Table Of Content

ProceedingsoftheSymposium:“Japan-GermanyWorkshopon AGNandthe X-rayBackground” heldin Tokyo,Japan,November1-3,1999, eds.T.Takahashiand H.Inoue, ISAS Report,pp.0-0 Warm Absorbers in Active Galactic Nuclei Stefanie Komossa Max–Planck–Institut fu¨r extraterrestrische Physik, Postfach 1603, 85740 Garching, Germany 0 email: [email protected] 0 0 2 n Abstract. We first provide a review of the properties of 2251-178(Halpern 1984). With the improvedspectral ca- a warm absorbers concentrating on what we have learned pabilities of ROSAT and ASCA, many more warm ab- J from ROSAT and ASCA. This includes dusty and dust- sorbers have been found1, and theoretical modeling of 4 free warm absorbers, non-X-ray emission and absorption plasmaunder‘warmabsorber’conditionshasbeenpushed 1 features of warm absorbers, and the possible warm ab- forward2. 1 sorber interpretation of the peculiar 1.1 keV features. Below, we first provide a review on what is known on v We then discuss facets of warm absorbers by a more warm absorbers after the ‘ROSAT and ASCA epoch’ of 3 6 detailedinvestigationofindividualobjects:Inafirstpart, observations and theoretical modeling (Sect. 2). We then 2 we discuss several candidates for dusty warm absorbers. discussinmoredetailthe possibilitythatseveral(butnot 1 In a second part, we review and extend our earlier study all) warm absorbers contain dust (Sect. 3). In Sect. 4 we 0 ofa possible relationbetweenwarmabsorberandCLRin investigate the possible relation of WA and optical/UV 0 NGC4051, and confirm that both components are of dif- emission line regions with focus on NGC4051. Sect. 5 is 0 ferentorigin(theobservedcoronallinesareunderpredicted concerned with the warm absorber in MR2251-178. In / h by the models, the warm absorber is too highly ionized). Sect. 7 we explore whether and under which conditions a p We then suggest that a potential overprediction of these warm absorber provides an explanation for the dramatic - o lines in more lowly ionized absorbers can be avoided if X-ray spectral variability of RXJ0134-4258. r thesewarmabsorbersaredusty.Inathirdpart,wepresent t s first results of an analysis of a deep ROSAT PSPC ob- a servation of the quasar MR2251–178, the first one dis- : v covered to host a warm absorber. Finally, we summa- i X rizeourscrutinyunder whichconditionsawarmabsorber 1 e.g., MCG6-30-15 (Nandra & Pounds 1992, Fabian et al. r could account for the dramatic spectral variability of the 1994, Otani et al. 1996b, Reynolds et al. 1997, Orr et al. a Narrow-line Seyfert1 galaxy RXJ0134-4258. 1997, and references therein), NGC3783 (Turner et al. 1993, George et al. 1995, 1998c), 3C212 (Mathur 1994), 3C351 (Fiore et al. 1993, Mathur et al. 1994, Nicastro et al. 1999b), NGC3227 (Ptak et al. 1994, Komossa & Fink 1997b, George 1. Introduction et al. 1998b), NGC4051 (e.g., Pounds et al. 1994, Mihara et al. 1994, McHardy et al. 1995, Guainazzi et al. 1996, Ko- Warm absorbers (WAs) are an important new diagnostic mossa & Fink 1997a, Guainazzi et al. 1998, and references of the physical conditions within the central regions of therein), NGC 4151 (Weaver et al. 1994, Warwick et al. 1995, active galaxies. They have been observed in 50% of the 1996), IC 4329A (Cappi et al. 1996, Perola et al. 1999), NGC ∼ well-studied Seyfert galaxies and have also been detected 3786(Komossa&Fink1997c),NGC5548(Nandraetal.1993, in quite a number of Narrow-line Seyfert1 galaxies. So Mathur et al. 1995, 1999, Done et al. 1995), NGC3516 (Kriss far, they revealedtheir existence mainly in the soft X-ray etal.1996a,Mathuretal.1996),Mrk290(Turneretal.1996), spectralregionwheretheyshowupbytheircharacteristic Mrk1298(Wangetal.1996,1999,Komossa&Fink1997d,Ko- metalabsorptionedges.Thestudy ofthe ionizedmaterial mossa & Meerschweinchen 2000), IRAS13349+2438 (Brandt et al. 1996, 1997, Komossa et al. 1999b, Siebert et al. 1999), provides a wealth of information about the nature of the IRAS17020+4544(Leighlyetal.1997,Komossa&Bade1998), warm absorber itself, its relation to other components of PG1114+445 (George et al. 1997), 4C +74.26 (Brinkmann the active nucleus, and the intrinsic AGN X-ray spectral et al. 1998, Komossa & Fink 1998); see Reynolds (1997) and shape, and leads to a better understanding of AGN in Georgeetal.(1998a)forasystematicstudyofWAswithASCA general. 2 e.g., Halpern 1984, Krolik & Kallman 1984, Netzer 1993, The presence of an ionized absorber was first discov- 1996, Krolik & Kriss 1995, Murray et al. 1995, Reynolds & ered based on Einstein observations of the quasar MR Fabian 1995, Komossa & Bade 1998, Nicastro et al. 1999a 2 St. Komossa: Warm Absorbers in AGN 2. Basic properties of warm absorbers: dusty where d is the distance between observer and warm ab- WAs, dust-free WAs, peculiar 1.1keV features sorber. The mass of the warm absorber is approximately given by 2.1. Warm absorber models and definitions The active nucleus is thought to contain an accreting su- Mwa ≃4πr2Nwmpη =0.02r126N22η M⊙ , (4) permassive black hole (SMBH), surrounded by the pu- with r =1016r cm and N =1022N cm−2. 16 w 22 tative molecular torus and two emission line regions, the From direct spectral fits, one first determines U and broadlineregion(BLR)andthenarrowlineregion(NLR). N .Iftheintrinsiccontinuumturnsouttobevariable,or w The physical properties of the two emission line regions if there is evidence that the warmmaterialcontains dust, arequitewellknown,sinceawealthofinformationcanbe important constraints on the density and distance of the extracted from emission line intensities, line profiles, line absorberfrom the nucleus canbe obtainedfromreaction- shifts,andlinevariability(seePeterson1993forareview). timescalearguments,anddustsurvivalarguments,respec- Concerning the nature and location of the warm ab- tively. In particular, the recombinationtimescale t of a rec sorber,severaldifferent models havebeen suggested:(i) a warm absorber is given by relationoftheWAtotheBLR(ahigh-densitycomponent 1 n 1 T of the inner BLR, a BLR confining medium, winds from n i ( )X (5) e ≈ t n A 104 bloated stars, or a matter bounded BLR