arXiv:astro-ph/0310027v1 1 Oct 2003 A Complete Sample of Soft X-ray Selected AGN: I. The Data 1 Dirk Grupe 2,3 Astronomy Department, Ohio State University, 140 W. 18th Ave., Columbus, OH-43210, U.S.A. [email protected]Beverley J. Wills 4 Astronomy Department, University of Texas at Austin, RLM 15.308, Austin, TX 78712, U.S.A. [email protected]Karen M. Leighly 5 Dept. of Physics and Astronomy, University of Oklahoma, 440 W. Brooks St., Norman, OK 73019, U.S.A. [email protected]Helmut Meusinger 6 Th¨ uringer Landessternwarte Tautenburg, Sternwarte 5, D-07778 Tautenburg, Germany ABSTRACT We present the optical spectra and simple statistical analysis for a complete sample of 110 soft X-ray selected AGN. About half of the sources are Narrow-Line Seyfert 1 galaxies (NLS1s), which have the steepest X-ray spectra, strongest FeII emission and slightly weaker [OIII]λ5007 emission than broad line Seyfert 1s (BLS1s). Kolmogorov Smirnov tests show that NLS1s and BLS1s have clearly different distributions of the X-ray spectral slope α X , X-ray short-term variability, and FeII equivalent widths and luminosity and FeII/Hβ ratios. The differences in the [OIII]/Hβ and [OIII] equivalent widths are only marginal. We found no significant differences between NLS1s and BLS1s in their rest frame 0.2-2.0 X-ray luminosities, rest frame 5100 ˚ A monochromatic luminosities, bolometric luminosities, redshifts, and their Hβ equivalent widths. Please note: this is a special version for astro-ph that does not contain the optical and FeII subtracted spectra. The complete paper including the spectra can be retrievd from http://www.astronomy.ohio-state.edu/∼dgrupe/research/sample paper1.html Subject headings: galaxies: active - quasars:general 1 Based in part on observations at the European South- ern Observatory La Silla (Chile) with the 2.2m telescope of the Max-Planck-Society during MPI and ESO time, and the ESO 1.52m telescope during ESO time in September 1995 and September 1999. 2 Guest observer, McDonald Observatory, University of Texas at Austin 1. Introduction With the launch of the X-ray satellite ROSAT (Tr¨ umper (1982)) a new chapter in the history of astronomy was written. With the spectral sensitivity of the Position Sensitive Proportional Counter (PSPC, Pfeffermann et al. (1987)) to en- ergies as low as 0.1 keV it was possible for the first time to study the soft X-ray properties of a large number of AGN. In the first half year of its mission 1
21
Embed
A Complete Sample of Soft X-ray Selected AGN: I. The Data · PDF fileA Complete Sample of Soft X-ray Selected AGN: I. The Data1 Dirk Grupe2,3 Astronomy Department, Ohio State University,
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
arX
iv:a
stro
-ph/
0310
027v
1 1
Oct
200
3
A Complete Sample of Soft X-ray Selected AGN: I. The Data1
Dirk Grupe2,3
Astronomy Department, Ohio State University, 140 W. 18th Ave., Columbus, OH-43210, U.S.A.
We present the optical spectra and simple statistical analysis for a complete sample of 110 softX-ray selected AGN. About half of the sources are Narrow-Line Seyfert 1 galaxies (NLS1s), whichhave the steepest X-ray spectra, strongest FeII emission and slightly weaker [OIII]λ5007 emissionthan broad line Seyfert 1s (BLS1s). Kolmogorov Smirnov tests show that NLS1s and BLS1s haveclearly different distributions of the X-ray spectral slope αX , X-ray short-term variability, andFeII equivalent widths and luminosity and FeII/Hβ ratios. The differences in the [OIII]/Hβand [OIII] equivalent widths are only marginal. We found no significant differences betweenNLS1s and BLS1s in their rest frame 0.2-2.0 X-ray luminosities, rest frame 5100A monochromaticluminosities, bolometric luminosities, redshifts, and their Hβ equivalent widths.
Please note: this is a special version for astro-ph that does not contain the opticaland FeII subtracted spectra. The complete paper including the spectra can be retrievd fromhttp://www.astronomy.ohio-state.edu/∼dgrupe/research/sample paper1.html
Subject headings: galaxies: active - quasars:general
1Based in part on observations at the European South-ern Observatory La Silla (Chile) with the 2.2m telescopeof the Max-Planck-Society during MPI and ESO time, andthe ESO 1.52m telescope during ESO time in September1995 and September 1999.
