.~M.,

795: Kraemer S.~B., Boggess, A. Maran, S.~P., Mushotzky R.~F., Wu C.-C.\

796: 1999, \apj, 516, 750

797: \bibitem[\protect\citeauthoryear{Crenshaw et al.}{2003}]{crenshaw_review} Crenshaw D.~M., 

798: Kraemer S.~B., George I.~M.\ 2003, \araa, 41, 117  

799: \bibitem[\protect\citeauthoryear{Crenshaw et al.}{2003}]{cren03} Crenshaw, D.~M., et 

800: al.\ 2003, \apj, 594, 116 

801: \bibitem[\protect\citeauthoryear{Crummy et al.}{2006}]{crummy} Crummy J., Fabian A.~C., Gallo L., Ross R.~R.\ 2006, \mnras, 365, 1067 

802: \bibitem[\protect\citeauthoryear{den Herder et al.}{2001}]{2001A&A...365L...7D} den Herder J.~W., 

803: et al.\ 2001, \aap, 365, L7 

804: \bibitem[\protect\citeauthoryear{Elvis et al.}{1989}]{elvis89} Elvis M., Wilkes B.~J., Lockman F.~J.\ 1989, \aj, 97, 777 

805: \bibitem[\protect\citeauthoryear{Ferland et al.}{2002}]{ferl02} Ferland G.~J., Martin P.~G., van Hoof P.~A.~M., Weingartner J.~C.\ 2002, X-ray Spectroscopy 

806: of AGN with Chandra and XMM-Newton, 103 

807: \bibitem[\protect\citeauthoryear{Ferland}{2004}]{fer98} Ferland G.~J.\ 2004, Bulletin 

808: of the American Astronomical Society, 36, 1574 

809: \bibitem[\protect\citeauthoryear{Gallo et al.}{2004}]{luigi04} Gallo L.~C., Boller T., 

810: Brandt W.~N., Fabian A.~C., Vaughan S.\ 2004, \aap, 417, 29  

811: \bibitem[\protect\citeauthoryear{Gallo}{2007}]{} Gallo L.~C., Brandt W.~N., Costantini E., Fabian ~A.C., Iwasawa ~K. 

812: Papadakis ~I.E.\ 2007a, \mnras, accepted, astro-ph/0610283 (Paper~I)

813: \bibitem[\protect\citeauthoryear{Gallo}{2007}]{} Gallo L.~C., Brandt W.~N., Costantini~E. Fabian~A.C. \ 2007b, \mnras, 

814: submitted

815: \bibitem[\protect\citeauthoryear{George et al.}{1998}]{george98} George I.~M., Turner 

816: T.~J., Mushotzky R., Nandra K., Netzer H.\ 1998, \apj, 503, 174 

817: \bibitem[\protect\citeauthoryear{Guilbert et al.}{1983}]{guilbert} Guilbert P.~W., 

818: McCray R., Fabian A.~C.\ 1983, \apj, 266, 466 

819: \bibitem[\protect\citeauthoryear{Halpern Oke}{1987}]{halpern87} Halpern J.~P., Oke J.~B.\ 1987, \apj, 312, 91 

820: \bibitem[\protect\citeauthoryear{Kaastra Barr}{1989}]{kaastrabarr} Kaastra J.~S., 

821: Barr P.\ 1989, \aap, 226, 59 

822: \bibitem[\protect\citeauthoryear{Kaastra et al.}{2002}]{kaastra02} Kaastra J.~S. et al.\ 2002, \aap, 386, 427

823: \bibitem[\protect\citeauthoryear{Kaspi et al.}{2002}]{kaspi02} Kaspi S., et al.\ 2002, 

824: \apj, 574, 643 

825: \bibitem[\protect\citeauthoryear{Kraemer et al.}{2001}]{kraemer01} Kraemer S.~B., 

826:  Crenshaw D.~M., Gabel J.~R.\ 2001, \apj, 557, 30 

827: \bibitem[\protect\citeauthoryear{Kraemer et al.}{2002}]{kraemer02} Kraemer S.~B., 

828: Crenshaw D.~M., George I.~M., Netzer H., Turner T.~J., Gabel J.~R.\ 

829: 2002, \apj, 577, 98 

830: \bibitem[\protect\citeauthoryear{Kraemer et al.}{2005}]{kraemer05} Kraemer S.~B., et al.\ 

831: 2005, \apj, 633, 693 

832: \bibitem[\protect\citeauthoryear{Kriss}{2002}]{kriss02} Kriss G.~A.\ 2002, ASP 

833: Conf.~Ser.~255: Mass Outflow in Active Galactic Nuclei: New Perspectives, 

834: 255, 69 

835: \bibitem[\protect\citeauthoryear{Krolik et al.}{1981}]{krolik81} Krolik J.~H., McKee C.~F., Tarter C.~B.\ 1981, \apj, 249, 422 

