On the nuclear obscuration of H2O maser galaxies
aa r X i v : . [ a s t r o - ph . C O ] D ec On the nuclear obscuration of H O maser galaxies
J. S. ZhangCenter for astrophysics, GuangZhou university, GuangZhou, 510006, China [email protected] andC. HenkelMax-Planck-Institut f¨ur Radioastronomie, Auf dem H¨ugel 69, D-53121 Bonn, GermanyandQ. Guo, H. G. Wang, J. H. FanCenter for astrophysics, GuangZhou university, GuangZhou, 510006, ChinaReceived ; accepted 2 –
ABSTRACT
To shed light onto the circumnuclear environment of 22 GHz ( λ ∼ O maser galaxies, we have analyzed some of their multi-wavelength properties,including the far infrared luminosity (FIR), the luminosity of the [O III] λ Kα emission line (EW (K α )). Our statistical analysis includes a total of85 sources, most of them harboring an active galactic nucleus (AGN). There arestrong anti-correlations between EW ( K α ) and two “optical thickness parame-ters”, i.e. the ratios of the X-ray luminosity versus the presumably more isotrop-ically radiated [O III] and far infrared (FIR) luminosities. Based on these anti-correlations, a set of quantitative criteria, EW ( K α ) >
300 eV, L − keV < L [O III] and L FIR > L − keV can be established for Compton-thick nuclear regions.18 H O maser galaxies belong to this category. There are no obvious correla-tions between the EW (K α ), the [O III] luminosity and the isotropic H O maserluminosity. When comparing samples of Seyfert 2s with and without detectedH O maser lines, there seem to exist differences in EW (K α ) and the fraction ofCompton-thick nuclei. This should be studied further. For AGN masers alone,there is no obvious correlation between FIR and H O maser luminosities. How-ever, including masers associated with star forming regions, a linear correlation isrevealed. Overall, the extragalactic FIR-H O data agree with the correspondingrelation for Galactic maser sources, extrapolated by several orders of magnitudeto higher luminosities.
Subject headings:
Masers – galaxies: active – galaxies: nuclei – galaxies: Comptonthickness : galaxies – X-rays: galaxies
1. Introduction
Thanks to a large number of dedicated surveys during the last 15 years, the numberof galaxies known to host 22 GHz ( λ ∼ O masers has increased tenfold and 85sources at distances larger than those of the Magellanic Clouds have been reported sofar to exhibit H O maser emission (e.g., Braatz & Gugliucci 2008, Darling et al. 2008,Greenhill et al. 2008). Some of them are AGN-related, while others are found in off-nuclearstar forming regions. Among the masers clearly identified as AGN-related, a large fraction( ∼ O maser spots locate preferentially inthe nuclear regions of Seyfert 2 or LINER galaxies and most of them are heavily obscured( N H > cm − ; Braatz et al. 1997, Madejski et al. 2006, Zhang et al. 2006, Greenhill etal. 2008). The nuclear X-ray source is generally believed to heat the gas to temperaturessuitable for 22 GHz H O maser emission (Neufeld et al. 1994), which is supported by arelation between maser luminosity, the unabsorbed intrinsic nuclear X-ray luminosity, andthe mass of the black hole (Kondratko et al. 2006, Su et al. 2008).For the obscured nuclear regions H O masers and X-rays provide, unlike optical data,deeply penetrating views. The X-ray absorption along the line-of-sight to the nucleuscan provide through spectral model fitting important information on the nature of the 4 –circumnuclear environment. With modern X-ray telescopes, high quality X-ray spectrahave been obtained for more than 30 H O maser galaxies. Based on an analysis of suchspectra, the nuclear column densities of maser host galaxies were investigated by Zhang etal. (2006) and Greenhill et al. (2008). While these studies have shown that nuclear H Omasers are mostly found in environments with high column density, it is still open whetherH O masers preferentially arise from Compton-thick ( N H > cm − ) AGN.To reduce limitations and uncertainties when modeling X-ray spectra, here we tryto provide additional constraints to evaluate line-of-sight column densities. The iron K α emission line ( ∼ α )) of the iron line were foundin highly absorbed sources and are generally used to identify Compton-thick nuclei (e.g.,Matt 1997). However, currently there are no quantitative criteria on EW (K α ) yet as aprobe of the gas absorption.Unlike the X-ray absorption, [O III] λ λ λ λ O maser galaxieswhich are located farther than the Magellanic Clouds and which are known to host 22 GHzH O masers. The data are analyzed to determine gas column densities along the line ofsight toward the AGN and to derive properties related to activity in highly obscured andtherefore particularly elusive Compton-thick nuclear environments.
2. Data
Data related to the galaxies with detected H O maser emission, located at largerdistances than the Magellanic Clouds, are compiled in Table 1. Among the 85 publishedH O maser sources, there are 66 AGN masers (63 “megamasers” with isotropic luminosities L H O > L ⊙ and 3 “kilomasers” with L H O < L ⊙ ). The megamasers are classified assuch because of their luminosity. While most of them have not yet been studied in detail,all thoroughly investigated megamasers are not separated by more than a few pc fromthe line of sight to the AGN of their parent galaxy. Ten kilomasers are related to star 6 –formation regions and another nine kilomasers are still awaiting interferometric observationsto investigate their nature. All activity types of these H O maser galaxies are listed inTable 1 and their dominant types (following Bennert et al. 2009) are used for the statistics(for details, see Table 2). Most of the AGN maser sources are Seyfert 2 galaxies or LINERs. 7 –Table 1. Physical Parameters of Extragalactic H O Maser Sources ∗ . Source Type Tel. EW ( K α ) F X Ref. F
F IR H α /H β F [OIII] Ref.
NGC 17
Sy2,LIRG, H II 1.33NGC 23 LINER,LIRG, H I 1.09 5.89 37.9 Ho97
IC 10 +68 −
482 Awa00 0.34 6.02 177 Dah88G 230 +120 − NGC 598
H II 67.71 4.55 0.24 Ho97
NGC 591
Sy2 X 2200 +700 −
20 Gua05b 0.28 178 Whi92NGC 613 Sy2,H II 2.6IC 0184 Sy2,H IINGC 1052 Sy2,LINER B 180 +80 −
400 Ter02 0.14 2.82 13.3 Dah88
NGC 1068
Sy2,Sy1 X 1200 ±
500 462 Cap06 25.01 7.00 6780 Dah88A 1210 +260 −
350 Bas99NGC 1106 Sy2 0.19Mrk 1066 Sy2 C 1120 +850 −
23 Shu07 1.14 8.51 514 Whi92
NGC 1320
Sy2 X 2200 +440 − NGC 1386
Sy2 X 1800 +400 − ± +8900 −
20 Bas99IRAS03355+0104 Sy2 0.13
IC 342
Sy2,H II 8.29 7.69 3.4 Ho97MG J0414+0534 QSO1
UGC 3193 +30 −
590 Bia05 0.57 6.67 4610 Whi92B 650 +182 −
650 Cap99A 997 +300 −
650 Bas99
NGC 2146
H II 12.54 11.1 30.47 Ho97VII ZW 73 Sy2 0.21
Source Type Tel. EW ( K α ) F X Ref. F
F IR H α /H β F [OIII] Ref.NGC 2273 Sy2 X 2200 +400 − +16 − Gua05b 0.7 6.92 164 Whi92A 1040 +440 −
125 Ter02
UGC 3789
He 2-10
SBG C 9.56 +0 . − . J¨ur05 2.40
Sy2 0.17Mrk 1210 Sy2,Sy1 C ∼
188 840 Zha09 0.36 5.20 580 Ter91X 130 970 Gua02B 108 +50 −
930 Ohn04A 820 +360 −
160 Awa00
NGC 2639
Sy1.9 A 1490 +11110 − ∼
30 Zha06 0.91 6.67 62.19 Ho97NGC 2824(Mrk 394) Sy? 0.12SBS 0927+49 LINER 0.29
NGC 2960
LINER 0.24UGC 5101 Sy1.5,LINE R,LIRG X 410 +270 − NGC3034
SBG,H II 96.49 25.0 1615.17 Ho97
NGC 3079
Sy2,LINER X 1480 ±
500 33 Cap06 5.5 25.0 92 Ho97B 2400 +2900 − ± Mrk 34
Sy2 0.43 10.5 204.19 Dah88
NGC 3359
H II 0.70
IC 2560
Sy2 X 2320 +180 − +1 . − . Til08 4.29 >
40 Ris99C 2770 ±
490 38.4 +21 . − . Mad06
NGC 3393
Sy2 X 1400 ±
800 9 +6 − Gu05a 0.33 4.12 316 Dia88A 3500 ± +262 − +236 − Del02NGC 3735 Sy2 1.02 6.31 33 Ho97
Source Type Tel. EW ( K α ) F X Ref. F
F IR H α /H β F [OIII] Ref.
Antennae
SBG C
NGC 4051
Sy1.5 X 240 ±
40 627 Cap06 1.32 3.33 59.99 Ho97NGC 4151 Sy1.5 X 300 ±
30 4510 Cap06 1.11 3.45 1695.39 Ho97A 101 ± ∼ NGC 4214
SBG C 243 Har04 1.81
NGC 4258
Sy1.9,LINER X 27 ±
20 837 Cap06 5.08 9.12 262 Hec80A 250 ±
100 300 Bas99NGC 4293 LINER 0.58 7.69 5.95 Ho97
NGC 4388
Sy2 X 440 ±
90 762 Cap06 1.44 5.50 374 ±
50 Bas99A 732 +243 − ESO 269-G012
Sy2 0.23NGC 4922 Sy2,LINER 0.58 7.14 33.03 Kim95
NGC 4945
Sy2 C 1300 500 Don03 41.38 >
40 Ris99B ∼ ±
160 350 Bas99NGC 5194 Sy2,H II X 986 ±
210 48 Cap06 6.62 8.33 228 Ho97A 910 +350 − NGC 5253
SBG,H II C 29.9 J¨ur05 3.16Mrk 266(NGC 5256) Sy2,LIRG,SBG B 575 56 Ris00 0.84 5.92 44.33 Dah88NGC 5347 Sy2 C 1300 ±
500 22 Lev06 0.27 114 Tra01
NGC 5495
Sy2,H II? 0.27
Circinus
Sy2 C 2250 +260 − +24 − ±
30 8380 Bas99NGC 5643 Sy2 X 500 84 Gua04 2.59 6.40 662 Whi92A 1800 +800 −
130 Bas99
NGC 5728
Sy2,H II C 1100 +320 −
133 Shu07 0.96 5.96 761 Sto95C 1130 Zha06
UGC09618NED02
LINER,H II 0.92
NGC 5793
Sy2 0.6NGC 6240 Sy2,LINER C 2400 +800 −
170 Pta03 2.18 17.2 135 ±
20 Kim95
10 –Table 1—Continued
Source Type Tel. EW ( K α ) F X Ref. F
F IR H α /H β F [OIII] Ref.A 1580 +380 −
190 Bas99
NGC6264
Sy2NGC 6300 Sy2 X 148 ±
18 2160 ±
100 Mat04b 2.34 320 Lum01
NGC 6323
Sy2ESO 103-G035 Sy2,Sy1 A 173 +50 −
907 Tur97 0.37 6.31 112 Po196IRAS F19370-013 Sy2,H II 0.26
3C 403
FRII C 244 ±
20 Kra05 0.1
NGC 6926
Sy2,H II 0.65 15.6 21.98 Kim95AM 2158-380NED02 Sy2,RG
TXS 2226-184
LINERNGC 7479 Sy2,LINER 0.83 10.0 37.9 Kim95IC 1481 LINER 0.12Note. — 85 published extragalactic H O maser sources with available physical parameters are listed (10 masers arise in starforming regions marked by italics and 29 out of 66 AGN-masers are potential disk-masers, their source names are presented inboldface). Recently reported masers, not being part of the 78 sources listed by Bennert et al. (2009), are one type I quasarfrom Impellizzeri et al. (2008), two sources (NGC 17 and NGC 1320) from Greenhill et al. (2008), and four new masers relatedwith star formation (He 2-10, Antennae, NGC 4214, NGC 5253) from Darling et al. (2008). C olumn Extragalactic H O maser host galaxies; C olumn Type of nuclear activity. SBG: StarBurst Galaxy; Sy1, Sy1.5, Sy1.9, Sy2: Seyfert types; LINER: Low-IonizationNuclear Emission Line Region; LIRG: Luminous-Infrared Galaxy; FR II: Fanarov-Riley Type II radio galaxy; NLRG: Narrow-Line Radio Galaxy; RG: Radio Galaxy; H II: classified as a H II region; QSO1 and QSO2: type 1 and 2 Quasars. References:Zhang et al. (2006); Kondratko et al. (2006) and NED; C olumn X-ray telescope–A:
ASCA; B: BeppoSAX ; C:
Chandra ; X:
XMM-Newton ; C olumns The EW ( K α ) of the Fe line (eV) and the 2-10 keV observed X-ray flux (in units of 10 − erg s − cm − ); C olumn References for Col. 4&5 — Awa00: Awaki et al. 2000; Bal04: Ballo et al. 2004; Bas99: Bassani et al. 1999; Bec04:Beckmann et al. 2004; Bia03: Bianchi et al. 2003; Bia05: Bianchi et al 2005; Cap99: Cappi et al. 1999; Cap06: Cappi et al.2006; Del02: Della Ceca et al. 2002; Dia88: Diaz et al. 1988; Don03: Done et al. 2003; Gua00: Guainazzi et al. 2000a; Gua02:Guainazzi et al. 2002; Gua04: Guainazzi et al. 2004; Gua05a: Guainazzi et al. 2005a; Gua05b: Guainazzi et al. 2005b; Ima03:Imanishi et al. 2003; Iwa02: Iwasawa et al. 2002; Iyo01: Iyomoto et al. 2001; Jen04: Jenkins et al. 2004; Kra05: Kraft et