2Guest observer, McDonald Observatory, University ofTexas at Austin
1. Introduction
With the launch of the X-ray satellite ROSAT(Trumper (1982)) a new chapter in the historyof astronomy was written. With the spectralsensitivity of the Position Sensitive ProportionalCounter (PSPC, Pfeffermann et al. (1987)) to en-ergies as low as 0.1 keV it was possible for the firsttime to study the soft X-ray properties of a largenumber of AGN. In the first half year of its mission
ROSAT performed, for the first time, an all-skysurvey (RASS, Voges et al. (1999)) in the 0.1-2.4keV energy band. This survey led to the discov-ery of a large number of previously unknown softX-ray sources (Thomas et al. (1998); Beuermannet al. (1999); Schwope et al. (2000)), about 1/3of them AGN. Many AGN show a strong excess insoft X-rays. Most of their bolometric luminosity isemitted in the energy range between the UV andsoft X-ray energies. It is commonly believed thatthis ’Big Blue Bump’ emission is produced by anaccretion disk surrounding the central black hole(e.g. Shields (1978); Malkan & Sargent (1982);Malkan (1983); Band & Malkan (1989)). Thesoft X-ray emission can be explained by Comp-ton scattering of thermal UV photons in a layerof hot electrons above the disk (e.g. Czerny &Elvis (1987); Laor & Netzer (1989); Ross et al.(1992); Mannheim et al. (1995)). The closer tothe Eddington limit the black hole accretes, thesofter the X-ray spectrum is expected to become(e.g. Ross et al. (1992); Pounds et al. (1995)).Alternatively, the soft X-ray emission may also beresult in an optically thick wind from the blackhole region (King & Pounds (2003)).
In the days before ROSAT the study of strongsoft X-ray AGN depended on serendipitous ob-servations, e.g. by EINSTEIN (Cordova et al.(1992); Puchnarewicz et al. (1992)), and ob-servations of AGN selected at optical wave-lengths. Stephens (1989) noticed in a sampleof EINSTEIN-selected AGN that more than 25%of her sources belonged to the Seyfert 1 (sub)classof Narrow-Line Seyfert galaxies (NLS1s, Oster-brock & Pogge (1985)), while in optically se-lected samples only about 10% of the sources areNLS1s (Osterbrock & Pogge (1985); Osterbrock(1987); Williams et al. (2003)) This higher frac-tion of NLS1s among X-ray selected sources wasconfirmed by Puchnarewicz et al. (1992) for asample of 52 EINSTEIN-detected AGN: 9 of their17 Seyfert 1 galaxies were NLS1s. Grupe (1996);Grupe et al. (1999b), and Edelson et al. (1999)found that in soft X-ray selected ROSAT AGNsamples up to 40% were NLS1s. NLS1s show ex-treme properties, such as steep X-ray spectra (e.g.Boller et al. (1996); Grupe (1996); Williams etal. (2003)), strong optical FeII and weak emis-sion from the Narrow-Line Region (e.g. Boroson& Green (1992); Boroson (2002); Grupe (1996);
Laor et al. (1997); Grupe et al. (1999b)).
We have studied the continuum and emissionline properties of a sample of 76 soft X-ray se-lected ROSAT AGN (Grupe (1996); Grupe et al.(1998a, 1999b)). However, that sample was in-complete lacking a significant number of sourcesfor which optical spectra were not obtained at thattime. Our new sample containing 110 sources iscomplete following the criteria in § 2. For eachsource X-ray and optical spectra exist that haveenough quality to allow for a detailed analysis oftheir X-ray and optical properties. We performeda detailed study of the X-ray properties of thecomplete soft X-ray selected AGN sample (Grupeet al. (2001a)). Here we describe the sampleselection (§ 2), the observations (§ 3), and data re-duction (§ 4). The FeII subtraction and the linemeasurements are described in § 5. We presentan analysis of the distributions of continuum andemission line properties of NLS1s and BLS1s in § 6.The results will be discussed in § 7. Previously un-published optical spectra are presented at the endof the paper. In a second paper (Grupe (2003),Paper II) we will present direct correlations and aPrincipal Component Analysis.
Throughout the paper spectral slopes are de-fined as energy spectral slopes with Fν ∝ ν−α. Lu-minosities are calculated assuming a Hubble con-stant of H0 =75 km s−1Mpc−1 and a decelerationparameter of q0 = 0.0.
2. Sample Selection
Our AGN sample was selected from the brightsoft X-ray sample presented by Thomas et al.(1998), using the following criteria:
• Mean RASS PSPC count rate ≥0.5 cts s−1
• Hardness ratio3 < 0.00
• Galactic latitude |b| > 20◦
A count rate threshold of 0.5 was chosen toensure sufficient X-ray photons during a typicalRASS time coverage of about 200-400s to performspectral analysis. The hardness ratio criterion en-sures a soft X-ray spectrum and the galactic lat-itude criterion ensures no hardening of the X-ray
3Hardness ratio = (hard-soft)/(hard+soft) with the soft en-ergies = 0.1-0.4 keV and hard energies = 0.5-2.0 keV.