component; see, rec i+1 e.g., Reynolds & Fabian 1995, Shields et al. 1995, Netzer (Krolik&Kriss1995)wheren/n isthe ionabundance i i+1 1996), (ii) an accretion disk wind (e.g., Königl & Kartje ratio of the metal ions dominating the cooling of the gas, 1994,Murrayetal.1995),(iii)arelationtothetorus(e.g., n the electron density, and the last term is the corre- e Reynolds1997,Komossa&Fink1997b),or(iv)arelation sponding recombination rate coefficient α−1 (Shull & i+1,i to the NLR (see, e.g., the two-component WA model of VanSteenberg1982;forcoefficientsA,X seeAldrovandi& Otani et al. 1996b). The reason for the large variety of Pequignot1973).Availableobservationslocatemostwarm models discussed is that not all physical properties of the absorbers outside the bulk of the BLR. warmabsorber(itsdensityn,columndensityN ,covering w factor η = ω/4π, distance r from the nucleus, elemental 2.2. X-ray appearance of warm absorbers abundances Z, its velocity field, and the shape of the il- luminating continuum) can be directly determined from 2.2.1. Dust-free warm absorbers X-rayspectralfits,butonlycertaincombinationsofthese parameters, defined below. Sofar,ionizedabsorbershavebeenmainlyobservedatX- rayenergies;therefore,mostofourknowledgeonthisma- The two most important quantities are the column terialcomesfromtheanalysisofX-rayspectra.Warmab- density N and ionization parameter U of the warm ab- w sorbersshowupinthesoftX-raybandbytheircharacter- sorber. N is given by w istic X-ray absorption edges which are created by K-shell photoionization of highly ionized metal ions. The warm N = n dx , (1) w Z H wheren isthetotalhydrogendensity,andxthethickness H of the absorber.The ionization parameter U is defined as Q U = . (2) 4πr2n c H Here, Q relates to the spectral energy distribution (SED) which illuminates the warm absorber, and is the number rate of photons above the Lyman limit3 (ν = 1015.512 0 Hz), given by ∞ f Q=4πd2 νdν , (3) Z hν ν0 Fig.1. X-ray absorption structure of a typical warm absorber. The thin solid line corresponds to the intrinsic X- 3 Inrelation towarmabsorbers, someauthorsusean‘X-ray ray spectrum, the thick solid line shows the spectrum af- ionization parameter’, counting only X-ray photons, because terpassageofthe ionizedabsorber.The oxygenOVIIand theionization state of thewarm material is dominated by the soft X-ray,not theEUV-partof thespectrum. OVIII edges are marked. St. Komossa: Warm Absorbers in AGN 3 material is thought to be photoionized by emission from he took as reddening indicator. Except for two outlyers, the central continuum source in the active nucleus (for a most ofthe objects of his sample showlow reddening and detailed model invoking shocks (Meier & Demmig 1997) weak OVII absorption,whereas four objects show the op- in additionto photoionizationsee Contini & Viegas 1999; posite (high reddening, a larger optical depth of OVII) seealsothediscussioninNicastroetal.1999a).Itsdegree and these latter are suggested to likely host dusty warm of ionization is higher than that of the bulk of the BLR absorbers. No such trend is found for OVIII. and the gas temperature is typically of order 105 K. Fig. 1 shows the X-ray spectrum of a typical warm absorber. 2.2.3. Peculiar 1.1 keV features Themostprominentabsorptionedgesaremarked.Incase of high coveringfactor of the warmabsorber,X-rayemis- Warm absorbers have also been observed in several sion and reflection becomes significant and will partially Narrow-line Seyfert1 galaxies (NLSy1s hereafter; e.g., fill up the edges (e.g., Netzer 1993 (his Fig. 1-3), 1996). Vaughan et al. 1999, Komossa & Meerschweinchen 2000, andreferencestherein).Inadditiontothe‘standard’warm absorbers with strong edges of OVII and OVIII around 2.2.2. Dusty warm absorbers 0.7-0.8 keV, peculiar absorption features have been re- Warm absorbers have been mostly modeled using the ported around 1 keV (e.g., Brandt et al. 1994, Otani ∼ codesCloudy(Ferland1993,Ferlandetal.1998)andION etal.1996a,Ulrich-Demoulin&Molendi1996,Hayashida (Netzer 1993, 1996). For several years, the ionized ab- 1997, Leighly et al. 1997, Matt 1998, Ulrich et al. 1999, sorberswereassumedtobe dust-free;partlybecausethey Vaughan et al. 1999). were observed in bright, optically unreddened Seyfert1 Mostinterpretationsofthesefeaturefocussedonwarm galaxies, and there was no necessity at all to suspect this absorbers,albeitinverydifferentways.Thesuggestedex- highly ionized material to contain dust. planations are: (i) Non-solar metal abundances: Absorp- The situation changed with the cases of 3C 212 tion around1.1keVcouldbe due to Ne-Kabsorptionbut (Mathur 1994), IRAS13349+2438 (Brandt et al. 1996), would require a rather peculiar abundance ratio of Ne/O and NGC3227 (Komossa & Fink 1996, 1997a): 3C 212 forthe Ne absorptionto be muchstrongerthanthe O ab- is a quasar with a heavily reddened optical spectrum. Its sorption.Inaddition,the strongestNe edgeinmostmod- X-ray spectrum was successfully fit by several models in- els is located at 1.36 keV at somewhat higher energies cluding a powerlaw plus excess cold absorption (Elvis et al. 1994). Since other multi-wavelength observations did not indicate a coldabsorberof high columndensity along the line of sight, Mathur (1994) suggested that the dust that reddens the optical spectrum is mixed with a warm absorber, instead. She also showed that this material can consistentlyaccountfortheobservedMgIIUVabsorption of 3C 212. In the cases of IRAS 13349 and NGC3277 the largecolumndensityN ofdustygasinferredfromopti- opt cal observations is inconsistent with the one (N ) derived x fromX-rayspectralfitsinthesensethatN N (here- x opt ≪ after referred to as ‘N N discrepancy’). The incon- opt x − sistency disappears,if the dust is accompaniedby ionized gas. Further candidates for dusty warm absorbers which show the same N N discrepancy emerged quickly: opt x − NGC3786,MCG6-30-15,IRAS17020+4544,andpossibly 4C74.26 (see Tab. 1). Warm absorber models taking into account the pres- Fig.2. Influence of dust mixed with the warm absorber ence of dust were first presented by Komossa & Fink on the