The rise of an ionized wind in the Narrow Line Seyfert 1 Galaxy Mrk 335 observed by XMM-Newton and HST

1.4 MB·English

by  A. L. Longinotti

The rise of an ionized wind in the Narrow Line Seyfert 1 Galaxy
Mrk 335 observed by XMM-Newton and HST
3 A.L Longinotti1,2, Y. Krongold3, G. Kriss4,9, J. Ely4, L. Gallo7, D. Grupe6, S. Komossa5, S.
1 Mathur8, A. Pradhan8
0
2 1 European Space Astronomy Centre of ESA, Madrid, Spain
2 MIT Kavli Institute, 77 Massachusetts Avenue, 02139 Cambridge, USA
n
a 3 Universidad Nacional Autonoma de Mexico (UNAM), Mexico
J 4 Space Telescope Science Institute, 3700 S. Martin Drive, Baltimore, MD 21218, USA

9 5 Max Planck Institut fuer Radioastronomie, Auf dem Huegel 69,53121 Bonn, Germany
2 6 Department of Astronomy and Astrophysics The Pennsylvania State University, USA

7 Department of Astronomy and Physics, Saint Mary’s University, Halifax, Canada
]
O 8 Department of Astronomy, Ohio State University, 140 West 18th Avenue, Columbus, Ohio 43210-1173
C 9 Department of Physics and Astronomy, The Johns Hopkins University, Baltimore, MD 21218, USA
.
h
p
-
o
r ABSTRACT
t
s
We present the discovery of an outflowing ionized wind in the Seyfert 1 Galaxy Mrk 335.
a
[ Despite having been extensively observed by most of the largest X-ray observatories in the last

decade, this bright source was not known to host warm absorber gas until recent XMM-Newton
2
observationsincombinationwithalong-termSwiftmonitoringprogramhaveshownextremeflux
v
3 and spectral variability. High resolution spectra obtained by the XMM-Newton RGS detector
6 revealthat the wind consists of three distinct ionizationcomponents,all outflowingat a velocity
4 of∼5000km/s. Thiswindisclearlyrevealedwhenthe sourceisobservedatanintermediateflux
5 state(2-5×10−12ergscm−2s−1). Theanalysisofmulti-epochRGSspectraallowedustocompare
.
1 the absorber properties at three very different flux states of the source. No correlation between
0 thewarmabsorbervariabilityandtheX-rayfluxhasbeendetermined. Thetwohigherionization
3 components of the gas (logξ∼2.3 and 3.3) may be consistent with photoionization equilibrium,
1
but we can exclude this for the only ionization component that is consistently present in all flux
:
v states(logξ∼1.8). We haveincludedarchival,non-simultaneousUVdata fromHST (FOS,STIS,
i
X COS) with the aim of searching for any signature of absorption in this source that so far was
known for being absorption-free in the UV band. In the COS spectra obtained a few months
r
a after the X-ray observations we found broad absorption in CIV lines intrinsic to the AGN and
blueshifted by a velocity roughly comparable to the X-ray outflow. The global behavior of the
gas in both bands can be explained by variation of the covering factor and/or column density,
possibly due to transverse motion of absorbing clouds moving out of the line of sight at Broad
Line Region scale.
Subject headings: ActiveGalaxies: general — ActiveGalaxies: Mrk 335
1. Introduction of many AGN both atUV and X-raywavelengths
(see Crenshaw et al. 2003 for a review).