2
spectrum due to extinction. Using these criteria,Thomas et al. (1998) found 397 sources of which113 turned out to be AGN (Thomas et al. (1998);Grupe et al. (2001a)), excluding BL Lac objectsthat have different emission mechanisms. We havealso excluded the three known transient sourcesfrom the present sample (IC 3599, Brandt et al.(1995); Grupe et al. (1995a); WPVS007, Grupeet al. (1995b), and RX J1624.9+7554, Grupe etal. (1999a)), because these sources were observedto be bright in X-rays only once and at least inIC 3599 and RX J1624.9+7554 the X-ray emis-sion might be due to a dramatic accretion event(e.g. Gezari et al. (2003)). The nature of theX-ray transience in WPVS 007 is still unclear..
3. Observations
3.1. X-ray data
In addition to the RASS data, available for allsources, about 50 have pointed PSPC observationsand are available from the ROSAT public archiveat MPE Garching. A detailed description of theX-ray observations and analysis is given in Grupeet al. (2001a).
3.2. Optical spectroscopy
Optical spectroscopy data in the rest frame Hβregion were collected over a period of 10 years us-ing various observatories and telescopes. Table 1summarizes the observations. The table containsthe coordinates of the X-ray position of the sourcein Equinox J2000, a common name, the observ-ing date, the telescope and instrument, the ob-servation time and comments. The comments listreferences to spectra that have been already pub-lished and weather conditions. In the followingwe describe the observations by telescope as theyappear in Table 1.
3.2.1. ESO 2.2m and 1.52 m telescopes
All observations of objects with southern dec-lination prior to 1995 were observed with theMPI/ESO 2.2m telescope at La Silla (ESO2.2 inTable 1). At that time, the 2.2m was equippedwith the ESO Faint Object Camera and Spectro-graph (EFOSC), which had a selection of differentgrisms for spectroscopy. Table 1 lists the numberof the grism(s) used. The grisms had the following
resolutions and wavelength coverages:
• #4: 2.2 A/pix ≈ 7 A FWHM resolution,4650 – 6800 A
• #8: 1.3 A/pix ≈ 4 A FWHM resolution,4640 – 5950 A
• #9: 1.1 A/pix ≈ 4 A FWHM resolution,5875 – 7020 A
• #10: 1.2 A/pix ≈ 4 A FWHM resolution,6600 – 7820 A
Slit widths were usually 1.′′
5 or 2′′
and the slitorientation was always in E-W direction.
The observations in 1995 and 1999 were per-formed with ESO’s 1.52m telescope equipped withthe Boller & Chivens spectrograph (B&C). Atboth times, the grating #23 with 600 groovesmm−1 resulting in a dispersion in first order of126 A mm−1 or 1.89 A Pixel−1. The resolutionwas about 6A FWHM. For all the spectra we useda slit width of 2
′′
.1. All observations with the ESO1.52m telescope were performed at parallactic an-gle.
3.2.2. McDonald Observatory 2.1m and 2.7m
telescopes
The telescope used for most northern hemi-sphere observations was the 2.1m Otto Struvetelescope at McDonald Observatory. The ES2spectrograph, which has a similar design tothe Boller & Chivens spectrograph, used grat-ing #22 during the 1994 run, giving a disper-sion of 112A mm−1 (≈4A FWHM resolution).During the 1995 to 1999 runs, grating #4 with222A mm−1 (≈8A FWHM resolution) was used.The slit widths for the 1995 to 1999 observing runswere either 1
′′
.6 or 2′′
.0 and the slit orientationwas in E-W direction for all observations.
A number of sources were observed with theLarge Cassegrain Spectrograph (LCS) on the 2.7mHarlan-Smith Telescope at McDonald Observa-tory. Several of these sources are from an over-lapping program of studies of PG quasars (Willset al. (2000); Shang et al. (2003)). The slit widthof the PG quasars was 1
′′
and 2′′
for all other ob-jects observed with the 2.7m telescope. The slitorientation was in E-W direction. The instrumen-tal resolution was ≈7A(Shang et al. (2003)) for
3
the PG quasars and 14 for the other sources. Oneof the PG quasars, PG 1626+554, was observedwith the Hubble Space Telescope (HST) using theFaint Object Spectrograph (FOS) with a 0.86
′′
cir-cular width.
3.2.3. CTIO and Tautenburg
A few objects were observed with the 4mBlanco telescope at the Cerro Tololo InternationalObservatory in Chile (CTIO4.0 in Table 1). Theinstrument used was the R-C spectrograph withthe KPGL3 grating having a dispersion of 116A mm−1 (4.3A FWHM resolution) with a slitwidth of 2
′′
. The slit was oriented in E-W direc-tion.
Four sources were observed with the 2.0m tele-scope of the Thuringer Landessternwarte (TLS)Tautenburg, Germany, (TLS2.0 in Table 1) duringthe test stage of the Nasmyth Focal Reducer Spec-trograph (NFRS) which was constructed and builtat TLS. The V200 grism, with 300 grooves mm−1
corresponding to a dispersion of 225 A mm−1 or3.38 A px−1, was used. The grism provides awavelength coverage from 4 500 to 9 000 A. Theslit width of 1′′ for B2 1128+31 and 2′′for theother objects yields a resolution of 7A or 14A,respectively. Slit orientation was fixed (N-S direc-tion).
