Emission and Absorption Properties of Low-Mass Type 2 Active Galaxies with XMM-Newton
aa r X i v : . [ a s t r o - ph . C O ] O c t Accepted for publication in ApJ
Emission and Absorption Properties of Low-Mass Type 2 Active Galaxieswith
XMM-NEWTON Carol E. Thornton , Aaron J. Barth , Luis C. Ho , and Jenny E. Greene , . ABSTRACT
We present
XMM-Newton observations of four low-redshift Seyfert galaxies se-lected to have low host luminosities ( M g > -
20 mag) and small stellar velocity disper-sions ( σ ∗ <
45 km s - ), which are among the smallest stellar velocity dispersions foundin any active galaxies. These galaxies show weak or no broad optical emission linesand have likely black hole masses . M ⊙ . Three out of four objects were detectedwith > σ significance in ∼
25 ks exposures and two observations had high enoughsignal-to-noise ratios for rudimentary spectral analysis. We calculate hardness ratios( - .
43 to 0 .
01) for the three detected objects and use them to estimate photon indicesin the range of Γ = 1 . - .
8. Relative to [O
III ], the type 2 objects are X-ray faint incomparison with Seyfert 1 galaxies, suggesting that the central engines are obscured.We estimate the intrinsic absorption of each object under the assumption that the [O
III ]emission line luminosities are correlated with the unabsorbed X-ray luminosity. Theresults are consistent with moderate ( N H ∼ cm - ) absorption over the Galacticvalues in three of the four objects, which might explain the non-detection of broad-lineemission in optical spectra. One object in our sample, SDSS J110912.40+612346.7,is a near identical type 2 counterpart of the late-type Seyfert 1 galaxy NGC 4395.While the two objects have very similar [O III ] luminosities, the type 2 object has anX-ray/[O
III ] flux ratio nearly an order of magnitude lower than NGC 4395. The mostplausible explanation for this difference is absorption of the primary X-ray continuum Based on observations obtained with XMM-Newton, an ESA science mission with instruments and contributionsdirectly funded by ESA Member States and NASA. Department of Physics & Astronomy, 4129 Frederick Reines Hall, University of California, Irvine, Irvine, CA92619-4575; [email protected] The Observatories of the Carnegie Institution for Science, 813 Santa Barbara Street, Pasadena, CA 91101. Department of Astrophysical Sciences, Princeton University, Princeton, NJ 08544 Hubble Fellow and Princeton-Carnegie Fellow L bol ∼ - erg s - . Subject headings: galaxies: active — galaxies: dwarf — galaxies: nuclei — galaxies:Seyfert — X-rays: galaxies
1. Introduction
Due to the advent of large area surveys in the past few years, extensive progress has beenmade in the search for low-mass active galactic nuclei (AGNs) with estimated black hole masses M BH . M ⊙ . Through searches for galaxies with low stellar velocity dispersions or weak broad-line emission in the Sloan Digital Sky Survey (SDSS), the number of candidate broad-line type1 (Greene & Ho 2004, 2007b) and narrow-line type 2 (Barth et al. 2008) low-mass AGNs hasincreased to number in the hundreds. Multi-wavelength studies (e.g. Gallo et al. 2008, Satyapal etal. 2008) have begun to provide new avenues for finding these low-mass and low-luminosity AGNsthat would not typically be identified in the optical, allowing us to observe a larger portion of thetotal energy output. X-ray observations are of key importance for detecting the primary ionizingcontinuum of the AGN and determining the total luminosity as well as constraining the obscurationtoward the central engine.Current unification schemes explain the observational differences between type 1 and type2 AGNs by the viewing angle from which we observe the central engine. For a type 2 ob-ject, an obscuring torus blocks the light coming from the innermost region, which contains thebroad emission-line and X-ray-emitting regions. Therefore in the classical AGN unification pic-ture (Antonucci & Miller 1985), a type 2 object will show the underlying properties of a type 1object if the effects of the obscuration are removed. Hence, type 1 and 2 objects with comparableblack hole masses and luminosities should also show similar optical narrow emission-line spectra,since the narrow-line emitting region remains largely unobscured. By systematically searchingfor galaxies of both types 1 and 2 with similar stellar velocity dispersions and narrow emission-line luminosities, one can develop comparable samples of type 1 and type 2 objects with similarblack hole masses to test if the unified model is applicable across the full range of black holemasses found in AGNs, or if the lack of broad emission lines in low-mass type 2 objects resultsfrom changes in the structure of AGNs at low bolometric luminosities (Nicastro 2000; Laor 2003;Elitzur & Shlosman 2006) rather than due to absorption.The two best studied AGNs with M BH < M ⊙ are located in the late-type spiral NGC 4395(Filippenko & Sargent 1989) and in the dwarf elliptical POX 52 (Kunth et al. 1987; Barth et al. 3 –2004), respectively. NGC 4395 has been shown to vary rapidly in the X-ray (Iwasawa et al. 2000;Shih et al. 2003; Moran et al. 2005), including dramatic changes in spectral slope ( Γ ≈ . - . < XMM-Newton and
Chandra have been used to investigate the X-ray properties of low-mass type 1 AGNsamples (Greene & Ho 2007a; Desroches et al. 2009; Miniutti et al. 2009) showing that they seemto be scaled down versions of their more massive counterparts with similar hardness ratios andphoton indices. The X-ray properties of the type 2 counterparts of these AGNs have not previouslybeen studied systematically.We present
XMM-Newton observations of four galaxies selected from the low-mass Seyfert 2sample of Barth et al. (2008) to have the lowest stellar velocity dispersions in the sample in orderto quantify the absorption and emission properties of this population. Our goal is to investigatewhether obscuration can explain the lack of broad-line emission in low-mass Seyfert 2 galaxies.From measurements of X-ray luminosities and spectral fitting, we estimate intrinsic absorbing col-umn densities, bolometric luminosities, and corresponding Eddington ratios in order to investigatethe differences between type 1 and type 2 objects in this mass range. Throughout this paper weassume H = 70 km s - Mpc - , Ω m = 0.3, and Ω Λ = 0 .
7. All estimates of Galactic foreground col-umn densities are calculated based on Galactic H I maps (Dickey & Lockman 1990; Kalberla et al.2005), using the HEASARC online N H calculator .