Outflowingphotoionizedgasisawell-established
There are several reasons for keeping our in-
feature of over half of all Seyfert Galaxies. The
terest high in the study of warm absorbers. The
presence of this gas is revealed as a series of
observedvelocityshift(almostalwaystotheblue)
blueshifted absorption signatures in the spectra
1
provides evidence that material is traveling out- The latest XMM-Newton and Hubble Space
ward from the central region of AGN. If this ma- Telescope (HST) observations of the Narrow Line
terial eventually leaves the AGN, then outflows Seyfert 1 Mrk 335 (z = 0.025785, Huchra et al.
might carrysignificant mass out of the AGN and, 1999) presented in this paper resolve a potential
as a consequence, give a substantial contribution challenge to this paradigm.
to the chemical enrichment of the intergalactic Mrk 335 has a long X-ray history. It has been
medium (IGM) (Furlanetto et al. 2001, Cavaliere observed by all the largest facilities throughout
et al. 2002, Germain et al. 2009, Hopkins et al. the years,with the exception of Chandra. The X-
2010, Barai et al. 2011). The outflow can also ray spectrum has always shown a standard AGN-
impact the development of the host galaxy itself. shape: mildly steep power law in the 2-10 keV
If it is as strong as 0.5–5% of the Eddington lu- band, a strong soft X-ray excess and a complex
minosity of the AGN, then the feedback from the and prominent Fe K line (see Larssonet al. 2007,
AGN can regulate the growth of the galaxy and Longinotti et al. 2007, O’Neill et al. 2007).
the growth of the central black hole as well (Di
Intrinsic soft X-ray absorption was never de-
Matteo et al. 2005,Hopkins et al. 2010).
tected with high confidence. In the ASCA survey
ThefactthatX-rayabsorbersarealwaysfound of warm absorbers presented by Reynolds (1997),
in sources with UV absorption (Crenshaw et al. thissourcewasnotamongthelistoftheonespre-
2003, Kriss 2006) indicates that there is some in- sentingthecharacteristicabsorptionedgesthatre-
terplay between the two phenomena and this in vealedthepresenceofionizedgasalongthelineof
turn implies that photoionized gas must be dis- sight. Likewise,Mrk335hasnevershownanyUV
tributed with high coveragein the AGN. absorptioneither (Zheng etal. 1995,Crenshawet
Several open questions are still awaiting solu- al. 1999,Dunn et al. 2007).
tion. In many cases, high quality observations of In2007,Mrk335underwentanabruptdecrease
bright sources have allowed us to relate kinemat- influx(Grupeetal. 2007,2008)anditgaveusthe
ically blue-shifted X-ray and UV absorption lines opportunitytoobserveandstudyapreviouslyun-
(e.g. Kaspi et al. 2002,Kaastra et al. 2002) lead- known emission-line component that revealed the
ing to the idea of an outflowing wind that spans presenceofphotoionizedgasatBroadLineRegion
a wide range of ionization states with a common scales (Longinotti et al. 2008). The broadband
velocity pattern. In contrast, other cases have (CCD) spectrum of this peculiar low flux state of
shown no obvious relation between UV and X- Mrk 335 was explained by Grupe and collabora-
ray absorption components (Kriss et al. 2011). tors by an intervening partial covering absorber,
The ionization structure of the wind was some- although a competing and alternative scenario in
times found to be a continuous distribution, as which the low state is produced by disk reflection
proposedfor NGC 5548(Steenbrugge et al. 2005) could not be rejected. Noticeably, soft X-ray ion-
or made by discrete, separate ionization compo- ized absorption was not observed in the high res-
nents, sometimes in pressure equilibrium (Kron- olution data ofany flux state (O’Neill et al. 2007,
goldetal. 2003),andothertimesnot(Detmerset Longinotti et al. 2008).
al. 2011). The location of the photoionized gas is
In 2009 Mrk 335 was observed for 200 ks by
alsoasubjectofdebate. The NarrowLine Region
XMM-Newton as part of a trigger program based
onkpcscales(Kinkhabwalaetal. 2002),theinner
onSwiftlongtermmonitoring(Grupeetal. 2012).
edgeofthemolecularAGNtorusonpcscale(Kro-
The source was caught for the first time at a flux
lik&Kriss2001)andtheaccretiondiskonsub-pc
state intermediate between the deep minimum of
distances (Elvis 2000, Krongold et al. 2007) have
2007 and the “standard” bright flux in which it
all been put forward as possible regions of origin
was always observed. These data present very
for these winds.
different characteristics with respect to previous
Onesoundpieceofevidencethatsofarhasbeen XMM-Newton observations. In particular they
fully supportedobservationallyisthat, inthevast clearly show intrinsic soft-ray absorption in the
majority of sources -practically 100% - intrinsic RGS band. This finding seems to question the
absorptionispresentinbothX-rayandUVbands one-to-one correspondence reported between X-
(Crenshaw et al. 2003, Kriss 2006). ray and UV absorbers.
