FAR-UV EMISSION PROPERTIES OF FR1 RADIO GALAXIES

The nucleus of the FR 1 prototype M87 with its famous optical/radio/X-ray jet has been observed intensely over the last two decades with HST including previous UV and optical spectra obtained with the Faint Object Spectrograph (FOS) by Tsvetanov et al. (1998, 1999) and Sankrit et al. (1999). These FOS spectra of the nucleus cover approximately the same wavelength range as the COS/G130M data, including the Lyα emission in M87 at 1220.8 Å. In both of the Sankrit et al. (1999) observations and our own, the intrinsic flux and line shape of Lyα are difficult to determine due to the presence of the Galactic damped Lyα absorption (DLA) as well as probable absorption in M87. Regardless of the corrections that must be made due to the Galactic DLA and other absorbers in M87, Sankrit et al. (1999) find that the Lyα emission line possesses a ∼2000 km s⁻¹-wide core and a redshifted tail that extends at least another 800 km s⁻¹ to the red. However, the M87 absorption has unknown ${N}_{{\rm{H}}{\rm{I}}}$, which makes it difficult to remove in order to determine the intrinsic Lyα emission-line properties. Absorption has been detected in two narrow components (${cz}\approx 980$ and 1330 km s⁻¹) in Ca ii H&K and Na i D by Carter & Jenkins (1992) and Carter et al. (1997) and in several other ions (Na i D, Mg i, Mg ii, Ca ii H&K, Mn ii and Fe ii) in one broad component (${cz}=1134\pm 22$ km s⁻¹) by Tsvetanov et al. (1999) with FOS. The variable absorption velocities reported as well as the partial covering of the source suggested by the Ca ii and Na i absorption, which have 1:1 doublet ratios but low optical depths (Carter et al. 1997), are strong indicators that variable absorption exists in the nucleus of M87, perhaps due to cloud proper motions.

On what is likely a slightly larger scale, HST/FOS off-nuclear optical spectra and HST/Faint Object Camera (FOC) optical imaging discovered an ionized gas disk with Keplerian motion (Ford et al. 1994; Ford & Tsvetanov 1999) that connects to non-Keplerian blueshifted emission-line filaments primarily to the NW of the nucleus. Ford & Tsvetanov (1999) speculate that the observed nuclear absorption lines found by Tsvetanov et al. (1998) are kinematically related to this filamentary gas since these are blueshifted with respect to the nucleus as well. Macchetto et al. (1997) used the spectroscopic capabilities of the FOC on HST to determine the rotation curve of the [O ii]-emitting gas in the nuclear disk, which better determined the mass of the SMBH. Relevant to this study, no [O ii] forbidden-line emission was found within 007 (5 pc) of the nucleus. Similar long-slit, optical spectra obtained with the STIS were obtained by Walsh et al. (2013) to refine the SMBH mass estimate.

Due to all these potential complications and uncertainties, we elected to deconvolve the emission and absorption components in Lyα using an a priori approach. The Lyα profile is characterized by a strong emission peak at ${v}_{\mathrm{LSR}}=1260$ km s⁻¹ with a dereddened peak amplitude of 8 × 10⁻¹⁴ $\mathrm{erg}\,{\mathrm{cm}}^{-2}\,{{\rm{s}}}^{-1}\,{\mathring{\rm A} }^{-1}$. This is ∼6× brighter than observed with FOS by Sankrit et al. (1999) on 1997 January 23, and nearly $10\times$ brighter than on 1997 January 18 (although these new observations are through a larger aperture; 25 with COS compared to 026 with FOS; see Section 3.1.2). Weaker emission can be seen extending over the range $1207\lt \lambda \lt 1232$ Å and thus blended with the Milky Way DLA. The emission-line profiles in the three COS spectra shown here, especially the Lyα line in M87, are far too broad to be affected by any spectral smearing caused by the marginal spatial resolution in the COS aperture (see Section 3.4).

