DISSECTING THE POWER SOURCES OF LOW-LUMINOSITY EMISSION-LINE GALAXY NUCLEI VIA COMPARISON OF HST-STIS AND GROUND-BASED SPECTRA^*

We have obtained high signal-to-noise ratio (S/N) spectrophotometry of 16 galaxy nuclei with the 6.5 m MMT, on the nights of 2008 February 13, November 5, and November 6. We employed the Blue Channel spectrograph with the 500 groove mm⁻¹ grating, used in first order with the 1'' slit to cover 3800–7000 Å with 3.6 Å resolution; the slit was at the parallactic angle (Filippenko 1982). This setup allows us to measure the red nebular lines ([O i] λ6300, Hα, [N ii] λ6583, and [S ii] λλ6716, 6731) that are of primary interest for comparison with the HST measurements, as well as the blue part of the spectrum covering Hβ and [O iii] λλ4959, 5007, which allow for a complete spectroscopic classification. The resolution is similar to that employed in the Palomar spectroscopic survey, which is sufficient to model and obtain a relatively clean separation of the continuum starlight and nebular emission (i.e., for typical velocities in the centers of galaxies, the stellar and nebular features are at least partially resolved). The seeing was ∼10–12 during all three nights. The total exposure times varied between 25 and 40 minutes. Each exposure was broken up into two subexposures to allow for the removal of cosmic rays (CRs).

Standard reductions including bias subtraction, flat-field correction, wavelength calibration, and flux calibration were performed using IRAF.⁷ For flux calibration, the Kitt Peak IRS spectroscopic standard stars G191B2B and BD +17°4708 were observed in February, while BD +17°4708, HD 19445, and HD 84937 were chosen during the November run; they were observed at the beginning, middle, and end of each night. Spectra of He–Ne–Ar comparison lamps were obtained before and after each observation to calibrate the wavelength scale. The one-dimensional (1D) spectra were extracted with a 1'' × 4'' effective aperture for consistency with the Palomar survey. We corrected for the telluric oxygen absorption lines near 6280 and 6860 Å through division by normalized, intrinsically featureless spectra of the standard stars; differences in airmass were scaled appropriately. This reduction procedure also corrected for continuum atmospheric extinction.

The emission-line fluxes in the MMT spectra were measured after removal of the underlying stellar continuum, which is significant in the ground-based aperture data. Host-galaxy starlight was modeled with six synthesized galaxy spectral templates, which were built from the library of stellar populations of Bruzual & Charlot (2003) using the method described by Tremonti et al. (2004). Each template was broadened by convolution with a Gaussian to match the stellar velocity dispersion of the host galaxy. The final extracted nuclear MMT spectra corrected for telluric absorption are shown in Figure 1, along with their best-fitting multiple-stellar-population models. Many nuclei display obvious nebular emission; however, especially in the blue part of the spectrum (e.g., [O iii] λλ4959, 5007; Hβ), the emission lines are mostly buried in the stellar continuum.

Zoom In Zoom Out Reset image size

We note that while the spectral fits are generally satisfactory for the continuum shape, the reddest part of the spectrum that hosts the [S ii] emission doublet fails to match the theoretical model for about half of the MMT sample (see Figure 1). In particular, the synthetic stellar continuum cuts through this doublet with the effect of artificially diminishing the total [S ii] fluxes and increasing the [S ii] λ6716/λ6731 ratios measured in the continuum-subtracted spectra. Because there is no significant stellar absorption feature that affects this emission doublet, we decided to use the measurements of the [S ii] features in the original flux-calibrated spectra without subtraction of the underlying continuum. For completeness, for the aperture comparison tests involving this particular emission doublet, we have compared the results considering measurements with and without the continuum subtraction, as well as with results obtained by simply removing these objects from the comparison, and the conclusions remained consistent within the errors.

The continuum-subtracted emission-line spectra have been modeled and measured with specfit (Kriss 1994), a method that employs line and continuum spectral fitting via an interactive χ² minimization. The formal uncertainties in the line fluxes include errors in the continuum subtraction and flux calibration, and they have been propagated to the line ratios. The continuum was approximated by a linear fit, while the emission features were modeled by Gaussian profiles. The fit was performed simultaneously for all lines, assuming common velocity fields and widths for all features of the NLR. For deblending the Hα + [N ii] emission region, we followed the technique of Ho et al. (1997a, 1997b) and assigned to both the [N ii] lines and the narrow Hα emission the line profile that was predetermined by fitting the [S ii] lines. When the [S ii] features presented visibly strong broad wings, the narrow components were modeled with multiple Gaussians, in order to determine the necessity of an additional broad component for the Balmer Hα. This procedure is in principle able to account for potential asymmetries, velocity blending, or multiple velocity peaks in the narrow lines, as well as for wings in their bases (see also Section 2.4). Line doublets were forced to share the same velocity widths. The flux ratio for the [N ii] doublet was constrained to the value determined by the branching ratio. As in Ho et al. (1997a), the measurements of the line flux ratios involving Hα employed the Gaussian components describing only the narrow-line emission. The fluxes of the main narrow emission features measured in the MMT spectra are listed in Table 2.

Table 2.  MMT Fluxes of Narrow Emission Lines for MMT Objects

Object	Hβ	[O iii]	[O i]	Hα	[N ii]	[S ii]	[S ii]
Name	(Narrow)	λ5007	λ6300	(Narrow)	λ6583	λ6716	λ6731	${f}_{\lambda }$(6563 Å)
NGC 193	183 ± 4	340 ± 4	148 ± 4	626 ± 5	979 ± 5	286 ± 5	267 ± 5	211 ± 2
NGC 383	241 ± 5	426 ± 5	254 ± 24	594 ± 70	1233 ± 82	400 ± 26	382 ± 24	127 ± 15
NGC 541	179 ± 3	120 ± 3	149 ± 4	621 ± 4	403 ± 4	83 ± 4	70 ± 4	226 ± 2
NGC 1300	241 ± 5	426 ± 5	260 ± 49	1326 ± 124	1780 ± 111	428 ± 56	374 ± 44	173 ± 16
NGC 1497	330 ± 5	947 ± 5	507 ± 13	1002 ± 57	1888 ± 62	416 ± 29	203 ± 21	174 ± 10
NGC 2179	543 ± 4	964 ± 4	199 ± 4	727 ± 45	945 ± 42	638 ± 35	387 ± 25	317 ± 20
NGC 2329	862 ± 6	1294 ± 6	193 ± 7	440 ± 20	945 ± 24	196 ± 28	188 ± 11	484 ± 22
NGC 2892	148 ± 6	188 ± 6	55 ± 24	68 ± 25	230 ± 35	188 ± 23	139 ± 21	45 ± 17
NGC 3801	21 ± 4	87 ± 4	64 ± 4	358 ± 4	352 ± 4	130 ± 4	78 ± 4	111 ± 1
NGC 3862	251 ± 4	212 ± 4	208 ± 46	704 ± 40	1098 ± 45	212 ± 15	137 ± 46	193 ± 11
NGC 4335	125 ± 5	333 ± 5	221 ± 111	657 ± 39	1144 ± 41	132 ± 5	180 ± 3	249 ± 15
NGC 7052	236 ± 4	241 ± 4	133 ± 4	1087 ± 5	1264 ± 4	267 ± 4	249 ± 4	252 ± 1
UGC 1841	618 ± 6	1133 ± 6	358 ± 21	687 ± 35	1476 ± 39	339 ± 21	213 ± 21	246 ± 13
UGC 2847	6221 ± 3	1518 ± 3	980 ± 3	43225 ± 4	18570 ± 3	3234 ± 3	3336 ± 3	1017 ± 1
UGC 7115	270 ± 6	284 ± 6	377 ± 7	455 ± 38	954 ± 43	219 ± 55	155 ± 13	182 ± 15
UGC 12064	152 ± 5	170 ± 5	102 ± 5	645 ± 5	818 ± 5	162 ± 6	112 ± 6	145 ± 1

Note. The emission-line fluxes are in units of 10⁻¹⁷ erg s⁻¹ cm⁻² and represent the observed values, not corrected for reddening.