2
Mrk 335 EPIC−pn
Table 1: XMM-Newton Observation log. 1) Gon-
doin et al. 2002; 2) Longinotti et al. 2007; 3) O’ Neill et 0306870101 − 03/01/2006
al. 2007; 4) Grupe et al. 2008; 5) Longinotti et al. 2008; 0101040101 − 25/12/2000
0600540501 − 13/06/2009
6)Grupeetal. 2012;7)Galloetal. 2013 m−2 0.01 0600540601 − 11/06/2009
OBSID Date Exp Flux state Ref V c−1 0510010701 − 10/07/2007
e
0101040101 2000-12-25 (3k7s) Hi-gh 1-,2 nts s k−1 10−3
u
0306870101 2006-01-03 133 High 3 co
d
00561000051400760011 22000079--0066--1111 12332 LMoiwd 64,,75 malize 10−4
or
0600540501 2009-06-13 82 Mid 6,7 n
0.5 1 2 5
Energy (keV)
Fig. 1.— Multi-epoch EPIC pn spectra of Mrk
The present paper presents the first high-
335. The effective area has been accounted for,
resolution study of the warm absorber in Mrk
thereforethedifferencebetweenthevariousobser-
335 and it is based on the comparison of all the
vationshastobeattributedtointrinsicchangesin
data obtained up to the year 2009 by the RGS
the spectral shape.
spectrometer onboard XMM-Newton.
The X-ray analysis presented here is com-
plemented by the inclusion of recent and non- (0101040101, 0600540501, 0600540601), Small
simultaneous HST-COS spectra that were ob- window mode (0306870101) and Large window
tained a few months after the X-ray observations. mode (0510010701). Data reduction was per-
formed by using the standard tasks epproc and
2. Observations and data reduction rgsproc, for pn and RGS data respectively.
Screening of high particle background was ap-
2.1. XMM-Newton data
plied to the pn data so as to maximise the signal-
Mrk335wasobserved5timesbyXMM-Newton to-noiseratioaccordingtotheproceduredescribed
(see Table 1 for the log of observations). All the inPiconcellietal. (2004). Thespectralextraction
datasetswerereducedwithSAS11.0.01. Forade- regions in the pn camera is a circle of 40 arcsec of
taileddescriptionofthevariousdatasets,wedefer radius for source and background spectra, in all
the reader to the numerous prior publications on observations.
this source that are listed in Table 1.
Thefocusofthepresentpaperistheanalysisof 2.1.1. The EPIC pn data
the warm absorber features in the high-resolution
The CCD data of the XMM-Newton campaign
spectra obtained by the Reflection Grating Spec-
arethesubjectoftwomorepublicationspresented
trometer(RGS,denHerderetal. 2001),therefore
by our group (Grupe et al. 2012 and Gallo et
wedonotinclude CCDdatafromthe EPICMOS
al. 2013). The former paper reports on the long
cameras, which are extensively described in pre-
termvariabilityofMrk335,whichhasbeenmoni-
vious publications. However, the EPIC pn data
toredbytheSwiftX-raytelescopesince2007,and
will be shown and used for measuring fluxes and
the application of the partial covering model to
constraining models, therefore we briefly describe
explain the variation observed by XMM-Newton.
their data reduction in the following.