X-ray absorption structure (logU = 0.5,logNw − (1996,1997a)andapplied to the ROSAT X-ray spectrum = 21.6). The straight solid line corresponds to the unab- of NGC4051 (which turned out to be likely dust-free, see sorbed intrinsic spectrum, the heavy solid line to a dust- Sect. 4 below), followed by NGC3227, NGC3786 (which freewarmabsorber(WA)withsolarabundances,thedot- are well fit by dusty warm absorbers; Komossa & Fink ted line to a dust-free WA with the same depleted abun- 1997b,c), and MCG6-30-15 (which likely possess a two- dances as in the dusty model, andthe long-dashedline to component WA one of which is dusty; Reynolds et al. thechangeinabsorptionstructureafteraddingdusttothe 1997). WA (other model parameters are fixed). Clear changes in Inanalternativeapproach,Reynolds(1997)correlated the absorption structure are revealed for the models that the optical depths of OVII and OVIII of a sample of include dust. The abscissa brackets the ROSAT energy Seyferts with the ‘steepness’ of the UV continuum which range (0.1-2.4 keV); some prominent edges are marked. 4 St. Komossa: Warm Absorbers in AGN than observed; Komossa & Fink 1998, Ulrich et al. 1999. galaxies (Crenshaw et al. 1999)suggests a common or re- Alternatively, it could be caused by Fe-L absorption (ap- lated origin of UV and X-ray absorbing material. plied to PG1404+226this requires an Fe overabundances of > 25 solar; Ulrich et al. 1999). (ii) A warm absorber × 2.4. Non X-ray line emission from warm absorbers in relativistic outflow; the 1 keV feature is then caused ∼ by highly blueshifted OVII or OVIII edges (Leighly et al. Examples of UV-EUV lines that become strong under 1997; see also Brandt et al. 1994 and Otani et al. 1996a). warm absorber conditions were given in, e.g., Krolik & (iii) Asequence ofresonance absorption lines incombina- Kriss 1995 (their Fig. 4), Shields et al. 1995, Netzer 1996 tion with a steep X-ray spectrum (Nicastro et al. 1999c). (his Fig. 10), and Komossa & Fink 1997d(their Fig.6), It is important to keep in mind, though, that the ob- 1997e. served features seem to occur always close to the posi- Observationally, the EUV emission line NeVIIIλ774 tionwherethedecliningsoftexcessemissionintersectsthe was detected in the spectra of high-redshift quasars powerlawcomponent,andthepossibilityremainsthatthe (Hamann et al. 1995b,c,1998)and was suggested to orig- absorption feature is only mimicked by a more complex inate from a warm absorber. (X-ray spectra of these ob- spectralshape.Further,inthe caseofArk564the feature jects are not yet available for a check and more detailed isbetterdescribedbyanemissionlinethananabsorption modeling). complex (Vaughan et al. 1999, Turner et al. 1999). A more detailed approach then was to select individual, well observed AGN and to investigate whether the physical conditions in one of the observed optical-UV 2.3. Relation to UV absorbers emission line regions and the X-ray warm absorbers are The ‘X-rays-only’approachto study warmabsorbers suf- identical, such that the ionized X-ray absorber could be fersfromthe stilllimitedspectralresolutionofcurrentX- directlyidentifiedwithoneoftheknown(high-ionization) raydetectorsmakingmeasurementsof,e.g.,red/blueshifts emission line regions. Mathur et al. (1994) presented de- of the absorption edges difficult. A powerful approach to tailed modeling of the BELR of the quasar 3C351 which constrain further the properties of warm absorbers is to also shows an X-ray warm absorber, and concluded that combine the X-ray with UV observations, and to search high-ionization BELR and warm absorber of 3C351 are forthe signatureofthe warmabsorberssimultaneouslyin not one and the same component. Komossa & Fink (e.g., both spectral regions. 1997a,b)studied the casesofNGC4051(summarizedand The relation between UV and X-ray (cold) absorbers extended in Sect. 4 below) and NGC3227. They first de- was examined by Ulrich (1988). Mathur and collabora- rivedallwarmabsorberpropertiesthatcanbedetermined tors(e.g.,Mathur1994,Mathuretal.1994,Mathur etal. fromX-rayspectralfits,thenvariedtheremainingfreepa- 1995) then presented several good cases (3C 212, 3C351, rameters (like metal abundances), predicted for each case NGC5548) where UV and X-ray warm absorber are very the optical/UV line emission from the X-ray absorbing likely one and the same component (see also deKool & gas and compared with optical observations. Again, they Meurs 1994 on PG1416-129). These absorbers are out- concluded that warm absorber and CLR in NGC4051, flowing with velocities of 1000-3000km/s and are lo- andWAandemission-linecomponentsinNGC3227show ∼ cated outside the BLR. Three further good candidates different physical conditions. Reynolds et al. (1997) fol- for a common UV–X-ray warm absorber are the high- lowed a similar approach in application to MGC6-30-15 redshift quasar PKS 2351-154 (Schartel et al. 1997), the andfindthat[FeX]and[FeXIV] couldbe explainedbyan Sy1 galaxyNGC3783(Shields &Hamann 1997),andthe (outer) warm absorber of covering factor 0.08, but that ∼ quasar PG 1114+445 (Mathur et al. 1998). Hamann et this wouldheavilyunderpredictthe observed[FeXI]emis- al. (1995a) reported the detection of NeVIIIλ774 absorp- sion. Contini & Viegas (1999) presented a detailed multi- tion in the quasar UM675, and noted that this gas could cloud model in which both, photoionization and shocks act as X-ray warmabsorber.Telfer et al. (1998) analyzed contribute to the heating and ionization of the clouds, the very highly ionized BAL system of the quasar SBS and applied their model to NGC4051.They find that the 1542+541 and concluded that it is likely associated with observed X-ray spectrum and the NLR line emission can an X-ray warm absorber. be successfully reproduced by an ensemble of clouds of Sometimes, though, the UV absorptionis rather com- different densities. [Predicted emission line intensities for plex andshows the presence ofdifferent components with coronal lines have not yet been reported.] differentdegreesofionizationand/orvelocities(e.g.,NGC 3516,Krissetal.1996b,Crenshawetal.1998;NGC5548, 3. Dusty warm absorbers Mathuretal.1999;NGC7469,Krissetal.1999),andonly one component(e.g.,NGC5548)ornone (e.g,NGC7469) Below, we describe in more detail the properties of dusty may be identified with the X-ray warm absorber. warm material, and provide a summary of the sug- Inanycase,thenearone-to-onematchofthe presence gested candidates for dusty WAs including explicit spec- ofUV-andX-raywarmabsorbersinasampleofSeyfert1 tral model fits. St. Komossa: Warm Absorbers in AGN 5 Table 1. Summaryofthe candidatesfordustywarmabsorbers,listedinthechronologicalorderthey weresuggested. The column “sug.” gives those references that proposed the presence of a dusty warm absorber, whereas the column “mod.” lists the works that first modelled the X-ray data in terms of a dusty warm absorber. Values of the ionization parameter U (Eq. 2) givenhere and elsewhere in the text refer to a continuum spectrum with α = 1.4 (between EUV − Lyman-limit and 0.1 keV) and Γ as listed. x object type warm absorber fit references comments Γixntr log U log Nw sug. mod. 3C212 ‘red’ quasar [1] dusty WA not yet fit df –2.4 –0.5 21.9 [1] best-fit parameters of dust-free WA IRAS13349+2438 NL quasar –2.9 –0.4 21.21 [2], [3] [10], [11] df –2.2 0.7 22.7 dust-free WA for comparison NGC3227 Sy 1.5 –1.9 –0.3 21.8 [4], [5] [4] NGC3786 Sy 1.8 –1.9 –0.8 21.7 [6] [6] MCG6-30-15 Sy 1 –2.2 21.7 [7] [7] (>)2-comp. WA IRAS17020+4544 NLSy1 –2.8 0.7 21.61 [8] [12] sil –2.4 1.0 21.61 [12] silicate species only 4C+74.26 radio quasar –2.2 –0.1 21.6 [9] [9] excess cold absorption possible (1) fixedtothevalue Nopt determinedfromoptical reddening.References:[1] Mathur1994, [2,3]Brandt et al. 1996, 1997, [4] Komossa& Fink 1997b(seealsoKomossa&Fink1996),[5]Georgeetal.1998b,[6]Komossa&Fink1997c,[7]Reynoldsetal.1997(thepossibilityofadustyWA in this galaxy was earlier brieflymentionedin Reynolds1997), [8] Leighly et al. 1997, [9] Komossa & Fink 1998, [10] Komossa & Greiner 1999, Komossaetal.1999b,[11]Siebertetal.1999,[12]Komossa&Bade1998. 3.1. Modifications of the X-ray absorption structure al. 1993) is only of the order of a few eV (Greaves et al. 1984).] Several changes occur in the presence of dust mixed with – There is a stronger temperature gradient across the thewarmabsorber:(i)Gas-dustinteractionsinfluencethe absorber, with more gas in a colder state. This and thermal conditions in the gas and change its ionization the previous effects lead to an effective flattening of state(e.g.,Draine&Salpeter1979).Inparticular,forhigh the X-ray spectrum (Fig. 2) ionization parameters dust effectively competes with the – The increasedsensitivity of dust to radiationpressure gas in absorbing photons (Laor & Draine 1993). (ii) Dust (e.g., Chang et al. 1987, Binette et al. 1997) likely scatters and absorbsthe incident radiationand X-rayab- causes the gas to outflow (if not otherwise confined) sorption edges are created by inner-shell photoionization leading to temporal changes of the absorber proper- of metals bound in the dust (see Figs 4,5 of Martin & ties. [Indeed, strong variability in X-ray count rate of Rouleau 1991). factors 10-15has been detected from the dusty-warm- Underthefollowingassumptionswe(Komossa&Fink absorber candidates NGC 3227 and NGC 3786 (Ko- 1997b,Komossa&Bade1998)havecalculatedasequence mossa & Fink 1997b,c.)] of dusty warm absorber models using Ferland’s (1993) – The warm absorber emission in certain optical/UV code Cloudy: (i) The dust composition (graphite and as- lines, in particular the iron coronal lines, is much re- tronomicalsilicate)andgrainsizedistributionwerechosen ducedbecauseofthedepletionofironintodustgrains. like in the Galactic diffuse interstellar medium (Mathis [This likely also alleviates the problem of the recently et al. 1977) if not mentioned otherwise, as incorporated reported potential over-prediction of these lines from in Cloudy, (ii) the metal abundances were depleted corre- some warm absorbers; see Sect. 4 below]. spondingly(ameanofCowie&Songaila1986;seeFerland 1993, Baldwin et al. 1991 for details), (iii) the warm material was assumed to be of constant density and ionized 3.2. Results from X-ray spectral fits: List of by a ‘mean Seyfert’ IR to gamma-ray continuum (with suggested/modeled candidates αuv−x=–1.4 in the EUV). Themainconsequencesofthepresenceofdustinternal We have investigated most suggested dusty-WA candi- to the warm absorber can be summarized as follows: dates with our models. We get successful X-ray spectral fits.Insomecasesthebest-fitparameterscandeviatesub- – Absorption edges in the X-ray spectrum from neutral stantially fromthose obtained by fitting a dust-free warm metals bound in the dust, like a carbonCI edge (from absorber, though. Table 1 gives a summary of the sug- the graphite species) and an oxygen OI edge (from gested dusty warm absorber candidates and the model silicate) are predicted. [The shift in edge energy due results. The ionization parameters are found to be some- to solid state effects (e.g., Martin 1970, Adamietz et what lower than those of dust-free warm absorbers, with 6 St. Komossa: Warm Absorbers in AGN the exceptionofIRAS17020.Itis interestingto note that 3.4. Model uncertainties the dusty warm absorbers show a rather narrow range in The line of arguments in favor of dusty warm absorbers columndensitiesN whichdifferbylessthanafactor 2. w ∼ (reviewed in Sect. 2.2.2; see Brandt et al. 1996 and Ko- Thisislikelypartlycausedbyselectioneffects:Firstly,X- mossa & Fink 1997b for many further details) relies on ray warm absorbers of lower column density are difficult severalassumption.Firstly,opticalandX-rayobservations to detect presently, and small deviations of the Balmer were not performed simultaneously. Variability in the op- decrement from the recombination value are less straight tical reddening on the timescale of years is possible in forwardly interpreted in terms of reddening. Secondly, in Seyferts (e.g., Goodrich 1989,1995),but we note that for case of higher column density the dusty gas becomes op- both, IRAS17020 and IRAS13349 optical spectra taken tically thick, and the concept of the standard extinction at many different epochs are available and they do not curve can no longer be applied. show evidence for changes in reddening (e.g., Siebert et al. 1999). Secondly, optical-UV and X-ray photons may not travel along the same l.o.s, and they may therefore 3.3. Origin of the dusty material ‘see’differentamountsofdustandgas.This wouldbe the case if the X-rays were seen mainly in scattered light, or In case dust is mixed with the warm absorber, impor- if there was a strong starburst contribution. Since most tant constraints on its density and locations can be ob- of the Seyferts with dusty warm absorber candidates are tainedfromtherequirementofdustsurvival:Firstly,dust rapidlyvariableinX-rays(e.g.,afactorof 2within800s isheatedanddestroyedbytheradiationfieldofthecentral ∼ and a factor 10 within a few years in case of NGC3227; source if at distances less than rev √L46 (e.g., Netzer ∼ ≈ Komossa& Fink 1997b)these twopossibilities canbe ex- 1990;see Laor & Draine 1993for a dependence ofthis re- cluded. lationongrainsize).Secondly,dustisheatedbycollisions As far as the modeling of dusty warm absorbers is with gas particles (e.g., Baldwin et al. 1991)to a temper- concerned, one uncertainty is the applicability of the ature thatexceeds the evaporationtemperature ifthe gas codeCloudy. Moreprecisely,radiationpressuremaydrive density is high enough. Both approaches locate the dusty strong outflows of the dusty gas. In case of large velocity warm absorbers outside the BLR. gradients,thetreatmentofoverlappingresonancelinesbe- One component that suggests itself to be identified comes inaccurate (Ferland 1993),and the heating-cooling with the dusty warm absorber is an atmosphere of the balance might be disturbed by expansional cooling of the molecular torus which plays an important role in unifi- gas. cation schemes. The torus has the advantage that it is Another uncertainty are the dust properties: Is the ‘known’ to be present (no need to invent a new compo- dust in Seyfert galaxies sufficiently similar to Galactic nent),andwouldprovideanaturalreservoirfordustclose ISM-like dust (and is the latter really ‘MRN’-like in na- to the central power source (but far enough away to en- ture4) ? Firstly, galactic ISM-like dust was also assumed suredustsurvival).Oneisthenleadtoaviewinggeometry to estimate the amount of dust from the optical spec- such that our line of sight grazes the torus atmosphere in tra. Therefore, making the same assumption in modeling those Seyferts that show dusty warm absorbers (see also the X-ray data is a good starting point. Secondly, obser- the discussion in Brandt et al. 1996, Reynolds 1997, and vations by ISO suggest that the dust/gas ratio in other Komossa & Fink 1997b,c). galaxies does not deviate much from the Galactic value (e.g., Bianchi et al. 1999). Thirdly, even for other dust LookingatTab.1,onequestionarisesimmediately:the compositionsorstructures,theX-rayabsorptionstructure dustywarmabsorbersseemtocomeinall typesofSeyfert wouldnotchangestrongly,becausethe X-rayspectrumis galaxies, which is not what would be expected in a sim- notdominatedbycomplexmoleculartransitionsandsolid ple viewing-geometry-scenario. The sample is still small, state andchemicaleffects (these mayhavesome influence though, and it might be interesting to re-investigate this on the heating-cooling balance, though), but by K-shell possibility once the dusty warm absorbers are confirmed, photoionization of metals bound in the dust. and further ones are known. The other already earlier suggested dusty component 3.5. Future perspectives of the active nucleus, and thus potential dusty warm ab- sorbercounterpart,isthe dusty transitionregionbetween Given the presence of X-ray absorption edges of neutral BLR and NLR ‘invented’ and studied by Netzer & Laor dust-bound metals like the carbon edge, dusty warm ab- (1993). The ionization parameter of this region should sorbers,if confirmed5 with XMM, AXAF and ASTRO-E, show a smooth transition to that of the BLR and NLR, though; whereas the ionization parameters of warm ab- 4 see, e.g., Laor & Draine 1993, and the recent review by sorbers are typically much higher. Witt 1999 5 EvenifthedustyX-raywarmabsorbersarenotconfirmed, thislikelytellsusalotaboutthedustinthesegalaxies, which St. Komossa: Warm Absorbers in AGN 7 will play an important role not only in probing compo- ability arguments we derive a limit on the density of the nents of the active nucleus, like the dusty torus, but also warm absorber, n < 3 107cm−3, which translates into H are they a very useful diagnostic of the (otherwise diffi- a distance from the∼nucl×eus of r >3 1016 cm. cult to determine) dust properties and dust composition ∼ × in other galaxies. 4.3. Prediction of coronal lines We then checked, for this best-fit warm absorber model, 4. The warm absorber in NGC4051, and the the strengths ofthe optical/UVlines originatingfromthe (missing) connection between WA and CLR ionized material; in particular, the Fe lines [FeVII]λ6087, 4.1. Introduction [FeX]λ6374,[FeXI]λ7892,and [FeXIV]λ5303 (see Tab. 2). We find that all of them are much weaker than observed! Given the several good candidates for dusty warm ab- We then variedthose parametersthat notstronglyin- sorbers the question emerges whether all warm absorbers fluence the X-rayabsorptionstructure,but couldhavean contain dust. We think this is unlikely because (i) warm effectonthepredictedFelinestrengths;namely:themetal absorbers have also been observed in several Seyfert1 abundances, the gas density, the EUV continuum shape, galaxies that do not show evidence for optical reddening, the IR continuum strength. We find that in all cases, the and (ii) in some cases the modifications in the X-ray ab- lines [FeVII]–[FeXI] remain much weaker than observed sorptionspectrumduetothepresenceofdustaresostrong byseveralordersofmagnitude.Thereasonforthisisthat that no successful X-ray spectral fit is possible. the warmabsorberis alwaystoo highly ionized, with a to- Dust-free warm absorbers might contribute signifi- tally negligible amountofFe9+ andFe10+ ions inthe gas. cantly to the emission in optical/UV iron coronal lines. Therefore,changesincollisionalstrengthsforthe relevant In the following we shall examine in detail the warm ab- Fe transitions, which are still poorly known, are not ex- sorberintheNLSy1galaxyNGC4051withrespecttothe pected to alter this result. We conclude that, for the case questionwhetherthisionizedmaterialcanaccountforthe of NGC4051, warm absorber and coronal line region are opticalcoronallinesobservedinthisgalaxy.Thisisanup- not one and the same component, but of different origin. date of our earlier work on this subject (see Komossa & This is consistent with the recent findings of Nagao et al. Fink 1997a;KF97a hereafter). (2000) that the [FeXI] emission of NGC4051 is not confined to the nucleus, but widely extended (out to at least 4.2. Observed properties of NGC4051 150 pc). ∼ We first give a brief summary of previous observational properties of NGC4051 relevant for the present study 4.4. Comparison with subsequent studies (takenfromKF97):(i)Tomodelthewarmabsorberandto Recently Porquet et al. (1999; P99 hereafter) presented predictitslineemission,wecompiledtheobservedSEDof a parameter space study of the strengths of Fe coronal thegalaxyfromtheliterature(seeFig.1ofKF97a);based lines that originate from warm absorbers. They conclude on photon counting arguments, we found evidence for an that Fe lines in low-density absorbers (they studied the EUV bump in the spectrum of NGC4051. (ii) Intensity density range logn = 8–12) are over-predicted for part ratiosofcoronallinesweretakenfromthe literature(e.g., H of the parameter space. Comparing our earlier results on de Robertis & Osterbrock 1984, Penston et al. 1984, Pe- NGC4051 (Komossa & Fink 1997a) with their recent re- terson et al. 1985,Wilson & Nath 1990,Osterbrocket al. sults, we find both to be consistent: For warm absorbers 1990,Erkens et al. 1997,Nagao et al. 2000).(iii) A warm dominated by OVIII absorption and high ionization pa- absorber fit to our ROSAT PSPC spectrum of NGC4051 rameters, no overprediction in line emission occurs (their yieldslogU =0.4andlogN =22.7.6 Further,fromvari- w Tab. 1). thenwouldhaveratherdifferentpropertiesfromGalacticISM- The questionremains whetherthe ‘OVII absorbers’of likedust,namelyadifferentsizedistributionwithadominance P99 do indeed overpredict the Fe lines and thus are in of small grains (which more effectively scatter, thus more ef- conflict with observations. We suggest that most of the fectively redden theoptical/UV spectrum),or averydifferent strong OVII absorberslikely contain dust (which was not gas/dust ratio ! included in the models of P99), as proposed by Reynolds 6 We have explored the possibility that the warm absorber (1997). The strong depletion of Fe into dust grains then inNGC4051containsdust.TheopticalspectrumofNGC4051 results in weaker gas phase emission in the Fe coronal is not heavily reddened, but the Balmer decrement of broad lines. Hα/Hβ deviates from the recombination value. If we try to modeltheX-rayspectrumofNGC4051withaone-component dusty warm absorber we do not get a successful spectral fit. The possibility remains, though, that there is a second, dusty free warm absorber. Since these could also be of instrumental warm absorber of low column density present. In fact, there origin, higher spectral resolution data are required to search aresomeresidualsaroundtheCarbonedgeafterfittingadust- for a second, dusty,warm absorber in NGC4051. 8 St. Komossa: Warm Absorbers in AGN Table 2. Predicted iron lines (Komossa & Fink 1997a) compared to the observed ones of NGC4051 assuming a covering factor of the warm absorber of 50%; all lines are given relative to the observed Hβ luminosity of NGC4051. Somenotes:(i)ThestrongestpredictedamongtheFelinesinmostmodelsisFe19λ1118;(ii)modelresultsarerobust againstthesmalluncertaintiesinthereddeningcorrection(e.g.,weassumedA =0.7mag,Nagaoetal.(2000)usedA v v =1mag);(iii)allnon-‘standard’modelsequenceswerefirstre-fittothedata,toderivethenewbest-fitWAparameters (which sometimes slightly differed from the oldones) before the lines were predicted [if this is notdone, the predicted Fe lines change more strongly!], (iv) recent NLR(and torus-)related multi-component photoionization models of Sy1s can successfully account for [FeVII] (e.g., Komossa & Schulz 1997, Murayama & Taniguchi 1998a,b). model line predictions comments [FeVII],[FeX],Fe[XI]/Hβobs [Fe XIV]/Hβobs ‘standard’ <510−5, ≪obs. 0.004 observed:Fe14/Hβ <0.06 (Nagao et al. 2000) var.abundances ′′ change≤ fact.2-4 0.3≤Z ≤1 HeIIλ4686 becomes strongest opt. line var.density ′′ ′′ 4 ≤lognH ≤9.5 ′′ ′′ var.IR conti. up to theobserved IR data points (from host galaxy) → increased f-f heating ′′ ′′ var.EUV conti. additional soft excess to account for LHβ logTbb=5,Qbb=Qpl ′′ ‘91 McHardy data 0.02 logU =0.2, logNw =22.47 (our re-fit of thedata originally presented by McHardy et al. 1995) 4.5. A warm absorber relation to any optical/UV sample study of ASCA observationsof AGN reported the emission line component ? detection of OVII and OVIII absorption edges in the X- ray spectrum of MR2251. Finally,we note thatourmore generalsearchfora warm- Here, we present first results from a detailed analysis absorber origin of of one of the emission-line regions ob- of a deep archival ROSAT PSPC (Tru¨mper 1983, Briel served in the UV/optical spectra of Seyfert galaxies did et al. 1994) observation of this source performed in Nov. notgiveasingleexamplewithaone-to-onematch(seeKo- 1993 with a duration of 18 ksec. mossa & Fink 1997e for a brief overview); emission lines from warm absorbers are always weaker than observed lines. However,the possibility of a (weak) warm-absorber 5.2. X-ray analysis: results contribution to individual emission lines (e.g., CIV in the case of NGC3227; Komossa & Fink 1997b) remains. X-raydatareductionwascarriedoutinastandardmanner (see Komossa 2000, in prep.). In total, 36 X-ray sources were detected with a likelihood 10 within the field of 5. The warm absorber of MR2251–178 ≥ view. The positions of those in the vicinity of MR2251 5.1. Introduction areshowninFig.5,overlaidonanopticalimagefromthe digitized Palomar sky survey. MR2251-178 at redshift z=0.068 was the first quasar to bediscoveredinX-rays,inthecourseoftheArielVall-sky The meansourcecountratewas3.1cts/s,afactor of3 survey (Cooke et al. 1978, Ricker et al. 1978). It turned stronger than during the ROSAT all-sky survey observa- out to be an outstanding object in many respects. It has tion performedin 1990.The X-ray lightcurve is displayed a high ratio of L /L , is surrounded by a giant HII en- in Fig. 6. x opt velope observed in [OIII] emission, and was the first ob- First,we fit asingle powerlawto the totalX-rayspec- ject in which a warm absorber was discovered (Halpern trum. Cold absorptionwas fixed to the Galactic value to- 1984).BasedonEXOSATandGinga observationsPanet wards MR2251, N = 2.77 1020 cm−2 (Lockman & Gal × al.