836: \bibitem[\protect\citeauthoryear{Krolik Kriss}{1995}]{kk95} Krolik J.~H., \&

837: Kriss G.~A.\ 1995, \apj, 447, 512

838: \bibitem[\protect\citeauthoryear{Krolik Kriss}{2001}]{kk01} Krolik J.~H., 

839: Kriss G.~A.\ 2001, \apj, 561, 684 

840: \bibitem[\protect\citeauthoryear{Laor et al.}{1997}]{laor97} Laor A., Jannuzi B.~T., 

841: Green R.~F., Boroson T.~A.\ 1997, \apj, 489, 656 

842: \bibitem[\protect\citeauthoryear{Lawrence et al.}{1997}]{lawrence97} Lawrence A., Elvis 

843: M., Wilkes B.~J., McHardy I., Brandt N.\ 1997, \mnras, 285, 879 

844: \bibitem[\protect\citeauthoryear{Leighly}{1999}]{leighly99} Leighly K.~M.\ 1999, \apjs, 

845: 125, 317 

846: \bibitem[\protect\citeauthoryear{Mason et al.}{2001}]{mason} Mason K.~O., et al.\ 2001, \aap, 365, 36 

847: \bibitem[\protect\citeauthoryear{Mathur et al.}{1995}]{smita95} Mathur S., Elvis M., 

848: Wilkes B.\ 1995, \apj, 452, 230 

849: \bibitem[\protect\citeauthoryear{Netzer}{1996}]{netzer96} Netzer H.\ 1996, \apj, 473, 

850: 781 

851: \bibitem[\protect\citeauthoryear{Netzer et al.}{2002}]{netzer02} Netzer H., Chelouche 

852: D., George I.~M., Turner T.~J., Crenshaw D.~M., Kraemer S.~B., 

853: Nandra K.\ 2002, \apj, 571, 256 

854: \bibitem[\protect\citeauthoryear{Nicastro et al.}{1999}]{nicastro99} Nicastro F., Fiore 

855: F., Perola G.~C., Elvis M.\ 1999, \apj, 512, 184  

856: \bibitem[\protect\citeauthoryear{Otani et al.}{1996}]{otani96} Otani C., et al.\ 1996, 

857: \pasj, 48, 211 

858: \bibitem[\protect\citeauthoryear{Proga Kallman}{2004}]{progak04} Proga D., 

859: Kallman T.~R.\ 2004, \apj, 616, 688 

860: \bibitem[\protect\citeauthoryear{Proga}{2005}]{proga05} Proga D.\ 2005, \apjl, 630, L9 

861: \bibitem[\protect\citeauthoryear{Rees}{1987}]{rees87} Rees M.~J.\ 1987, \mnras, 228, 

862: 47P 

863: \bibitem[\protect\citeauthoryear{Reynolds}{1997}]{reynolds97} Reynolds C.~S.\ 1997, 

864: \mnras, 286, 513 

865: \bibitem[\protect\citeauthoryear{Ross Fabian}{2005}]{ross05} Ross R.~R., Fabian 

866: A.~C.\ 2005, \mnras, 358, 211 

867: \bibitem[\protect\citeauthoryear{Sargent}{1968}]{sargent68} Sargent W.~L.~W.\ 1968, 

868: \apjl, 152, L31 

869: \bibitem[\protect\citeauthoryear{Steenbrugge et al.}{2005}]{steen05} Steenbrugge K.~C.,

870: et al.\ 2005, \aap, 434, 569

871: \bibitem[\protect\citeauthoryear{Str{\" u}der et al.}{2001}]{2001A&A...365L..18S} Str{\" u}der L., et al.\ 2001, \aap, 365, L18

872: \bibitem[\protect\citeauthoryear{Tananbaum et al.}{1979}]{tan79} Tananbaum H., et 

873: al.\ 1979, \apjl, 234, L9 

874: \bibitem[\protect\citeauthoryear{Turner et al.}{2001}]{2001A&A...365L..27T} Turner M.~J.~L., et al.\ 2001, \aap, 365, L27 

875: \bibitem[\protect\citeauthoryear{Turner et al.}{2005}]{turner05} Turner T.~J., Kraemer

876: S.~B., George I.~M., Reeves J.~N., Bottorff M.~C.\ 2005, \apj, 618,

877: 155

878: \bibitem[\protect\citeauthoryear{Vestergaard Wilkes}{2001}]{vest_wilkes} Vestergaard M., 

879: Wilkes B.~J.\ 2001, \apjs, 134, 1 

880: 

881: 

882: \end{thebibliography} 

883: 

884: 

885:  \vfill\eject

886: \end{document}

887: 

888: 

889: 

890: