ABSTRACT
We study the correlation between the [O iii] λ5007 and X-ray luminosities of local active galactic nuclei (AGNs), using a complete, hard X-ray (>10 keV) selected sample in the Swift/BAT 9-month catalog. From our optical spectroscopic observations at the South African Astronomical Observatory and the literature, a catalog of [O iii] λ5007 line flux for all 103 AGNs at Galactic latitudes of is compiled. Significant correlations with intrinsic X-ray luminosity () are found for both observed () and extinction-corrected () luminosities, separately for X-ray unabsorbed and absorbed AGNs. We obtain the regression form of and from the whole sample. The absorbed AGNs with low (<0.5%) scattering fractions in soft X-rays show on average smaller and ratios than the other absorbed AGNs, while those in edge-on host galaxies do not. These results suggest that a significant fraction of this population is buried in tori with small opening angles. By using these versus correlations, the X-ray luminosity function (LF) of local AGNs (including Compton-thick AGNs) in a standard population synthesis model gives much better agreement with the [O iii] λ5007 LF derived from the Sloan Digital Sky Survey than previously reported. This confirms that hard X-ray observations are a very powerful tool to find AGNs with high completeness.
Export citation and abstract BibTeX RIS
1. INTRODUCTION
In order to reveal the growth history of supermassive black holes (SMBHs) in galactic centers, it is crucial to completely survey all types of active galactic nuclei (AGNs) in the universe. According to the unified scheme of AGNs (Antonucci 1993), an SMBH is surrounded by a bagel-shaped, dusty torus and only the viewing angle determines the observable nature of an AGN; one sees type 1 and type 2 AGNs when the line of sight is unblocked and blocked by the torus, respectively, which causes dust extinction of optical lights from the accretion disk and broad-line region and photoelectric absorption (plus Compton scattering) of the primary X-ray emission. Basically, the unified scheme seems fairly successful to explain many aspects of AGN phenomena. The spectrum of the X-ray background indicates that a dominant population of AGNs are type 2 (obscured) AGNs (e.g., Ueda et al. 2014). Hence, surveys only using the broad emission lines or soft X-rays could easily miss the main population of AGNs.
Emission lines from the narrow-line region (NLR), which is located outside the inner torus region, should be observable from both type 1 and type 2 AGNs unless the SMBH is entirely surrounded by the torus. Thus, as long as the unified scheme holds, narrow emission lines induced by an AGN, such as [O iii] λ5007, have been considered to be a useful indicator of the AGN luminosity, even in Compton-thick AGNs whose "observed" X-ray flux below 10 keV is significantly attenuated (e.g., LaMassa et al. 2009). If, however, there is a wide scatter between the line luminosity and the intrinsic AGN luminosity, surveys based on the narrow lines may be subject to strong selection effects. Also, optical and ultraviolet lines are very sensitive to extinction by interstellar dust in the host galaxy and by circumnuclear dust that may be present around the NLR. Note that contamination from star-forming activities in the host galaxy may make it difficult to compose a clean AGN sample based on the [O iii] λ5007 flux (Simpson 2005; Toba et al. 2014).
Hard X-ray observations at rest-frame energies above 10 keV are able to provide the least biased AGN samples against obscuration thanks to their strong penetrating power, except for heavily Compton-thick AGNs with column densities of log ≳ 25 (Tueller et al. 2008). From these surveys, AGNs with very low scattering fractions in soft X-rays have been discovered (Ueda et al. 2007), many of which were missed in previous optical surveys because of their weak [O iii] λ5007 emission. It has been suspected that the AGNs might be buried in very geometrically thick tori, although Hönig et al. (2014) suggest that a part of them may be subject to interstellar absorption by the host galaxy. In geometrically thick tori with small opening angles, the AGN should have fainter intrinsic [O iii] λ5007 luminosity relative to the hard X-ray luminosity compared with classical Seyfert 2 galaxies, because much less of the nuclear flux leaks out to ionize the NLR. An extreme case can be found in ultraluminous infrared galaxies that contain buried AGNs almost entirely surrounded by Compton-thick matter (Imanishi et al. 2007; Ichikawa et al. 2014).
Thus, AGN selections using [O iii] λ5007 line and hard X-rays are considered to be complementary to each other in detecting obscured populations. It is therefore very important to study the correlations between [O iii] λ5007 and hard X-ray luminosities so that we can compare the statistical quantities of AGNs (such as luminosity function [LF]) obtained from these different surveys and evaluate the completeness and cleanness of each selection. For this study, we need to use a statistically complete sample of all types of AGNs with well-known properties.
Following early works by Mulchay et al. (1994) for Seyfert 1s and 2s and by Polletta et al. (1996) and Alonso-Herrero et al. (1997) for Seyfert 2s, several authors have studied correlations between [O iii] λ5007 and hard X-ray luminosities, using various samples of local AGNs (e.g., Heckman et al. 2005; Netzer et al. 2006; Panessa et al. 2006; Meléndez et al. 2008; Lamastra et al. 2009). Heckman et al. (2005) and Meléndez et al. (2008) use observed [O iii] λ5007 luminosities (hereafter ), while the others use those corrected for extinction (hereafter ). Among these works, only Meléndez et al. (2008), who focused more on the [O iv] 25.89 μm line, used an AGN sample based on hard X-ray surveys above 10 keV, although the sample is not statistically complete and is limited in number (40). The quantitative results of the [O iii] λ5007 and X-ray luminosity correlation obtained so far have been a little puzzling. From combined samples of type 1 and type 2 AGNs, Panessa et al. (2006) obtained a regression of the form whereas Lamastra et al. (2009) found an almost linear correlation of
In this paper, we investigate the correlation between the [O iii] λ5007 and X-ray luminosities, using a complete sample consisting of 103 objects at Galactic latitudes of in the Swift/BAT 9-month hard X-ray survey (Tueller et al. 2008). To follow up sources in the southern hemisphere, many of which did not have optical spectra, we conducted systematic optical spectroscopic observations at the SAAO. Then we complement it with a compilation from the literature, including Winter et al. (2010), where the optical spectra of Swift/BAT 9-month AGNs in the northern sky are analyzed. Section 2 describes the sample, optical observations, and data reduction and presents the catalog of the [O iii] λ5007 flux, together with those of narrow Hα and Hβ lines whenever available. In Section 3, we present the results of correlation analysis between the [O iii] λ5007 (or narrow Hα) luminosity and intrinsic (de-absorbed) X-ray luminosity for different types of AGNs. We then discuss the origin of these correlations and compare [O iii] λ5007, Hα, and X-ray LFs of local AGNs in Section 4. The conclusions are summarized in Section 5. Throughout the paper, we adopt H0 = 70 km s−1 Mpc−1, ΩM = 0.3, and ΩΛ = 0.7.
2. THE OPTICAL SPECTROSCOPY DATA
2.1. Parent Sample
For our study, we utilize the Swift/BAT 9-month catalog (Tueller et al. 2008) to define a complete sample of hard X-ray selected AGNs in the local universe. The Tueller et al. (2008) catalog contains 137 AGNs in total, excluding blazars at a flux limit of 2 × 10−11 erg cm−2 s−1 in the 14–195 keV band with detection significance above 4.8σ. To minimize the effects of extinction by Galactic interstellar medium, we limit the sample to those located at high Galactic latitudes of for our optical spectral studies. We exclude Cen A, which is a very nearby object, and SWIFT J0350.1–5019, which likely is confused by two AGNs, PGC 13946 and ESO 201-IG 004 (C. Ricci et al. 2015, in preparation). These selections leave 103 AGNs that constitute our "parent" sample (hereafter "Sample A").
The Swift/BAT AGNs are extensively followed up by X-ray observatories covering below 10 keV, such as Swift/XRT, XMM-Newton, Suzaku, and Chandra. Key spectral parameters in our study are the absorption column density () and the fraction of scattered component () for absorbed AGNs, which are often obtained by utilizing a partially covered absorber (or its equivalent) model. Because there can be other soft X-ray components that are spatially unresolved from the AGN emission, the value determined in this way is an upper limit to the true scattering fraction. Here we basically adopt the results of spectral analysis summarized in Table 1 of Ichikawa et al. (2012), which was largely based on Winter et al. (2009a) and was revised from (then) available Suzaku results for some targets. In our paper, we further revised their table by referring to later papers utilizing Suzaku data for more objects. Furthermore, for sources whose spectral parameters were not well constrained by using only the Swift/XRT data in Winter et al. (2009a), we update their spectral parameters according to C. Ricci et al. (2015, in preparation), who perform uniform broadband spectral analysis in the 0.3–150 keV band by including Swift/BAT spectra for the whole AGN sample of the Swift/BAT 70-month catalog. We also utilize the 70-month averaged, de-absorbed 2–10 keV flux of the primary continuum listed in the C. Ricci et al. (2015, in preparation) catalog, as well as the 9-month averaged 14–195 keV flux in the original Tueller et al. (2008) catalog.
We divide the sample into two types, X-ray unabsorbed AGNs (hereafter "X-ray type 1 AGNs") and absorbed AGNs ("X-ray type 2 AGNs"), which have absorptions of log < 22 cm−2 and log ≥ 22, respectively. Among X-ray type 2 AGNs, we call those with < 0.5% the low scattering fraction AGNs (so-called new type AGNs), a putative population of AGNs deeply buried by geometrically thick tori. Owing to our revision of the X-ray spectral parameters in the original Swift/BAT 9-month catalog, the sample of low scattering fraction AGNs has also been updated14 from that originally defined in Ichikawa et al. (2012).
Table 1 lists the targets of Sample A with their basic X-ray properties: source number in Tueller et al. (2008) (** are attached to the low scattering fraction AGNs), source name, redshift, , , observed luminosity in the 14–195 keV band (9-month average), absorption-corrected 2–10 keV luminosity (70-month average), and reference for the X-ray spectral parameters. Though not listed in Table 1, we also compile the information on the inclination angle of the host galaxy, , using the HyperLeda database, which is available for 98 AGNs.15 In addition, the black hole mass (and hence an estimate of Eddington ratio) is available for 99 AGNs16 from Winter et al. (2009a).
Table 1. X-Ray and Optical Emission Line ([O iii] λ5007, Hα, Hβ) Properties of AGNs in the 9-Month Swift/BAT Catalog
No. | Object | z | References | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
(erg s−1) | (erg s−1) | (cm−2) | (erg s−1) | (erg s−1) | (erg cm−2s−1) | (erg cm−2 s−1) | (erg cm−2 s−1) | (X-ray) | (Optical) | ||||
(1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) | (9) | (10) | (11) | (12) | (13) | (14) |
1 | NGC 235A | 0.0222 | 43.56 | 43.22 | 23.50 | 0.003 | 42.17 ± 0.43 | 41.23 ± 0.10 | −12.82 ± 0.10 | −13.17 ± 0.10 | −13.96 ± 0.10 | (xa) | (oa) |
2 | Mrk 348 | 0.0150 | 43.68 | 43.30 | 23.20 | 0.004 | 41.95 ± 0.01 | 41.07 ± 0.01 | −12.64 ± 0.01 | −13.02 ± 0.01 | −13.80 ± 0.01 | (xb) | (ob) |
3 | Mrk 352 | 0.0149 | 43.27 | 42.74 | 20.75 | ⋯ | ⋯ | 40.39 ± 0.01 | −13.31 ± 0.01 | ⋯ | ⋯ | (xc) | (oc) |
4 | NGC 454 | 0.0121 | 42.88 | 42.17 | 23.30 | 0.030 | >40.87 | 40.41 ± 0.20 | −13.11 ± 0.20 | −13.13 ± 0.20 | <−13.77 | (xa) | (oa) |
5 | Fairall 9 | 0.0470 | 44.39 | 44.15 | 20.36 | ⋯ | ⋯ | 42.15 ± 0.20 | −12.56 ± 0.20 | ⋯ | ⋯ | (xc) | (oa) |
6 | NGC 526A | 0.0191 | 43.63 | 43.07 | 22.18 | ⋯ | 41.44 ± 0.43 | 41.32 ± 0.10 | −12.59 ± 0.10 | −13.08 ± 0.10 | −13.59 ± 0.10 | (xd) | (oa) |
7 | NGC 612 | 0.0298 | 43.81 | 43.45 | 24.05 | 0.006 | 40.09 ± 0.92 | 40.09 ± 0.13 | −14.22 ± 0.13 | −13.96 ± 0.11 | −14.41 ± 0.29 | (xe) | (od) |
8 | ESO 297-G018 | 0.0252 | 43.85 | 43.67 | 23.81 | 0.003 | 41.43 ± 1.02 | 40.84 ± 0.04 | −13.33 ± 0.04 | −13.44 ± 0.21 | −14.12 ± 0.27 | (xf) | (od) |
9 | NGC 788 | 0.0136 | 43.39 | 43.22 | 23.67 | 0.007 | 40.91 ± 0.49 | 40.91 ± 0.18 | −12.71 ± 0.18 | −13.29 ± 0.01 | −13.03 ± 0.16 | (xd) | (oe) |
10 | Mrk 1018 | 0.0424 | 44.17 | 43.61 | 0.00 | ⋯ | 41.71 ± 0.65 | 41.71 ± 0.09 | −12.91 ± 0.09 | −13.46 ± 0.20 | −13.91 ± 0.09 | (xa) | (oe) |
12 | Mrk 590 | 0.0264 | 43.77 | 42.71 | 20.43 | ⋯ | 42.07 ± 0.12 | 41.74 ± 0.04 | −12.46 ± 0.04 | −13.00 ± 0.01 | −13.59 ± 0.04 | (xc) | (oe) |
15 | NGC 931 | 0.0167 | 43.66 | 43.41 | 21.56 | ⋯ | 41.91 ± 0.03 | 41.14 ± 0.01 | −12.66 ± 0.01 | −13.07 ± 0.01 | −13.81 ± 0.01 | (xd) | (od) |
16 | NGC 985 | 0.0430 | 44.21 | 43.78 | 21.59 | ⋯ | 42.75 ± 1.99 | 41.92 ± 0.01 | −12.72 ± 0.01 | −13.00 ± 0.01 | −13.75 ± 0.68 | (xc) | (od) |
17 | ESO 416-G002 | 0.0592 | 44.42 | 43.57 | 20.43 | ⋯ | ⋯ | 41.63 ± 0.20 | −13.29 ± 0.20 | ⋯ | ⋯ | (xc) | (oa) |
18 | ESO 198-024 | 0.0455 | 44.27 | 43.50 | 21.00 | ⋯ | ⋯ | 41.22 ± 0.20 | −13.46 ± 0.20 | ⋯ | ⋯ | (xc) | (oa) |
20 | NGC 1142 | 0.0289 | 44.17 | 43.88 | 23.80 | 0.003 | 42.03 ± 0.03 | 41.07 ± 0.01 | −13.21 ± 0.01 | −13.30 ± 0.01 | −14.10 ± 0.01 | (xf) | (od) |
23 | PKS 0326–288 | 0.1080 | 44.84 | 44.42 | 23.82 | 0.009 | 43.58 ± 0.85 | 42.52 ± 0.20 | −12.95 ± 0.20 | −13.15 ± 0.20 | −13.98 ± 0.20 | (xa) | (oa) |
24 | NGC 1365 | 0.0055 | 42.67 | 42.03 | 24.65 | 0.260 | 40.98 ± 0.01 | 39.62 ± 0.01 | −13.21 ± 0.01 | (8.7) | ⋯ | (xg) | (of) |
25 | ESO 548-G081 | 0.0145 | 43.19 | 42.91 | 20.00 | ⋯ | 41.94 ± 0.08 | 40.99 ± 0.01 | −12.69 ± 0.01 | −12.65 ± 0.01 | −13.45 ± 0.03 | (xd) | (od) |
28 | 2MASX J03565655–4041453 | 0.0747 | 44.51 | 43.70 | 22.66 | 0.032 | 42.56 ± 0.43 | 42.18 ± 0.10 | −12.95 ± 0.10 | −13.29 ± 0.10 | −13.90 ± 0.10 | (xa) | (oa) |
29 | 3C 105 | 0.0890 | 44.83 | 44.32 | 23.75 | 0.003 | 42.79 ± 0.10 | 41.59 ± 0.01 | −13.70 ± 0.01 | −14.10 ± 0.01 | −14.99 ± 0.03 | (xa) | (oe) |
31 | 1H 0419–577 | 0.1040 | 44.91 | 44.60 | 24.31 | ⋯ | 43.55 ± 0.43 | 43.16 ± 0.10 | −12.28 ± 0.10 | −12.64 ± 0.10 | −13.26 ± 0.10 | (xh) | (oa) |
32 | 3C 120 | 0.0330 | 44.45 | 43.98 | 21.20 | ⋯ | ⋯ | 41.86 ± 0.01 | −12.54 ± 0.01 | ⋯ | ⋯ | (xd) | (og) |
34 | MCG –01-13-025 | 0.0159 | 43.41 | 42.54 | 19.60 | ⋯ | 40.95 ± 0.03 | 40.73 ± 0.01 | −13.02 ± 0.01 | −13.21 ± 0.01 | −13.76 ± 0.01 | (xc) | (oe) |
36 | XSS J05054–2348 | 0.0350 | 44.24 | 43.49 | 23.47 | 0.009 | 41.65 ± 0.43 | 41.42 ± 0.10 | −13.03 ± 0.10 | −13.15 ± 0.10 | −13.71 ± 0.10 | (xf) | (oa) |
38 | Ark 120 | 0.0323 | 44.11 | 43.79 | 20.30 | ⋯ | ⋯ | 41.35 ± 0.01 | −13.03 ± 0.01 | ⋯ | ⋯ | (xd) | (oc) |
39 | ESO 362-G018 | 0.0126 | 43.26 | 42.88 | 23.43 | 0.087 | ⋯ | 40.88 ± 0.20 | −12.67 ± 0.20 | ⋯ | ⋯ | (xd) | (oa) |
40 | PICTOR A | 0.0351 | 43.80 | 43.45 | 20.78 | ⋯ | ⋯ | 41.57 ± 0.20 | −12.89 ± 0.20 | ⋯ | ⋯ | (xd) | (oa) |
45 | NGC 2110 | 0.0078 | 43.54 | 43.17 | 22.45 | 0.048 | 41.64 ± 0.01 | 40.36 ± 0.01 | −12.77 ± 0.01 | −12.66 ± 0.01 | −13.57 ± 0.01 | (xd) | (ob) |
47 | EXO 055620–3820.2 | 0.0339 | 44.14 | 43.07 | 22.41 | 0.034 | ⋯ | 41.46 ± 0.10 | −12.97 ± 0.10 | ⋯ | ⋯ | (xd) | (oa) |
49 | ESO 005-G004 | 0.0062 | 42.56 | 41.93 | 24.06 | 0.003 | ⋯ | <38.63 | <−14.30 | −13.80 ± 0.01 | <−15.39 | (xf) | (oh) |
50 | Mrk 3 | 0.0135 | 43.61 | 43.35 | 24.04 | 0.009 | 43.33 ± 0.01 | 42.31 ± 0.01 | −11.31 ± 0.01 | −11.65 ± 0.01 | −12.47 ± 0.01 | (xi) | (oe) |
51 | ESO 121-IG028 | 0.0403 | 44.03 | 43.63 | 23.31 | 0.004 | >40.86 | 40.86 ± 0.24 | −13.72 ± 0.24 | −13.69 ± 0.10 | <−13.76 | (xa) | (oa) |
53 | 2MASX J06403799–4321211 | 0.0610 | 44.40 | 43.51 | 23.00 | <0.011 | >41.74 | 41.74 ± 0.20 | −13.21 ± 0.20 | −13.19 ± 0.20 | <−13.33 | (xa) | (oa) |
55 | Mrk 6 | 0.0188 | 43.72 | 43.09 | 20.76 | ⋯ | 42.38 ± 1.56 | 42.38 ± 0.01 | −11.52 ± 0.01 | −11.97 ± 0.01 | −12.41 ± 0.53 | (xa) | (oe) |
56 | Mrk 79 | 0.0222 | 43.72 | 43.19 | 19.78 | ⋯ | 41.96 ± 0.12 | 41.96 ± 0.02 | −12.09 ± 0.02 | −12.66 ± 0.01 | −13.11 ± 0.04 | (xc) | (oe) |