(1990)andMineo&Stewart(1993),again,interpreted Savage 1995), since it is underpredicted otherwise. This and modeled the data in terms of the presence of a warm gives Γ = 2.3 and clearly is a very poor description of x − absorber. In contrast, Walter & Courvoisier (1992), in an the data (χ2 =5.4;Fig. 7). A Raymond-Smith emission red independent analysis explained the observations in terms model does not fit the spectrum at all, even if temper- of a variable powerlaw spectral shape, and concluded the ature, normalization, metal abundances and the amount presence of a warm absorber was unnecessary to explain of cold absorption are all treated as free parameters. We the X-ray data. Reynolds (1997), in the course of a large then tried two-componentspectral fits involving a power- St. Komossa: Warm Absorbers in AGN 9 lawplussoftexcessparametrizedbydifferentmodels.The the RASS observation,and has (nearly) disappeareddur- quality of the fit remains unacceptable, though. ing the later PSPC observation (see KM2000 for details). The presence of a warm absorber markedly improves In summary, we find that a warm absorber fits well the fit. Performing a two-edges fit with edge energies the ROSAT survey observation,and residuals around the fixed at the theoretical values of OVII and OVIII we expectedlocationofabsorptionedgesarestillpresentdur- obtain τ = 0.26 0.12, τ = 0.20 0.12, and ing our later pointed PSPC observation.7 In addition, OVII OVIII ± ± Γ = 2.20 0.02. The not yet totally satisfactory qual- the presence of high-ionization UV absorption lines in x − ± ityofthefit,χ2 =1.6,canbetracedbacktoadeviation this object was recently reported by Goodrich (talk at red of the low energy part of the spectrum (between 0.1–0.18 the Narrow-line Seyfert1 workshop in Bad Honnef, Dec. keV) from the model prediction which already caused an 1999),andtherecentstudybyCrenshawetal.(1999)sug- underprediction of the galactic N in the pure powerlaw gests a near one-to-one match of the presence of UV and H fit. If this partofthe spectrum is excludedfromthe spec- X-ray warm absorbers. tral fitting, we obtain χ2red = 0.9, τOVII = 0.22± 0.11, We conclude, that the presence of a time-variable τOVIII =0.24 0.12, andΓx = 2.21 0.02.We conclude warm absorber is a viable explanation for all available X- ± − ± that a warm absorber is clearly present in this source. ray observations of RXJ0134-4258. High-spectral resolution data taken during episodes of repeated spectral vari- 6. The X-ray transient AGN RXJ0134-4258 ability of this source are required to distinguish this from other possible scenarios like real changes in the steepness 6.1. Summary of the observations ofthepowerlaworapowerlawplussoftexcessdescription The Narrow-line Seyfert1 galaxy RXJ0134-4258is one of (see Komossa & Meerschweinchen 2000). the rare sources which showeddramatic spectral variability. Its spectrum changed from ultrasoft (Γ = 4.3) in Acknowledgements. ItisapleasuretothankGaryFerland for x the ROSAT all-sky survey (RASS) to flat (Γ =− 2.2) providing Cloudy, Smita Mathur for very useful and pleas- x − ant discussions, and the organizers for carrying out this ex- in our pointed PSPC observation taken two years later citing workshop. This workshop was funded by the Inter- while its countrate remained nearly constant (Greiner Research Centers Cooperative Program of the JSPS and the 1996, Komossa & Fink 1997d, Komossa & Greiner 1999, Deutsche Forschungsgemeinschaft DFG. The ROSAT project Komossaetal.1999b,Grupeetal.2000,Komossa&Meer- hasbeensupportedbytheGermanBundesministeriumfu¨rBil- schweinchen 2000). dung,ForschungundTechnologie(BMBF/DLR)andtheMax- Onepossibleexplanationforthiskindofspectralvari- Planck-Society. ability is the presence of a warm absorber. In particu- Preprintsof this andrelated papers can beretrieved from our lar, we have studied the following two scenarios (details webpage at http://www.xray.mpe.mpg.de/∼skomossa/ are given in Komossa & Meerschweinchen 2000; KM2000 hereafter): References Adamietz P., Binetruy P., Girardi G., Grimm R., 1993, Nucl. 6.2. Explanations invoking a warm absorber Phys.B 401:(1-2), 257 Changes in ionization state of a warm absorber: We find AldrovandiS.M., Pequignot D., 1973, A&A25, 137 thata warmabsorberofionizationparameterlogU 0.5 Baldwin J.A., Ferland G.J., Martin P.G. et al.1991, ApJ374, ≃ andcolumndensitylogN 23fitswelltheRASSobser- 580 w vation ofthis source (χ2 =≃0.6;Fig. 8). To accountthen BianchiS.,AltonP.B.,DaviesJ.I.,1999,in‘ISOBeyondPoint red Sources’ (Vilspa, Sept. 1999); astro-ph/9910011 for the much flatter spectrum during the later PSPC ob- BinetteL.,Wilson A.S.,RagaA.,Storchi-BergmannT.,1997, servationrequiresachangeinionizationstateofthewarm A&A327, 909 absorber. Since the intrinsic luminosity of the source did not vary strongly from one to the other observation, it is 7 Theflat-statespectrumshowstwokindsofresidualswhen then required that the ionization state of the warm ab- fit by a single powerlaw (cf Fig. 9): Deviations of the lowest sorber reflects the (unobserved) history of the variability energybinsnear0.1keVsuggestingthepresenceofaverysoft inthe intrinsicluminosity(seeKM2000fordetails).After excess (or instrumental features; they are similar to the ones