We model the Lyα emission profile as three emission components ("blue," "main," and "red"), a damped Galactic Lyα absorber at ${v}_{\mathrm{LSR}}\sim 0$, a weak, broad absorption component to accomodate possible absorption in M87, and a linear FUV continuum flux. The detailed flux model includes 15 fitting parameters (three for each Gaussian emission component, three for the M87 absorber, two each for the Galactic DLA and the nonthermal continuum of M87), each allowed to vary around a reasonable range. Despite this complexity, a remarkably robust fit is determined (${\bar{\chi }}^{2}=0.92$) even when fit parameters are allowed to vary over a broad range of plausible values. Best-fit quantities and $1\sigma$ fitting uncertainties are given in Table 2 and shown overplotted on the data in the left-hand panels of Figure 1.

Zoom In Zoom Out Reset image size

Table 2.  M87 Emission-line Fits

Quantity	Blue Comp.	Main Comp.	Red Comp.	Unit
H i Lyα 1215 Å
centroid	1211.93 ± 0.47	1220.79 ± 0.05	1228.00 ± 0.07	Å
	−923 ± 116	1263 ± 12	3040 ± 17	km s−1
FWHM	6.43 ± 0.68	5.04 ± 0.08	3.30 ± 0.15	Å
	1600 ± 170	1241 ± 18	813 ± 37	km s−1
${I}_{\mathrm{Ly}\alpha }$	22 ± 5	597 ± 16	14.0 ± 0.7	${10}^{-15}\,\mathrm{erg}\,{\mathrm{cm}}^{-2}\,{{\rm{s}}}^{-1}$
${L}_{\mathrm{Ly}\alpha }$	0.18 ± 0.15	5.06 ± 0.35a	0.12 ± 0.02	${10}^{39}\,\mathrm{erg}\,{{\rm{s}}}^{-1}$
continuum	⋯	1.20 ± 0.03	⋯	${10}^{-15}\,\mathrm{erg}\,{\mathrm{cm}}^{-2}\,{{\rm{s}}}^{-1}\,{\mathring{\rm A} }^{-1}$
N v 1238, 1242 Å
centroid	⋯	1226 ± 30	⋯	km s−1
FWHM	⋯	5.5 ± 0.8	⋯	Å
	⋯	1443 ± 100	⋯	km s−1
${I}_{{\rm{N}}{\rm{V}}1238}$	⋯	4.9 ± 0.9	⋯	${10}^{-15}\,\mathrm{erg}\,{\mathrm{cm}}^{-2}\,{{\rm{s}}}^{-1}$
${I}_{{\rm{N}}{\rm{V}}1242}$	⋯	3.7 ± 0.7	⋯	${10}^{-15}\,\mathrm{erg}\,{\mathrm{cm}}^{-2}\,{{\rm{s}}}^{-1}$
doublet ratio	⋯	1.3 ± 0.3	⋯
continuum	⋯	1.27 ± 0.02	⋯	${10}^{-15}\,\mathrm{erg}\,{\mathrm{cm}}^{-2}\,{{\rm{s}}}^{-1}\,{\mathring{\rm A} }^{-1}$
C ii 1334 Å
centroid	⋯	1500 ± 40	⋯	km s−1
FWHM	⋯	6.76 ± 0.88	⋯	Å
	⋯	1500 ± 100	⋯	km s−1
${I}_{{\rm{C}}{\rm{II}}}$	⋯	4.59 ± 0.59	⋯	${10}^{-15}\,\mathrm{erg}\,{\mathrm{cm}}^{-2}\,{{\rm{s}}}^{-1}$
continuum	⋯	1.20 ± 0.02	⋯	${10}^{-15}\,\mathrm{erg}\,{\mathrm{cm}}^{-2}\,{{\rm{s}}}^{-1}\,{\mathring{\rm A} }^{-1}$

Note.

^aAssuming D = 16.7 Mpc.