2. Sample Selection
Barth et al. (2008, hereafter BGH08) searched SDSS for nearby low-mass active galaxies withabsolute magnitudes fainter than M g = -
20 mag and emission-line ratios consistent with a Seyfert2 classification (Ho et al. 1997; Kauffmann et al. 2003; Kewley et al. 2006). They obtained stellarvelocity dispersions from high-resolution Keck spectra and found 39 < σ ∗ <
95 km s - for thesample of 29 galaxies. These type 2 galaxies have a similar range of [O III ] line luminosities andstellar velocity dispersions as the sample of type 1 objects found by Greene & Ho (2004, 2007b).We selected the four objects from the BGH08 sample with the lowest stellar velocity dispersionsand therefore the lowest estimated black hole masses for X-ray observations.
SDSS J011905.14+003745.0 ( z = 0 . σ ∗ = 39 ±
8, in the BGH08 sample. Optical spectropolarimetry data obtainedby BGH08 found no polarized emission lines or a polarized continuum. http://heasarc.gsfc.nasa.gov/cgi-bin/Tools/w3nh/w3nh.pl SDSS J103234.85+650227.9 ( z = 0 . α emission. Soft X-ray emission from this objecthas been previously detected by ROSAT (Boller et al. 1992; Moran et al. 1996). Seth et al. (2008)also note it as an example of a galaxy with both an AGN and a nuclear star cluster. BGH08 didnot list a stellar mass for this galaxy, so using the prescription of Bell et al. (2003) and the SDSScatalogue Petrosian magnitudes ( u = 14 . , r = 12 .
88 and z = 12 . M ⋆ / M ⊙ ) ≈ . SDSS J110912.40+612346.7 ( z = 0 . M ∗ / M ⊙ ) = 8 . M g = - . u = 16 . , r = 15 .
21 and z = 15 . M ⋆ / M ⊙ ) ≈ .
69. BGH08 were unable to measure σ ∗ for this galaxy, but they demonstrate that the well-established correlation between σ ∗ and [O III ]linewidth for Seyfert 2 galaxies (Nelson & Whittle 1996) holds for galaxies with stellar velocitydispersions as low as σ ∗ ∼ -
80 km s - . From the measured FWHM([O III ]) = 66 ± - ,this correlation suggests a stellar velocity dispersion of σ ∗ ∼ FWHM([O
III ]) / .
35 = 28 km s - .BGH08 also obtained spectropolarimetry data for this object, finding a significant polarized con-tinuum component, but no polarized emission-lines from a hidden broad-line region. In the sameobservation, the blue spectrum shows [Ne V ] λ SDSS J144012.70+024743.5 ( z = 0 . ROSAT
All-Sky Survey and a high-resolution Keck spectrum showsevidence for possible, but very weak broad H α emission (BGH08). Its spectrum is also very similarto those of NGC 4395 and POX 52, with similar narrow-line ratios and detected high-ionizationlines, but much weaker broad H α emission. We again use the prescription of Bell et al. (2003) andthe SDSS catalogue Petrosian magnitudes ( u = 18 . , r = 16 .
40 and z = 15 .
91) to estimate a stellarmass of log ( M ⋆ / M ⊙ ) ≈ .
85 for this object. We note that the AGN light is relatively bright inthis object and therefore our stellar mass estimate should reflect an upper limit to the actual stellarmass of the galaxy. 5 –
3. Observations and Data Reduction
Each object was observed using the EPIC instrument on
XMM-Newton for ∼
25 ks. Due tothe low signal-to-noise (S/N) of our observations, we focus our attention on the data taken withthe EPIC-pn instrument, which has a higher quantum efficiency than the EPIC-MOS instruments.Observation dates and actual exposure times (with soft proton flares excluded) for each object canbe found in Table 1. At most, 32 .
5% of the exposure time was lost due to soft proton flaringin SDSS J011905, 25 .
3% was lost in SDSS J144012, 19 .
4% was lost in SDSS J103234, and notime was lost in SDSS J110912. All data were reduced using the Science Analysis System (SAS,version 7.1.0) and XSPEC (version 12.4.0aa) following the guidelines of the SAS Cookbook andSAS ABC-Guide. Each object was extracted in a circular region with a radius of 30 ′′ . Due tothe proximity of all four objects to the edge of their respective chips, background regions freeof sources were used with radii of 45 ′′ , located on the same chip an average of 107 ′′ away fromthe source. Only events corresponding to patterns 0 - -
12 (single, double, triple and quadruple pixel events) in order to maximize the S/N. In additionto pattern filtering, events were excluded that occurred next to the edges of the CCD or next to badpixels, and all event files were restricted to an energy range of 0 . - .
4. Results
No counts above background are detected within the extraction region of SDSS J011905.Assuming Possion statistics we estimate a 3 σ upper limit of < . > σ significance in the 0 . -
10 keV energy range, with the net counts in therange of 27 -
163 for each detected object.
We use the SAS task calview to determine the EPIC-pn point spread function (PSF) at the timeand CCD chip location of each of the three detected objects. For each object, we plot the encircledenergy curve for the
XMM-Newton
PSF along with the fraction of encircled energy as a functionof radius (Figure 1). The radius that encircles 50% of the flux from a point source is ∼ ′′ , whichtranslates to a physical diameter of ∼ ∼
12 6 –Fig. 1.— Fraction of encircled energy as a function of radius. The solid line shows the encircledenergy curve for the
XMM-Newton
PSF. The data points are normalized such that 87% of the totalenergy is encircled at a radius of R = 30 ′′ . 7 –kpc for SDSS J144012. This covers a considerable portion of the disk of SDSS J103234, morethan half of the disk of SDSS J110912, and the entirety of SDSS J144012. We see no evidencefor extended emission outside of the physical extent of the PSF, except in object SDSS J110912where we see a strong mismatch between the fractional encircled energy of the object and what isexpected from a point source. The slope of the data is much steeper than that expected from thePSF, suggesting that we see only extended emission and not a PSF-dominated source. This maybe due to high obscuration of the central engine and a few X-ray binaries within the source regioncontaminating our detection of the central engine. Separating the events by energy, we investigate the hard ( C H , . - . C S , . - . C H - C S ] / [ C H + C S ]). SDSS J144012 showedthe softest spectrum in the sample, with HR = - . ± .
05. The other two objects, SDSS J103234and SDSS J110912, each had similar hard and soft count rates, resulting in HR = 0 . ± .
06 andHR = - . ± .