11 – al. 2005; Lev06: Levenson et al. 2006; Mad06: Madejski et al. 2006; Mat01: Matt et al. 2001; Mat04a: Matt et al. 2004a;Mat04b: Matsumoto et al. 2004b; Ohn04: Ohno et al. 2004; Pta03: Ptak et al. 2003; Ris00: Risaliti et al. 2000; Smi96: Smith& Done 1996; Smi01: Smith & Wilson 2001; Ter02: Terashima et al. 2002; Tur97: Turner et al. 1997; Wea01: Weaver et al.2001; Zha06: Zhang et al. 2006; Zha09: Zhang et al. 2009; C olumn FIR flux in units of 10 − erg s − cm − ; C olumn The Balmer increment line intensity ratio H α /H β ; C olumn Extinction corrected [O III] λ − erg s − cm − ; C olumn References for Col. 8&9 — Bas99: Bassani et al. 1999; Dah88: Dahari & De Robertis 1988; Hec80: Heckman etal. 1980; Ho97: Ho et al. 1997; Kim95: Kim et al. 1995; Lum01: Lumsden et al. 2001; Pol96: Polletta et al. 1996; Ris99:Risaliti et al. 1999; Shu07: Shu et al. 2007; Sto89: Storchi-Bergmann & Pastoriza 1989; Sto95: Storchi-Bergmann et al. 1995;Ter91: Terlevich et al. 1991; Tra01: Tran 2001; Whi92: Whittle 1992;
12 –Besides the type of nuclear activity, we collected multi-wavelength data and parametersfor all of the 85 extragalactic H O maser sources, including the EW ( K α ) of the ironemission line, the observed X-ray (2-10 keV) flux F X , the FIR flux F F IR , the observedH α /H β line intensity ratio and the [O III] λ [O III] . These data are alsopresented in Table 1.The X-ray data of our sample are collected from the literature, based on observationsfrom ASCA , BeppoSAX , Chandra , and the
XMM-Newton satellite. For some sources morethan one value were reported in the literature for parameters such as the EW (K α ) ofthe iron line and the observed X-ray flux, either due to intrinsic variability of the sourceor due to a different modeling of the spectra. For comparison, all results available fromthe literature are listed in Table 1. For our statistical analyses, observational results weretaken from XMM-Newton and
Chandra whenever possible. Otherwise the most recentmeasurements were used.For the [O III] λ λ H α /H β ) = 3, the intrinsic [O III]line fluxes were derived from the formula F [O III] , cor = F [O III] , obs [( H α /H β ) obs / ( H α /H β ) ] . (Bassani et al. 1999). The [O III] λ µ m) for our H O maser host galaxies. Following themethod of Wouterloot & Walmsley (1986), the infrared flux (6 < λ < µ m) was derived 13 –by extrapolating flux densities beyond 12 and 100 µ m and assuming a grain emissivityproportional to frequency ν . In this way, infrared luminosities could be determined for76 H O maser galaxies. Throughout the paper, the luminosity distance was derived usingCalculators I provided by the NASA Extragalactic Database (NED), assuming Ω M = 0.270,Ω vac = 0.730, and H =70 km s − Mpc − (e.g., Spergel et al. 2007).
3. Results3.1. Isotropic indicators of the intrinsic nuclear power
For our H O maser galaxies with available data, the luminosity distributions of bothisotropic indicators, the [O III] line and the FIR luminosities, are presented in Fig. 1. Theleft panel shows the histogram of the infrared luminosities for all 76 H O maser galaxieswith available data ( logL
F IR , hereafter luminosity in logarithmic scale and in units oferg s − ). Log L
F IR ranges from 41.8 to 45.5, with a mean value of 44.1 ±
Log L
F IR and
Log L [O III] , the ratio was calculated and thedistribution of the ratio is shown as a histogram in Fig. 2 (left panel). No significantdifference can be found for the distribution of the ratio between our H O maser galaxies 14 –
41 42 43 44 45 4605101520253035 log L
FIR N u m be r o f ga l a x i e s %
38 39 40 41 42 43 440510152025 log L [OIII]
Fig. 1.— Distributions of the FIR luminosity ( log L
F IR , 76 sources, left panel) and the[OIII] λ log [ O III ], 46 sources, right panel) for H O maser sources (solidlines), in units of erg s − . For comparison, the distributions are also given for a Seyfert 2sample without known H O maser emission (dashed lines, 38 sources from Mulchaey et al.1994). All numbers are plotted on a percent scale (%). 15 –
35 36 37 38 39 40 41 42 43 4441.542.042.543.043.544.044.545.045.546.01 2 3 4 5 6 7024681012141618202224262830323436 N u m be r o f ga l a x i e s ( % ) log(L FIR /L OIII ) l og L F I R ( e r g s s - ) log L OIII (ergs s -1 ) Fig. 2.— Left panel: The distribution (on a percent scale) of the ratio between the FIR andthe [O III] line luminosity, corrected for extinction, of H O maser galaxies on a logarithmicscale. For comparison, the distribution is also given for a Seyfert 2 sample without knownH O maser emission (dashed lines, 38 sources from Mulchaey et al. 1994). Right panel: theFIR versus extinction-corrected [O III] luminosity. The straight line shows a linear fit to theunweighted data. 16 –and the Seyfert 2 galaxy sample not containing galaxies with known masers (Mulchaeyet al. 1994). This is in agreement with the fact that most H O maser sources have beenfound in Seyfert 2 systems. The good agreement between Seyfert 1 and Seyfert 2 galaxies(Mulchaey et al. 1994) with respect to this parameter, in spite of different viewing angles,lends further support to an approximately isotropic emission of both tracers. In orderto further compare those two isotropic tracers, the FIR luminosity is plotted against theextinction-corrected [O III] luminosity of our maser sample in the right panel of Fig. 2. Acorrelation is found, log L
F IR =(30.86 ± ± [O III] with a Spearman’s rankcorrelation coefficient r=0.66 and a chance probability < × − . Assuming the corrected[O III] line luminosity to be a good isotropic tracer (Bassani et al. 1999), the strongcorrelation between FIR luminosity and [O III] line luminosity may suggest that the FIRluminosity is to some extent also an indicator of intrinsic nuclear activity, even though theFIR flux may be contaminated by a spatially extended starburst component. As mentioned above, nuclear absorbing columns can be obtained by analyzing X-rayspectra. These analyses are model dependent. Comparing the observed 2-10 keV X-rayemission from the nuclear region, absorbed by the obscuring material along the line-of-sight,with the intrinsic nuclear power provides another method for evaluating the absorbingcolumn density. The ratio of the observed 2-10 keV luminosity and the luminosity of thenuclear isotropic indicators was assumed to represent the optical thickness parameter,which allows us to diagnose the gas absorption (Bassani et al. 1999). Here the L X /L [O III] and L X /L F IR ratios are used as optical thickness parameters. In addition, high EW ( K α )values of the iron emission line are considered as a qualitative feature indicating a heavilyobscured nucleus (e.g., Maiolino & Risaliti 2007). Combining the EW ( K α ) and the optical 17 – -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.51.52.02.53.03.5 -4.5 -4.0 -3.5 -3.0 -2.5 -2.0 -1.5 -1.01.52.02.53.03.5 NGC2639 l og E W ( F e ) [ e V ] log T Mrk3 NGC4945 NGC4945 Mrk3 l og E W ( F e ) [ e V ] log T NGC2639