Thelatterpaperconcentratesinsteadontheanal-
For all 5 data sets, the EPIC pn camera ysisofthediskreflectioncomponentintheXMM-
(Struder et al. 2001) was used as prime instru- Newton 2009 data.
ment, and it was operated in Full Frame mode
This complementary work provides the first
detailed description of the line-of-sight gas in
1http://xmm.esac.esa.int/sas/current/documentation/sas concise.shtml
Mrk 335 as seen by a high resolution spectrome-
3
vation on 12 February 2010 used gratings G130M
Table 2: X-ray fluxes (in units of
and G160M, again with four central wavelength
10−11 erg cm−2 s−1) measured by fitting the
settingsforeachgratingandFPPOS=3. Thefour
EPIC pn data with a broken power law model
different central wavelength settings provide full
with break energy fixed at 2 keV.
coverage across the wavelength gap between seg-
ments A and B of the FUV detector. They also
OBSID Flux(0.3-2keV) Flux(2-10keV)
- - placethegrid-wireshadowsandotherdetectorar-
0101040101 2.60±0.01 1.37±0.01
0306870101 3.32±0.01 1.77±0.01 tifacts at independent places along the spectrum
0510010701 0.20±0.01 0.34±0.01 so that they can be more easily removedfrom the
0600540601 0.36±0.01 0.48±0.04
0600540501 0.52±0.02 0.51±0.05 data without gaps in wavelength coverage. We
used v12.17.6 of the COS calibration pipeline to
process and combine the data. Our processing
included a 1-dimensional flat-field correction that
eliminates the grid-wire shadowsand other detec-
ter, and, as to the X-rayband, it is entirely based
tor artifacts. The combined G160M spectra cover
on the RGS data. However, consistency between
the wavelength range 1400–1770 ˚A; the G130M
spectral fits performed on the RGS and the EPIC
spectrum covers 1135–1440 ˚A. The G160M spec-
pn data is explored for completeness.
traeachhaveamediansignal-to-noiseratio(S/N)
To generally illustrate the dramatic change in of ∼ 17 per resolution element. For G130M, the
thespectralshapeofMrk335,wehaveincludedall median S/N is ∼33 per resolution element.
the EPIC-pnspectra of Mrk 335in Figure 1. The
Although the COS line-spread function (LSF)
0.3–10 keV figure shows very effectively that the
has broad wings that can fill in narrow absorp-
spectral changes are concentrated below 4-5 keV,
tion features (Ghavamian et al. 2009), we do not
and that the difference at higher energy is merely
detect any intrinsic lines in Mrk 335 that would
attributed to a normalization effect. Continuum
necessitate deconvolving the effects of the LSF as
fluxes extracted from these spectra are listed in
in Kriss et al. (2011). Ghavamian et al. (2009)
Table 2.
note that the broad LSF has negligible impact
on spectral features with Doppler widths exceed-
2.2. HST Ultraviolet Spectra
ing ∼ 100 kms−1. All spectral features of inter-
A few months after the XMM-Newton obser- est directly relatedto Mrk 335 have much greater
vations of Mrk 335 in 2009, the Cosmic Origins widths.
Spectrograph (COS) team observed Mrk 335 as After processing through the pipeline, we mea-
part of their program to probe warm and hot gas sured the locations of strong interstellar lines in
in and near the Milky Way using AGN as back- the spectra to determine a zero-point correction
groundsources(ProposalID11524,JamesGreen, for the wavelength scale using the H i velocity of
PI). Green et al. (2012) describe the key char- V = −11 km s−1 along the Mrk 335 sight-
LSR
acteristics of the design and performance of the line (Murphy et al. 1996). Overall uncertainties
COS instrument on the Hubble Space Telescope in the COS wavelength scale limit our knowledge
(HST). We have retrieved these spectra from the ofthe absolutewavelengthscale to sim15kms−1.
HST archive to see if any trace of the unexpected Figure 2 shows the calibrated COS spectrum of
X-rayabsorptionwe see in the RGS spectra has a Mrk335,withtheG130MobservationfromFebru-
UVcounterpart. Table3summarizestheobserva- ary 2010 merged with the G160M observation of
tional details. October 2009.