allowingfornon-equilibriumeffectsintheabsorberand/or seen in MR2251, Sect. 5.2), and deviations near the expected locationsoftheWAabsorptionedges.Nosingle-componentfit a range in densities, such a warm absorber is consistent canremoveallresiduals.Thesimultaneouspresenceofboth,a with the long- and short-timescale variability behavior of soft excess and a warm absorber is well possible, but such fits RXJ0134-4258.Wedidnotfavorthispossibilityinourear- would be an over-representation of the PSPC spectral resolu- lier work, because it introduces a new level of complexity tion. Since both features, very soft excess and warm absorber (more free parameters) as compared to the simpler case (note the redshift of RXJ1242) are outside or very close to of an absorber in equilibrium. theborderoftheASCAsensitivityrange,furtherobservations Cloud passing our line-of-sight: Alternatively, a cloud of withthenewgenerationX-raytelescopesarerequiredtosearch ionized material may have passed our line of sight during for thesefeatures. 10 St. Komossa: Warm Absorbers in AGN Brandt W.N., Fabian A.C., Nandra K., Reynolds C.S., Hayashida K., 1997, in ‘Emission Lines in Active Galaxies - Brinkmann W., 1994, MNRAS 271, 958 NewMethodsandTechniques’,B.M. Petersonetal.(eds), Brandt W.N., Fabian A.C., Pounds K.A., 1996, MNRAS 278, ASPconf. ser. 113, 40 326 Königl A.,Kartje J.F., 1994, ApJ 434, 446 Brandt W.N., MathurS., ReynoldsC.S., Elvis M., 1997, MN- Komossa S., Bade N., 1998, A&A331, L49 RAS 292, 407 Komossa S., Fink H., 1996, Astron.Ges.Abstr.Ser. 12, R. Briel U., Aschenbach B., Hasinger G., et al., 1994, ROSAT Schielicke(ed.), 228 user’s handbook,MPE: Garching Komossa S., Fink H., 1997a, A&A 322, 719 (KF97) Brinkmann W., Otani C., Wagner S., Siebert J., 1998, A&A Komossa S., Fink H., 1997b, A&A 327, 483 330, 67 Komossa S., Fink H., 1997c, A&A327, 555 Komossa S., Fink H., 1997d, in ‘Accretion Disks – New As- CappiM.,MiharaT.,MatsuokaM.,etal.,1996,ApJ458,149 pects’,E.Meyer-Hofmeister,H.Spruit(eds),LectureNotes ChangC.A., SchianoA.V.B., WolfeA.M., 1987, ApJ322, 180 in Physics 487, 250 (also available as astro-ph/9612185) Contini M., Viegas S.M., 1999, ApJ523, 114 Komossa S., Fink H., 1997e, in ‘Emission Lines in Active CookeB.A.,RickettsM.J.,MaccacaroT.,etal.,1978,MNRAS Galaxies: New Methods and Techniques’, B.M. Peterson 182, 489 et al. (eds), ASPconf. ser. 113, 246 Cowie L.L., Songaila A., 1986, ARA&A,24, 499 Komossa S., Fink H., 1998, in ‘Highlights in X-rayastronomy CrenshawD.M., Maran S.P.,MushotzkyR.F.,1998, ApJ496, inhonourofJoachimTru¨mper’s65thbirthday’,B.Aschen- 797 bach&M.J.Freyberg(eds.),MPEReport272,147 (also Crenshaw D.M., Kraemer S.B., Boggess A., et al., 1999, ApJ available as astro-ph/9808205) 516, 750 KomossaS.,GreinerJ.,1999,in‘HighEnergyProcessesinAc- deKool M., Meurs E.J.A., 1994, A&A 281, L65 cretingBlack Holes’, J.Poutanen,R.Svensson(eds),ASP DeRobertis M.M., Osterbrock D.E., 1984, ApJ286, 171 conf. ser. 161, 228 (also available as astro-ph/9810105) DoneC.,PoundsK.A.,NandraK.,FabianA.C.,1995,MNRAS Komossa S., SchulzH., 1997, A&A 323, 31 275, 417 KomossaS.,BreitschwerdtD.,GreinerJ.,MeerschweinchenJ., Draine B.T., Salpeter E.E., 1979, ApJ231, 77 1999a, in ‘Astrophysical Dynamics’, D. Berry et al. (eds), Elvis M., Fiore F., MathurS., Wilkes B., 1994, ApJ425, 103 Ap&SS,in press; astro-ph/9906330 ErkensU., Appenzeller I., Wagner S.,1997, A&A 323, 707 Komossa S., Breitschwerdt D., Meerschweinchen J., 1999b, in Ferland G.J., 1993, University of Kentucky, Physics Depart- ‘AstrophysicalDynamics’,D.Berryetal.(eds),Ap&SS,in ment, InternalReport press; astro-ph/9906377 Ferland G.J., Korista K.T., Verner D.A., et al., 1998, PASP Komossa S., Meerschweinchen J., 2000, A&A in press; astro- 110, 761 ph/9911429 (KM2000) Fabian A.C., KuniedaH.,InoueS.,et al., 1994, PASJ46, L59 Kriss G.A., Krolik J.H., Otani C., et al., 1996a, ApJ 467, 629 Fiore F., Elvis M., Mathur S., Wilkes B.J., McDowell J.C., Kriss G.A., Espey B.R., Krolik J.H., et al., 1996b, ApJ 467, 1993, ApJ415, 129 622 George I.M., TurnerT.J., Netzer H., 1995, ApJ 438, L67 Kriss G.A., Peterson B.M., Crenshaw D.M., Zheng W., 1999, George I.M., NandraK., Laor F., et al., 1997, ApJ491, 508 ApJin press; astro-ph/9912204 George I.M., Turner T.J., Netzer H., et al., 1998a, ApJS 114, Krolik J.H., Kallman T.R., 1984, ApJ 286, 366 73 Krolik J.H., Kriss G.A., 1995, ApJ447, 512 GeorgeI.M.,MushotzkyR.,TurnerT.J., 1998b,ApJ509,146 Laor A.,Draine B.T., 1993, ApJ 402, 441 Leighly K.M., Mushotzky, R.F., Nanda K., Forster K., 1997, George I.M., Turner T.J., Mushotzky R., Nandra K., Netzer ApJ489, 25 H., 1998c, ApJ503, 147 Leighly K.M.,KayL.E., WillsB.J., WillsD.,GrupeD.,1997, Goodrich R.W., 1989, ApJ340, 190 ApJ489, L137 Goodrich R.W., 1995, ApJ440, 141 Lockman F.J., Savage B.D., 1995, ApJS 97, 1 Greaves G.N., Binstead N., Henderson C.M.B., 1984, in ‘EX- Martin P.G., 1970, MNRAS149, 221 AFSandnearedgestructureIII’,K.O.Hodgsonetal.(eds), Martin P.G., Rouleau F., 1991, in ‘Extreme UV Astronomy’, 297 R.F. Malina, S. Bowyer (eds), 341 GreinerJ.,1996,in‘SupersoftX-raySources’,J.Greiner(ed.), Mathis J.S., RumplW., Nordsieck K.H., 1977, ApJ 217, 425 Lect. Notesin Phys. 472, 285 Mathur S., 1994, ApJ431, L75 GrupeD., et al., 2000, A&A,submitted Mathur S., Wilkes B., Elvis M., Fiore F., 1994, ApJ434, 493 GuainazziM.,MiharaT.,OtaniC.,MatsuokaM.,1996,PASJ Mathur S., Elvis M., Wilkes B., 1995, ApJ452, 230 48, 781 Mathur S., Wilkes B., Aldcroft T., 1996, ApJ478, 182 GuainazziM.,NicastroF.,Fiore,etal.,1998,MNRAS301,L1 Mathur S., Wilkes B., Elvis M., 1998, ApJ503, L23 Halpern J.P., 1984, ApJ281, 90 Mathur S., Elvis M., Wilkes B., 1999, ApJ519, 605 HamannF.,BarlowT.A.,BeaverE.A.,etal.,1995a,ApJ443, Matt G., 1998, in ‘Science with XMM’, M. Dahlem (ed.) 606 McHardyI.M.,GreenA.R.,DoneC.,etal.,1995,MNRAS273, Hamann F., ZuoL., Tytler D., 1995b, ApJ 444, L69 549 Hamann F., Shields J.C., Ferland G.J., Korista K.T., 1995c, MeierU.,DemmigF.,1997,in‘ProceedingsoftheJSMECen- ApJ 454, 688 tennial Congress’, (Tokio, Japan) Hamann F., Cohen R.D., Shields J.C., et al., 1998, ApJ 496, MiharaT.,MatsuokaM.,MushotzkyR.,etal.,1994,PASJ46, 761 L137