4. Data reduction
All spectra, except for the ones taken at Taut-enburg (see below), were bias and flat-field cor-rected and were wavelength- and flux-calibratedby taking spectra of calibration lamps and fluxcalibration standard stars. All data were reduced,except for the McDonald 2.7m and CTIO data,with ESO’s MIDAS data reduction package. TheMcDonald 2.7m and CTIO data were reduced us-ing IRAF. The 1D-spectra were extracted from thetwo-dimensional spectra using the optimal extrac-tion algorithm as described by Horne (1986).
The data taken at TLS Tautenburg requiredspecial treatment because no calibration files wereobtained. The bias was estimated from unexposedparts of the CCD. No flatfield correction was ap-plied.
The TLS Tautenburg is located about 10 kmnorth-east of the city of Jena. Usually, light pol-lution from cities can severely harm astronomical
observations. However, in our case the illumina-tion of the sky by Jena’s street lamps gave usa perfect night sky wavelength calibration spec-trum. Osterbrock & Martel (1992) describe howthe light pollution at Lick Observatory can be usedfor wavelength calibration. We identified the emis-sion lines of the street lamps of Jena in the ob-served spectra and used these for the wavelengthcalibration.
The flux calibration was more challenging, be-cause no standard star was observed. However, bychance a star was in the slit of three exposures atα2000= 10h 40m 01.0s, δ2000 = 21◦08
′
34′′
. Fromthe colors derived from the Automatic Plate Mea-suring (APM) scans of the Palomar ObservatorySky Survey plates (POSS) and 2 Micron All-SkySurvey (2MASS) we could identify the star as aK3V star. The known spectral shape of a K3V starallowed us to correct for the detector/telescope re-sponse and atmospheric transmission. The fluxdensity was then determined from the APM mag-nitudes of the K3V star, using the K3V star RXJ1320.7+0701 (Thomas et al. (1998)) as a refer-ence.
Figure 1 shows the spectrum of RX J1304.2+0205taken with the 2.7m at McDonald Observatoryand the TLS Tautenburg. The figure demon-strates how well the calibration of the Tautenburgdata matches the McDonald Observatory data.
5. Line measurements
5.1. FeII subtraction
In general, Seyfert 1 galaxies show FeII emis-sion in their optical spectra. The strength ofthe FeII emission is correlated with the softnessof the X-ray spectrum and anti-correlated withthe strength of the [OIII] emission (e.g. Boroson& Green (1992); Boroson (2002); Grupe et al.(1999b). In order to accurately measure the [OIII]and Hβ emission lines, the FeII emission must beremoved from the spectrum. This is especiallyimportant for sources with weak [OIII] emission,which are the ones with the strongest FeII emis-sion.
We adopt the method described by Boroson &Green (1992), using the FeII template of I Zw1given in their paper for the wavelength range ∼4400-6000A. In order to correct also for the bluerpart of the spectrum, this template was extended
4
towards shorter wavelengths using the relativeFeII line intensities given by Phillips (1978a,b).The whole template was wavelength-shifted ac-cording to the redshift of the object’s spectrumand the individual FeII lines were broadened to theFWHM of the broad Hβ line by using a Gaussianfilter. The template was scaled by eye to matchthe line intensities of the object spectrum and thensubtracted.
The FeII rest frame equivalent width and fluxwere measured in the rest frame range between4430-4700A(λ4570A blend) from the redshiftedand scaled template. This wavelength range al-lows a direct comparison with the EW(FeII) givenin Boroson & Green (1992). By measuring theFeII flux and equivalent width from the templateinstead of the source spectrum, we avoid contami-nation of the HeIIλ4686 line to the measurements.Note that the FeII/Hβ and EW(FeII) values givenfor the old sample (Grupe et al. (1999b)) werebased on the flux in the entire template between4250A-5880A.
The method of subtracting an FeII template ofthe NLS1 I Zw 1 from the AGN spectrum workswell for most of the sources (e.g. RX J2242.6–3845 or MS 23409–1511). However, in some casesthe FeII subtracted spectrum shows large residu-als. Measurements of the FeII equivalent widthsand fluxes from these sources have to be takenwith caution. This is the case for the followingobjects: Mkn 335, Mkn 1044, RX J1005.7+4332,Mkn 141, Mkn 142, NGC 4051, QSO 1421–0013,and NGC 7214.
The reason for these discrepancies is that thetemplate used was derived from one galaxy (I Zw1) which is likely to have different physical con-ditions (temperature, electron density, ionizationspectrum). The strengths of the different FeIIatomic transitions depend strongly on these con-ditions (e.g. Sigut & Pradhan (2003) and Verneret al. (2003)). If the conditions in the BLR ofour sources deviate from those in I Zw 1, we canexpect that the template will not completely workto subtract and describe the FeII emission in thissource.