60 | Mrk 18 | 0.0111 | 42.93 | 41.82 | 23.26 | 0.030 | 40.56 ± 0.28 | 40.56 ± 0.11 | −12.88 ± 0.11 | −12.72 ± 0.01 | −12.98 ± 0.09 | (xd) | (oe) |
61 | 2MASX J09043699+5536025 | 0.0370 | 44.03 | 43.31 | 20.78 | ⋯ | 41.91 ± 0.09 | 41.63 ± 0.03 | −12.87 ± 0.03 | −12.95 ± 0.01 | −13.52 ± 0.03 | (xd) | (oe) |
62 | 2MASX J09112999+4528060 | 0.0268 | 43.69 | 43.16 | 23.52 | 0.006 | 40.98 ± 0.11 | 39.68 ± 0.01 | −14.54 ± 0.01 | −14.49 ± 0.01 | −15.41 ± 0.04 | (xd) | (oe) |
64 | 2MASX J09180027+0425066 | 0.1560 | 45.31 | ⋯ | 23.05 | 0.013 | 42.59 ± 0.01 | 42.22 ± 0.01 | −13.60 ± 0.01 | −14.08 ± 0.01 | −14.68 ± 0.01 | (xd) | (oe) |
65 | MCG –01-24-012 | 0.0196 | 43.60 | 43.24 | 22.81 | 0.005 | 41.10 ± 0.15 | 41.10 ± 0.08 | −12.83 ± 0.08 | −13.14 ± 0.01 | −13.46 ± 0.05 | (xa) | (oe) |
66 | MCG +04-22-042 | 0.0323 | 43.99 | 43.46 | 20.59 | ⋯ | 42.12 ± 0.62 | 42.12 ± 0.20 | −12.26 ± 0.20 | −12.57 ± 0.01 | −12.72 ± 0.20 | (xc) | (oe) |
67 | Mrk 110 | 0.0353 | 44.19 | 43.86 | 20.20 | ⋯ | 42.71 ± 0.59 | 42.29 ± 0.19 | −12.17 ± 0.19 | −12.47 ± 0.01 | −13.09 ± 0.19 | (xd) | (oe) |
68 | NGC 2992 | 0.0077 | 42.94 | 41.93 | 22.08 | 0.524 | 42.51 ± 0.43 | 40.76 ± 0.10 | −12.36 ± 0.10 | −12.48 ± 0.10 | −13.55 ± 0.10 | (xd) | (oa) |
69 | MCG –05-23-016 | 0.0085 | 43.55 | 43.21 | 22.20 | ⋯ | >41.18 | 40.64 ± 0.10 | −12.56 ± 0.10 | −12.85 ± 0.10 | <−13.50 | (xd) | (oa) |
70 | NGC 3081 | 0.0080 | 43.09 | 42.96 | 23.99 | 0.006 | 41.66 ± 0.43 | 41.32 ± 0.10 | −11.83 ± 0.10 | −12.38 ± 0.10 | −12.97 ± 0.10 | (xe) | (oa) |
71 | NGC 3227 | 0.0039 | 42.63 | 42.05 | 22.24 | 0.148 | 41.26 ± 0.01 | 40.44 ± 0.01 | −12.09 ± 0.01 | −12.54 ± 0.01 | −13.29 ± 0.01 | (xd) | (ob) |
72 | NGC 3281 | 0.0107 | 43.27 | 42.69 | 23.94 | 0.019 | 41.42 ± 0.85 | 41.05 ± 0.20 | −12.36 ± 0.20 | −12.71 ± 0.20 | −13.31 ± 0.20 | (xd) | (oa) |
75 | Mrk 417 | 0.0328 | 43.95 | 43.73 | 23.93 | 0.002 | 41.19 ± 0.05 | 40.92 ± 0.01 | −13.48 ± 0.01 | −13.76 ± 0.01 | −14.33 ± 0.02 | (xd) | (oe) |
77 | NGC 3516 | 0.0088 | 43.26 | 42.72 | 21.55 | ⋯ | 41.54 ± 0.01 | 40.92 ± 0.01 | −12.32 ± 0.01 | (4.9) | ⋯ | (xd) | (oc) |
78 | RX J1127.2+1909 | 0.1055 | 44.79 | 43.84 | 0.00 | ⋯ | 43.02 ± 0.14 | 43.02 ± 0.11 | −12.43 ± 0.11 | −13.03 ± 0.01 | −13.42 ± 0.03 | (xa) | (oe) |
79 | NGC 3783 | 0.0097 | 43.53 | 43.30 | 21.76 | 0.278 | ⋯ | 41.47 ± 0.10 | −11.85 ± 0.10 | ⋯ | ⋯ | (xd) | (oa) |
80 | SBS 1136+594 | 0.0601 | 44.33 | 43.82 | 0.00 | ⋯ | 42.70 ± 0.06 | 42.55 ± 0.01 | −12.39 ± 0.01 | −12.92 ± 0.01 | −13.45 ± 0.02 | (xa) | (oe) |
81 | UGC 06728 | 0.0065 | 42.54 | 41.94 | 20.00 | ⋯ | 40.22 ± 0.01 | 40.22 ± 0.01 | −12.75 ± 0.01 | −12.34 ± 0.01 | −12.80 ± 0.01 | (xd) | (oe) |
82 | 2MASX J11454045–1827149 | 0.0330 | 43.98 | 43.64 | 0.00 | ⋯ | 42.19 ± 0.43 | 42.19 ± 0.10 | −12.21 ± 0.10 | −12.91 ± 0.10 | −13.36 ± 0.10 | (xa) | (oa) |
83 | CGCG 041–020 | 0.0360 | 43.88 | 43.39 | 23.03 | 0.009 | 41.17 ± 0.07 | 40.38 ± 0.01 | −14.10 ± 0.01 | −13.96 ± 0.01 | −14.70 ± 0.02 | (xd) | (oe) |
85 | NGC 4051 | 0.0023 | 41.74 | 41.33 | 20.46 | ⋯ | 40.41 ± 0.56 | 40.20 ± 0.18 | −11.87 ± 0.18 | −11.97 ± 0.01 | −12.52 ± 0.18 | (xc) | (oe) |
86 | Ark 347 | 0.0224 | 43.42 | 42.90 | 23.36 | 0.016 | 41.53 ± 0.22 | 41.53 ± 0.19 | −12.52 ± 0.19 | −13.08 ± 0.01 | −13.31 ± 0.04 | (xa) | (oe) |
87 | NGC 4102 | 0.0028 | 41.62 | 41.41 | 24.30 | ⋯ | 40.72 ± 0.02 | 38.78 ± 0.01 | −13.46 ± 0.01 | −12.48 ± 0.01 | −13.62 ± 0.01 | (xj) | (od) |
88 | NGC 4138 | 0.0030 | 41.62 | 41.23 | 22.90 | 0.012 | 38.95 ± 0.01 | 38.95 ± 0.01 | −13.35 ± 0.01 | −13.33 ± 0.01 | −13.72 ± 0.01 | (xd) | (od) |
89 | NGC 4151 | 0.0033 | 42.96 | 42.58 | 22.73 | 0.041 | 41.87 ± 0.24 | 41.87 ± 0.01 | −10.51 ± 0.01 | −11.12 ± 0.01 | −11.40 ± 0.08 | (xd) | (oe) |
90 | Mrk 766 | 0.0129 | 42.94 | 42.67 | 21.72 | ⋯ | 42.03 ± 0.13 | 41.79 ± 0.01 | −11.78 ± 0.01 | −12.10 ± 0.04 | −12.66 ± 0.02 | (xc) | (oe) |
91 | NGC 4388 | 0.0084 | 43.60 | 43.17 | 23.53 | 0.011 | 41.29 ± 0.13 | 41.29 ± 0.11 | −11.90 ± 0.11 | −12.33 ± 0.01 | −12.80 ± 0.03 | (xk) | (oe) |
92 | NGC 4395 | 0.0011 | 40.81 | 40.64 | 22.52 | 0.322 | 39.02 ± 0.01 | 38.93 ± 0.01 | −12.49 ± 0.01 | −12.81 ± 0.01 | −13.32 ± 0.01 | (xd) | (oe) |
94 | NGC 4507 | 0.0118 | 43.78 | 43.54 | 23.54 | 0.029 | 42.12 ± 0.43 | 41.76 ± 0.10 | −11.73 ± 0.10 | −12.10 ± 0.10 | −12.70 ± 0.10 | (xd) | (oa) |
95 | ESO 506-G027 | 0.0250 | 44.28 | 43.95 | 23.92 | 0.002 | >41.14 | 40.96 ± 0.20 | −13.19 ± 0.20 | −13.67 ± 0.20 | <−14.20 | (xl) | (oa) |
96 | XSS J12389–1614 | 0.0366 | 44.26 | 43.35 | 22.63 | 0.043 | 41.88 ± 0.85 | 41.87 ± 0.20 | −12.63 ± 0.20 | −12.81 ± 0.20 | −13.29 ± 0.20 | (xa) | (oa) |
97 | NGC 4593 | 0.0090 | 43.21 | 42.81 | 20.49 | ⋯ | 41.04 ± 0.01 | 40.50 ± 0.01 | −12.76 ± 0.01 | (4.6) | ⋯ | (xd) | (oc) |
100 | SBS 1301+540 | 0.0299 | 43.72 | 43.09 | 20.60 | ⋯ | 41.25 ± 0.21 | 41.25 ± 0.01 | −13.06 ± 0.01 | −13.54 ± 0.07 | −13.87 ± 0.01 | (xc) | (oe) |
102 | NGC 4992 | 0.0251 | 43.83 | 43.17 | 23.75 | <0.003 | 39.85 ± 0.16 | 39.85 ± 0.01 | −14.30 ± 0.01 | −14.50 ± 0.02 | −14.82 ± 0.05 | (xm) | (od) |
103 | MCG –03-34-064 | 0.0165 | 43.46 | 43.95 | 23.61 | 0.039 | 42.55 ± 0.43 | 42.21 ± 0.10 | −11.57 ± 0.10 | −12.13 ± 0.10 | −12.73 ± 0.10 | (xd) | (oa) |
105 | MCG –06-30-015 | 0.0077 | 43.00 | 42.74 | 21.28 | ⋯ | ⋯ | 40.23 ± 0.20 | −12.89 ± 0.20 | ⋯ | ⋯ | (xd) | (oa) |
106 | NGC 5252 | 0.0230 | 43.90 | 43.39 | 22.64 | 0.038 | 40.79 ± 0.01 | 40.37 ± 0.01 | −13.71 ± 0.01 | −13.32 ± 0.01 | −13.94 ± 0.01 | (xd) | (oe) |
108 | IC 4329A | 0.0160 | 44.24 | 43.84 | 21.79 | ⋯ | 41.62 ± 0.19 | 41.02 ± 0.04 | −12.74 ± 0.04 | −13.09 ± 0.04 | −13.77 ± 0.04 | (xd) | (od) |
109 | Mrk 279 | 0.0304 | 43.97 | 43.42 | 20.11 | ⋯ | 41.76 ± 0.21 | 41.68 ± 0.01 | −12.65 ± 0.01 | −13.01 ± 0.01 | −13.51 ± 0.07 | (xd) | (od) |
110 | NGC 5506 | 0.0062 | 43.30 | 42.91 | 22.44 | 0.011 | 41.89 ± 0.11 | 41.03 ± 0.08 | −11.90 ± 0.08 | −11.99 ± 0.01 | −12.76 ± 0.03 | (xd) | (oe) |
112 | NGC 5548 | 0.0172 | 43.59 | 43.14 | 20.85 | ⋯ | 42.13 ± 0.06 | 42.13 ± 0.02 | −11.69 ± 0.02 | −12.37 ± 0.01 | −12.74 ± 0.02 | (xd) | (oe) |
113 | ESO 511-G030 | 0.0224 | 43.73 | 43.41 | 20.99 | ⋯ | ⋯ | 40.62 ± 0.20 | −13.44 ± 0.20 | ⋯ | ⋯ | (xd) | (oa) |
115 | NGC 5728 | 0.0093 | 43.23 | 43.03 | 24.14 | 0.007 | 41.96 ± 0.43 | 41.52 ± 0.10 | −11.76 ± 0.10 | −12.23 ± 0.10 | −12.85 ± 0.10 | (xm) | (oa) |
116 | Mrk 841 | 0.0364 | 44.20 | 43.87 | 21.34 | ⋯ | 41.64 ± 0.04 | 41.64 ± 0.01 | −12.85 ± 0.01 | −13.37 ± 0.01 | −13.79 ± 0.01 | (xc) | (od) |
117 | Mrk 290 | 0.0296 | 43.79 | 42.93 | 21.18 | ⋯ | 41.71 ± 0.50 | 41.59 ± 0.01 | −12.71 ± 0.01 | −13.20 ± 0.01 | −13.72 ± 0.17 | (xd) | (od) |
118 | Mrk 1498 | 0.0547 | 44.50 | 44.05 | 23.10 | 0.016 | 42.43 ± 0.31 | 42.43 ± 0.20 | −12.42 ± 0.20 | −13.18 ± 0.01 | −13.05 ± 0.08 | (xf) | (oe) |
120 | NGC 6240 | 0.0245 | 43.81 | 44.16 | 24.25 | <0.005 | 42.12 ± 0.12 | 40.71 ± 0.02 | −13.43 ± 0.02 | −12.73 ± 0.01 | −13.69 ± 0.04 | (xa) | (oe) |
124 | 1RXS J174538.1+290823 | 0.1113 | 45.09 | 44.37 | 0.00 | ⋯ | ⋯ | 42.75 ± 0.03 | −12.75 ± 0.03 | ⋯ | ⋯ | (xa) | (oe) |
125 | 3C 382 | 0.0579 | 44.81 | 44.67 | 20.11 | ⋯ | 42.31 ± 0.08 | 41.70 ± 0.01 | −13.20 ± 0.01 | −13.36 ± 0.01 | −14.04 ± 0.03 | (xc) | (od) |
126 | ESO 103-035 | 0.0133 | 43.58 | 43.38 | 23.33 | 0.001 | 42.20 ± 0.85 | 40.87 ± 0.20 | −12.73 ± 0.20 | −12.80 ± 0.20 | −13.73 ± 0.20 | (xd) | (oa) |
127 | 3C 390.3 | 0.0561 | 44.88 | 44.52 | 21.08 | ⋯ | 42.96 ± 0.38 | 42.72 ± 0.01 | −12.15 ± 0.01 | −12.51 ± 0.13 | −13.06 ± 0.02 | (xd) | (od) |
129 | NGC 6814 | 0.0052 | 42.57 | 42.22 | 20.76 | ⋯ | ⋯ | 40.17 ± 0.10 | −12.61 ± 0.10 | ⋯ | ⋯ | (xc) | (oa) |
133 | NGC 6860 | 0.0149 | 43.39 | 42.89 | 21.00 | ⋯ | ⋯ | 40.93 ± 0.10 | −12.76 ± 0.10 | ⋯ | ⋯ | (xn) | (oa) |
136 | 4C +74.26 | 0.1040 | 45.14 | 44.87 | 21.25 | ⋯ | 43.37 ± 0.61 | 43.13 ± 0.17 | −12.31 ± 0.17 | −12.83 ± 0.01 | −13.39 ± 0.20 | (xo) | (oe) |
137 | Mrk 509 | 0.0344 | 44.43 | 44.08 | 20.18 | ⋯ | ⋯ | 42.17 ± 0.20 | −12.26 ± 0.20 | ⋯ | ⋯ | (xd) | (oa) |
138 | IC 5063 | 0.0114 | 43.31 | 43.08 | 23.40 | 0.009 | 42.26 ± 0.43 | 41.58 ± 0.10 | −11.88 ± 0.10 | −12.20 ± 0.10 | −12.91 ± 0.10 | (xh) | (oa) |
139 | 2MASX J21140128+8204483 | 0.0840 | 44.80 | 44.35 | 0.00 | ⋯ | >43.60 | 42.89 ± 0.01 | −12.35 ± 0.01 | −12.54 ± 0.04 | <−13.25 | (xa) | (od) |
144 | UGC 11871 | 0.0266 | 43.80 | 43.26 | 22.32 | 0.016 | 42.56 ± 0.01 | 41.45 ± 0.01 | −12.76 ± 0.01 | −12.25 ± 0.01 | −13.10 ± 0.01 | (xa) | (oe) |
145 | NGC 7172 | 0.0087 | 43.32 | 42.90 | 22.91 | 0.001 | 39.94 ± 0.85 | 39.94 ± 0.20 | −13.28 ± 0.20 | −13.55 ± 0.20 | −13.75 ± 0.20 | (xd) | (oa) |
146 | NGC 7213 | 0.0058 | 42.59 | 41.85 | 20.40 | ⋯ | 40.23 ± 0.85 | 40.23 ± 0.20 | −12.64 ± 0.20 | −12.39 ± 0.20 | −12.83 ± 0.20 | (xd) | (oa) |
147 | NGC 7314 | 0.0048 | 42.45 | 41.96 | 21.60 | ⋯ | 40.25 ± 0.43 | 39.78 ± 0.10 | −12.93 ± 0.10 | −13.21 ± 0.10 | −13.85 ± 0.10 | (xd) | (oa) |
148 | NGC 7319 | 0.0225 | 43.68 | 43.40 | 23.82 | 0.004 | 40.72 ± 0.20 | 40.72 ± 0.03 | −13.34 ± 0.03 | −13.68 ± 0.01 | −13.86 ± 0.07 | (xa) | (oe) |
149 | 3C 452 | 0.0811 | 44.73 | 43.83 | 23.36 | 0.064 | 42.09 ± 0.09 | 40.98 ± 0.01 | −14.23 ± 0.01 | −14.22 ± 0.01 | −15.07 ± 0.03 | (xd) | (oe) |
151 | MR 2251–178 | 0.0640 | 45.03 | 44.60 | 21.45 | ⋯ | 42.87 ± 0.43 | 42.33 ± 0.10 | −12.66 ± 0.10 | −12.89 ± 0.10 | −13.55 ± 0.10 | (xd) | (oa) |
152 | NGC 7469 | 0.0163 | 43.70 | 42.97 | 20.61 | ⋯ | 43.14 ± 0.01 | 41.70 ± 0.01 | −12.08 ± 0.01 | −11.89 ± 0.01 | −12.85 ± 0.01 | (xc) | (ob) |
153 | Mrk 926 | 0.0469 | 44.45 | 44.19 | 20.54 | ⋯ | 42.66 ± 0.03 | 42.66 ± 0.01 | −12.05 ± 0.01 | −12.68 ± 0.01 | −13.05 ± 0.01 | (xd) | (oe) |
154 | NGC 7582 | 0.0052 | 42.61 | 42.65 | 23.80 | 0.033 | 41.26 ± 0.43 | 40.38 ± 0.10 | −12.39 ± 0.10 | −12.10 ± 0.10 | −12.88 ± 0.10 | (xp) | (oa) |
Note. This table summarizes X-ray and optical emission line ([O iii] λ5007, Hα, Hβ) properties of 95 Swift/BAT 9-month AGNs in Tueller et al. (2008) excluding Cen A, blazars, and those at low Galactic latitudes (). Columns: (1) Source no. in Tueller et al. (2008); (2) object name; (3) redshift; (4) 9-month averaged 14–195 keV luminosity calculated from the observed flux; (5) absorption-corrected 2–10 keV luminosity of the transmitted component averaged for 70 months; (6) X-ray absorption hydrogen column density (0.00 means NH = 0); (7) soft X-ray scattering fraction; (8) [O iii] λ5007 luminosity corrected for extinction based on the Balmer decrement; (9) observed [O iii] λ5007 luminosity (with no extinction correction); (10) [O iii] λ5007 flux; (11) narrow Hα flux (number in parentheses refers to Hα/ Hβ flux ratio); (12) narrow Hβ flux; (13) reference for the X-ray spectra (columns (6) and (7)): (xa) C. Ricci et al. (2015, in preparation), (xb) Noguchi et al. (2010), (xc) Tueller et al. (2008), (xd) Winter et al. (2009a), (xe) Eguchi et al. (2011), (xf) Eguchi et al. (2009), (xg) Risaliti et al. (2009), (xh) Turner et al. (2009), (xi) Awaki et al. (2008), (xj) González-Martín et al. (2011), (xk) Shirai et al. (2008), (xl) Winter et al. (2009b), (xm) Comastri et al. (2010), (xn) Winter & Mushotzky (2010), (xo) Ballantyne (2005), (xp) Bianchi et al. (2009); (14) reference for the optical line fluxes (columns (8)–(12)): (oa) this work, (ob) Dahari & De Robertis (1988), (oc) Mulchay et al. (1994), (od) M. Koss et al. (2015, in preparation), (oe) Winter et al. (2010), (of) Bassani et al. (1999), (og) Xu et al. (1999), (oh) Landi et al. (2007). Columns (1)–(4) are taken from Tueller et al. (2008) except for the revised redshift of no. 53 (2MASX J06403799–4321211). Column (5) is taken from C. Ricci et al. (2015, in preparation). All luminosities are calculated from the redshift given in column (3) with (H0, Ωm, Ωλ) = (70 km s−1 Mpc−1, 0.3, 0.7).
Figure 1 plots the host inclination against log for Sample A. In all plots of our paper, the diagonal crosses correspond to X-ray type 1 (X-ray unabsorbed) AGNs and the filled circles to X-ray type 2 (X-ray absorbed) AGNs, among which the open circles denote those with low scattering fractions. As noticed, 9 objects out of 10 with > 85° are X-ray type 2 AGNs, rejecting the null hypothesis that X-ray absorption is independent of the host inclination at >98% confidence level. This is expected as galactic interstellar matter could produce an X-ray absorption of log > 22 when viewed edge-on, and it is in agreement with the deficiency of nearly edge-on Seyfert 1 galaxies reported by Keel (1980). Except for that, there is no correlation between and . These results are consistent with the random distribution of the orientation angle of the torus (or the accretion disk) with respect to that of the galactic plane, confirming previous findings (Schmitt et al. 2001). We note that the distribution of the AGNs with low scattering fractions in our sample is not concentrated at large values; more than half of this population is free from absorption by interstellar matter along the galactic disk, which therefore cannot account for their observed low scattering fractions. Indeed, a K-S test for the distribution between the low scattering fraction AGNs and the rest of X-ray type 2 AGNs in Sample A yields a matching probability of 0.53. By considering the small sample size, this does not necessarily contradict the statistical result by Hönig et al. (2014); they obtain a more edge-on dominant distribution of the same population based on a slightly larger sample collected from the literature, although some of their sample is revised in our paper (see Section 2.1). We can conclude that there are at least two origins for their low scattering fractions: (1) intrinsic nature of the nucleus and (2) interstellar absorption in the host galaxy.
2.2. Optical Observations at SAAO and Data Reduction
We performed optical spectroscopic observations of Swift/BAT AGNs visible in the southern sky (δ < −10°) by using the SAAO 1.9 m telescope with the Cassegrain spectrograph during four observation runs: 2007 July, 2008 January, 2008 August, and 2009 February, each consisting of roughly 14 nights. In this paper, we focus on sources in the Swift/BAT 9-month catalog, although our observation targets at the SAAO also include those in the Swift/BAT 22-month catalog, whose results will be reported in M. Koss et al. (2015, in preparation). In total, the spectra of 38 AGNs have been analyzed in this work.
We used the 300 line mm−1 grating, blazed at 6000 Å, covering about 4400–7600 Å, with a 2'' slit width placed on the center of each galaxy, producing a spectral resolution of ≈5 Å. The integration is split into a series of 150 s exposures, added up to a total integration time ranging from 750 to 3600 s. Wavelength calibration of the spectra was obtained from CuAr arc lamp exposures taken during the same night. A flux calibration was obtained from long-slit (with 6'' slit width) observations of spectrophotometric standard stars. To derive the sensitivity curve, we fit the observed spectral energy distribution of the standard stars with a low-order polynomial.
The spectral line flux of [O iii] λ5007 and those of narrow components of Hα and Hβ were measured using the IRAF task splot from the co-added, dispersion-corrected, and flux-calibrated spectra. If the lines were not significantly detected, we then estimated their upper limits (3σ) from the fluctuation of the noise level. The line fluxes are corrected for reddening from the Milky Way, by using the map by Schlafly & Finkbeiner (2011) and the reddening curve by Cardelli et al. (1989) with RV = 3.1. Finally, we approximately corrected these fluxes for the slit loss in the following way. For each dispersed spectrum, we projected the 4500–5500 Å region onto the spatial axis and measured its spread by fitting with a Gaussian. We then calculated the fraction contained within the slit by assuming that the image is axisymmetric. By comparing the results of the same target taken on different days when available, we estimate that the flux uncertainties are typically of 0.1–0.2 dex, depending on the quality of the spectrum. This is similar to general errors in the [O iii] λ5007 fluxes reported by Whittle (1992) when they are measured with small (2–4'') apertures. In some cases of broad-line AGNs, we were unable to reliably measure the fluxes (or the upper limits) of the narrow components of Hα and Hβ lines by separating them from the broad components.
2.3. Catalog
To complement the results from the SAAO observations, we gather [O iii] λ5007, Hα, and Hβ fluxes in the literature for AGNs in the northern sky. We mainly adopt the results summarized by Winter et al. (2010), except for those with too large uncertainties, and refer to other references (Mulchay et al. 1994; Bassani et al. 1999; Xu et al. 1999; Landi et al. 2007; M. Koss et al. 2015, in preparation) for the rest. We obtain constraints on the [O iii] λ5007 flux for all 103 AGNs (48 X-ray type 1 and 55 X-ray type 2 AGNs) of the parent sample defined in Section 2.1 (Sample A), where one object (No. 49) does not show detectable [O iii] λ5007 emission and hence has only an upper limit. Among them, 77 objects (31 X-ray type 1 and 46 X-ray type 2 AGNs) have reliable flux measurements (not upper limits) of both narrow Hα and Hβ emission lines, FHα and FHβ, or their flux ratios, constituting "Sample B."
Table 1 lists the observed [O iii] λ5007 luminosity () along with the fluxes of [O iii] λ5007, narrow Hα, and narrow Hβ lines for Sample A with the reference of the optical spectroscopic data. For Sample B, we also calculate an extinction-corrected luminosity of [O iii] λ5007 () from the Balmer decrement as
following Bassani et al. (1999). When the FHα/FHβ ratio is smaller than 3.0, we do not apply any correction. As discussed in Hao et al. (2005), however, the intrinsic flux ratio between Hα and Hβ in the NLR of an AGN could be different from the value assumed here, being subject to the gas density and radiative transfer effects. Also, there is an uncertainty in the correction because the spatial distributions of the [O iii] λ5007 and Balmer line emitting regions may not be the same owing to the clumpiness of the NLR (see Section 4.1). Thus, we should regard these corrections only as an approximation.
3. CORRELATIONS BETWEEN X-RAY AND OPTICAL LINE LUMINOSITIES
3.1. Regression Analysis between X-Ray and [O iii] λ5007 Luminosities
Figure 2 plots the correlation of the observed [O iii] λ5007 luminosity () against (a) the luminosity in the 14–195 keV band (logL14–195) or (b) that in the 2–10 keV band (logL2–10), using Sample A. Figure 3 shows the same but for the [O iii] λ5007 luminosity corrected for extinction (), using Sample B. For each plot, we evaluate the strength of the luminosity–luminosity and flux–flux correlations separately for X-ray type 1, X-ray type 2, and all (X-ray type 1 + type 2) AGNs; the resultant Spearman's rank coefficients and Student's t-null significance levels are summarized in Table 2. We also calculate the ordinary least-squares bisector regression lines of the luminosity–luminosity correlation with the form of , where Y is either or and X is either log L14–195 or log L2–10. The parameters and their 1σ errors are listed in Table 2. The best-fit lines obtained from all AGNs are plotted in Figures 2 and 3. When Sample A is used, we ignore the two objects whose [O iii] λ5007 luminosities are upper limits and restrict the luminosity range above log L2–10 > 41.
Download figure:
Standard image High-resolution imageDownload figure:
Standard image High-resolution imageTable 2. Correlation Properties between Different Luminosities
Y | X | Sample | N | PL | Pf | a | b | ||
---|---|---|---|---|---|---|---|---|---|
(1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) | (9) | (10) |
All | 101 | 0.672 | 0.419 | −11.0 ± 2.9 | 1.20 ± 0.07 | ||||
Type 1 | 48 | 0.781 | 0.240 | −8.0 ± 3.8 | 1.13 ± 0.09 | ||||
Type 2 | 53 | 0.461 | 0.548 | −10.3 ± 3.7 | 1.18 ± 0.09 | ||||
All | 100 | 0.630 | 0.375 | −10.0 ± 2.9 | 1.18 ± 0.07 | ||||
Type 1 | 48 | 0.761 | 0.239 | −5.5 ± 3.6 | 1.08 ± 0.09 | ||||
Type 2 | 52 | 0.447 | 0.488 | −10.2 ± 4.0 | 1.19 ± 0.10 | ||||
All | 76 | 0.594 | 0.475 | −9.8 ± 4.0 | 1.18 ± 0.09 | ||||
Type 1 | 31 | 0.705 | 0.341 | −5.4 ± 4.2 | 1.08 ± 0.10 | ||||
Type 2 | 45 | 0.421 | 0.571 | −12.0 ± 5.5 | 1.23 ± 0.13 | ||||
All | 75 | 0.619 | 0.516 | −8.5 ± 3.7 | 1.16 ± 0.09 | ||||
Type 1 | 31 | 0.716 | 0.463 | −1.6 ± 3.5 | 1.01 ± 0.09 | ||||
Type 2 | 44 | 0.503 | 0.552 | −12.8 ± 5.3 | 1.26 ± 0.13 | ||||
All | 82 | 0.608 | 0.352 | −3.0 ± 3.1 | 1.02 ± 0.08 | ||||
Type 1 | 32 | 0.739 | 0.128 | 3.9 ± 3.9 | 0.87 ± 0.09 | ||||
Type 2 | 50 | 0.542 | 0.514 | −4.0 ± 4.2 | 1.04 ± 0.10 | ||||
All | 82 | 0.934 | 0.851 | −7.6 ± 2.1 | 1.19 ± 0.05 | ||||
Type 1 | 32 | 0.937 | 0.799 | −6.2 ± 2.7 | 1.16 ± 0.07 | ||||
Type 2 | 50 | 0.886 | 0.877 | −7.6 ± 3.1 | 1.03 ± 0.11 |
Note. Columns: (1) Y variable; (2) X variable; (3) AGN type; (4) number of sample; (5) Spearman's rank coefficient for luminosity–luminosity correlation (ρL); (6) Spearman's rank coefficient for flux–flux correlation (ρf); (7) Student's t-null significance level for luminosity–luminosity correlation (PL); (8) Student's t-null significance level for flux–flux correlation (Pf); (9) regression intercept (a) and its 1σ uncertainty; (10) slope (b) and its 1σ uncertainty. Equation is represented as
Download table as: ASCIITypeset image
As shown in Table 2, we find significant correlations at confidence levels of >99% between all combinations of the [O iii] λ5007 and X-ray luminosities for any AGN types. The flux–flux correlations are weaker but significant at >90% confidence levels; relatively weak correlation is obtained for the X-ray type 1 AGN sample, most probably as a result of the narrow X-ray flux range (FX ≃ 2 × 10−11–3 × 10−10 erg cm−2 s−1 in the 14–195 keV band).
From the luminosity correlations for the entire AGN sample, we obtain b ≈ 1.2 in the regression line, which is significantly (>1σ) different from 1. This result is confirmed by the recent work based on a larger but less complete sample of Swift/BAT AGNs at z > 0.01 by Berney et al. (2015). We find that the slope for the X-ray type 1 AGNs is smaller than that for X-ray type 2 AGNs, although consistent within errors, given the large scatter of the correlations. The correlations with respect to L14–195 and those to L2–10 are found to be similar except for the normalizations. This is expected because absorption has a small effect on the observed hard X-ray luminosity (L14–195) except for heavily Compton-thick AGNs, and L2–10 is corrected for absorption through the X-ray spectral analysis.
We compare our results on the –L2–10 correlation obtained from X-ray type 2 AGNs with previous works. The slope we obtain, b = 1.26 ± 0.13 (i.e., ), is somewhat larger than that of Lamastra et al. (2009), who derive b = 0.98 ± 0.06 from a sample consisting of X-ray and optically selected Seyfert 2 galaxies. To check the effects of sample incompleteness of Sample B, we perform regression analysis with the asurv software (Isobe et al. 1986), by considering the lower limits of to be for the objects excluded in Sample B. We find that the slope b changes only by ∼0.01 compared with the case obtained from Sample B. Hence, the sample incompleteness cannot explain the difference of our result from Lamastra et al. (2009). The reason behind the discrepancy is not clear, but could be due to the different sample selections and luminosity ranges. Lamastra et al. (2009) include a sample compiled by Panessa et al. (2006) from the Palomar optical spectroscopic survey, which covers a lower luminosity range (L2–10 < 1042 erg s−1) than our sample. In fact, Panessa et al. (2006) obtain a much smaller slope, b = 0.75 ± 0.09, from their Seyfert 2 sample including Compton-thick AGNs, whose intrinsic X-ray luminosities are simply estimated by multiplying by a constant factor. The slope flatter than ours would be explained if contamination of [O iii] λ5007 from star formation in the host galaxy is more significant in lower-luminosity AGNs (see Section 4.1). Another possibility is enhanced past activity in the low-luminosity AGNs, which are left with a higher [O iii] λ5007 luminosity with respect to the current low X-ray activity.
3.2. Averaged [O iii] λ5007 to X-Ray Luminosity Ratio
We calculate the error-weighted mean value of the [O iii] λ5007 to X-ray luminosity ratio and its standard deviation for different AGN types. Here we consider a systematic error of 0.2 dex in and 0.5 dex in in addition to the errors listed in Table 1. The results are summarized in Table 3. Although we find that the best-fit regression line is not linear (b > 1), its effect can be checked by calculating an averaged value in each sample, which is also listed in Table 3. In fact, we confirm that it little affects the following discussions.