Download table as:  ASCIITypeset image

We classified spectroscopically the MMT objects based on their locations in line diagnostic diagrams using the criteria of Ho et al. (1997a), and the resulting respective classes are recorded in Table 1. For these 16 objects we also list, for comparison, the spectral classes based on the Kewley et al. (2006) criteria; the consistency is reasonably good. The uncertain classifications refer to objects for which the sets of conditions involving the low-ionization lines ([O i], [N ii], [S ii]) do not hold simultaneously for the three Veilleux & Osterbrock (1987) diagnostic diagrams; in these cases, to remain consistent with Ho et al. (1997a), the [O i]/Hα ratio was given the largest weight, and the corresponding class is listed first, followed by all the classes that may be consistent with the data. The MMT sources are spectrally classified as mostly L nuclei (of Type 2, with the exception of NGC 383, where a broad Hα component is needed for the best spectral fit; see Table 4 for measurements), thus providing a welcome increase in the number statistics for this spectral class; the MMT sample also adds three T nuclei and one Seyfert galaxy to the Palomar sample.

The STIS spectra used for this study were originally acquired mainly for the purpose of deriving central black hole masses and cover the Hα and adjacent strong nebular features. In all cases, the grating employed was G750M centered at 6581 or 6768 Å, which makes available for measurements [N ii] λλ6548, 6583, [S ii] λλ6716, 6731, and at the 6581 Å setting, [O i] λ6300.

For 10 of the Palomar objects (NGC 3675, 3953, 4258, 4321, 4435, 4486, 4527, 4826, 5879, and 7331), STIS spectra have been obtained via two different programs. For completeness, Table 3 presents our measurements of the fluxes of the strong emission lines, as described in Section 2.3, for all of the STIS spectra of these objects, except for the SUNNS spectra (of NGC 4321 and 4435). We excluded from analysis observations of 18 spectra from program IDs (PID) 8228⁸ (five H ii, four T, and one Seyfert), 8607 (five T), and 12187 (two H ii and one T) owing to extremely low S/N. Two of these sources, NGC 4527 and 7331, are common to two PIDs; for NGC 4527 we could extract a 1D spectrum for one data set (PID 8607), while for NGC 7331 none of the programs offered sufficiently good data. The STIS spectrum of NGC 3516 exhibits an unusual shape that proves unsuitable for spectral modeling using [S ii] as an empirical template for the [N ii] and the narrow Hα emission (see Section 2.3), and therefore we do not record any measurements for this object. Thus, STIS data for 17 objects remain unusable for this project. The final count of 80 sources with Palomar and STIS spectroscopy, including the SUNNS sources, consists of 20 H ii, 22 T, 14 L2, 13 L1, and 11 S nuclei (both Type 1 and 2). With the addition of the MMT objects, the breakdown of the total sample by the ground-based spectral type consists of 20 H ii, 25 T, 26 L2, 13 L1, and 12 S nuclei (both Type 1 and 2), thus providing quite balanced statistics of bona fide AGNs (L1 and S), H ii, and ambiguous sources of both L2 and T flavors.

We retrieved from the archive all on-the-fly calibrated frames with the intention to use the fully rectified and flux- and wavelength-calibrated 2D images to further extract the 1D spectra. For the majority of objects the S/N is low, and/or CRs are abundant. We obtained the best results by performing our own CR correction to the archival flat-fielded images before feeding these frames back into the HST pipeline software (the x2d task under the stsdas package, IRAF) to generate the calibrated images.

Incident CRs and bad pixels have been removed using the ccdclip algorithm under the imcombine task; after the images to be combined are realigned to account for the offsets parallel to the slit, pixels with values outside a determined range are assigned the median value. The range of acceptable pixel values, and thus the parameters involved in this procedure (e.g., hsigma, mclip, nkeep), are adjusted for each individual case after close comparisons of the central 10 rows (of interest for the 1D extraction) in both the initial and the corrected image spectra, to ensure that valid emission features were not damaged. For the observations that consist of only one nuclear image spectrum, and which have a high CR density, we interpolated over affected pixels using an automated detection algorithm for the strongest events and did manual intervention for weaker events comparable in strength to real emission features.

The 1D object spectra were extracted using the apall procedure. As indicated in Table 1, the majority of objects were observed through a slit with a 02 width. Part of these observations were acquired with a plate scale of 005 pixel⁻¹, while for others the measurements have been binned to 01 pixel⁻¹ at readout. For the unbinned spectra we extracted the central five rows (pixels), while for the cases where data have been rebinned we extracted only the central 2.5 rows, to preserve the angular scale. However, since the galaxies are at different distances/redshifts, the metric scale mapped out by these observations is not conserved. For simplicity, the object spectra acquired with the 01 slit width (which were all unbinned spatially, so the plate scale is 005 pixel⁻¹) were also generated using 5-pixel-wide extractions.

The STIS aperture is much smaller (two orders of magnitude in area) than that typically used in ground-based measurements of galaxy nuclei. Hence, the corresponding contamination by the host-galaxy starlight is significantly reduced. However, the stellar component is not negligible and it is important to correct for it, as it can introduce additional structure in the continuum and thus might influence the appearance of the emission features. Fitting a template galaxy stellar contribution that would eventually be subtracted from the nuclear spectrum (as we did for the MMT ground-based data) is, however, very difficult for most of these HST observations because of the modest S/N and the relatively small wavelength coverage. The strongest effect of starlight contamination of the nuclear emission is expected to be in the Hα emission, owing to the presence of the Balmer absorption line. A potentially useful alternative to stellar background light subtraction is to adopt reasonable values for the strength of the Hα absorption, using information offered by stellar population synthesis models, and correct for it.

A typical value of the equivalent width (EW) of the Hα absorption line can be estimated from measuring all the templates contained in the Bruzual & Charlot (2003) library of synthetic galaxy spectra. For three different models of star formation history (instantaneous starburst, constant star formation, and exponentially decaying star-forming history), each equally represented for three different metallicity values (subsolar, supersolar, and solar, with Z = 0.004, 0.05, and 0.02, respectively, where Z is the total mass fraction of metals), and ages that range from 0.005 to 12 Gyr, the EW of the Hα absorption line varies between 2 and 6 Å, with most of the cases concentrating around 4 Å. Moreover, studies of the properties of the nuclear stellar populations (e.g., Sarzi et al. 2005) show that, in a large majority of nearby nuclei (∼80%), the age is at least 5 Gyr and the chemical enhancement of the emitting gas is at least solar, narrowing down the possible range in Hα absorption EW to 2.5–4.5 Å. Therefore, for the corrections used in the narrow Hα line flux and the line flux ratio comparison involving this feature, we used for the EW of the Hα absorption line the value of 3.5 ± 1 Å. The adopted error translates into a typical additional uncertainty of ∼10% in the resulting line flux ratios, but with considerably larger error bars for the weak-lined objects.

To understand the consequences of our adopted absorption correction, we repeated the correction with different values of the absorption EW in the range 2.5–4.5 Å. We found that the distributions of line ratios overall and within object subclasses did not change significantly, and the resulting conclusions were similarly robust. Nevertheless, because this method of correcting for stellar absorption does not account for the inherent variety in the true strength of the stellar absorption in Hα, we present the results of this study for narrow Hα fluxes with and without the correction described above.

The line fluxes of the emission features observed in the STIS spectra were measured via spectral modeling following the same technique employed for the stellar-continuum-subtracted MMT spectra, as discussed in Section 2.1. To recap, the spectra were fit with multiple Gaussians to match both broad and narrow components of all observed lines, while describing the continuum with a first-order polynomial. The [S ii] emission was used as an empirical template for the profiles of the [N ii] doublet and the narrow component of Hα. To account for the contribution of blueshifted or redshifted wings, in an attempt to assess the presence or absence of a broad Hα component, we employed multiple Gaussians for fitting the emission profile of the NLR. The resulting measurements of the HST fluxes of the strong emission lines, without correction for stellar absorption, are recorded in Table 3 for all objects included in this study.

It is important to note that the procedure described above does not account for the possible effects of stellar continuum structure on other emission features considered in this study (e.g., the [S ii] doublet), or for its effect on the shape of the Hα emission line. In these particular features, however, these effects are in general expected to be much smaller than in typical ground-based observations of these nuclei. Figure 2 provides an illustration of the degree to which the contrast between the nuclear emission and the continuum is increased in the HST observations; we show this comparison for EWs measured both with and without correction for the Hα stellar absorption, as described above. The plots show the ratio of the EWs of the narrow Hα line emission in the two apertures separately for all individual spectral types; for the objects observed at two epochs with HST (see Table 1 for the Prop IDs), we include both measurements, which are identified by the same value on the abscissa. Their corresponding Hα fluxes differ by ∼0.2 dex when different aperture sizes are employed (02 vs. 01) and are consistent within the errors when the same aperture size is used. It is readily apparent that the Hα line is stronger in the HST spectra of most Seyferts and LINERs, and of many transition objects as well. For the rest of the T and H ii nuclei, the uncertainties are large and are dominated by the limited data quality and structure in the continuum; it appears thus that our correction for starlight in these sources does not affect the final conclusions of this study.

Zoom In Zoom Out Reset image size

One of the clearest indications of accretion-powered activity is the presence of broad permitted lines (FWHM intensity of a few thousand kilometers per second). The strongest broad component at optical wavelengths is expected to be Hα, and thus detection of a broad Hα component is a promising method of constraining the origin of the nuclear activity. The results of the Palomar survey suggest that broad Hα is quite common in local Seyferts and a significant fraction of LINERs; it is found in at least 10% of all nearby luminous galaxies (Ho et al. 1997b). The true incidence of broad-lined sources must, however, be higher; especially in the weakly active nuclei, ground-based observations are unable to distinguish a possibly diluted (and weak) BLR from the structure present in the continuum produced by starlight. HST spectroscopy that employs small-aperture observations has the power to eliminate a significant fraction of contaminating starlight and hence to increase the sensitivity to broad wings of line emission. Thus, an important question to address is whether, at HST resolution, the fraction of nuclei classified as broad lined is larger than that emerging from ground-based data.

Ho et al. (1997b) describe the set of objects that show definite or probable evidence of broad Hα emission in their survey. Our sample contains 15 of the nuclei with unambiguous broad Hα in the Palomar spectra, and the STIS observations confirm the presence of the broad Balmer feature in these sources (Table 1). In Table 4 we summarize for each of these nuclei several parameters concerning the broad Hα component. The fractional contribution of the broad component of Hα to the entire Hα + [N ii] complex, f_blend, is given in column (2), while column (3) indicates its fractional contribution to the whole Hα (narrow + broad) emission, f_Hα. The FWHM of broad Hα and its velocity shift relative to the narrow component are recorded in columns (4) and (5), respectively. Columns (6) and (7) give the observed (not corrected for reddening) flux and luminosity of broad Hα in units of erg s⁻¹ cm⁻² and erg s⁻¹, respectively, where we used the distances given by Ho et al. (1997a) to ease the comparison. We show in parentheses the corresponding Ho et al. (1997b) parameters, when available.

Table 4.  Galaxies with Broad Hα Emission

Object	${f}_{\mathrm{blend}}$	${f}_{{\rm{H}}\alpha }$	FWHM	${\rm{\Delta }}v$	log $F({\rm{H}}\alpha )$	log $L({\rm{H}}\alpha )$	Obs. yr.
(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Palomar Sample
NGC 315	0.25	0.59	2870	1190	−14.52	39.19	06/2000
	(0.29)	(0.64)	(2000)		(−13.86)	(39.85)	(10/1985)
NGC 1052	0.34	0.54	2800	10	−13.25	39.33	01/1999
	(0.32)	(0.55)	(1950)		(...)	(...)	(02/1984)
NGC 2985	0.44	0.81	2250	210	−13.78	38.96	07/2011
	(0.22)	(0.37)	(2050)		(...)	(...)	(02/1984)
NGC 3031a	0.72	0.90	3680	360	−12.15	38.22	07/1999
	(0.57)	(0.86)	(2650)		(−12.02)	(39.17)	(06/1986)
NGC 3227	0.20	0.44	2680	0	−11.78	40.93	02/1999
	(0.71)	(0.86)	(2950)		(...)	(...)	(02/1986)
NGC 3245	0.47	0.79	4300	420	−14.39	38.38	02/1999
	(...)	(...)	(...)		(...)	(...)	(02/1986)
NGC 3642a	0.79	0.85	1650	450	−13.81	39.15	10/2000
	(0.35)	(0.51)	(1250)		(...)	(...)	(02/1984)
NGC 3998a	0.59	0.77	6320	320	−12.65	40.09	04/2002
	(0.37)	(0.59)	(2150)		(−12.58)	(40.16)	(02/1984)
NGC 4036	0.24	0.61	2440	730	−14.59	38.27	04/1999
	(0.14)	(0.39)	(1850)		(−13.69)	(39.17)	(02/1984)
NGC 4051a	0.89	0.96	6500	0	−12.40	40.14	03/2000
	(0.74)	(0.84)	(1000)		(...)	(...)	(02/1984)
NGC 4258	0.50	0.70	1470	90	−13.45	38.29	08/2000
NGC 4258	0.25	0.41	1310	−40	−13.86	37.88	03/2001
	(0.49)	(0.67)	(1700)		(−13.09)	(38.65)	(02/1984)
NGC 4278	0.41	0.67	2880	50	−14.03	38.02	05/2000
	(0.19)	(0.38)	(1950)		(−13.07)	(38.98)	(02/1984)
NGC 4429	0.28	0.75	2650	820	−14.60	37.93	05/2001
	(...)	(...)	(...)		(...)	(...)	(01/1985)
NGC 4579a	0.70	0.92	6590	1280	−12.92	39.61	04/1999
	(0.21)	(0.48)	(2300)		(−13.08)	(39.45)	(02/1984)
NGC 4594	0.29	0.77	4000	910	−13.59	39.09	01/1999
	(...)	(...)	(...)		(...)	(...)	(02/1984)
NGC 4636	0.76	0.98	2330	260	−14.38	38.12	04/2001
	(0.27)	(0.56)	(2450)		(−14.18)	(38.36)	(02/1985)
NGC 4736	0.80	0.93	1570	140	−13.38	37.96	06/2002
	(...)	(...)	(...)		(...)	(...)	(02/1984)
NGC 5005	0.17	0.58	2610	5	−14.41	38.32	12/2000
	(0.33)	(0.79)	(1650)		(−12.69)	(40.05)	(02/1984)
NGC 5077	0.34	0.58	2570	550	−14.24	39.05	03/1998
	(0.12)	(0.26)	(2300)		(−14.06)	(39.23)	(02/1984)
NGC 5921	0.20	0.45	2280	40	−14.82	38.06	05/2000
	(...)	(...)	(...)		(...)	(...)	(10/1999)
MMT Sample
NGC 193	0.66	0.95	2460	550	−13.98	39.68	10/1999
NGC 383	0.63	0.86	2710	480	−12.99	40.79	10/2000
	(0.56)	(0.83)	(2780)	(411)	(−13.57)	(40.21)	(11/2008)
NGC 4335	0.39	0.79	3020	370	−14.16	39.55	09/1999
UGC 1841	0.43	0.81	3040	410	−13.71	40.27	11/1999

Notes. Col. (1): Galaxy name. When the objects are observed in different programs, the rows showing the results are listed in increasing order of the PID number, as in Table 1. Col. (2): fractional contribution of the broad component of Hα to the Hα + [N ii] blend. Col. (3): fractional contribution of the broad Hα to the total (narrow + broad) Hα emission. Col. (4): FWHM of broad Hα in units of km s⁻¹. Col. (5): velocity shift of the broad Hα component relative to the narrow Hα line; the units are km s⁻¹. Col. (6): observed (not corrected for reddening) flux of broad Hα, in units of erg s⁻¹ cm⁻²; Palomar fluxes are available only for objects observed under photometric conditions. Col. (7): observed (not corrected for reddening) luminosity of broad Hα, in units of erg s⁻¹, assuming distances from Ho et al. (1997a). Col. (8): the year of observation with STIS. Cols. (2)–(4) and (6)–(8) show in parentheses the corresponding parameters measured from the ground with the Palomar (Ho et al. 1997b) and MMT.

^aNon-Gaussian profile of the broad component; FWHM and velocity shift are flux-weighted averages of the respective values in all Gaussians used to fit the broad Hα.

Download table as:  ASCIITypeset image

While most of the time the detection and measurement of a broad Hα feature are fairly obvious, there are cases where this process remains ambiguous and therefore debatable. For example, NGC 4036 is one of the nuclei with broad Hα detections in the Palomar survey, for which our spectral fits of the 1D extraction of the central five rows of the STIS spectrum reveal a broad component, but for which Walsh et al. (2008) do not find the need for it in any of the individual central rows of the 2D spectrum derived from the same observational data set. The disagreement with our findings may be the result of the increased signal, and thus an increased contrast with the continuum, that we obtain by adding the flux in the central few rows. Nevertheless, our use of a wider extraction would also decrease the contrast between the broad line and the continuum since the wider aperture adds more continuum. We will revisit this issue later in the section, where we discuss the new detections of broad Hα in the HST data.

Of the objects that are ambiguous in terms of the presence/detection of broad Hα in the Palomar survey, three nuclei are included in this study. While NGC 4501 (a Seyfert included in the SUNNS sample) is claimed to lack the broad emission in the STIS aperture (S07), NGC 3245 (viewed from the ground as a narrow-lined T nucleus) requires for a good fit of the Hα + [N ii] doublet the presence of a broad component, with a contribution to the total blend of f_blend ≈ 47%. The detection significance is tested by fitting the feature with models based on both one- and two-Gaussian component templates for the [S ii] lines; f_blend remains basically unchanged in both cases. We show in Figure 3, among other sources for which we find new evidence for broad emission in the STIS spectra, the individual Gaussian components, the broad Hα feature, and the final spectral fit for both the Hα + [N ii] line blend and the [S ii] λλ6716, 6731 doublet for NGC 3245. The existence of the broad Balmer feature in this nucleus has also been suggested by Barth et al. (2001b), based on this same data set. The third object with tentative broad Hα emission in the Palomar spectrum, NGC 2985, remains, as in the ground-based observations, rather elusive: a moderate (f_blend ≈ 44%) broad component may be necessary to account for the entire flux of the blend after the narrow components have been modeled based on [S ii] templates; however, the data are of low quality across all features, and we cannot rule out the possibility that the excess emission originates from weak, intrinsically broad wings in the narrow-line profile. We decided to include the detection and measurements of this broad feature in our analysis, for the purpose of comparison with the ground-based data.

Zoom In Zoom Out Reset image size

In the same Ho et al. (1997b) study that presents the search for "dwarf" nuclei with broad emission, the authors also list the sources where the total Hα + [N ii] blend hints at a probable presence of weak broad wings of Hα, but for which a careful profile fitting analysis indicates null detections. Among those objects, seven sources are included in our study. For six of them (NGC 4314, 4321, and 4698 [included in the SUNNS sample] and NGC 1961, 2911, and 6500), the profile fits of the HST spectra do not require any broad component. The seventh object is NGC 4594, which has also been previously observed with the Faint Object Spectrograph (FOS) aboard HST using a 086-diameter circular aperture. In these data, the Hα + [N ii] complex shows evidence for broad wings; attributing them to a broad Balmer component, however, remains a matter of debate (Kormendy et al. 1996; Nicholson et al. 1998). This nucleus exhibits severely blended velocity structure in the STIS slit (01), which is almost one order of magnitude narrower than that used in the FOS observation. While the optical Hα + [N ii] emission remains difficult to interpret, we find that a good fit can be obtained only by adding a moderately strong broad component (f_blend ≈ 29%; Figure 3, Table 4).

The STIS observations of our MMT sample have been previously presented and discussed by Noel-Storr et al. (2003) for all but NGC 1497, which is discussed by Walsh et al. (2008). Both of these studies emphasized measurements of the gas kinematics in the vicinity of the supermassive black holes in the centers of these galaxies and performed spectral fits and measured the nebular emission in individually extracted rows. As described above, we have performed our own data reduction and extraction of the 1D spectra for all 16 MMT objects. Thus, the line fluxes and other parameters (e.g., evidence for Hα broad emission) differ somewhat from those listed in Noel-Storr et al. (2003) and Walsh et al. (2008). For example, Noel-Storr et al. (2003) suggest the possible presence of a broad Hα component, as measured in the single central row, in six of our MMT sources and two Palomar objects (NGC 4261 and NGC 7626); we find that including a broad feature in the spectral fit does not improve the resulting χ² per degree of freedom (${\chi }_{\nu }^{2}$) significantly enough to warrant its physical existence in NGC 315, 4261, 7626, and UGC 12064, while we confirm its presence in NGC 193, 383, 4335, and UGC 1841. As for NGC 1497, our measurements agree with the findings of Walsh et al. (2008) that a broad Hα line does not improve the final spectral fit; however, this data set has rather poor S/N, and the decision for the best spectral fit is basically the most conservative choice that allows for the smallest number of free parameters. For some of the galaxy nuclei for which we decided against the presence of the broad feature, apparent broad wings in the Hα + [N ii] blend can be fit with two Gaussians for the [S ii] lines and for all the other narrow features (see Table 1). We adopted the same method when fitting the host-galaxy-corrected (pure emission-line) spectra acquired from the ground with the MMT, and we found that in only one of these systems (NGC 383) a broad Hα component is needed for the best spectral fit (Table 4).

Arguments against detecting broad Hα emission, or for mismatches in the measurements of this component, could also stem from employing different spectral fitting methods. For example, in a recent discussion of the relative efficiency of HST versus ground-based spectra for detecting BLRs in low-luminosity AGNs, Balmaverde & Capetti (2014) suggested that one should use the [O i] profile (which is usually of larger width than that of the [S ii] lines that we use as templates) to model the [N ii] and the narrow Hα contributions to the blend, because the stratification in density and ionization within the NLR causes differences in the location of the emitting region of the various lines and thus in their line profiles; when the [O i] profile is used, these authors find that there is no more need for a broad component in nine of the objects we discuss here (NGC 315, 1052, 2985, 3642, 4036, 4258, 4278, 5005, and 5077). Nevertheless, it is important to note that over the spatial scales of the HST spectra, the line width versus n_crit relationship is spatially resolved, and thus the NLR stratification should no longer be an issue; the use of the [O i] emission as a template is not justified, as its critical density is much higher than that of the [N ii] and [S ii] lines.

2.4.1. New Evidence for Broad Hα in the STIS Spectra

2.4.2. Decade-scale Variability in the BLR?

The overall characteristics of the broad Hα emission measured in the STIS aperture are somewhat surprising, especially when compared to those measured from the ground. The fractions f_blend and f_Hα and the FWHM (broad Hα) are usually larger in the HST spectra, as expected if the narrow emission is spatially resolved and a significant amount of its light is not included in the small aperture.

Nevertheless, with very few exceptions, the STIS broad Hα fluxes are smaller than those measured in the ground-based aperture; in median, the reduction in the broad Hα flux exceeds a factor of eight. A similar trend was noticed by S07 for the subsample of six SUNNS broad-lined objects. Since the broad feature is expected to arise from subparsec scales, it should not suffer from aperture losses; thus, S07 interpreted this result as possibly arising from variability in the broad emission, on timescales that are shorter than a decade (i.e., the interval between the Palomar and HST observations).

The reason why variability can explain these differences comes from the different detection thresholds for a BLR component in the ground-based and space-based data. Since the larger aperture employed in the ground-based observations includes a higher fraction of the stellar light surrounding the nucleus, there is a weaker contrast between the BLR and the associated continuum and thus a lower likelihood to detect the broad Balmer line as a separate component in the Hα + [N ii] emission blend. Thus, while the chance for the luminosity to increase or decrease is the same in a (simple) variability scenario, AGNs in which the BLRs are low in the ground-based data and high in the space-based data would simply be less likely to provide us with a broad Hα detection and hence with a flux measurement in the low state (i.e., from the ground). Similarly, AGNs in which the BLRs are high in the ground-based data and low in the space-based data would exhibit a lower flux in the broad component measured by HST and hence a negative flux offset similar to what we see in the comparison of the Palomar and the HST nebular emission.

With the increased statistics offered by the present sample we can, probably for the first time, analyze broad Hα variation in more objects and over longer intervals. In addition to simply comparing the flux values from the ground-based and HST data, it is of interest to examine their ratio as a function of time between these observations, to get some sense of the timescales of variability. If the scatter in the observed ratio increases significantly with increasing time interval, we can conclude that the variability spectrum has significant power on timescales extending beyond a decade.

Figure 4 presents the change in log F_Hα and in f_blend as a function of time between the Palomar and the STIS observations, for all galaxies with broad-line emission detected from the ground. There is evidence of a weak trend between the amplitude of variation and time: an increasingly negative difference with increasing time interval between observations. There is no apparent trend in f_blend; as expected based on the increased contrast with the continuum, f_blend remains generally higher when measured in the small HST apertures, regardless of the time interval between observations. The MMT object for which we have the broad Hα comparison data is too isolated from the rest of the sample in its time interval (Δt ≈ −8 yr) and is not plotted here. A more definitive quantitative analysis of the variability statistics in such objects will require larger samples or additional epochs of observation.

Zoom In Zoom Out Reset image size

Of course, the variability scenario also predicts new detections of broad-line sources in the STIS spectra of objects that in the Palomar survey would have been in a low-flux state and therefore appeared as Type 2 nuclei owing to sensitivity limits. The SUNNS sample does not fulfill this variability prediction; however, in our much larger data set there are potentially eight new broad Hα detections (Section 2.4.1), thereby bringing additional supporting evidence for the validity of this result.