Fig. 3.— EW ( K α ) of the iron line versus the thickness parameters L − keV /L [O III] (T1,left panel) and L − keV /L F IR (T2, right panel). Square and pentacle symbols representCompton-thick and -thin sources as identified by modeling their X-ray spectra (e.g., Zhanget al. 2006). The bold squares show those Compton-thick sources, which have high EW(K α ) and low T values. These are located in the upper-left regions of both panels. Three exceptional sources NGC 2639, Mrk 3, and NGC 4945 are marked. For NGC 4945, the lowerlimit of its [O III] luminosity is taken. Details are given in Sect. 3.2. 18 –thickness parameters, we probe the circumnuclear environment of the maser sources.For the maser galaxies with available data (EW ( K α ), L X , L [O III] and L F IR , in total31 AGN maser sources), the EW (K α ) of the iron lines is plotted against the opticalthickness parameters in Fig. 3. The prominent feature of the figure is the existence of ananti-correlation, which is similar to that obtained for the Seyfert 2 sample of Bassani et al.(1999), which is predominantly containing targets without known maser lines. In the leftpanel (EW ( K α ) v.s. L X /L [O III] ), a least-squares fit shows log (EW ( K α ))=(2.74 ± ± X /L [O III] , with Spearman’s rank correlation coefficient R=-0.57 and achance probability of P ∼ α ) and low L X /L [O III] values. This area is marked by dashed lines.The approximate boundaries between Compton-thick and -thin sources, the latter shownby pentacles, are log (L X /L [O III] ) ∼ L − keV ∼ L [O III] ) and log EW ( K α ) ∼ ∼
300 eV). For comparison with Seyfert 2 galaxies without detected H O maseremission, the Seyfert 2 sample of Bassani et al. (1999) is used, excluding those objects withmaser emission. The trend is the same for this sample (here not shown). Compton-thicksources fall again into the upper left region with high EW (K α ) and low L X /L [O III] values,while Compton-thin sources are located in the lower right. The right panel of Fig. 3 showsa similarly clear anti-correlation between EW (K α ) and L X /L F IR . Linear fitting results inlog (EW ( K α ))=(1.71 ± ± X /L F IR , with R=-0.61 and P ∼ × − .Most of those sources with high EW (K α ) and low L X /L F IR value (upper-left region, insquare symbols) are Compton-thick sources, while those sources with low EW (K α ) andhigh L X /L F IR value (in pentacles) are Compton-thin as determined from X-ray spectralfitting. The approximate boundaries between Compton-thick and -thin environments arein this case log L X /L F IR ∼ -2.75 (i.e., L FIR ∼ L − keV ) and log EW ( K α ) ∼ ∼
300 eV. 19 –We like to emphasize that these boundaries are not arbitrary. Among the 20 Compton-thick candidate sources from our sample, classified by conventional X-ray spectroscopy, wefind 19 in each of our limited “Compton-thick” regions, related either to L X /L [O III] or toL X /L F IR . 18 of these sources are identical (see also Sect. 4).The results obtained so far could be affected by systematic errors in the measurements.Assuming for the [O III] values uncertainties of ∼
20% (Dahari & De Robertis 1988) resultsin L X /L [O III] errors of order 0.08 dex. Obviously, this does not affect the robust fundamentaltrend in our diagnostic Fig. 3. Compton-thick sources are still placed in the upper left andCompton-thin sources in the lower right. The anti-correlation between the EW ( K α ) andthe optical thickness parameters are readily explained. With an increase of the absorbingcolumn density, the X-ray luminosity will decrease so that L X /L [O III] and L X /L F IR arereduced with respect to L [O III] and L
F IR . On the other hand, the EW ( K α ) values willincrease, since these are measured against a reduced 6.4 keV continuum level.With Compton-thick galaxies being located in the upper left part of the panels in Fig.3, we find for our H O maser galaxies three criteria hinting at a Compton-thick nuclearenvironment: EW ( K α ) >
300 eV, L − keV < L [O III] and L FIR > L − keV . These areindependent of the detailed shape of the X-ray spectrum. O maser emission
The search for new extragalactic H O masers is ongoing and important with respect toseveral key aspects of modern astrophysics (see, e.g., Sect. 1 and Braatz et al. 2009). Herewe therefore investigate possible indicators of H O maser emission. We analyze relationshipsbetween H O maser luminosity and the iron line EW (K α ), FIR, and [O III] luminosity.For 33 maser sources with available EW ( K α ) of the iron line in Table 1, we obtain 20 –Table 2: Activity types of H O maser host galaxies (79 sources with available type)
Type a ) SF-masers b ) Kilomasers of unknown origin c ) AGN-masers d ) the whole sampleSeyfert 2 0 4 45 49LINER 0 2 8 10Inter Sy. 0 1 2 3Seyfert 1 0 0 2 2SBG 5 1 0 6H II 4 1 1 6FR II 0 0 1 1Quasar 0 0 2 2a) Types from NED: Inter Sy., intermediate Seyfert types; LINER, low-ionization nuclear emission line region; SBG, StarBurstGalaxy; FR II, Fanaroff-Riley type II radio galaxy; H II, classified as H II region galaxy; b) SF-masers, H O masers associatedwith off-nuclear star formation regions; c) kilomasers without known type; d) AGN-masers, H O masers associated with AGN. N u m be r o f ga l a x i e s % LogEW K