Table 3. Summary of Luminosity Ratios
Luminosity | Sample | N | σ | ||
---|---|---|---|---|---|
Ratio | (1) | (2) | (3) | (4) | (5) |
All | 102 | −2.45 ± 0.06 | 0.60 ± 0.05 | 43.71 | |
Type 1 | 48 | −2.27 ± 0.07 | 0.47 ± 0.05 | 43.85 | |
Type 2 | 54 | −2.62 ± 0.10 | 0.66 ± 0.07 | 43.60 | |
Type 2 () | 12 | −3.13 ± 0.10 | 0.33 ± 0.08 | 43.92 | |
Type 2 () | 10 | −2.43 ± 0.13 | 0.41 ± 0.10 | 43.24 | |
All | 101 | −1.99 ± 0.07 | 0.63 ± 0.05 | 43.24 | |
Type 1 | 48 | −1.78 ± 0.08 | 0.51 ± 0.06 | 43.36 | |
Type 2 | 53 | −2.19 ± 0.10 | 0.67 ± 0.07 | 43.14 | |
Type 2 () | 12 | −2.85 ± 0.09 | 0.31 ± 0.07 | 43.62 | |
Type 2 () | 10 | −2.01 ± 0.17 | 0.53 ± 0.13 | 42.81 | |
All | 77 | −1.94 ± 0.08 | 0.69 ± 0.06 | 43.65 | |
Type 1 | 31 | −1.89 ± 0.09 | 0.48 ± 0.07 | 43.79 | |
log | Type 2 | 46 | −1.98 ± 0.12 | 0.80 ± 0.09 | 43.56 |
Type 2 () | 9 | −2.44 ± 0.28 | 0.82 ± 0.21 | 43.87 | |
Type 2 () | 7 | −1.63 ± 0.23 | 0.59 ± 0.17 | 43.06 | |
All | 76 | −1.48 ± 0.08 | 0.69 ± 0.06 | 43.18 | |
Type 1 | 31 | −1.38 ± 0.10 | 0.51 ± 0.07 | 43.28 | |
log | Type 2 | 45 | −1.55 ± 0.12 | 0.78 ± 0.09 | 43.11 |
Type 2 () | 9 | −2.16 ± 0.24 | 0.71 ± 0.18 | 43.58 | |
Type 2 () | 7 | −1.20 ± 0.29 | 0.76 ± 0.22 | 42.62 | |
All | 83 | −2.18 ± 0.07 | 0.61 ± 0.05 | 43.22 | |
Type 1 | 32 | −1.95 ± 0.10 | 0.54 ± 0.07 | 43.36 | |
Type 2 | 51 | −2.35 ± 0.09 | 0.60 ± 0.07 | 43.14 | |
Type 2 () | 13 | −2.97 ± 0.08 | 0.28 ± 0.06 | 43.49 | |
Type 2 () | 11 | −2.28 ± 0.18 | 0.57 ± 0.13 | 42.73 | |
All | 83 | 0.23 ± 0.04 | 0.32 ± 0.03 | 41.06 | |
Type 1 | 32 | 0.32 ± 0.05 | 0.27 ± 0.04 | 41.40 | |
Type 2 | 51 | 0.16 ± 0.05 | 0.34 ± 0.04 | 40.85 | |
Type 2 () | 12 | 0.15 ± 0.09 | 0.30 ± 0.07 | 40.63 | |
Type 2 () | 10 | 0.26 ± 0.06 | 0.17 ± 0.05 | 40.52 |
Note. Columns: (1) luminosty ratio; (2) AGN type; (3) number of objects; (4) average; (5) standard deviation; (6) mean luminosity value of the denominator in the sample.
Download table as: ASCIITypeset image
As noticed from Table 3, we find that the mean ratio of observed [O iii] λ5007 luminosity () to X-ray luminosity is significantly smaller in X-ray type 2 AGNs than in X-ray type 1 AGNs, by ≈0.4 dex, using Sample A. This trend remains the same for the extinction-corrected [O iii] λ5007 luminosity () obtained from Sample B, although the difference between X-ray type 1 and X-ray type 2 AGNs is reduced to 0.1–0.2 dex. The "reduction" is consistent with the fact that the mean extinction correction factor is larger in X-ray type 2 AGNs ( / = 0.59 ± 0.09) than in X-ray type 1 AGNs ( / = 0.29 ± 0.11). This indicates a higher degree of obscuration toward the NLR in X-ray type 2 AGNs, consistent with previous results (e.g., Dahari & De Robertis 1988; Mulchay et al. 1994; Meléndez et al. 2008). It may be explained if the torus, or its extended structure such as dusty outflow (Hönig et al. 2012), is large enough to block a part of the NLR.
[To investigate the nature of AGNs with low scattering fractions, we calculate the mean [O iii] λ5007 to X-ray luminosity ratios for two subsamples of X-ray type 2 AGNs: (1) those with < 0.5%, and (2) those hosted by edge-on galaxies ( > 80°). The results are also listed in Table 3. We find that the mean extinction-corrected [O iii] λ5007 to X-ray luminosity ratio of the low scattering fraction AGNs is much smaller than that of the total X-ray type 2 AGN sample, while that of the edge-on galaxies does not differ from it within uncertainties. A simple χ2 test shows that the difference of the mean value of between the low scattering fraction AGNs and the other X-ray type 2 AGNs is significant at >99.9% confidence level. This is also noticeable from Figure 4(a), where we plot the ratio against log for Sample B.
Download figure:
Standard image High-resolution imageThese results suggest that a significant fraction of low scattering fraction AGNs are indeed buried in a torus with very small opening angles as originally proposed by Ueda et al. (2007). This population of AGNs could contribute to reducing the averaged ratio in the total X-ray type 2 AGN sample compared with that of X-ray type 1 AGNs, because they are predominantly identified as X-ray type 2 AGNs owing to the large covering fraction by the torus. We can rule out the possibility that their low scattering fractions are merely the result of deficiency of scattering gas in the NLR. If this were the case, we should observe a similar fraction of low objects among the X-ray type 1 AGN sample. Figure 4(b) plots the ratio against "X-ray Eddington ratio" (the 2–10 keV luminosity divided by the Eddington luminosity) using objects with available black hole masses in Winter et al. (2009a). No clear correlation is noticeable for the whole sample. The low scattering fraction AGNs do not always have high Eddington ratios, while Noguchi et al. (2010) report a possible negative correlation between the scattering fraction and Eddington ratio. Theoretically, deeply buried AGNs would be expected in the early growth phases of SMBHs with relatively small masses (hence with low luminosities). Thus, to further investigate the natures of this population, we need a larger sample of low-luminosity AGNs.
3.3. Correlations with Narrow Hα Line Luminosity
In AGNs, intense narrow Hα and Hβ lines are also produced from the NLR. Hence, we also perform regression analysis between the narrow Hα and X-ray luminosities, and that between the narrow Hα and [O iii] λ5007 luminosities, in the same way as done in Section 3.1. For each analysis, we utilize objects in Sample A that have available flux measurements of Hα or [O iii] λ5007. The correlation plots are displayed in Figures 5 and 6, respectively, together with the best-fit linear regression forms, which are given in Table 2. We also calculate the mean and standard deviation of the log(LHα/L2–10) ratio and the log(L[O iii]/LHα) ratio, which are summarized in Table 3.
Download figure:
Standard image High-resolution imageDownload figure:
Standard image High-resolution imageWe find that, for all AGNs, (1) LHα with a similarly large scatter (≈0.6 dex) to that seen in the versus L2–10 correlation, and (2) with a much smaller scatter (≈0.3 dex). The slope of the LHα versus L2–10 correlation obtained from the X-ray type 2 AGNs, 1.04 ± 0.10, is larger than that obtained by Panessa et al. (2006) from their sample of 34 Seyfert 2s, 0.78 ± 0.09, which covers a lower luminosity range (L2–10 < 1042 erg s−1) than ours. Even though here we use only the luminosity of the "narrow" component of Hα, the regression slope and scatter between LHα and L2–10 are similar to those found between "total" LHα (i.e., that including the broad component) and L2–10 (e.g., Ho 2008).
4. DISCUSSION
4.1. Origin of Correlation and Scatter between [O iii] λ5007 and Hard X-Ray Luminosities
Using so far the largest (N > 100) statistically complete sample of hard X-ray (E > 14 keV) selected AGNs in the local universe, we determine the statistical properties between [O iii] λ5007 and hard X-ray luminosities with the best accuracy. The linear regression form of ( ) is obtained from the whole sample (see Table 2). These results can be used as the reference for AGNs in the luminosity range of log L2–10 = 41–46. We also find that the mean luminosity ratio between and L2–10 of X-ray type 2 AGNs is significantly smaller than that of X-ray type 1 AGNs. The difference is largely contributed by a population of low scattering fraction AGNs. Another important result is the very large variance in the log(L[O iii]/L2–10) ratio, corresponding to its standard deviation of ∼0.5 in X-ray type 1 AGNs and ∼0.7 in X-ray type 2 AGNs (see Table 3).
The nonlinear correlation (i.e., ) between and may be explained by a combination of multiple effects. The first effect is the luminosity dependence of the AGN spectral energy distribution. The luminosity of the narrow lines is predominantly determined by the continuum flux of ultraviolet photons responsible for photoionization of the NLR gas, rather than the X-ray flux. Thus, if the spectral slope α (for the flux density ) between UV and hard X-rays above 2 keV is larger in more luminous AGNs as suggested by Scott & Stewart (2014), it works to make the [O iii] λ5007 to X-ray luminosity correlation steeper. Second, according to the luminosity-dependent unification model (Ueda et al. 2003; Ricci et al. 2013), the opening angle of a torus increases with luminosity, thus making the angular spread of the "NLR cone" larger in more luminous AGNs. This also leads to increased b. The third effect is due to the luminosity dependence of the NLR size in the radial direction, which is proportional to L0.33±0.04 (Schmitt et al. 2003). Thus, the actual size of the NLR might be saturated in very luminous AGNs if the outer radius exceeds the scale height of the host galaxy (Netzer et al. 2004). The fourth effect is the contamination of from star formation in low-luminosity AGNs, as mentioned in Section 3.1. The last two effects make the regression slope flatter than unity.
To better understand the origin of the observed luminosity correlations and scatters between (or ) and , comparison with the LHα and correlation is useful. The Balmer lines are emitted by recombination as the result of photoionization, whereas the [O iii] λ5007 line is emitted via collisional excitation in the heated gas. Thus, the intensity ratio between Hα and [O iii] λ5007 depends on the physical parameters, such as the ionization parameter and density (Ferland & Netzer 1983). Also, the [O iii] λ5007 line comes preferentially from gas with a density of ∼106 cm−3 unlike the Balmer lines, which come from a wide range of densities. In fact, detailed images of the NLR with Hubble Space Telescope for a few low-z objects (e.g., Evans et al. 1991; Fischer et al. 2013) show that much of the [O iii] λ5007 flux comes from clumpy structures. The effects of dust extinction inside the NLR, which may not be correctly measured with the Balmer decrement, make it even more complex. Hence, depending on how the NLR gas and dust is distributed, nonlinear correlation, as well as a significant scatter in the flux ratio between the Balmer lines and [O iii] λ5007 line, would be also expected.