We also note that our working sample is not subject to biases with respect to the broad-line detection. The sample selection is based purely on the availability of high-quality G750M STIS spectra, with no a priori knowledge of the existence or detectability of a broad Hα emission feature. Nevertheless, one might worry that because the ratios presented in Figure 4 can be computed only for those cases where we detect broad Hα in both apertures, the results would be biased against those objects in which the broad Hα significantly increased over the years (from nondetection in the Palomar sample to detection in HST data). To address this, we compare in Figure 5 the overall distributions of the broad Hα flux and f_blend in both apertures, where we show separately the whole sample of objects with HST BLR detection, as well as the subsample of objects previously known to exhibit broad emission from the ground. We find that, on average, the HST fluxes are lower than those measured from the ground, regardless of whether we consider all of the HST detections or just those previously known to be of Type 1. By the same token, the BLR component accounts for a higher fraction of the total Hα emission in the higher spatial resolution (small aperture) data, which argues again for a clear improvement in the contrast with the continuum as the HST observations resolve the nuclear emission. One might also think that the diminution of the broad Hα in the small apertures could be caused by cutting out some BLR light from larger scales, and thus that the BLR is not spatially constrained into subarcsecond regions, as has been previously assumed. The increase in f_blend, however, argues against this possibility.

Zoom In Zoom Out Reset image size

Because of the fixed slit size, the metric aperture depends on the distance to the source, and thus the measured flux ratios reflect nebular properties of different spatial extents within the galaxy centers. Such an effect may influence the comparisons of flux ratios; for example, Storchi-Bergmann (1991) reported an inverse correlation between [N ii]/Hα and distance for a sample of (distant and relatively bright) Seyfert 2s and LINERs that can be attributed to inclusion of increasing amounts of circumnuclear H ii region emission in more distant sources. We investigate the potential influence of this effect in our sample of nearby nuclei in Figure 6, where both this line flux ratio as measured with HST and the [N ii]/Hα ratio measured in the small and large (ground-based) apertures are shown as a function of galaxy distance. The different spectral types, as defined by ground-based observations, are illustrated by different symbol types (Seyferts include both Type 1 and Type 2). The Hα flux is not corrected for the absorption from starlight in the host galaxies (see Section 2.3). The distribution in distances, and consequently in the metric scales that the HST nuclear observations encompass, is heavily concentrated around ∼16 Mpc,⁹ with outliers extending to 96 Mpc (aperture radius of ∼55 pc). The LINER 2 nuclei are the only class that shows a large dispersion in distance (6–96 Mpc). No obvious trend in line ratio with source distance is present in either the full sample or the subgroups of specific spectral types, suggesting that distance-dependent aperture effects are not a major concern.

Zoom In Zoom Out Reset image size

For assessing orientation-biased selection effects, we also illustrate in Figure 6 the dependence on the inclination angle (i) of the host galaxies of both the HST [N ii]/Hα flux ratio and the ratio of this line ratio measured in the small and large apertures. We have calculated i based on the isophotal axial ratio R₂₅ = log(a/b), where a and b are the major- and minor-axis diameters measured at a B surface brightness level of 25 mag arcsec⁻², as in Ho et al. (1997a). Visual inspections and two-sample Kolmogorov–Smirnov (K-S) tests indicate that, with the exception of T and L2 nuclei, all other types of objects present uniform distributions in their host inclinations. As found in other studies (e.g., Ho et al. 1997b), galaxies with T nuclei might be marginally more inclined than those with L2 nuclei. Similarly to the above discussion regarding the possible trends of the [N ii]/Hα flux ratio with distance, the more edge-on systems would be expected to include more circumnuclear H ii emission in the aperture as well, and consequently to reveal smaller [N ii]/Hα ratios. However, within the whole sample and the individual subsets of different spectral types, there is no obvious inverse correlation between the [N ii]/Hα ratio and the galaxy inclination. There is also no statistically supported evidence that [N ii]/Hα ratios are smaller in T than in L2 nuclei. This analysis suggests that inclination biases are not significant in our sample.

Because for elliptical galaxies the axis ratio does not indicate an inclination angle, we also show in Figure 6 the dependence on the host morphological type of the HST [N ii]/Hα flux ratio, as well as the ratio of this line ratio measured in the two apertures. We use here the morphological classification given by the mean numerical Hubble type index as given in the Third Reference Catalog of Bright Galaxies (RC3; de Vaucouleurs et al. 1991). It is apparent that the sample contains virtually every morphological type and hence provides a good representation of the general galaxy population. Ellipticals compose 16% of the sample, while lenticulars and spiral systems make up 25% and 60%, respectively. Barred galaxies (AB and B) contribute 24% to the disk systems (S0–Sm) and 34% to the spiral galaxies. Once again, we find no apparent influence on the distributions of the line ratios as a function of the host morphology, both for the whole sample of galaxy nuclei and for the subsets of different spectral types.

Of great interest is the degree to which the AGN phenomenon extends to luminosities fainter than those of the well-established accretion-like sources (Seyferts and LINER 1s). Figure 7 illustrates the range in narrow Hα luminosity spanned by these nearby nuclei in the HST aperture, separately for each spectral category. On average, L1 galaxies are the most luminous among all objects; the more luminous L2 nuclei share a similar distance range with that of the L1s and are not among the most distant galaxies in our sample; the ambiguous T objects are generally less luminous than the S and L1 nuclei and cover the widest range in the observed power; and the H ii galaxies span a similar range in L(Hα) as the Seyferts.

Zoom In Zoom Out Reset image size

The broad-lined objects are of special interest, particularly in terms of potential statistical trends toward finding or missing them in regard to their total emitting power. We thus indicate the nuclei with broad Hα components in the STIS spectra (as described in Section 2.4) by large open circles; we also mark with large open squares the sources for which the narrow emission lines seem to have contributions from multiple kinematically distinct regions (i.e., multiple Gaussians are needed to fit their line profiles) that are interpreted in some instances by Véron et al. (1997) and Gonçalves et al. (1999) as evidence for composite S/L + H ii emission. We indicate these types of objects among the Seyferts as well. With only two exceptions, the nuclei with new evidence of broad Hα are among the most luminous ones; the same is true for the objects marked by squares. This pattern may reflect selection effects related to sensitivity; in the more luminous sources, the contrast between the nuclear and the surrounding emission is higher, thus implying a higher likelihood to detect a broad component.

An important issue in this comparative study is the degree to which the high spatial resolution HST observations are able to resolve the nuclear emission in these nuclei. Spectra obtained through a smaller aperture should be characterized by a reduction in the nebular flux. In particular, for extended sources of uniform surface brightness, the decrease in flux is expected to match the simple ratio of the aperture areas (∼−2.2 dex); smaller changes in the emission-line fluxes would indicate some degree of central concentration.

Figure 8 presents the ratios of Hα flux, in the narrow component only, measured in the HST spectra relative to that given by the Palomar and MMT observations. The results are again sorted according to the ground-based nuclear spectral classification as defined by Ho et al. (1997a). For almost all sample objects, the Hα flux is smaller by at least 0.5 dex in the high-resolution HST aperture, which means that, even for the most distant objects in the sample, the nuclear emission is resolved. The symbol size scales again with the galaxy distance, and it can be seen that the ratios do not show any trend with the sampled metric aperture.

Zoom In Zoom Out Reset image size

It is readily apparent that the smallest change in the flux ratio is found in the nuclei that are unambiguously powered by accretion, the Seyfert and the LINER 1 objects, indicating as expected the highest degree of central concentration of emission. The H ii, T, and L2 nuclei show lower values of the flux ratio, suggesting a shallower gradient in the surface brightness distribution, and thus a more distributed ionization source. The behavior of the T and L2 sources is very similar, arguing that the T nuclei are probably not more prone to contamination by emission from extranuclear star-forming regions than L2 nuclei, contrary to what has been suggested by Ho et al. (2003) based on indications that T nuclei might reside in more highly inclined host galaxies than those hosting LINERs. As noted in Section 3.1, our sample properties are not strongly influenced by projection effects, if present; hence, such biases should not play a major role in distinguishing between these two types of emission-line nuclei.

Some trends are particularly noteworthy. For 5 out of 21 H ii nuclei, the HST aperture spectra show Hα emission at levels that are even lower than those predicted by scaling down the ground-based value by the aperture ratio. These cases show that "nuclear" star formation can in reality span a significant range of radial scales and may not continue into the actual center of many galaxies; this is consistent with previous findings of, for example, Cid Fernandes et al. (2004) and González Delgado et al. (2004), who find that the star formation powering the centers of H ii galaxies resides on scales larger than tens of parsecs. The same trend is apparent among the most distant L2 and some T nuclei (not only the most distant), thus suggesting that the average surface brightness actually increases with radius in these objects. This could be caused by, for example, obscuration of a source in the center, or extended nebulosity like diffuse ionized gas (DIG) or warm ionized medium (WIM) that is actually brighter at larger radii. This behavior can have interesting consequences on the results of the nebular aperture comparison we conduct here, and we address them in the next section.