Fig. 4.— Distributions (on a percent scale) of the iron line EW ( K α ) (logarithmic scale, inunits of eV) for Seyfert 2s with detected H O maser emission (19 sources, solid lines) andnon-masing Seyfert 2s (34 sources from Bassani et al. 1999, dashed lines). 21 –an EW (K α ) mean value of 945 ±
135 eV. X-ray observations show that strong iron lineemission is common in the spectra of AGN hosting H O masers. It is interesting to checkif the 6.4 keV line can be used as a criterion to search for AGN masers. Have Seyfert 2swith detected maser emission higher EW (K α ) values than non-maser Seyfert 2s? Ourstatistical results show that the iron line EW ( K α ) of masing Seyfert 2 galaxies (mean value1063 ±
169 eV and median value ∼
800 eV for our 19 maser sources) is higher than that ofthe non-maser Seyfert 2 sample (mean value 375 ±
60 eV and median value ∼ K α ) values were taken almostexclusively (except four sources) from XMM-Newton or Chandra observations. The resultsfor Seyfert 2s without known masers (Bassani’s sample) come mostly from ASCA data oflower sensitivity, which might lead to an increase in the real average (only sources with astrong iron line could be detected). This strengthens our result and amplifies the differencebetween Seyfert 2s with and without detected 22 GHz H O maser. Nevertheless, we considerour result as tentative.In order to investigate possible correlations between H O maser and iron emissionlines, the EW ( K α ) of the Fe line was plotted against the isotropic H O luminosity in Fig. 5(upper panel). As already mentioned, H O maser emission can be produced by collisionalpumping in a dense molecular layer, which is heated by irradiated X-rays from the nucleus(Neufeld et al. 1994). Strong Fe K α emission is believed to be produced via X-ray reflectionby the cold iron in the circumnuclear region (e.g., Fabian et al. 2000). In those cases whereH O maser and iron line emission are detected, the nuclear X-ray emission plays a key role.Correlations between maser and iron line emission are therefore expected. However, our 22 – l og E W ( F e ) l og L [ O III] l og L F I R log L H2o
Fig. 5.— Upper panel: the EW ( K α ) of the iron line (logarithmic scale, in units of eV) versusisotropic luminosity of the H O maser emission (logarithmic scale, in L ⊙ ). Center: [O III] lineluminosity against L H O ; Bottom: FIR luminosity against L H O , empty circles show GalacticH O masers from Jaffe et al. (1981) and the line marks the correlation of L H O / L [FIR] ∼ − .Squares and triangles represent AGN-masers and masers in star formation regions respec-tively. Disk-masers, as a subsample of AGN-masers, for which maser emission is anticipatedto be particularly well connected with indicators of the intrinsic nuclear power, are markedby crosses over squares. 23 –results show no apparent trend between the iron line EW (K α ) and the isotropic H O maserluminosity, neither for the entire sample nor the two subsamples, AGN- and star-formingmasers (squares and triangles, respectively, in Fig. 5). In view of alternative H O excitationmechanisms (see, e.g., Lo 2005), the subsample of possible disk-masers (with detected highvelocity maser features) was analyzed separately (see the crosses in Fig. 5). However, evenin this case no significant correlation can be found.The [O III] and FIR luminosities were also plotted against H O maser luminosity inFig. 5 (central and bottom panel respectively). There is no significant correlation between L [O III] and L H O , although maser sources related to star formation seem to have lower[O III] luminosities than AGN masers. For FIR versus H O maser luminosities, thereappears to be a correlation, similar to that previously found by Henkel et al. (2005),Castangia et al. (2008), Bennert et al. (2009) and Surcis et al. (2009). The relationwas first found for galactic star forming regions by Jaffe et al. 1981, i.e., luminous H Omasers form in star formation regions with high FIR luminosity. For comparison, valuesof Galactic H O masers are also plotted in Fig. 5 (empty circles) and the line showsthe correlation of L H O / L [FIR] ∼ − from Jaffe et al. (1981). Apparently, there exists acorrelation between FIR and H O maser luminosity over many orders of magnitude. Whenconsidering AGN-masers only (squares in Fig. 5, lowest panel), the strongest masers appearto be overluminous with respect to the L F IR - L H O correlation. This is likely caused by thedifferent properties of AGN versus star-forming masers.