5.2. Emission line and continuum param-eters
Table 2 summarizes the redshifts, the FWHM,rest frame equivalent widths (EW), and the[OIII]/Hβ and FeII/Hβ line ratios measured fromthe optical spectra. The FWHM of the Hβ and[OIII]λ5007 lines were measured directly from thelines. For the Hβ line, FWHM, EW and fluxwere measured in the broad component after sub-tracting a narrow component. This narrow com-ponent was constructed from an appropriatelyscaled template of the [OIII]λ5007 line (Grupeet al. (1998b, 1999b)). All FWHM are givenin the rest frame and are corrected for instru-mental broadening assuming that FWHMtrue =√
FWHM2obs − FWHM2
instr. Due to the differ-ing weather conditions for the observations, no linefluxes are given, only line ratios. Note that the[OIII]/Hβ line ratio is the flux of the [OIII]λ5007line to the broad component of Hβ and not thenarrow component, as used in the diagrams ofVeilleux & Osterbrock (1987).
Errors in the FWHM are typically ≈ 10% forthe Hβ line, but are larger for the [OIII] line dueto the contamination of FeII and uncertainties incorrections for observational and instrumental res-olution. The errors in the equivalent widths arecritically dependent on determination of the con-tinuum and can be as much as 25%, and in somecases where the FeII subtraction was less satisfac-tory they can be even larger. Another source ofuncertainty in the equivalent width is in the con-tribution of star light from the host galaxy. Thisis more important for the low-luminosity than forthe high-luminosity AGN. Additional errors of theHβ FWHM and EW are the uncertainties of thesubtraction of a narrow Hβ component. Whileit is quite easy to subtract a narrow Hβ compo-nent in Seyfert 1.5 galaxies, such as Mkn 1048,it is more complicated in NLS1s. Adjusting the[OIII]λ 5007 template by eye usually results ina [OIII]/Hβnarrow ratio of about 3 while this ra-tio is typically 10-15 in Seyfert 1.5s (e.g. Co-hen (1983)). We therefore measured the Hβ linetwice, once with an [OIII]/Hβnarrow ratio deter-mined from the template adjusted by eye and sec-ond from a [OIII]/Hβnarrow=10 ratio of the tem-plate. This was used to estimate the error in theHβ FWHM and EW. The optical continuum andline luminosities for an individual source can be
5
in error by a factor of two to three due to non-photometric weather, bad seeing, and slit losses.
Table 3 lists the soft X-ray 0.2-2.0 keV spec-tral index αX (Grupe et al. (2001a)), the restframe 0.2-2.0 X-ray luminosity LX, the opticalmonochromatic luminosity at 5100A, λL5100, thebolometric luminosity Lbol, and the soft X-rayshort-term variability parameter χ2/ν (Grupe etal. (2001a)). The bolometric luminosity was es-timated from a combined powerlaw model fit withexponential cutoff to the optical-UV data and apowerlaw with absorption due to neutral elements,to the soft X-ray data (see Figure 2). Note that be-cause the EUV part of the spectral energy distri-bution of AGN is unobservable, there are large un-certainties in the bolometric luminosity (e.g. Elviset al. (1994)) and the values given here are onlyapproximate.
6. Results
We classified all objects with FWHM(Hβ)≤2000km s−1 as Narrow Line Seyfert 1s (NLS1s) and allsources with FWHM(Hβ)>2000 km s−1 as BroadLine Seyfert 1s (BLS1s) following the definition ofOsterbrock & Pogge (1985) and Goodrich (1989)regardless of subgroups such as Sy 1.5s. This re-sults in 51 NLS1s and 59 BLS1s.
6.1. Simple Statistics
Table 4 summarizes the mean, standard devia-tions, medians of the FWHM of Hβ and [OIII],the rest frame equivalent widths of Hβ, [OIII] andFeII, the [OIII]/Hβ and FeII/Hβ flux ratios, the X-ray slopes αX, the rest frame 0.2-2.0 keV X-ray lu-minosities, the 5100A rest frame monochromaticluminosities, the bolometric luminosities, and theredshifts of the whole sample of 110 AGN, the 51NLS1s, and the 59 BLS1s. NLS1s have in gen-eral the steepest X-ray spectra and strongest FeIIemission, shown both in the equivalent width ofFeII as well as in the FeII/Hβ line ratio. Thereare no differences between NLS1s and BLS1s withrespect to their mean equivalent width of Hβ andluminosities in X-rays and at 5100A. Kolmogorov-Smirnov KS tests for these properties show thatthe distributions are similar.
Figure 3 displays the distributions of the FWHMof Hβ and [OIII]. Due to their definition, theyhave different distributions of their FWHM(Hβ).