The first observation on 2009 October 31 used To place these most recent HST spectra in an
the Primary Science Aperture (PSA) and grating historicalcontext, we also retrievedarchivalspec-
G160Mwithfourdifferentcentralwavelengthset- tra obtained using the Faint Object Spectrograph
tings. All four observations used the same focal (FOS) in 1994, and more recent Space Telescope
plane position for the detector, FPPOS=3.ll four Imaging Spectrograph(STIS) spectra obtained in
observations used the same focal plane position 2004. The relevant data sets are also listed in Ta-
for the detector, FPPOS=3. The second obser- ble 3. For these data we performed no special
4
Table 3: HST Observations of Mrk 335
Data Set Name Instrument Grating/Tilt Date Start Time Exposure Time
(GMT) (s)
lb4q05010 COS G160M/1589 2009-10-31 13:40:19 408
lb4q05020 COS G160M/1600 2009-10-31 13:50:24 408
lb4q05030 COS G160M/1611 2009-10-31 14:00:29 408
lb4q05040 COS G160M/1623 2009-10-31 14:10:34 409
lb4q06010 COS G160M/1589 2010-02-08 07:34:24 302
lb4q06020 COS G160M/1600 2010-02-08 07:43:02 302
lb4q06030 COS G160M/1611 2010-02-08 07:51:40 302
lb4q06040 COS G160M/1623 2010-02-08 08:00:18 301
lb4q06050 COS G130M/1291 2010-02-08 08:58:05 608
lb4q06060 COS G130M/1300 2010-02-08 09:15:01 605
lb4q06070 COS G130M/1309 2010-02-08 09:28:52 604
lb4q06080 COS G130M/1318 2010-02-08 10:37:38 605
y29e0202t FOS H13 1994-12-16 05:43:31 1390
y29e0203t FOS H13 1994-12-16 07:01:06 770
y29e0204t FOS H19 1994-12-16 07:20:06 960
y29e0205t FOS H27 1994-12-16 07:40:07 60
y29e0206t FOS H27 1994-12-16 08:40:19 420
y2gq0301t FOS PRISM 1994-12-16 08:48:35 60
o8n505010 STIS E140M 2004-07-01 16:24:51 1945
o8n505020 STIS E140M 2004-07-01 17:44:50 2295
o8n505030 STIS E140M 2004-07-01 19:20:49 2875
o8n505040 STIS E140M 2004-07-01 20:56:49 2875
Fig. 2.— Calibrated COS spectrum of Mrk 335. Wavelengths shortwardof 1430˚A are from the G130Mob-
servationofFebruary2010;longerwavelengthsarefromtheG160MobservationinOctober2009. Prominent
emission features are indicated. Geocoronal emission in the center of the galactic Lyα absorption trough
is indicated with an Earth symbol. The narrow absorption features in the spectrum are either foreground
interstellar lines or intervening intergalactic Lyα absorbers.
5
processing other than to correct the zero point of asimplepowerlawandGalacticabsorptionmodel.
the wavelength scale of the FOS spectra to align The absorption around 16 ˚A, which is obvious in
the ISM absorption features with the H i veloc- the mid state spectrum observed in 2009, is un-
ity noted above. The STIS spectrum required no doubtedly less pronounced in the high state data
additional correction. Table 4 compares the con- of 2006. Analogously, the emission line around
tinuumfluxes forallthe HST observationsofMrk 19 ˚A, is more evident in the mid state data, due
335 that we discuss below. to the lower continuum flux with respect to the
2006 spectrum. After this first-order comparison,
3. X-ray spectral analysis we proceed to model the absorption observed in
the mid state spectra.