4. Discussion
Compton-thick nuclei are known to contribute a significant fraction of the hard X-raybackground. Their density as a function of redshift is also a relevant parameter whenstudying the evolution of the universe. Thus our newly defined criteria identifying such 24 –nuclei may be helpful when trying to reach this goal.Based on our new approach introduced in Sect. 3.2, 18 H O maser sources in theupper left part of the panels in Fig. 3 (see the solid squares) can be considered tobe Compton-thick. These are NGC 591, NGC 1068, Mrk 1066, NGC 1386, NGC 2273,UGC 5101, NGC 2782, NGC 3079, IC 2560, NGC 3393, Arp 299, NGC 5194, Mrk 266,NGC 5347, Circinus, NGC 5643, NGC 5728 and NGC 6240. Comparing this with Zhang etal. (2006), there are five new sources, NGC 591, UGC 5101, NGC 3072, IC 2560, Mrk 266.About 60% of the 31 AGN masers turn out to be Compton-thick. This is consistent withthe result found by Greenhill et al. (2008). For maser sources associated with Seyfert 2nuclei, 60% (15/25) are Compton-thick, which is, however, not significantly, higher thanthat of Seyfert 2 objects without detected maser emission ( ∼ exceptional sources, which show the limits of ourselection criteria. For NGC 4945, its low [O III] line flux places the source outside ourCompton-thick border (EW ( K α )- L X / L [O III] , see left panel in Fig. 3). However, the [O III]flux from Risaliti et al. (1999) only gives a very stringent lower limit to the actual intrinsic[O III] emission. The lack of a reliable [O III] flux is thought to be due to high absorptionin the edge-on galactic disk, instead of its intrinsic weakness. This is supported by itshard X-ray spectrum, which indicates that NGC 4945 hosts one of the brightest AGN inthe hard X-ray range ( >
100 keV) and can therefore be considered to contain a ‘bona-fide’Compton-thick nucleus (e.g., Guainazzi et al. 2000b). Similar to NGC 4945, Mrk 3 is alsoconsidered to be a ‘Bona-fide’ Compton-thick Seyfert 2 from its large brightness in the hardX-ray range (e.g., Cappi et al. 1999). However, the source is found outside our limited EW( K α )-L X /L F IR region (right panel in Fig. 3) for its relatively low value of the infrared flux.More constraints are desirable to probe its circumnuclear environment. The disk-maser 25 –galaxy NGC 2639 is located inside the required EW ( K α )-L X /L F IR region and outsidethe EW ( K α )-L X /L [O III] area. However, large uncertainties in the X-ray results (ASCAobservations) have to be noted. NGC 2639 is a weak X-ray source with an ASCA count rate < − . The ASCA data were analyzed by Wilson et al. (1998) and Terashima etal. (2002) and the source was considered to be Compton-thin ( N H ∼ × cm − ). Lowphoton statistics lead to uncertainties in the fitting models and Chandra observation aretherefore needed to investigate its highly obscured nucleus.We conclude with some cautionary notes. First, uncertainties are involved when usingthe [O III] λ Omegamaser galaxies may mainly arise from the AGN environment, but there is possiblecontamination from a starburst component and it is unclear how much it contributes.Second, the EW (K α ) of the iron line can be affected by other factors, such as the geometryof the accretion disk and the inclination angle at which the reflecting surface is viewed (e.g.,Fabian et al. 2000 , Bianchi et al. 2005). A high EW (K α ) of the iron line can also appearif the radiation is anisotropic or if there is a time lag between a drop in the continuum andthe line emission (Bassani et al. 1999). With future advanced X-ray telescopes, sensitiveobservations of more H O maser host galaxies at higher energies (above 10 keV) will provideimportant complementary information, further constraining nuclear column density.
5. Summary
In this paper, multi-wavelength data from the complete sample of galaxies ( D>
100 kpc)so far reported to host 22 GHz H O masers are analyzed, including the equivalent width 26 –(EW) of the iron K α line, the [O III] λ K α ) and the optical thickness parameters ( L X / L [O III] , L X / L F IR ) are combinedhere to probe the obscuration of maser host AGN. The main results are summarized below:(1) Our statistical analysis shows obvious anti-correlations between the EW ( K α ) of theFe emission line and the two optical thickness parameters. Without requiring a full X-rayspectrum, Compton-thick nuclear environments can be identified with these parametersand are found to be characterized approximately by EW ( K α ) >
300 eV, L − keV < L [O III] and L FIR > L − keV ;(2) 18 H O maser sources matching these criteria are identified to be Compton-thick. Acomparison with Zhang et al. (2006) shows, that among these there are five newly identifiedH O maser galaxies which are Compton-thick, i.e., NGC 591, UGC 5101, NGC 3072, IC 2560and Mrk 266. Masers associated with Seyfert 2 nuclei may be more likely Compton-thick(60%) than Seyfert 2s without detected maser emission ( ∼ O maser surveys, new ways to find extragalacticH O sources are also explored. H O maser sources may show larger EW (K α ) values thannon-maser Seyfert 2s, which, however, also needs further support. No significant correlationshave been found between EW (K α ), L [O III] and L H O . There appears a linear correlationbetween L F IR and L H O , which is consistent with the correlation found for Galactic H Omasers. However, the strongest H O masers appear overluminous with respect to their L F IR . This may be related to their different origin when compared with masers associatedwith sites of massive star formation well outside the nuclear region of their parent galaxy. 27 –We wish to thank the anonymous referee for many detailed and constructive commentsas well as P. Castangia for critically reading the manuscript. This work is supportedpartly by the National Natural Scientific Foundation of China (10633010) and GuangDongprovince Natural Science Foundation (8451009101001047). We made use of the NASA/IPACextragalactic Database (NED), High-Energy Astrophysics Science and Research Center(HEASARC) and the NASA Astrophysics Data System Bibliographic Services (ADS). 28 –
REFERENCES