NLS1s and BLS1s have similar distribution intheir FWHM([OIII]). Figure 4 shows the distribu-tions of the equivalent widths of Hβ, [OIII], andFeII. There are no differences in the distributionsof the EW(Hβ) (Figure 4a), but the EW(FeII)distributions (Figure 4c) are different at a levelP >99.99%. NLS1s dominate the group withsmall values of the EW([OIII]). KS tests indicatethat the EW([OIII]) distributions (Figure 4b) ofNLS1s and BLS1s are only marginally differentwith a probability of 99.1%
Figure 5 displays the distributions of the X-rayspectral index in the ROSAT PSPC energy range.NLS1s have, as expected, the steepest X-ray spec-tra while BLS1s show flatter X-ray spectra. Thereis a >99.99% probability that the distributions aredifferent.
The distributions in the rest frame 0.2-2.0 X-ray luminosities and the bolometric luminositiesare shown in Figure 6. There are no significantdifferences in any of the luminosity distributionsbetween NLS1s and BLS1s. Figure 7 displays theredshift distributions of the two samples. A KS-test shows that the distributions are similar.
Figure 8 displays the distributions of the [OIII]and FeII luminosities. A KS test shows thatNLS1s and BLS1s have different distributions intheir FeII luminosity (P >99.99%), but show sim-ilar distributions in their [OIII] luminosity. Fig-ure 9 shows the distributions of the [OIII]/Hβ andFeII/Hβ flux ratios. While a KS test shows thatthe distributions of FeII/Hβ of NLS1s and BLS1sare clearly different, there is a 3% chance that the[OIII]/Hβ distributions are similar.
Figure 10 shows the distributions of the short-term variability parameter χ2/ν of the RASS ob-servations (Grupe et al. (2001a)). A KS testshows that the distributions are different with aprobability of P>99.99%.
6.2. Spectra
Figure 11 displays all optical spectra in the ob-served frame that have not been published yet.For many of the sources, optical spectra are shownhere for the first time. In other cases, spectra havebeen published before as listed in Table 1, but thespectra shown here are either of better quality orshow a wider wavelength coverage. The spectrapreviously published by Grupe et al. (1999b) can
6
be accesses electronically at the CDS anonymousFTP site at ftp 130.79.128.5.
Figure 12 displays the FeII-subtracted spectra,the template used and the original spectra beforetemplate subtraction. The FeII subtracted spectraare shown with their calibrated fluxes. The FeIItemplate and the original spectra are offset.
Please note: this is a special version forastro-ph that does not contain the optical andFeII subtracted spectra due to disk space limita-tions. The complete paper including the spectracan be retrievd from http://www.astronomy.ohio-state.edu/∼dgrupe/research/sample paper1.html
7. Discussion
7.1. Statistics
The main aims of this paper are to present thespectra and a simple statistical analysis of oursample of 110 soft X-ray selected AGN. About halfof the sources of our complete sample of soft X-rayselected AGN turn out to be NLS1s (51). In a sec-ond paper (Grupe (2003)) a Principal ComponentAnalysis (PCA) will be shown. The NLS1s showthe well-known Boroson & Green (1992) Eigen-vector 1 relation between [OIII] and FeII. It is nota complete surprise that about half of our sourcesare NLS1s, because NLS1s do have the steepest X-ray spectra among AGN (e.g. Boller et al. (1996);Grupe et al. (1999b, 2001a); Grupe (2000)) andthe best way to find them is through a soft X-raysurvey.
It is surprising that there is no significant differ-ence between the rest frame 0.2-2.0 keV X-ray and5100A, and bolometric luminosities for NLS1s andBLS1s. Even though the sources are variable in X-rays (Grupe et al. (2001a)) the rest frame 0.2-2.0keV X-ray luminosity is a better measure of poweroutput of the nucleus than the optical luminosityhere because of the different weather conditions,it is impossible to get absolute flux calibratedspectra if the conditions were not photometric.The other reason is that most of the spectra havebeen taken with 2
′′
slit widths, which means thatthe contribution of galactic starlight can be sig-nificant, especially for the low-luminosity AGN.The contribution to the total X-ray luminosityof sources such as high-luminous X-ray binariesof supernova remnants, is on the order of about1031 − 1033 W, and is therefore negligible com-
pared with the X-ray power of the nucleus.
The usual interpretation of the steep X-rayspectra of NLS1s is that these sources accreteclose to their Eddington limit (e.g. Pounds et al.(1995)). If this is true and the distributions in lu-minosity of all types of sources in our sample arethe same we can interpret this result to imply thatin general NLS1s have smaller central black holemasses than BLS1s as suggested by Boller et al.(1996) and Wandel & Boller (1998) and shown forsome NLS1s by e.g. Onken et al. (2003). This as-sumption applies also for our sample. Grupe et al.(2003) found, that for the soft X-ray selected sam-ple presented here, for a given luminosity, NLS1shave smaller black hole masses than BLS1s whenthe black hole masses were determined by the re-lations given in Kaspi et al. (2000). It will alsobe shown in that paper, that NLS1s show a differ-ent MBH − σ relation (Magorrian et al. (1998);Tremaine et al. (2003)) than non-active galaxiesand broad-line Seyferts.