3.1. Strategy
3.2. The ionized gas in the 2009 mid state
ThespectralanalysisoftheXMM-NewtonRGS
spectra was carried out by using two indepen- We start by fitting the continuum in the RGS
dent photoionization codes. We applied the xabs band(7–30˚A)withapowerlawandablackbody
warmabsorbermodelincludedintheSPEXfitting componenttomimicthesoftexcess. TheGalactic
package (Kaastra, Mewe, Nieuwenhuijzen, 1996). column density along the line of sight to Mrk 335
In parallel, the spectra were analyzed following isalsoincludedandfixedto4×1020 cm−2 (Dickey
the same procedure by employing the photoion- & Lockman 1990).
ization code PHASE, developed by Krongold et
To this purely phenomenological model of the
al. (2003). All the results presented in this pa-
continuum we addedthe effect ofabsorptionfrom
per are extracted from the analysis carried out
ionized material. This effect is modeled by the
with the SPEX software. However, the warm ab-
xabs warm absorber code that reproduces the
sorberpropertiesderivedforthe 2009spectraand
transmission of the nuclear continuum through a
reported in the following section were fully con-
slabofionizedgasin photoionizationequilibrium.
firmed by the second analysis. In this way we are
Thefreeparametersinourfitsarethecolumnden-
confident to have checked the possible model de-
sity of the gas, the ionization parameter, the out-
pendency thatsometimesisintroducedbytheuse
flow velocity and the root mean square velocity
of different photoionization codes.
width ofthe absorptionlines. SolarSystemabun-
During the fitting procedure RGS data have dances from Lodders et al. (2009) are assumed.
been binned by a factor of 4. Whenever other
choices of grouping were taken (i.e. for plotting
10−10
purpose),itwillbespecified. χ2-statisticswasap-
plied and 1-sigma error bars are quoted through-
out the paper.
We highlight our strategy in the following. −2−1m s)
Saindceetatilheedmstauidnypoufrptohsee wofartmheapbrseosrebnetr wanodrkitiss E) (erg c 10−11
F(
multi-epoch behaviour, we start by analysing the E
data of 2009, where the presence of the ionized
absorber is very evident.
10−12
Once a complete model of the ionized gas is 0.001 0.010 0.100 1.000 10.000
Energy (keV)
established on sound bases for the mid state, we
extend it to the other data sets. In this way we
Fig. 4.— Observed spectral energy distribution
willtestthe behaviourofthe absorberatdifferent
of Mrk 335 in the 2009 mid state from XMM-
epochs and its relation to the corresponding flux
Newton.
levels that Mrk 335 has shown during the XMM-
Newton observations. Initially, the ionized absorber is assumed to be
Figure 3 displays the comparison of the RGS totally covering the source of radiation. Subse-
residuals of the mid and high state data fitted by quently, in the fitting procedure we will test for
6
Table 4: Continuum Fluxes for HST Observations of Mrk 335
Observatory Date F(1500 ˚A)
(10−14 ergscm−2s−1˚A−1)
HST/FOS 1994-12-16 6.0
HST/STIS 2004-07-01 3.6
HST/COS 2009-10-31 3.1
HST/COS 2010-02-08 3.1
the possibility of a more patchy geometry of the at the residuals, was tested and found not signifi-
gas. cant.
The absorption spectrum is calculated assum- The best fit parameters for the three warm ab-
ing the photoionization balance produced by the sorbers are summarised in Table 5. The separate
incoming ionizing radiation. contributionof eachabsorptioncomponentto our
Figure 4 shows the spectral energy distribu- globalmodelisillustratedinFigure6. The3mod-
tion (SED) of Mrk 335 including also the data els have been renormalized so as to highlight the
from the Optical Monitor (OM) onboard XMM- dominant features in each of them.
Newton. The OM observed Mrk 335 in 4 filters
centered at 2182, 2341, 2946 and 3481 ˚A, simul- 3.2.1. Consistency between the two mid state
spectra of 2009
taneously to the X-ray instruments. This allowed
us to add the information on the ultraviolet part
To carefully check the consistency between
of the SED for calculating the ionization balance
the two separate XMM-Newton observations
in our warm absorber models.