Figure 9 presents the "ratios of ratios" that compare R = [N ii]/Hα, [S ii] λλ6716, 6731/Hα, and [O i] λ6300/Hα measured from the HST data to those obtained from the ground-based spectra. For the sake of completeness, we show these ratios of ratios calculated with and without correction for stellar absorption (see Section 2.3) in the left and the right panels, respectively. We indicate in the right panel the systems with HST detections of a broad Hα line (large open circles), as well as those whose spectra are fitted with multiple Gaussian profiles to account for broad wings in the emission lines (large open squares), as in Figure 7.

Zoom In Zoom Out Reset image size

There is a general pattern that can be seen in all of these plots: for the majority of the L1 and S nuclei, these ratios show little scatter around unity, while H ii, T, and L2 nuclei reveal a wide range of values for these ratios. The L1 and S nuclei exhibit no clear average change in [N ii]/Hα between the two apertures, a systematic decrease in [S ii]/Hα in the small aperture, and a systematic increase in the [O i]/Hα line ratio in the more nuclear spectra. For the H ii, T, and L2 nuclei, the majority of sources have larger [N ii]/Hα, [S ii]/Hα, and possibly [O i]/Hα, in the small-aperture observations. As revealed in Figure 6, there is no obvious overall correlation with distance to the host galaxy, host inclination, or host morphology in either the [N ii]/Hα or R_HST/R_ground values that could skew this comparison, and we tested that this remains true for the other ratios.

Figure 10 illustrates the distributions in the [N ii]/Hα, [S ii]/Hα, and [S ii] λ6716/λ6731 line ratios measured in the small and large apertures (where we excluded the upper and lower limits); there is a clear increase in the median values of the forbidden to Hα line ratios for H ii, T, and L2 nuclei in the more central regions, with not much change between the small- and large-aperture measurements of S and L1 nuclei.

Zoom In Zoom Out Reset image size

The aperture effects on the ratio [S ii] λ6716/λ6731 that serves as a density indicator (Osterbrock 1989) are also illustrated in Figure 10, separately for each nuclear spectral type. It is quite apparent that in the more central nuclear regions mapped by the STIS observations the gas densities are larger (i.e., smaller [S ii] λ6716/λ6731 flux ratios), in all types of objects. A few exceptions are among the H ii nuclei, but there are generally large error bars associated with these measurements, so they are not necessarily excluded from the general trend. The T nuclei show a particularly pronounced difference in the gas density measured in the two apertures, with the largest fraction (∼90%) of objects exhibiting greater than 50% increase in their n_e.

It is noteworthy that the differences between the ground-based and HST ratios of forbidden to Hα lines are generally small. Although, as described in Section 3.3, the flux sampled in the HST apertures is typically a small fraction of that contributing to the Palomar and MMT spectra, and the spatial scale of the observed regions differs by at least an order of magnitude, there are very few sources that might indicate a change in their classification from ambiguous (narrow-lined) nuclei to a more secure AGN-like behavior—that is, a rightward shift in the line-diagnostic diagrams. The relatively small fraction of objects with newly detected broad Balmer lines is consistent with this result.

To aid in discussing and understanding the possible effects of extended DIG/WIM nebulosity at larger radii, we also mark with red open squares (only in the left panels of Figure 9) the peculiar sources for which the Hα flux ratio falls below the −2.2 dex line corresponding to the simple ratio of the aperture areas (Figure 8). As expected, these particular sources are generally among those exhibiting an increase in the forbidden to Hα ratios, owing to the drastic decrease in the Balmer line intensity; when the Hα emission is not corrected for stellar absorption (right panels), all of these objects show an increase in the flux ratios. Thus, especially for the more distant objects, the small-aperture data support the geometric stratification of the extended low-ionization emission regions, which was recognized as the dominant power source in a great majority of (early-type) galaxies spectrally classified as Type 2 LINERs (i.e., the LIERs, with LINER-like emission that is not nuclear; e.g., Sarzi et al. 2011; Yan & Blanton 2012) Nevertheless, these particular sources are not the only ones showing changes in the flux ratios, suggesting that the changes in the flux ratios are not entirely caused by this effect.

The objects with newly detected broad Hα emission exhibit an increase in the narrow-line ratios in spectra acquired at smaller scales. Thus, at least for these objects, the scenario in which the small-aperture emission may be dominated to a greater degree by a central accretion source seems applicable. Nuclei whose ground-based spectra indicate a potential mix of velocity fields (i.e., distinct multiple Gaussian profiles required to fit the narrow emission) behave similarly in both apertures, as their spectra remain consistent with very little or no change in the nebular line ratios; they are also among the most distant targets, which might mean that, if the "composite" picture holds, their HST nuclear observations could still include light from circumnuclear star-forming regions.

With the expectation of revealing an increasingly similar behavior of the T and L2 nuclei to that of bona fide AGNs (L1 and S nuclei) in their more nuclear spectra, we statistically compared the distributions in the line ratios, as well as in the Hα fluxes and luminosities, for all possible galaxy-type-pair combinations. For the ground-based measurements, the K-S probabilities are generally low, revealing potentially common parent populations only for L1–L2 systems (p_K-S ≲ 0.2), with basically no chance for commonality with T and S systems or between the latter two types (p_K-S ≪ 10⁻³ for all these pairs of spectral types). The K-S probabilities are, however, boosted to p_K-S ≈ 0.3–0.8 for the HST line-ratio distributions for all possible combination pairs of Type 1 and 2 LINERs, T nuclei, and S nuclei, with higher values for the [S ii]/Hα and the [S ii] λ6716/λ6731 ratios, especially for the S–T and L2–T pairs, suggesting an increased difficulty in ruling out the "null hypothesis" of a common parent sample.

The trends we measure in the ratios of collisionally excited forbidden lines to Hα are not particularly well mapped by current theoretical models for AGN ionization (e.g., by classical dust-free AGN photoionization codes, MAPPINGS III; Dopita & Sutherland 1996; Groves et al. 2004a; Allen et al. 2008). For example, they do not reproduce the high [S ii]/Hα ratios we measure in the HST observations, and thus remain powerless in addressing electron densities or gradients in n_e as high as we find here. The models predict a tight clustering of the line-ratio values for relatively broad ranges of variation in the parameters that can cause gradients like those we measure in our sample: for example, an increase by 0.5 dex in the average [S ii]/Hα or [N ii]/Hα measured for T nuclei is consistent with either (i) an increase in n_e by at least an order of magnitude (10³–10⁴), (ii) a factor of ∼3 increase in metallicity, (iii) a hardening of the continuum corresponding to a change of the spectral index α from 2 to −1.4, or (iv) a combination of these changes. While the degeneracy in identifying the principal cause for such gradients cannot be broken, our data probe a clear increase in n_e, which proves important for understanding of the nuclear emission in these nuclei. We explore in more detail the potential insights from these results in the following subsections.

That the emission in the small aperture is dominated to a greater degree by a central accretion-powered source is not, however, the only reason for the somewhat stronger forbidden lines relative to Hα in the more nuclear spectra. The enhancement in these ratios can also be simply a direct consequence of the trend toward larger electron densities at small radii. As revealed by the patterns in the [S ii] doublet ratio, this is happening for virtually all nuclear types. As long as the density remains below the critical value, n_crit, the line emissivity increases ∝n². Because large densities may lead to suppression of the infrared fine-structure line cooling, a boost may result in the optical forbidden lines like [N ii], [S ii], and [O i]. Interestingly, the [S ii]/Hα ratios generally decrease in the small aperture for L1 and S nuclei, suggesting that the densities in these sources extend to values beyond the [S ii] critical densities.

Looking closer, it becomes clear that there is varied behavior for the different line ratios studied here, which can be understood as a consequence of the different critical densities of the forbidden lines. For comparison, n_crit for [S ii], [N ii], and [O i] are (1.5–3.9) × 10³ cm⁻³, 8.7 × 10⁴ cm⁻³, and 1.8 × 10⁶ cm⁻³, respectively, thus spanning several orders of magnitude. While n_e generally increases in the HST aperture, it is possible that a significant fraction of objects reach n_e ranges similar to or greater than the (relatively small) n_crit of [S ii] λλ6716, 6731; this could act to collisionally suppress the [S ii]/Hα ratio in these small apertures, thereby leading to the measured decrease in this ratio relative to the one in ground-based spectra.