7.2. [OIII] and FeII strengths
There are two interesting things about the dis-tributions of [OIII] and FeII strengths: a) for theluminosity distributions NLS1 and BLS1s havesimilar FeII luminosity distributions but different[OIII] luminosity distributions, and b) surprisinglythe [OIII]/Hβ distributions are only slightly dif-ferent with a 3% change of being similar, but theFeII/Hβ distributions are clearly different. Fromthe Boroson & Green (1992) Eigenvector 1 re-lation, which we also see in our sample (PaperII), one would expect that also the FeII luminosityand [OIII]/Hβ distributions of NLS1s and BLS1swould be different. On the other hand, NLS1s andand BLS1s do show slightly different distributionsof their [OIII] equivalent widths and clearly dif-ferent distributions of their FeIIλ4560 equivalentwidths.
One has to keep in mind that due to the slitwidths of about 2
′′
, a contribution from circum-nuclear HII regions cannot be excluded. The con-tribution of hydrogen absorption lines of a mediumage stellar population can also be a problem. Thetypical contribution of this component is on theorder of 2-3 A in the equivalent widths of theBalmer lines (e.g. Shields & Searle (1978); Mc-Call et al. (1985); Ho et al. (1997); Popescu &Hopp (2000); Meusinger & Brunzendorf (2002).
7
The Hβ equivalent widths of the sources of oursample are much larger than this value (Tab 2 and4) and the effect of stellar absorption lines can beneglected.
8. Conclusions
We have presented the optical data of a com-plete sample of 110 soft X-ray selected AGN. Wefound that
• About half of the sources are NLS1s basedon their FWHM(Hβ)≤2000 km s−1.
• NLS1s and BLS1s show clearly different dis-tributions of their αX, rest frame equivalentwidths of FeII, their FeII/Hβ ratios, FeII lu-minosities, and soft X-ray variability. KStests also suggest slightly different distribu-tions of [OIII]/Hβ and EW([OIII]).
• NLS1s and BLS1s show similar distributionsin their redshifts, continuum luminosities,and equivalent widths of Hβ
• The similar continuum luminosities in bothsub-samples and the high accretion to massratios in NLS1s suggests that these havesmaller black hole masses for a given lumi-nosity than BLS1s.
Correlation analysis of the sample including aPrincipal Component Analysis will be presented ina second paper (Grupe (2003)). The black holemasses and their relation to the Magorrian et al.(1998) and Tremaine et al. (2003) Mbh − σrelation will be presented in Grupe et al. (2003).
We would like to thank Marianne Vestergaardfor comments and suggestions on the manuscript,the anonymous referee for a fast referee reportthat helped improving the paper, Hans-ChristophThomas for providing the spectrum of the K3Vstar RX J13020.7+0701, and Brad Peterson forproviding a spectrum of NGC 4593. We also wantto thank the night assistants and technical peopleat La Silla and McDonald observatory, namely,Hector Vega, Dave Doss, Jerry Martin and EarlGreen. Without their support this project wouldnot have been possible. This research has madeuse of the NASA/IPAC Extra-galactic Database
(NED) which is operated by the Jet Propul-sion Laboratory, Caltech, under contract withthe National Aeronautics and Space Administra-tion. The ROSAT project is supported by theBundesministerium fur Bildung und Forschung(BMBF/DLR) and the Max-Planck-Society.
REFERENCES
Beuermann, K., Thomas, H.-C., Reinsch, K., etal., 1999, A&A, 347, 47
Grupe, D., 2000, New Astronomy Reviews Vol.44, proceedings of the workshop on ’Observa-tional and Theoretical Progress in the Study ofNarrow-Line Seyfert 1 Galaxies, Eds Th. Boller,W.N. Brandt, K.M. Leighly, and M.J. Ward,p455
Grupe, D., 2003, AJ, to be submitted (Paper II)
Grupe, D., Beuermann, K., Mannheim, K., et al.,1995a, A&A, 299, L5
Grupe, D., Beuermann, K., Mannheim, K.,Thomas, H.-C., Fink, H.H., & de Martino, D.,1995b, A&A, 300, L21
Grupe, D., Beuermann, K., Thomas, H.-C., &Fink, H.H., 1998a, A&A, 330, 25
Grupe, D., Wills, B.J., Wills, D., Beuermann, K.,1998b, A&A, 333, 827
Grupe, D., Thomas, H.-C., & Leighly, K.M.,1999a, A&A, 350, L31
Grupe, D., Beuermann, K., Mannheim, K., &Thomas, H.-C., 1999b, A&A, 350, 805
Grupe, D., Thomas, H.-C., & Beuermann, K.,2001a, A&A, 367, 470
Grupe, D., Thomas, H.-C., Leighly, K.M., 2001b,A&A, 369, 450
Wills, B.J., Wills, D., Evans, N.J., Natta, A.,Thompson, K.L., Breger, M., & Sitko, M.L.,1992, ApJ, 400, 96
Wills, B.J., Netzer, H., & Wills, D., 1985, ApJ,288, 94
Wills, B.J., Shang, Z., & Yuan, J.M., 2000, NewAstronomy Reviews Vol. 44, proceedings of theworkshop on ’Observational and TheoreticalProgress in the Study of Narrow-Line Seyfert 1Galaxies, Eds Th. Boller, W.N. Brandt, K.M.Leighly, and M.J. Ward, p511
This 2-column preprint was prepared with the AAS LATEXmacros v5.0.