we applied the warm absorber model described
Figure1andTable 2showafew percentdiffer- above to the second mid state data set (OBSID
ence in the softX-rayflux ofthe two observations 0600540601). This spectrum is shown in the top
of 2009 (see also Grupe et al. 2012 for the light panel of Figure 3. In this fit, all the parame-
curve analysis). A visual inspection of the data ters are left free to vary, including those of the
also reveals little difference in the spectral shape baseline continuum model. The detailed results
of the two RGS spectra. We therefore decided to are reported in the second row of Table 5. We
start by carrying out the analysis of the two data concluded that the fit of the two RGS data sets
sets separately and check consistency in the spec- yieldparametersfullyconsistentwithintheerrors.
tralparametersduring the fitting process. All the Furthermore,wenotethatthesoftX-rayfluxvari-
fit statistics reported below refer to the data set ations between the two observations of 2009 (the
extracted from OBSID 0600540501. 2-10 keV flux being practically constant, see Ta-
A continuum model without any intrinsic ble 2), does not present variability higher than a
absorption consisting of Nh Gal *(power law 30% level, which is approximately the accuracy
+ black body) produces an unsatisfactory fit in the errorsof the detected column densities and
(χ2/d.o.f. = 1516/1052). ionizationparameters(Table 5). Therefore,warm
absorber variations induced by the observed flux
The addition of one warm absorber compo-
variability would not be detectable in these data
nent to this baseline model is highly significant
(χ2/d.o.f.=1435/1048) but still not sufficient to sets.
provide a good fit to the spectra. We added a Nonetheless, we need to consider the results of
secondcomponentandfoundanimprovedbestfit the variability behaviour of Mrk 335 reported by
model with χ2/d.o.f.=1334/1044. When the pres- Grupe et al. (2012). These authors have split the
enceofathirdwarmabsorberwastestedthe final dataof2009accordingtothepncountingrateand
fit statistics of χ2/d.o.f.=1307/1040 was reached. they have obtained two spectra corresponding to
The data fitted by this model are shown in Fig- “faint”and“bright”portionswithin the 2009mid
ure 5. The presence of an additional absorption
componentthatmightbe postulatedafterlooking
7
Table 5:Best fit parametersof the 3 warmabsorber model in the mid states of2009. The coveringfactor of
the absorbers is fixed to 1. The emission lines reported in Section 3.3 are included in these fits.
Dataset continuum WaPhase Logξ Columndensity vout vbroad χ2/d.o.f.
(TinkeV) - (ergcms−1) (cm−2) (kms−1) (kms−1)
0600540501 Γ=2.45+−00..1107 I 1.92+−00..0151 2.72+−11..1013×1021 4300+−915500 <30 1307/1040
kTbb=0.09±0.02 II 2.35+−00..2146 4.43+−21..0310×1021 5450+−117800 120+−15400
III 3.31+−00..0098 2.13+−61..2152×1022 4000+−213800 170+−8650
0600540601 Γ=2.45+−00..3543 I 1.75+−00..2315 3.27+−22..0311×1021 6000+−13400000 <70 1249/1033
kTbb=0.11±0.07 II 2.15+−00..1102 4.80+−22..3300×1021 4900+−116700 120+−6800
III 3.22+−00..5358 2.04+−410.82×1022 5200+−625500 <650
First80ks Γ=2.33+−00..3500 I 1.61±0.16 3.34+−11..6305×1021 4700+−815200 <20 1226/1033
0600540601 kTbb=0.10±0.08 II 2.16+−00..0099 4.07+−32..4005×1021 4900+−116705 <70
III 3.26+−00..1176 4.02+−138.00×1022 5200+−1200000 90+−7505
combinedRGS Γ=2.57+−11..7941 I 1.85+−00..0141 2.15+−00..3266×1021 4000+−178000 <10 2143/1678
kTbb=0.11±0.01 II 2.11+−00..0023 4.22+−00..4488×1021 5140+−16900 140±30
III 3.19+−00..0059 3.56+−11..0882×1022 5300+−91000 120+−4300
0600540601
(RGS+pn) Γ=1.76+−00..0022 I 1.82+−00..0067 2.86+−10..7880×1021 4700+−15700 <15 1704/1284
kTbb=0.11±0.01 II 2.15+−00..0032 3.62+−00..2124×1021 5100+−9900 120+−4200
III 3.31±0.13 6.22+−21..2152×1022 6200+−88000 30+−1105
8
Fig. 3.—Comparisonbetween residuals(in terms ofσ) ofthe mid state (top)and highstate (bottom) RGS
spectrum (OBSID 0600540601and 0306870101)fitted only by a power law and Galactic absorption model.