16, respectively.We use the
XMM-Newton response matrices, auxiliary response file (ARF) and redistributionmatrix file (RMF), to create model spectra in order to calculate a photon index ( Γ HR ) from the HRfollowing the method of Gallagher et al. (2005). These models assume the neutral absorber is setto the Galactic value in the direction of the object and the AGN is described by a simple power law.We caution that the intrinsic slope of the power-law continuum may not be well described by thissimplistic model if the spectrum contains more complex components, such as a high level of ab-sorption or a soft excess due to a thermal component. We include this analysis as a simple indicatorof spectral slope, particularly for those objects where detailed spectral modeling is not feasible dueto low S/N. SDSS J144012 has a photon index ( Γ HR = 1 . ± .
1) that is similar to many Seyfert1 galaxies, including POX 52, which had Γ HR = 1 . Chandra (Thornton et al. 2008). X-ray surveys of unobscured AGNs have found average power-law slopes of Γ = 1 . Γ HR = 1 . ± . Γ HR = 1 . ± .
2. All count rates, HR, and Γ HR can be found in Table 2. We estimate the 2 -
10 keV flux from the same model spectra used to derive the photon indicesfrom the HR. Again, these model spectra assume a neutral absorber set to the Galactic value and 8 –a power law with a slope derived as above and do not account for any additional componentspresent in the spectra. The X-ray luminosities ( L X ) are derived from these fluxes accounting forthe distance of the object and Galactic absorption corrections are negligible. These flux estimatesare consistent with those derived using the energy conversion factors calibrated by Hasinger et al.(2001) and found in the XMM-Newton User’s Handbook . Individual flux and L X estimates can beseen in Table 2. The two brightest objects in the sample, SDSS J103234 and SDSS J144012, have high enoughcount rates to search for temporal variations. Using 500 s time bins, we create light curves (Figure2) for these two objects, excluding any background flares present during the observations. Weattempt to quantify any variability in these sources using the normalized excess variance ( σ ) ofNandra et al. (1997a): σ = 1 N µ N X i =1 [( X i - µ ) - σ i ] , (1)where X i denotes the count rate of the i -th point in the light curve, σ i is its uncertainty, µ is themean of the X i values over the entire light curve, and N is the number of points in the light curve.The excess variance using 500 s time bins for SDSS J103234 and SDSS J144012 is σ nxs = - . ± .
76 and σ nxs = - . ± .
2, respectively. In order to increase the S/N in each datapoint, we enlarge the time bins to 1000 s and calculate an excess variance of σ nxs = 1 . ± σ nxs = 0 . ± . σ nxs isconsistent with zero, meaning that there is no evidence for intrinsic source variability.Previous ROSAT detections of SDSS J103234 and SDSS J144012 as part of the
ROSAT
AllSky Survey suggest a decrease in flux over a ∼
10 year time period. Assuming a photon index inthe range of Γ = 1 - . - . ROSAT is L X = (0 . - . × erg s - and L X = (1 . - . × erg s - for SDSS J103234 and SDSS J144012, respectively. XMM-Newton images of SDSS J103234 andSDSS J144012 show no other objects with similar or larger fluxes within the large (96 ′′ ) PSF ofthe ROSAT
All Sky Survey, so contamination of the
ROSAT luminosities is unlikely. This is upto an order of magnitude larger than the luminosities derived in this work and could be larger if asubstantial amount of absorbing material were present at the time the
ROSAT observations weretaken. Variations of this magnitude have been observed before and are typically explained byvariations in the absorbing material, especially at such soft energies. The Seyfert 2 galaxy NGC4388 was observed to have a factor of ∼
10 increase in flux due to an order of magnitude decreasein the absorbing column density, typically at N H ≈ × cm - (Elvis et al. 2004), over the 9 –Fig. 2.— Light curves of SDSS J103234 ( top panel ) and SDSS J144012 ( bottom panel ) binned by500 s each. 10 –course of a year. Similarly, NGC 4358 has shown order of magnitude variations in the 0 . - ∼
25% variations in the absorbing column density over a one month time period(Fruscione et al. 2005). This object has also seen factors of 2 - SDSS J103234 and SDSS J144012 are the only objects in the sample with high enough S/Nfor spectral fitting, albeit over a limited range of energy. In both objects, the source spectrumbecomes indistinguishable from that of the background at the high-energy end, due to low sourcecounts. We choose to minimize the Cash C statistic (Cash 1979) in order to optimize the spectralfits instead of the often used χ statistic because it does not require a minimum number of countsper bin and the results are independent of the bin size used (for further discussion, see Cash 1979).Therefore, the C statistic is more reliable than the χ statistic when fitting low S/N spectra. Due to low S/N, SDSS J103234 could only be fit over the 0 . - . < . > . ∼ N H = 1 . × cm - for all following model fits. The best fit ( C = 54 .
32 using18 PHA bins and 16 dof) was a Γ = 1 . ± . Γ = - . + . - . . Using thesame absorbed power law with a thermal blackbody model, we tested setting the power-law slope 11 –Fig. 3.— The spectrum of SDSS J103234 from 0 . - . black line ) with ( a ) an ab-sorbed power law with absorption set at the Galactic value, ( b ) an absorbed (Galactic value) powerlaw (photon index set to Γ = 1 .
1) with a thermal disk blackbody, and ( c ) an absorbed (Galacticvalue) power law ( Γ = 1 .
1) with a thermal disk blackbody and a partial covering absorption com-ponent. Residuals are calculated as Residual = (Data - Model)/Model. 12 –to Γ = 1 .
0, increasing this slope by 0 . Γ = 3 .
0, whileallowing the blackbody temperature to vary freely. This allowed us to test more complex modelsover our limited energy range, while still keeping the number of free parameters to a minimum.Using this technique, we found the model fit with the photon index set at Γ = 1 . Γ = 1 .
1, the value derived from the HR, forall further model fits. Fixing the photon index to this value in the absorbed power law and thermalblackbody model improves slightly ( C = 37 .
63 using 18 PHA bins and 15 dof) with a blackbodytemperature of kT = 0 . ± .
05 keV, but still slightly underpredicts the spectrum at energies of > . > . - XMM-Newton spectrum of POX 52 when it wasobserved to be in a partially-covered state (Thornton et al. 2008). With this in mind, we add apartial-covering component to our model and keeping the Galactic absorber fixed, we again testthe incremental photon index values, finding similar results as before. We therefore fixed the pho-ton index to value derived from the HR, Γ = 1 . C = 17 .
93 using 18 PHA binsand 13 dof) from this model includes a kT = 0 .
20 keV blackbody and an additional absorbing col-umn density of N H = 4 . × cm - covering 95% of the X-ray-emitting region. This model ismore complex than the previous models tested, but better fits the flattened region between energies1 - > The spectrum of SDSS J144012 extends from 0 . - . N H = 2 . × cm - ), withthe understanding that the best-fit power law may not accurately describe the intrinsic underlyingcontinuum over the full 0 . - . Γ = 2 . ± . C = 26 .
60 using 13 PHAbins and 11 degrees of freedom, dof). This power law is steeper than what is commonly seen inSeyfert 1 galaxies (Nandra et al. 1997b) and does not fit energies below 0 . . - . black line ) with ( a ) anabsorbed (set at the Galactic value) power law, and ( b ) an absorbed (Galactic value) power lawwith a Raymond-Smith plasma. Residuals are calculated as Residual = (Data - Model)/Model. 14 –tested allowing the column density to vary freely, but this did not improve the fit quality and sowe choose the Galactic value for simplicity. Therefore, the simple absorbed power law is a poormodel for describing the X-ray spectrum of this source.In order to better fit the soft-energy end of the spectrum, we added a thermal component to theabsorbed power law. We first added a blackbody component to the absorbed power law, allowingthe thermal temperature of the blackbody and the photon index of the power law to vary freely,while holding the column density fixed to the Galactic value, for the reasons discussed above.This model was unable to fit the peak in the spectrum, seen at E ∼ . C > z = 0 . C = 18 .
73 using 13 PHA bins and 9 dof; Figure 4) with a power-law photon index of Γ = 1 . + . - . and a plasma temperature of kT = 0 . + . - . keV. This temperature is within the range oftypical values of kT ∼ . - . C = 17 .
94 using 13 PHA bins and 8 dof), with a MEKAL plasma temperature of kT = 0 .
14 keV and a corresponding photon index of Γ = 1 . + . - . . We note that the photon indices ofboth of these models are not well constrained with lower bounds of Γ ≤
0. This is most likely dueto the very small energy range over which we are attempting to fit these models. Both model fits arenearly indistinguishable from each other, except that the MEKAL model uses one additional freeparameter. Therefore, we select the model containing the Raymond-Smith plasma as our best-fitmodel for simplicity.
5. Discussion5.1. L [O III] - L X Correlation
The relationship between the luminosity of the [O
III ] emission line at 5007 Å ( L [O III] ) andthe 2 -
10 keV luminosity has been studied for a range of objects, including type 1 and 2 Seyfertgalaxies and quasars, to determine if L [O III] / L X is similar among all Seyfert galaxies or whether therelationship has any additional dependence on properties such as accretion rate, luminosity, and 15 –Fig. 5.— Plot of L [O III] vs L -
10 keV with our sample ( red asterisks ) along with the objects ( black,filled squares ) from Panessa et al. (2006), which includes quasars and Seyfert 1 and 2 galaxies, allcorrected for X-ray absorption. The open symbols represent NGC 4395 ( green circle ; Panessa etal. 2006, corrected to the updated distance of 4.3 Mpc; Thim et al. 2004), POX 52 ( green triangle ;Barth et al. 2004, Thornton et al. 2008) and low-mass Seyfert 1 objects ( blue squares ; Greene & Ho2004, Desroches et al. 2009). The solid line represents the least-squares fit derived by Panessa et al.(2006) and the two dotted lines represent the 1 σ scatter of the Panessa et al. sample about the fit.Unless otherwise displayed, error bars for our sample are smaller than the plot symbol. 16 –black hole mass. Kraemer et al. (2004) investigated the range of X-ray and [O III ] luminositiesin both broad-line and narrow-line Seyfert 1 galaxies, finding little difference between the twopopulations. Heckman et al. (2005) followed this study with another, in which they included bothSeyfert 1 and 2 galaxies, specifically investigating if this relationship extended to Seyfert 2 galaxiesand whether or not L X needed to be corrected for absorption. They found that if the L X of a Seyfert 2galaxy was corrected for absorption, the correlation remained intact with minimal scatter added dueto the uncertainties involved in the absorption correction. Panessa et al. (2006) included a widervariety of objects to their sample in order to include a larger luminosity range than previous used,including objects with L X ∼ - erg s - , and found that the correlation remained approximatelythe same. An important outcome of this is that the optical [O III ] luminosity can be used as a tracerof the intrinsic X-ray luminosity, and therefore used to estimate the amount of absorption seen inthe X-ray. We note that both the Heckman et al. (2005) and Panessa et al. (2006) samples includea selection of radio-loud and radio-quiet AGN, while our four objects are defined as radio-quietusing the standard Kellermann et al. (1989) definition and the [O
III ] luminosity to infer an opticalluminosity.We plot the L [O III] measurements for our four objects against our estimates of L X along withthe L [O III] - L X relation derived by Panessa et al. (2006) in Figure 5. All four objects are X-rayweak with respect to the relation, although SDSS J103234 is within the 1 σ scatter of the relation-ship, which is 0.6 dex in L X at fixed L [O III] . SDSS J110912 is ∼ . > σ outliers, falling ∼ . . L [O III] - L X correlation from Panessa et al. (2006) and calculate the expected 2 -
10 keV luminosity from the observed L [O III] values. The [O III ] luminosities are determined fromthe SDSS spectra and corrected for Galactic extinction (Kauffmann et al. 2003, BGH08). Com-paring these values to the L X values measured from the XMM-Newton data, we can attempt toquantify the level of intrinsic absorption within each galaxy. The observed L X of SDSS J103234is consistent with the L X derived from L [O III] , so no additional absorption is needed to reconcilethe two L X values. SDSS J110912 needs an absorbing column density of N H = 8 . × cm - inorder to bring the observed L X to the value suggested by its [O III ] luminosity. Due to the low L X of the upper limit of SDSS J011905, a considerable amount of absorption, N H > . × cm - ,is needed to bring the observed and predicted L X values into agreement. However, without aproper detection of SDSS J011905, the magnitude of the intrinsic absorbing column density willremain unknown. SDSS J144012 also falls below the L [O III] - L X relation, suggesting that there isa considerable amount of absorption in this object. Using our measured L X , we estimate the ab-sorbing column density needed to account for its displacement from the L [O III] - L X correlation tobe N H = 2 . × cm - . 17 –It is unclear whether an absorbing column density of N H ∼ cm - is large enough tocompletely obscure all of the broad-line emission from a source, or even how the X-ray absorbingcolumn and optical extinction are related. X-ray and optical surveys of AGN find 10 -
20% ofobjects show the properties of one AGN type in the optical and another in X-ray, e.g. a narrow-lineAGN with no absorption in the X-ray (Perola et al. 2004; Silverman et al. 2005; Tozzi et al. 2006).Among well-studied bright Seyfert galaxies, there are examples of objects with substantial broad-line emission in the optical, but with moderate levels of absorption observed in the X-ray. NGC3227 is one of these objects with obvious broad-line emission and an X-ray absorbing columndensity of N H ≈ . × cm - (Gondoin et al. 2003, see Piconcelli et al. 2006 and Jiménez-Bailón et al. 2008 for further examples). NGC 3227 and galaxies with similar optical line ratiosare typically classified as intermediate Seyferts, due to the relative flux of the broad and narrowcomponents of the permitted lines. The difference in appearance between intermediate Seyfertsand more typical Seyfert 1 galaxies is often attributed to some amount of obscuration. Whetherthe level of X-ray absorption seen in galaxies like NGC 3227 is related to the amount of opticalextinction in other objects, such as those in our sample, remains uncertain.If we compare our measurements with other L [O III] - L X correlations, we find that the resultscan change substantially. For example, using the relationships derived by Heckman et al. (2005)or Netzer et al. (2006), we find that the expected X-ray luminosity calculated from the observed L [O III] values are an order of magnitude lower than if the Panessa et al. (2006) relationship is used,which would suggest little to no absorption is present in any of our objects. However, both theHeckman et al. (2005) and Netzer et al. (2006) samples contain objects more luminous than L X ∼ erg s - , up to 1-3 orders of magnitude larger than observed in our objects. Netzer et al. (2006)also found a luminosity dependance in the L [O III] / L X ratio, which would explain different resultsfor samples with substantially different ranges in luminosity. The Panessa et al. (2006) sampleincludes objects with the same range of X-ray and [O III ] luminosities as our sample, making it themore appropriate sample for comparisons.
X-ray observations of low-mass AGNs, including NGC 4395 (Moran et al. 2005), POX 52(Thornton et al. 2008) and SDSS-selected objects (Greene & Ho 2007a; Desroches et al. 2009;Miniutti et al. 2009), show that the type 1 population is predominantly low-luminosity, with typ-ical luminosities of L X ≈ × - erg s - . The X-ray luminosity is much lower for thepopulation of type 2 AGNs in this sample ( L X ≈ × - × erg s - ) and is close to thecollective luminosity one would expect from a population of X-ray binaries in a host galaxy. Thispossible contamination is amplified by the fact that the XMM-Newton
PSF covers a large fraction 18 –of each galaxy in our sample. We now consider how much of the observed X-ray flux might arisefrom non-nuclear sources in each of our host galaxies.Gilfanov (2004) examined the relationship between the X-ray luminosity and stellar mass ofold populations, and found a typical ratio of L X / M ⋆ = 8 . × erg s - M ⊙ - . Given the stellarmass range of our sample, log ( M ⋆ / M ⊙ ) = 8 -
10, it is unlikely that the observed X-ray fluxes aresignificantly contaminated by low-mass X-ray binaries associated with the old stellar population.However, high-mass X-ray binaries associated with recent star formation can result in a higherX-ray luminosity. Lehmer et al. (2008) examined the relationship between the 0 . - L X / M ⋆ =1 . × erg s - M ⊙ - for star-forming galaxies with stellar masses of log ( M ⋆ / M ⊙ ) = 9 - g - r ≈ . L X / M ⋆ ratio found by Lehmer et al. applies to this galaxy, then X-ray binariescould in principle account for the observed X-ray luminosity. However, the SDSS image showsthat most of the recent star formation in this galaxy is located in the spiral arms at distances of & ′′ from the nucleus. If the X-ray luminosity of this galaxy was dominated by high-mass X-raybinaries then it would be noticeably extended in the XMM-Newton image rather than compact, soit appears unlikely that high-mass X-ray binaries make a significant contribution to the observedX-ray flux. For SDSS J144012, the predicted luminosity due to high-mass X-ray binaries is a fac-tor of 3 smaller than the observed X-ray luminosity. Although we can not rule out the possibilitythat some observed properties, such as Γ and HR, might be affected by the presence of high-massX-ray binaries, we conclude that the observed X-ray flux is most likely dominated by emissionfrom the AGN. We investigate the accretion power of these objects by deriving estimates of their Eddingtonratios, L bol / L Edd . The M BH - σ ∗ relation allows us to estimate a black hole mass from the stellar ve- 19 –locity dispersion ( σ ∗ ) of the host galaxy (Gebhardt et al. 2000; Ferrarese & Merritt 2000). BGH08used their measurements of σ ∗ along with the M BH - σ ∗ relation of Tremaine et al. (2002) to esti-mate a black hole mass for each object, which includes an additional offset in black hole mass seenin other low-mass Seyfert 1 galaxies (Barth et al. 2005). This offset probably reflects a flatteningof the M BH - σ ∗ relation at low masses (Wyithe 2006; Greene & Ho 2006), but without a moredetailed study of the M BH - σ ∗ relation in this mass range, we follow the example of Barth et al.(2005) and assume a uniform offset.A common estimator of bolometric luminosity in the optical is based on the [O III ] λ L [O III] for each object using the bolometric correction of L bol / L [O III] = 3500 derivedby Heckman et al. (2004) from a low-redshift sample of Seyfert 1 galaxies and quasars.We also use a broad-band 2 -
10 keV X-ray bolometric correction from Vasudevan & Fabian(2007, 2009). Because a larger wavelength range is used to estimate L bol , X-ray bolometric correc-tions tend to produce better estimates of the intrinsic L bol than corrections based on narrow-bandluminosities, assuming the unabsorbed X-ray luminosity is used. The 2 -
10 keV energy rangealso probes the primary continuum emission from the central engine producing a better estimateof the bolometric luminosity of the object. Vasudevan & Fabian (2009) find that highly accretingobjects tend to require a larger X-ray bolometric correction than those accreting at a lower rate.For L bol / L X = κ , objects with an intrinsic L bol / L Edd > . κ ∼ . < L bol / L Edd < . κ ∼
45, and for those objects with intrin-sic L bol / L Edd < . κ ∼
20. We find that with this two-tier bolometric correction, no objects in oursample are accreting at a high enough rate to warrant the higher bolometric correction. Therefore,we apply the X-ray bolometric correction of L bol / L X = 20 to the unabsorbed X-ray luminosities ofeach of our objects to obtain an estimate of the bolometric luminosity. For each object, estimatesof stellar velocity dispersion, black hole mass, Eddington luminosity and bolometric luminositycan be seen in Table 3.The Eddington ratios from L bol (X-ray) and the absorption-corrected X-ray luminosities aresystematically lower than those determined from the [O III ] luminosity by a factor of ∼ - L [O III] bolometric correction of Heckman et al. (2004),finding that even low levels of extinction in the narrow-line region cause bolometric luminositiesto be systematically over-estimated. By correcting L [O III] for this extinction and including a de-pendance on luminosity, they find the L [O III] bolometric corrections to be over an order of mag-nitude lower than in Heckman et al. (2004). The resulting bolometric luminosities more closely 20 –resemble the bolometric luminosities derived using X-ray bolometric corrections. This luminosity-dependent [O III ] bolometric correction may resolve the discrepency observed between our bolo-metric luminosities derived from L [O III] and L X . Regardless of the bolometric correction used,SDSS J144012 has the highest Eddington ratio in the sample with L bol (X-ray)/ L Edd ≈ .
03. Acareful look of the spectral energy distributions of these objects using recently observed
Spitzer
IRS spectra along with existing multi-wavelength data should better determine the bolometric lu-minosities of these sources.
SDSS J110912 is an object of particular interest, not only for the strong similarity between itand NGC 4395, but also because it has the lowest detected X-ray luminosity in the sample. Theradial profile of SDSS J110912 shows possible evidence for extended emission and the X-ray lu-minosity of L X ∼ × erg s - is so low that it could be contaminated by a collection of X-raybinaries in the host galaxy. On average, low-mass X-ray binaries have photon indices in the rangeof Γ ≈ . - . Γ ≈ - Γ = 1 . ± . M BH - σ ∗ relation used to estimate theblack hole mass is poorly constrained at these low masses. Given the strong morphological andspectral similarities between NGC 4395 and SDSS J110912, it is conceivable that they also sharecomparable black hole masses. Then, the defining difference between these two objects wouldbe the lack of broad-line emission in SDSS J110912 (BGH08). If SDSS J110912 falls within thenormal scatter of the L [O III] - L X relation and therefore there is little internal absorption, one couldspeculate that SDSS J110912 represents a nearly identical “true” type 2 version of NGC 4395 withno broad-line region. There have been a few suggested cases of type 2 objects without a broad-lineregion (see Ghosh et al. 2007, Gliozzi et al. 2007, Bianchi et al. 2008), but Brightman & Nandra(2008) recently show that many of the candidate “true” type 2 objects can also be fit with morecomplex spectral models, including high amounts of absorption. 21 –Various models attempt to explain how an accreting black hole might fail to form a broad-lineregion, typically due to a combination of low AGN luminosity and black hole mass (for a review,see Ho 2008). Laor (2003) suggested that since the radius of the BLR decreases as the luminosityof an AGN decreases, there might exist a luminosity threshold below which the BLR would ceaseto exist. For a black hole with M BH ∼ M ⊙ , this luminosity would be L min = 6 × erg s - and L min = 6 × erg s - for a 10 M ⊙ black hole, both of which are 2 - L bol below which the radiation-driven windis unable to produce the broad-line emission. This suggests that at lower L bol , unobscured AGNswould be "true" type 2 objects. The calculations of Nicastro (2000) show that this threshold lies at L bol / L Edd . - , below which the objects would be unable to produce the broad-line region.Using the X-ray results, we can examine where these low-mass AGNs lie relative to the pre-dicted thresholds for BLR formation. We find that SDSS J103234 and NGC 4395 both have Ed-dington ratios close to the hypothetical threshold for BLR formation in the Nicastro model (NGC4395: L bol / L Edd ≈ - ; Peterson et al. 2005). However, optical spectra show definite broad-lineemission in NGC 4395 and weak broad-line emission in SDSS J103234 (BGH08). SDSS J110912has a very uncertain Eddington ratio of L bol / L Edd ≈ .