The results for all AGNs, LHα and show that the observed nonlinear correlation (b ≈ 1.2) between and cannot be simply explained by a single reason. In addition to the four possibilities listed above, it is found that the nonlinear correlation between and LHα, which is determined by plasma physics, also plays a role. The fact that the slope between LHα and L2–10 is close to unity (b = 1.02 ± 0.08) indicates that the third effect (luminosity dependence of the NLR physical size) and/or the fourth effect (contamination by star formation) must work to cancel the first and second effects. The fact that flatter slopes are found from the X-ray type 1 AGNs, which are dominant in the largest luminosity range, suggests that the third effect is more important.
The large variation between and may be explained because the optical emission lines from the NLR are a secondary indicator of the intrinsic AGN luminosity in that they do not directly come from near the SMBH and have strong dependence on the geometry and size of the NLR, its averaged density, clumpiness, and amount of dust. In fact, a significant scatter of ∼0.4 dex between the [O iii] λ5007 luminosity and the continuum luminosity at 5100 Å is also reported in the Sloan Digital Sky Survey (SDSS) quasar sample (Shen et al. 2011). The presence of the low scattering fraction AGNs accounts for the larger scatter of the ratio in X-ray type 2 AGNs (∼0.7 dex) than in X-ray type 1 AGNs (∼0.5 dex), which could be understood in terms of variation in the geometry (cone angle) of the NLR. Since the correlation between and LHα is found to be tighter than that between and , the clumpiness of the NLR gas and dust extinction effects would not be the prime cause of the – scatter. Another effect could be time variability; even though we utilize "70-month" averaged hard X-ray fluxes, the emission from the NLR reflects the past AGN power averaged over >102 yr.
4.2. Comparison of [O iii] λ5007, Hα, and X-Ray LFs
The LF is one of the most important statistical properties of AGNs. Utilizing an AGN sample selected from the SDSS, Hao et al. (2005) determined [O iii] λ5007 and Hα LFs of AGNs at z ≤ 0.15 (they adopt emission-line luminosities not corrected for extinction, and we follow the same procedure below). Heckman et al. (2005) then compared the SDSS [O iii] λ5007 LF with an X-ray LF in the 3–20 keV band derived from the RXTE Slew Survey by Sazonov & Revnivtsev (2004). They found that the X-ray LF significantly underpredicts the [O iii] λ5007 LF when the mean luminosity ratio between and obtained from the RXTE AGN sample is assumed without considering the scatter. On the basis of this result, they argue that X-ray surveys seem to miss a significant fraction of AGNs, particularly Compton-thick AGNs.
Recently, Ueda et al. (2014) determined the X-ray LF of AGNs, including Compton-thick AGNs over a redshift range of z = 0–5, using a highly complete sample of X-ray selected AGNs. The local AGN sample from the Swift/BAT survey is also utilized. Detection biases against (mildly) Compton-thick AGNs are taken into account to correctly estimate their intrinsic number. Because heavily Compton-thick AGNs with log = 25–26 are difficult to detect even in the E > 10 keV hard X-ray band, they assume that the fraction of AGNs with log = 25–26 is the same as those with log = 24–25. The X-ray LF and absorption distribution function are used as the basis of a standard population synthesis model of the X-ray background (Ueda et al. 2014). We note that the X-ray AGN LF by Sazonov & Revnivtsev (2004) may not be appropriate to adopt for direct comparison with LFs in other wavelengths because (1) the original X-ray LF by Sazonov & Revnivtsev (2004) was unfortunately affected by an error in the count-rate-to-flux conversion (by a factor of 1.4; see Sazonov et al. 2008; Ueda et al. 2011); (2) even after correcting for that error, it significantly underestimates other X-ray LFs of Compton-thin AGNs (Ueda et al. 2011); and (3) Compton-thick AGNs, which are difficult to detect in the 3–20 keV band, are not included.
Thus, it is very interesting to make comparison with the [O iii] λ5007 and Hα LFs with the most up-to-date X-ray LF of local AGNs, including Compton-thick AGNs, in order to understand the completeness and cleanness of AGN selections in these different wavelengths. The red curve in Figures 7(a) and (b) represent the best-fit [O iii] λ5007 and narrow Hα LFs in Hao et al. (2005) (two-power-law model, the sum of Seyfert 1s and 2s), after correcting both luminosity and space density for the difference of the adopted Hubble constant, from H0 = 100 km s−1 Mpc−1 (Hao et al. 2005) to H0 = 70 km s−1 Mpc−1 (our paper). The black curve in each figure is a prediction for the [O iii] λ5007 (or Hα) LF calculated from the Ueda et al. (2014) X-ray LF at z = 0. Here we convert L2–10 into (or Hα) with the best-fit linear regression form (Table 2) separately for X-ray type 1 and X-ray type 2 AGNs, and we also consider the scatter around it by assuming a Gaussian distribution with the standard deviation listed in Table 3. As the Hao et al. (2005) result is obtained from AGNs at z ≤ 0.15, we then multiply luminosity-dependent density evolution factors (Ueda et al. 2014) at the mean redshift. The black dashed curves denote the boundaries when both errors (1σ) in the mean and standard deviation of log(L[O iii]/L2–10) (or log(LHα/L2–10)) are taken into account. For comparison, we also plot the case when the standard deviation is set to be zero (i.e., no scatter is considered) with the blue, dot-dashed curve.
Download figure:
Standard image High-resolution imageAs noticed from Figure 7(a), the [O iii] λ5007 (red) and X-ray (black) LFs are roughly consistent with each other within a factor of ∼2 when we take into account the uncertainties in the to conversion. Thus, the systematic (≈4) underestimate of the [O iii] λ5007 LF by the X-ray LF over a wide range of luminosity reported by Heckman et al. (2005) is now resolved. Rather, at ≳ 40, the X-ray LF outnumbers the [O iii] λ5007 LF, while statistical uncertainties in the [O iii] λ5007 LF are large (a factor of >2) at ≳ 41.6 owing to the limited sample size in the SDSS. Figure 7(b) shows even better agreement between the Hα and X-ray LFs over a wider luminosity range, although a similar discrepancy is noticed at log LHα ≳ 41. We note that it is important to consider the scatter between the two luminosities when making the comparison of LFs, as seen in the difference between the black solid curve (with scatter) and blue dot-dashed curve (without scatter).
These results confirm that hard X-ray (>10 keV) observations are a very powerful tool to find AGNs with high completeness, not missing a dominant portion of the entire AGN population, once biases against Compton-thick AGNs are properly corrected (see, e.g., Malizia et al. 2009). For the correction, however, it is essential to obtain the broadband X-ray spectra covering up to, at least, a few tens of keV, with sufficiently good sensitivities. If the discrepancy between the [O iii] λ5007 (or Hα) and X-ray LFs at the high-luminosity range is true, this instead implies that the optical selection would miss some AGN populations. The selection based on emission-line diagrams could be incomplete for AGNs significantly contaminated by star formation; indeed, Winter et al. (2010) show that a nonnegligible fraction of hard X-ray selected AGNs could be optically classified as H ii galaxies, even though they are truly AGNs. Other candidates of "optically missing" AGNs are those deeply embedded in tori with almost spherical geometry, in which no or little NLR is formed. They may be similar to some of the low scattering fraction AGNs in our sample whose [O iii] λ5007 fluxes are very weak. If many of the heavily Compton-thick AGNs assumed in the Ueda et al. (2014) model correspond to this population, it would partially account for the mismatch between the optical and X-ray LFs.
5. CONCLUSIONS
From our observations at the SAAO and the literature, we have compiled a complete catalog of [O iii] λ5007 line flux for 103 hard X-ray selected AGNs in the local universe located at together with narrow Hα and Hβ line fluxes (or their ratio) for a large fraction (∼80%) of the sample. The main conclusions are summarized below.
- 1.We detect significant correlations between [O iii] λ5007 (without or with extinction correction) and X-ray luminosities independently from X-ray type 1 AGNs (log < 22) and X-ray type 2 AGNs (log ≥ 22), even though there is a large scatter in their luminosity ratio. The best regression forms obtained from the whole sample are and
- 2.Absorbed AGNs with low scattering fractions in the X-ray spectra show smaller and ratios than the other absorbed ones. This suggests that a significant number of low scattering fraction AGNs are buried in tori with small opening angles.
- 3.Significant correlations are also found between the Hα and X-ray luminosities. The [O iii] λ5007 and Hα luminosities are more tightly correlated than the [O iii] λ5007–X-ray luminosity correlation.
- 4.The X-ray LF of local AGNs in a standard population synthesis model shows much better agreement with the [O iii] λ5007 LF derived from the SDSS than previously reported. It rather predicts a larger number of AGNs than the [O iii] λ5007 selection at ≳ 40. This confirms that hard X-ray (>10 keV) observations are a very powerful tool to find AGNs with high completeness, once biases against Compton-thick AGNs are properly corrected on the basis of the broadband X-ray spectra.
This paper uses observations made at the South African Astronomical Observatory (SAAO). Part of this work was financially supported by Grants-in-Aid for Scientific Research 26400228 (Y.U.) and for JSPS Fellows for Young Researchers (K.I.) from the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan, and by the National Science Council of Taiwan under the grants NSC 99-2112-M-003-001-MY2 and NSC 102-2112-M-003-016 (Y.H.). P.V. and A.Y.K. acknowledge the support from the National Research Foundation (NRF) of South Africa.