The broad absorption feature produced by the Fe UTA transitions that is very prominent in the mid state
disappears almost completely in the high state data: residuals around 16˚A in the bottom panel are still
negative but at a lower significance.
state observations2. The “faint spectrum” corre-
100 High Ionization phase sponds to the first 80 ks of obs 0600540601and it
Med Ionization phase
is made by events with less than 3 counts per sec.
Low Ionization phase
In order to exclude that the warm absorber fit
m 10 is driven by the brightest part of the 2009 data,
m^2/s/Angstro wm80eodkhsea.lvteToahtpehpepliaeRdrGatmSheestpteehrcsrteoreufwmthariesmxtfiratabcastroeerdbreferrposombrteteshdtefisinet
Photons/ 1 the third row of Table 5. In this way we test for
possible variations of the warm absorber proper-
ties in correlation with the source flux within the
0.1 10 15 20 25 30 2009 data. The parameters derived in the “faint”
Wavelength (Angstrom) 80 ks spectrum are remarkably consistent with
those derivedwhenusingthe twointegratedspec-
Fig. 6.—Theplotshowsthe3(rescaled)warmab-
tra of obs 0600540601 and 0600540501 (first two
sorbercomponentsdetectedinthemidstatespec-
trumandreportedinTable5. Fromtop: medium,
high and low ionization. 2Inthiscontext“faintandbright”donothavereferenceto
themulti-epochbehaviourofthesource
9
Fig. 5.— The RGS spectrum of the 0600540501 mid state data fitted by the best fit model in Table 5.
The labels of the absorption lines (grey) are blueshifted by an average outflow velocity of 5000 km/s, the
labels for the emission lines (red) correspond instead to the laboratory values. Instrumental features can
be distinguished by their typical “squared” shape. The relative contribution of the three warm absorbers is
best seen in Figure 6.
rows in the table). 3.3. The emission lines in 2009
We conclude that the ionized gas does not
When Mrk 335 was observed at its lowest flux
present significant variation while XMM-Newton
state, Grupe et al. (2008) and Longinotti et
observedMrk335in2009,thereforeweappliedthe
al. (2008) discovered an important emission line
best fit model to the combined data sets of 2009
component. Residuals in Figure 3 show a clear
and from now on, we consider this globaldata set
emission line corresponding to the position of
as“themidstate”. Theresultsfromthislatestfit
OVIII Lyα.
are also included in Table 5.
ToreplicatetheanalysiscarriedoutbyLonginotti
Last, we have checked the effect of the broad-
et al. (2008), we fitted the mid state data by
band X-ray continuum on our warm absorber
adding to the three warm absorbers model a se-
model. Indeed, extending the spectral band up
ries of emission lines with 0-width at the wave-
to 10 keV might modify the parameters since the
lengths listed in Table 6. The wavelengths were
power law slope may not be correctly measured
kept frozen to the laboratory values in order to
giventhe factthatthe RGSbandis limitedtothe
allow a direct comparison to the fluxes reported
soft X-rays (below 2 keV). To this purpose, we
by Longinotti et al. 2008 in their Table 2. Final
haveappliedthewarmabsorbermodeltotheRGS
detections are listed in Table 6. Note that the
and pn data simultaneously. This test was done
warm absorber parameters reported in Table 5
for OBSID 0600540601. The resulting photon in-
and the model plotted in Figure 5 include these
dexisindeedmodified,beingΓ=1.74+0.05,butthe
−0.06 emission lines.
warmabsorberspropertiesarebasicallyconsistent
Figure 7 shows the fluxes of the emission lines
with the RGS-only based fit (see Table 5).
10