10
Fig. 1.— Combined spectrum of spectra of RXJ1304.2+0205 taken at McDonald Observatory(black) and the Thuringer Landessternwarte Taut-enburg (grey)
Fig. 2.— Powerlaw with exponential cutoff andpowerlaw with neutral absorption (solid line) tothe optical/UV (Shang et al. (2003)) and X-rayspectrum of PG 1115+407 as an example how thebolometric luminosity was determined. The dot-ted lines display the powerlaw with exponentialcutoff for the optical/UV part and the powerlawwith neutral absorption for the soft X-ray part sep-arately.
Fig. 5.— Distribution of the X-ray spectral slopeαX. The lines are the same as described in Fig-ure 3.
Fig. 7.— Distribution of the redshift z. The linesare the same as described in Figure 3.
11
Fig. 3.— Distribution of the FWHM of Hβ and [OIII] for the whole sample (dotted line), NLS1s (solid line),and BLS1s (short dashed line). FWHM have been corrected for instrumental resolution.
Fig. 4.— Distribution of the rest frame Equivalent widths of Hβ, [OIII], and FeII. The lines are the same asdescribed in Figure 3.
12
Fig. 6.— Distribution of the rest frame 0.2-2.0 keV X-ray luminosity LX and the bolometric luminosity Lbol.The lines are as described in Figure 3.
Fig. 8.— Distribution of the [OIII] (left) and FeII luminosities (right). The lines are the same as describedin Figure 3.
Fig. 9.— Distribution of the [OIII]/Hβ and FeII/Hβ flux ratios. The lines are the same as described inFigure 3.
13
Table 1
Summary of the optical spectroscopy of the soft X-ray selected sample
No. α2000 δ2000 Name UT date Tel. Instrument Texp comments
1 00 06 19.5 +20 12 11 Mkn 335 1999/09/13 ESO1.52 B&C 302 00 25 00.2 −45 29 34 ESO 242−G8 1993/09/14 ESO2.2 EFOSC 1,8,10 5,30,30 Grupe et al. (1999b)3 00 57 20.2 −22 22 57 Ton S 180 1992/10/17 ESO2.2 EFOSC 1,8,10 5,30,30 Grupe et al. (1999b)4 00 58 37.4 −36.06.05 QSO 0056−36 1993/10/12 ESO2.2 EFOSC 1,8,10 10,25,20 Grupe et al. (1999b)5 01 00 27.1 −51 13 54 RX J0100.4−5113 1993/10/12 ESO2.2 EFOSC 1,8,10 10,40,35 Grupe et al. (1999b)
Fig. 10.— Distribution of the X-ray variabilityparameter χ2/ν. Outside the plot are NGC 4051,Mkn 766, and RX J1304.2+0205 with χ2/ν=18.3,8.86, and 5.08, respectively. The lines are the sameas described in Figure 3.
15
Table 1—Continued
No. α2000 δ2000 Name UT date Tel. Instrument Texp comments
Fig. 11.— Optical spectra of the soft X-ray selected sample which have not been published yet (see Table1). The wavelength is given in units of A and the flux density is given in units of 10−19 W m−2 A−1. Allspectra are shown in the observed frame. Please note: this is a special version for astro-ph that does notcontain the optical and FeII subtracted spectra. The complete paper including the spectra can be retrievdfrom http://www.astronomy.ohio-state.edu/∼dgrupe/research/sample paper1.html
Fig. 12.— FeII-subtracted spectra, the wavelength is given in units of A and the flux density is givenin units of 10−19 W m−2 A−1. The dotted line marks the zero line of the FeII template. Pleasenote: this is a special version for astro-ph that does not contain the optical and FeII subtractedspectra. The complete paper including the spectra can be retrievd from http://www.astronomy.ohio-state.edu/∼dgrupe/research/sample paper1.html
20
Table 4
Mean, Standard deviation, and median of the whole sample (110 sources), NLS1s (51), andBLS1s (59).
all sources (110) NLS1s (51) BLS1s1 (59)PropertyMean σ Median Mean σ Median Mean σ Median