In this scenario, only a few objects would reach n_e values higher than n_crit of the [N ii] λλ6548, 6583, with even fewer reaching the ∼10⁶ cm⁻³ values of n_crit for [O i] λλ6300, 6364, leading to increasingly less suppression in these features; thus, there is an increasing fraction of objects with ratio of ratios R_HST/R_ground > 1 for [N ii] and [O i], respectively, which is what we find in our comparison. This effect has been seen directly in previous work illustrating measurements of the line flux ratios as a function of position along the slit; for example, Barth et al. (2001b) and Walsh et al. (2008) revealed for NGC 4579 and additional objects a strong trend where the [S ii] ratio rises toward the high-density limit closer to the nucleus, while the [O i]/Hα and [O i]/[S ii] ratios spike upward in that innermost region.

There is yet another physical explanation for the observed small change in the line ratios with distance from the center: the line ratios would be left unchanged if both the density of ionizing photons and that of electrons diminish with radius. A physical basis for such a balance has been described by Groves et al. (2004a, 2004b), who argue that the response of gas pressure to radiation pressure acting on dust in the emitting clouds will produce a local excitation source that becomes independent of the external flux and hence distance from a central radiation source. The results of this study for L1 nuclei and Seyferts are at least qualitatively in accord with their model.

Within this picture, it is then interesting to investigate to what degree this effect might contribute to the lack of change in the T and L2 nuclei as well, given that these sources' emission is, by definition, dominated by low-ionization regions. The T and L2 systems with new detections of broad components are not among those with the largest gradient in the gas densities. Thus, for these objects in particular, the potential counteracting changes in n_e are probably not very strong, and a significant aperture dependence is seen. On the other hand, the rest of the T nuclei, with the greatest changes in n_e among the entire sample, would be expected to show weak gradients, and this is in accord with the observations. This pattern does not seem to apply for L2 nuclei as well; the gradients in n_e are large among L2 galaxies, but the values of n_e in the HST apertures remain systematically lower than in the T, S, and L1 nuclei. Thus, gradients in n_e could be, at least in part, the cause for the similar line ratios in the small and large apertures.

We note that because the measured density is weighted by luminosity, the difference in density gradients between the different types of objects may not reflect a difference in the physical density gradients, but could be due to the difference in their luminosity profiles. As evidenced by the smaller difference between the two apertures, the Seyferts have a sharper luminosity profile; thus, the densities measured in both large and small apertures are more heavily weighted toward the central density. The T nuclei have a shallower luminosity profile, which might translate into a larger density difference. Nevertheless, this point does not invalidate any of the results presented here.

That Type 1 and 2 L nuclei do not show significantly greater similarity in their nuclear nebular emission brings back the question of whether these two object types are fundamentally the same and just appear different from various angles owing to obscuration, or have real intrinsic differences—for example, the BLR could be completely missing in the Type 2 systems, perhaps as a consequence of the evolution of the AGN environment as a function of the accretion rate.

To summarize our findings, the L1 and L2 nuclei span different ranges in narrow Hα luminosities and fluxes as measured with STIS (see Section 3). The forbidden to Balmer line-ratio distributions of these nuclei do not exhibit an increase in the K-S likelihood of being drawn from the same parent populations when measured in the more nuclear STIS spectra. The values of their ratio of ratios in [N ii]/Hα and [S ii]/Hα seem to follow different trends: they are almost scattered in opposite directions around the unit value, with L1 nuclei showing lower [S ii]/Hα values in the HST spectra relative to those of Palomar and MMT (Figure 9). Also, the average n_e values and general distributions remain dissimilar. While the largest gradients in n_e among all LINERs are registered in L2 nuclei, L1s exhibit higher gas densities than L2s in both ground-based and space-based observations: median n_e(L1) ≈ 8.5 × 10³ cm⁻³, n_e(L2) ≈ 5 × 10³ cm⁻³ in the HST spectra, and n_e(L1) ≈ 1.5 × 10³ cm⁻³, n_e(L2) ≈ 0.6 × 10³ cm⁻³ in the ground-based data.

A possible impediment to understanding these differences can be that the L2 nuclei employed in this study cover a wider distribution in galaxy distances than the L1, T, or S nuclei, which are more concentrated near 20 Mpc. In an attempt to match the two samples of LINERs in both mean/median and distribution of distances, we ran the K-S comparison tests on the line ratios using L2 subsamples that exclude some or all of the five most distant sources (see Figure 6). In all cases, the subsamples of L2 nuclei manifest the same behavior as the full L2 sample, which means that differences in metric aperture do not contribute significantly to the differences in the nebular emission behavior between L1 and L2 nuclei.

The differences we found between the L1 and L2 sources appear to support both an obscured BLR and the missing BLR interpretation for the L2 sources, with both of these phenomena being linked to the low luminosities and accretion levels in the latter systems.

On the one hand, our findings of lower average n_e in the more nuclear regions of L2s than in L1s, which could explain their positive gradients in [S ii]/Hα ratios that are in contrast to the negative gradients in L1s, offer support to the unified scheme for orientation-dependent obscuration in LINERs: an obscuring medium blocks the innermost (and densest) part of the NLR in addition to the BLR (e.g., Barth et al. 1999). Greater obscuration is expected for lower-luminosity accretion sources (i.e., the receding torus model; Lawrence 1991; Simpson 2003; Constantin 2004), which, again, is consistent with the lower Hα luminosities for the L2 nuclei than for the Type 1 LINERs. Nevertheless, as discussed in detail by Ho (2008), there remains very little observational evidence for any obscuration or absorption in these particular systems; thus, the idea that L2 nuclei are more obscured than the L1s remains speculative.

On the other hand, if differences between the two types of LINERs are linked to the disappearance of the BLR in the Type 2 systems, then we should also see differences in their nuclear activity; for example, this phenomenon is expected at very low luminosity levels (Nicastro 2000; Laor 2003; Nicastro et al. 2003) and extremely low accretion rates or efficiency in accretion (Elitzur & Ho 2009; Elitzur et al. 2014). To test this idea, we have compared L_Hα/σ⁴ for objects of different spectral types. We employ L_Hα/σ⁴ as a proxy for the Eddington ratio L/L_Edd, with L_Hα being the Hα line luminosity measured in the HST spectra, which we use as a gauge of the accretion luminosity, and σ⁴ as a measure of the black hole mass, via the black hole mass versus stellar velocity dispersion (M_bh–σ) relation that has been well measured for both inactive galaxies (e.g., Ferrarese & Merritt 2000; Gebhardt et al. 2000) and active systems (e.g., Bennert et al. 2015). We use the σ values from Ho et al. (2009) for the Palomar objects and from Beifiori et al. (2012) for the MMT galaxies. Figure 11 shows the distributions and average values of L_Hα/σ⁴ for S, L1, L2, and T nuclei. The distributions, average, or median values of the L1 and L2 nuclei are indeed different, with low K-S probabilities (p < 0.1) for being drawn from the same population; the L1 systems are at least 0.7 dex stronger in the median L_Hα/σ⁴ values, regardless of whether L_Hα values include or exclude the broad Hα emission. This comparison argues, then, for "BLR nakedness" in the L2 systems. For these galaxies, the lack of detectable dense gas at small radii could simply extend beyond the BLR to encompass the inner NLR.

Zoom In Zoom Out Reset image size

One could note that the flatter gradients in the line ratios of the L2 nuclei have another possible explanation that may have nothing to do with accretion power: if they are also dominated by some kind of stellar photoionization on the scales probed by HST, they will display flat gradients in line ratios. Nevertheless, the line gradients for this class of objects are only weak in [N ii]/Hα, while they are quite significant in [S ii]/Hα and especially in [O i]/Hα, arguing against dominant stellar photoionization at the parsec scales probed by the STIS observations, for at least half of the sample.

Thus, the parallel between L and S nuclei of Types 1 and 2 extends nicely to the smallest physical scales probed to date, from the morphology of the narrow-line nebular emission of both L2 and Seyfert 2 nuclei (Pogge 1989a) appearing less concentrated than their Type 1 counterparts (Figure 8), to a potentially strong relation between the strength of accretion and the existence (or disappearance) of a broad-line component. Although we do not argue for any particular physical model here, it is possible that the trends we see in the line-ratio gradients, the gradients in n_e, and the differences in L_Hα/σ⁴ for the L1 and L2 nuclei mirror the differences between the Type 1 and 2 Seyferts or quasars, which are likely caused by a systematic change in the shape of the accretion-disk continuum and its interplay with the ambient line-emitting regions (e.g., Shen & Ho 2014).