001 which lies at the threshold for BLR for-mation in the Nicastro model. Therefore, there exists the possibility that SDSS J110912 accretesat a low enough rate to explain the lack of broad-line emission. However, this seems unlikelygiven that SDSS J110912 and NGC 4395 have such similar narrow emission-line luminosities, andpresumably similar ionizing luminosities.Given that the observed L X of SDSS J110912 is an order of magnitude lower than that ofNGC 4395, but they have nearly identical [O III ] luminosities, we conclude that the most likelyexplanation is that the central engine of SDSS J110912 is obscured, although we are unable toaccurately determine the amount of X-ray absorption in SDSS J110912. If a substantial fractionof the observed L X is due to X-ray binaries, then that increases the discrepancy between L [O III] versus L X and strengthens the case for an obscured nucleus. Elitzur & Shlosman (2006) present amodel in which the radiation-driven wind produces the obscuring material, such that objects with L bol . erg s - would be unable to produce the obscuring torus. If the observed properties ofSDSS J110912 are indeed the result of obscuration, then this would indicate that the obscuringtorus could still persist even at AGN luminosities more than an order of magnitude below thethreshold suggested by Elitzur & Shlosman. 22 –
6. Summary and Conclusions
We find all four objects in the sample to be X-ray faint with X-ray luminosities approximatelyan order of magnitude lower than those seen in the Greene & Ho (2004) sample (Greene & Ho2007a), but only two objects, SDSS J011905 and SDSS J144012 show evidence of moderate tosubstantial absorption with estimated column densities of N H ∼ cm - . SDSS J103234 showslittle evidence of absorption which is consistent in the context of the unified model with the pre-vious detection of broad emission lines from high-resolution optical spectra. It is unclear withoutfurther observations whether SDSS J110912 is truly lacking a broad-line region or if it is absorbedand the emission detected is due to a combination of X-ray binaries in the host galaxy and weakemission from the AGN, but given the low observed X-ray luminosity, we believe SDSS J110912contains at least a moderate amount of absorption.Two objects had high enough S/N ratios for spectral fitting. SDSS J103234 has a spectrumsimilar to the partially absorbed spectrum of POX 52 discussed by Thornton et al. (2008), and iswell fit with a high partial-covering fraction of 95%, in the 0 . - . . - . L [O III] - L X correlation.By comparing type 1 and 2 objects with similar masses, we have the opportunity to furtherinvestigate the absorbing properties of low-mass AGNs and whether or not line-of-sight absorptioncan explain the presence or lack of broad emission lines, or if other processes, such as accretionrate, play a role. If high S/N observations are obtained, X-ray spectra are an excellent tool that canbe used to quantify the amount of absorption in an object. IR spectroscopy is another important toolused to investigate absorption and reprocessing of the AGN continuum. We plan to investigate bothlow-mass Seyfert 1 and 2 objects using Spitzer spectra in order to better constrain the bolometricluminosities of these objects and further investigate the presence of an obscuring torus.The analysis of
XMM-Newton data presented herein was supported by grant NNX06AF08Gfrom NASA. This work was also supported by the National Science Foundation under grant AST-0548198. This research has made use of the NASA/ IPAC Infrared Science Archive, which isoperated by the Jet Propulsion Laboratory, California Institute of Technology, under contract withthe National Aeronautics and Space Administration. 23 –
REFERENCES
Antonucci, R. R. J., & Miller, J. S. 1985, ApJ, 297, 621Barth, A. J., Greene, J. E., & Ho, L. C. 2005, ApJ, 619, L151Barth, A. J., Greene, J. E., & Ho, L. C. 2008, AJ, 136, 1179Barth, A. J., Ho, L. C., Rutledge, R. E., & Sargent, W. L. W. 2004, ApJ, 607, 90Bell, E. F., McIntosh, D. H., Katz, N., & Weinberg, M. D. 2003, ApJS, 149, 289Bianchi, S., Corral, A., Panessa, F., Barcons, X., Matt, G., Bassani, L., Carrera, F. J., & Jiménez-Bailón, E. 2008, MNRAS, 385, 195Boller, T., Meurs, E. J. A., Brinkmann, W., Fink, H., Zimmermann, U., & Adorf, H.-M. 1992,A&A, 261, 57Brightman, M., & Nandra, K. 2008, MNRAS, 390, 1241Cash, W. 1979, ApJ, 228, 939Desroches, L.-B., Greene, J. E., & Ho, L. C. 2009, ApJ, 698, 1515Dickey, J. M., & Lockman, F. J. 1990, ARA&A, 28, 215Elitzur, M., & Shlosman, I. 2006, ApJ, 648, L101Elvis, M., Risaliti, G., Nicastro, F., Miller, J. M., Fiore, F., & Puccetti, S. 2004, ApJ, 615, L25Ferrarese, L., & Merritt, D. 2000, ApJ, 539, L9Filippenko, A. V., & Sargent, W. L. W. 1989, ApJ, 342, L11Fruscione, A., Greenhill, L. J., Filippenko, A. V., Moran, J. M., Herrnstein, J. R., & Galle, E. 2005,ApJ, 624, 103Gallagher, S. C., Richards, G. T., Hall, P. B., Brandt, W. N., Schneider, D. P., & Vanden Berk,D. E. 2005, AJ, 129, 567Gallo, E., Treu, T., Jacob, J., Woo, J.-H., Marshall, P. J., & Antonucci, R. 2008, ApJ, 680, 154Gebhardt, K., et al. 2000, ApJ, 539, L13Gehrels, N. 1986, ApJ, 303, 336 24 –Ghosh, H., Pogge, R. W., Mathur, S., Martini, P., & Shields, J. C. 2007, ApJ, 656, 105Gilfanov, M. 2004, MNRAS, 349, 146Gliozzi, M., Sambruna, R. M., & Foschini, L. 2007, ApJ, 662, 878Gondoin, P., Orr, A., Lumb, D., & Siddiqui, H. 2003, A&A, 397, 883Greene, J. E., & Ho, L. C. 2004, ApJ, 610, 722Greene, J. E., & Ho, L. C. 2006, ApJ, 641, L21Greene, J. E., & Ho, L. C. 2007a, ApJ, 656, 84Greene, J. E., & Ho, L. C. 2007b, ApJ, 670, 92Hasinger, G., et al. 2001, A&A, 365, L45Heckman, T. M., Kauffmann, G., Brinchmann, J., Charlot, S., Tremonti, C., & White, S. D. M.2004, ApJ, 613, 109Heckman, T. M., Ptak, A., Hornschemeier, A., & Kauffmann, G. 2005, ApJ, 634, 161Ho, L. C. 2008, ARA&A, 46, 475Ho, L. C., Filippenko, A. V., & Sargent, W. L. W. 1997, ApJS, 112, 315Irwin, J. A., Athey, A. E., & Bregman, J. N. 2003, ApJ, 587, 356Iwasawa, K., Fabian, A. C., Almaini, O., Lira, P., Lawrence, A., Hayashida, K., & Inoue, H. 2000,MNRAS, 318, 879Jiménez-Bailón, E., Krongold, Y., Bianchi, S., Matt, G., Santos-Lleó, M., Piconcelli, E., & Schar-tel, N. 2008, MNRAS, 391, 1359Kaastra, J. S. 1992, An X-Ray Spectral Code for Optically Thin Plasmas (Internal SRON-LeidenRep.) (updated version 2; Leiden: Space Res. Organization Netherlands)Kalberla, P. M. W., Burton, W. B., Hartmann, D., Arnal, E. M., Bajaja, E., Morras, R., Pöppel,W. G. L. 2005, A&A, 440, 775Kauffmann, G., et al. 2003, MNRAS, 346, 1055Kellermann, K. I., Sramek, R., Schmidt, M., Shaffer, D. B., & Green, R. 1989, AJ, 98, 1195Kewley, L. J., Groves, B., Kauffmann, G., & Heckman, T. 2006, MNRAS, 372, 961 25 –Kraemer, S. B., George, I. M., Crenshaw, D. M., & Gabel, J. R. 2004, ApJ, 607, 794Kunth, D., Sargent, W. L. W., & Bothun, G. D. 1987, AJ, 93, 29Lamastra, A., Bianchi, S., Matt, G., Perola, G. C., Barcons, X., & Carrera, F. J. 2009, A&A, inpressLaor, A. 2003, ApJ, 590, 86Lehmer, B. D., et al. 2008, ApJ, 681, 1163Leighly, K. M. 1999, ApJS, 125, 317Liedahl, D. A. 1992, Ph.D. thesis, Univ. California, BerkeleyMatsumoto, H., Koyama, K., Awaki, H., Tsuru, T., Loewenstein, M., & Matsushita, K. 1997, ApJ,482, 13McHardy, I. M., Koerding, E., Knigge, C., Uttley, P., & Fender, R. P. 2006, Nature, 444, 730Mewe, R., Gronenschild, E. H. B. M., & van den Oord, G. H. J. 1985, A&AS, 62, 197Mewe, R., Lemen, J. R., & van den Oord, G. H. J. 1986, A&AS, 65, 511Miniutti, G., Ponti, G., Greene, J. E., Ho, L. C., Fabian, A. C., & Iwasawa, K. 2009, MNRAS, 394,443Moran, E. C., Eracleous, M., Leighly, K. M., Chartas, G., Filippenko, A. V., Ho, L. C., & Blanco,P. R. 2005, AJ, 129, 2108Moran, E. C., Halpern, J. P., & Helfand, D. J. 1996, ApJS, 106, 341Murray, N., & Chiang, J. 1997, ApJ, 474, 91Nandra, K., George, I. M., Mushotzky, R. F., Turner, T. J., & Yaqoob, T. 1997a, ApJ, 476, 70Nandra, K., George, I. M., Mushotzky, R. F., Turner, T. J., & Yaqoob, T. 1997b, ApJ, 477, 602Nandra, K., et al. 2005, MNRAS, 356, 568Nelson, C. H., & Whittle, M. 1996, ApJ, 465, 96Netzer, H., Mainieri, V., Rosati, P., & Trakhtenbrot, B. 2006, A&A, 453, 525Nicastro, F. 2000, ApJ, 530, L65 26 –Nicastro, F., Martocchia, A., & Matt, G. 2003, ApJ, 589, L13Panessa, F., Bassani, L., Cappi, M., Dadina, M., Barcons, X., Carrera, F. J., Ho, L. C., & Iwasawa,K. 2006, A&A, 455, 173Perola, G. C., et al. 2004, A&A, 421, 491Peterson, B. M., et al. 2005, ApJ, 632, 799Piconcelli, E., Jiménez-Bailón, E., Guainazzi, M., Schartel, N., Rodríguez-Pascual, P. M., &Santos-Lleó, M. 2005, A&A, 432, 15Piconcelli, E., et al. 2006, A&A, 453, 839Raymond, J. C., & Smith, B. W. 1977, ApJS, 35, 419Sasaki, M., Pietsch, W., & Haberl, F. 2003, A&A, 403, 901Satyapal, S., Vega, D., Dudik, R. P., Abel, N. P., & Heckman, T. 2008, ApJ, 677, 926Seth, A., Agüeros, M., Lee, D., & Basu-Zych, A. 2008, ApJ, 678, 116Shih, D. C., Iwasawa, K., & Fabian, A. C. 2003, MNRAS, 341, 973Silverman, J. D., et al. 2005, ApJ, 618, 123Thim, F., Hoessel, J. G., Saha, A., Claver, J., Dolphin, A., & Tammann, G. A. 2004, AJ, 127, 2322Thornton, C. E., Barth, A. J., Ho, L. C., Rutledge, R. E., & Greene, J. E. 2008, ApJ, 686, 892Tozzi, P., et al. 2006, A&A, 451, 457Tremaine, S., et al. 2002, ApJ, 574, 740Vasudevan, R. V., & Fabian, A. C. 2007, MNRAS, 381, 1235Vasudevan, R. V., & Fabian, A. C. 2009, MNRAS, 392, 1124Wyithe, J. S. B. 2006, MNRAS, 365, 1082
This preprint was prepared with the AAS L A TEX macros v5.2.
27 –Table 1. Observation DetailsGalaxy Obs. ID Obs. Date Exp. Time Cor. Exp.(UT) (s) Time (s)SDSS J011905.14+003745.0 0400570301 2006 Jul 26 26,739 18,040SDSS J103234.85+650227.9 0400570401 2006 May 6 24,013 19,354SDSS J110912.40+612346.7 0400570201 2006 Nov 25 23,613 23,613SDSS J144012.70+024743.5 0400570101 2006 Aug 8 22,915 17,120Note. — Cor. Exp. Time is the exposure time of each observation, corrected for the time lostdue to background flares.
Table 2. X-ray Parameters
Galaxy X-ray Net Count Rate ( s - ) C S ( s - ) C H ( s - ) HR Γ HR Flux (erg cm - s - ) L X (erg s - )offset ( ′′ ) CountsSDSS J011905 · · · < < · · · · · · · · · · · · < . × - < . × SDSS J103234 2.28 140.8 0 . ± . . ± . . ± . . ± .
06 1 . ± . . ± . × - (2 . ± . × SDSS J110912 1.83 26.7 0 . ± . . ± . . ± . - . ± .
16 1 . ± . . ± . × - (4 . ± . × SDSS J144012 0.64 163.2 0 . ± . . ± . . ± . - . ± .
05 1 . ± . . ± . × - (3 . ± . × Note. — X-ray offset is the positional difference between the optical and X-ray positions. Total counts is the number of counts in the 0 . - . C S isthe count rate from 0 . - . C H is the count rate from 2 . - . L X are based on the 2 -
10 keV energy range and are inferred from a power law withphoton index Γ HR and Galactic absorption. Table 3. Derived Parameters
Galaxy σ ∗ (km s - ) M BH ( M ⊙ ) L Edd (erg s - ) L bol ([OIII]) (erg s - ) L bol ([OIII])/ L Edd L bol (X-ray) (erg s - ) L bol (X-ray)/ L Edd
SDSS J011905 39 ± . × . × . × < . × < . ± . × . × . × . × ± . × . × . × . × ± . × . × . × . × σ ∗ = FWHM([OIII]) / . L bol ([OIII]) was estimated from the [O III ] luminosity using the bolometric correction of L bol / L [OIII] = 3500. L bol (X-ray) was estimated using thebolometric correction L bol / L X = κ , where κκ