The prevailing paradigm of what T nuclei are involves a weak central AGN with compact circumnuclear star formation that all falls in the aperture of a typical ground-based measurement (e.g., Kennicutt et al. 1989). The composite model (e.g., Ho et al. 1993) predicts that the small aperture should help isolate the accretion-powered component such that the small-aperture spectrum is more AGN-like. The results of the aperture comparisons presented here show unprecedented evidence for the fact that T objects behave more like the accretion-type sources in their more nuclear regions. We have found for T nuclei (i) a threefold increase in the incidence of broad-lined nuclei, (ii) the largest of the predicted gradients in the line ratios (Figure 9), and (iii) a shift in their overall line-ratio distributions that supports a closer similarity of their nebular properties with those of the S, L1, and L2 nuclei in the small aperture. The K-S tests suggest a less likely probability of these systems not sharing a common parent population. All of these trends are consistent with the composite picture for T nuclei.

The central AGN-like source in the proposed models has been generally considered to be a LINER; however, previous studies did not have the sensitivity or the statistical power to investigate in detail the nature of the central object in the composite model. This work offers new quantitative insights into its nature. We found an increase in similarity between the properties of T and L2 nuclei in the ∼10 pc scale regions; there are comparable distributions of fluxes, line ratios, and their gradients, and possibly accretion rates as well, as probed by the L_Hα/σ⁴ parameter (Figure 11). Nevertheless, the transition to a more AGN-like behavior in the more nuclear regions is different in the T and L2 galaxies: the T nuclei show the strongest line-ratio gradients, thus exemplifying most positively the expected results of the aperture comparison test proposed here. Interestingly, when we compared the stellar velocity dispersions (σ∗), as a measure of M_bh, for our samples of S, L1, L2, and T galaxies, we found that T and S nuclei show consistent (low) average values for σ∗, and thus for M_bh, while the L1 and L2 galaxies exhibit increasingly larger ones (Figure 12). These characteristics suggest greater likeness in the T nuclear environments with Seyferts than with L2 systems.

Zoom In Zoom Out Reset image size

A possible account for these trends is provided by the idea that T nuclei are the intermediate phase in the evolution from a star-formation-dominated nucleus (i.e., an H ii galaxy) toward an unveiled compact accretion source (i.e., a Seyfert), where the accretion rate rises toward the values seen in S galaxies, while the L2 systems are the manifestation of the post-Seyfert (and possibly post-L1 phase; e.g., Elitzur et al. 2014), where the accretion wanes down to eventual quiescence (Schawinski et al. 2007, 2010; Constantin et al. 2008, 2009). This scenario supports (i) the closeness of the T systems with the H ii systems in the ground-based observations, as traces of surrounding stellar activity should be more pronounced than in L2s, while (ii) exhibiting increasing AGN-like nature in their most nuclear regions, as evidenced via both new detections of broad Hα lines and strong gradients in the line ratios.

If T nuclei are an intermediate phase in the evolution from a star-formation-dominated nucleus toward a dominance of Seyfert-like excitation, it is thus possible that our observations catch these systems in various evolutionary stages. This would naturally explain the wide range in their nuclear properties, and in particular, the span of the line ratios, as well as the small geometrical segregation between H ii and AGN-like emission when more than 90% of the nebulosity is resolved as the aperture is reduced by a factor of ∼100 in area and the measured flux drops by an order of magnitude or more (i.e., Figures 2, 8, and 9).

One could argue that if a galaxy has a cold gas supply, its nucleus could go through H ii, LINER, and Seyfert phases episodically, in whatever order. Correlations between these spectral types and the stellar population of the host galaxy could then simply be a reflection of differences in gas supply and thus different AGN duty cycles in various galaxy types. Nevertheless, there is a systematic trend in the masses of the black holes in these galaxies, suggesting that the black hole grows along a T $\to$ S $\to$ L sequence, which is readily apparent even within our small, inhomogeneous, and statistically incomplete sample (Figure 12).¹⁰ Markedly, the timescales necessary to transform from one galaxy type to the next are not unreasonable: considering a radiative efficiency value ≈ 10⁻⁶ (e.g., Rees et al. 1982; Narayan & Yi 1994; Ho 2008), the e-folding time in the BH mass is ${t}_{e}\approx 4.5\times {10}^{8}\displaystyle \frac{\epsilon }{\lambda (1-\epsilon )}\approx 4.5\times {10}^{2}/\lambda$; with λ = L/L_Edd = 5 × 10⁻⁴, a growth in mass from 2 × 10⁷ ${M}_{\odot }$ (for H ii galaxies) to 2 × 10⁸ ${M}_{\odot }$ (for LINERs) is a few Myr. Thus, such an evolutionary scenario is physically possible.

Further constraints on this evolutionary scenario could come from investigations of the amount and properties of stellar light in the very central regions, as probed by the HST apertures, in comparison with those measured at ∼100 pc scales. Unfortunately, high-quality spectroscopy at high (i.e., HST) spatial resolution for a sample that is as large and homogeneously distributed in spectral types as the one discussed in this study is not yet available for a rigorous investigation of these issues. The HST data presented here have low S/N that does not allow accurate fits of the underlying starlight and thus secure identification of their nuclear stellar properties. Also, the lack of [O iii]/Hβ measurements at high spatial resolution precludes a full account for a possible reclassification of the T into S nuclei in the small aperture.

Another way of testing the dominant power source in these low-luminosity AGNs can be an assessment of whether AGN photoionization can power the measured emission-line luminosities, for both the broad and the narrow Hα components. Following EHF, and using the data therein, we run an energy-budget test via a direct comparison of the Hα luminosity and count rate with the ionizing luminosity ${L}_{i}={L}_{1\;\mathrm{Ry}-100\;\mathrm{keV}}$ and the ionizing photon rate ${Q}_{i}={Q}_{1\;\mathrm{Ry}-100\;\mathrm{keV}}$, which are obtained by integrating the observed spectral energy distributions (SEDs) or by scaling a template quasar SED.

Based on photoionization models (Cloudy, v94.0; Ferland et al. 1998) computed by Lewis et al. (2003) for a wide range of ionization parameters, densities, and metallicities, EHF find that energy balance in a line-emitting nebula requires L_i > (18 ± 2) L_Hα/f_c, where f_c is the covering factor, or the fraction of the ionizing luminosity of the AGN that is absorbed by the line-emitting gas. Thus, with only a fraction 10% of ionizing photons being absorbed by the line-emitting gas, the minimum energy balance condition for AGN ionization is given by L_i/L_Hα > 180, for both narrow and broad emission features. At the same time, a minimum requirement for photon balance can be quite different for the two emission regions: for the narrow-line component Q_i > 2.2 Q_Hα, corresponding to Case B recombination (i.e., one Hα photon is emitted for every 2.2 recombinations); for the denser and much closer (to the center) BLR, the number of Hα photons that can be produced for each ionization can be at least 7–8 times higher than the standard Case B estimate, or Q_i ≳ 0.25 Q_Hα (i.e., potential contributions to the Hα emission via collisional excitation should be minimal for the NLR, but important for the broad component; e.g., Osterbrock 1989).

${Q}_{1\;\mathrm{Ry}-100\;\mathrm{keV}}$ and ${L}_{1\;\mathrm{Ry}-100\;\mathrm{keV}}$ measurements from EHF are available for 16 of our objects, and we present them in Figure 13 as a function of Q_Hα and L_Hα calculated separately for the narrow (gray) and broad (when detected; black) components, from both ground-based and space-based spectra. The sample includes two T, one S, five L1, and eight L2 nuclei, as classified from the ground. The ground-based narrow Hα measurements reflect merely a reiteration of the results of the EHF analysis and are generally consistent with a deficit of ionizing photons. Nevertheless, the small-aperture narrow nebular emission clearly moves away from the deficit, higher than the lines corresponding to the various energy-budget limits discussed above, showing that the (weak) AGN is definitely capable of powering the line emission within the compact region immediately surrounding it. As probably expected, the energy balance is satisfied for the broad Hα emission in all cases; all of the points are above the line corresponding to the minimum requirement for AGN photoionization.

Zoom In Zoom Out Reset image size

These results support with new and homogeneous data similar trends presented previously by EHF. For 10 of the objects included in our sample, EHF also illustrated a comparison of the ionizing photon rate with the rate required to power the Hα emission in the compact regions mapped by HST, and their conclusion was that the "little monster" is generally capable of powering the line emission close to it. It should be noted, however, that their measurements consisted of a mix of narrowband imaging and spectroscopy, and that the recorded Hα emission included both the narrow and the broad components. The separation of the broad and narrow emission is important, as the energy-balance mechanisms operating in these two regions are expected to be qualitatively different, and we are able to pursue this with our data for the first time.