Download table as:  ASCIITypeset image

Blue and red wings are apparent in the emission profile and these are best fit with two additional Gaussian components with centroids at 1211.93 ± 0.47 Å and 1228.00 ± 0.07 Å, respectively, modified by the Galactic DLA. If Lyα emission, these correspond to ${v}_{\mathrm{LSR}}=-922$ km s⁻¹ and ${v}_{\mathrm{LSR}}=+3040$ km s⁻¹. We note that the blue and red wings of the profile cannot be fitted with a single very broad emission component at ${v}_{\mathrm{LSR}}\sim 1200$ km s⁻¹; two Gaussian profiles offset from the main, central component by $\sim 1000\mbox{--}2000$ km s⁻¹ are required. Thus, if the blue and red emission are both Lyα, they are approximately symmetrically placed around the main Lyα line. We interpret the red component as Lyα emission redshifted by ∼1800 km s⁻¹ with respect to the main Lyα emission in agreement with Sankrit et al. (1999). The maximum redshifted velocity observed in this broad component is ∼3000 km s⁻¹ relative to the main Lyα line.

The blue-side emission is possibly either Si iii 1206 Å emission at ${v}_{\mathrm{LSR}}=+1540\pm 140$ km s⁻¹ or Lyα emission blueshifted at −922 km s⁻¹ LSR or −2185 km s⁻¹ with respect to the main Lyα emission peak. The larger uncertainty in the velocity centroid of this line is due to the presence of the Galactic DLA, making this component's position and profile very dependent upon the exact assumed DLA column density (see below).

The Si iii interpretation is consistent with the velocity and line width of emission seen in C ii (see below), albeit with significant uncertainty. However, the velocity centroid of the proposed Si iii emission is almost 300 km s⁻¹ greater than the strong Lyα emission centroid, casting doubt on this identification. Further, if this emission is Si iii, it would be much stronger than what is seen in any other AGN. In a typical Seyfert like Mrk 817 (Winter et al. 2011), a Si iii line as luminous as the blue component in M87 would be visible as an asymmetry in the line profile of Lyα. This is not observed. Further, the FUV spectum of the prototypical narrow-line Seyfert 1 galaxy, I Zw 1, has only a reported marginal detection of Si iii emission at a level 10 times weaker than the feature we observe in M87. We conclude that this feature is unlikely to be Si iii.

If this emission is interpreted as Lyα, it has a velocity of −922 km s⁻¹ LSR and a blueshifted velocity of −2185 km s⁻¹ relative to the Lyα emission peak, making it reasonably symmetrical in both velocity offset and luminosity to the redshifted Lyα emission line, which is not affected by the Galactic DLA and has been reported previously.

A dip near the peak of the main component of the Lyα emission line is well fit as a very broad Lyα absorption line ($b\approx 250$ km s⁻¹) at ${v}_{\mathrm{LSR}}=1490$ km s⁻¹ ($v=+230$ km s⁻¹ relative to the main line centroid). While similar to the broad, low-ion detections seen by Tsvetanov et al. (1999) this absorption component is at a very different heliocentric velocity ($\approx +350$ km s⁻¹ relative to the Tsvetanov et al. 1999). This absorption lies on the red side of the Lyα peak compared to the blue-side absorption seen in the epoch 1997 FOS spectrum.

The Galactic DLA absorption we fit has a derived $\mathrm{log}{N}_{{\rm{H}}{\rm{I}}}({\mathrm{cm}}^{-2})=20.14$, quite close to but slightly less than the Galactic $\mathrm{log}{N}_{{\rm{H}}{\rm{I}}}({\mathrm{cm}}^{-2})=20.34\mbox{--}20.53$ value inferred by Sankrit et al. (1999). This value is also close to but slightly larger than the total N_H found by Di Matteo et al. (2003) from a continuum fit to the X-ray emission at the M87 nucleus. If we constrain the Galactic DLA ${N}_{{\rm{H}}{\rm{I}}}$ to be between the two Sankrit values, our best-fit solution is driven to the minimum value (i.e., $\mathrm{log}{N}_{{\rm{H}}{\rm{I}}}({\mathrm{cm}}^{-2})=20.34$) and the intrinsic main Lyα emission component would increase by a factor of 8% in flux. However, the blue emission is much more affected by the Galactic DLA increasing; its line flux increases by nearly a factor of two and its emission centroid shifts to 1214.04 ± 0.16 Å. At nearly 1.5 Å redward of its location in the unconstrained model (Table 2, Figure 1), it is even more unlikely to be Si iii 1206 Å and much more likely to be a blueshifted Lyα velocity component at ${v}_{\mathrm{lsr}}=-400$ km s⁻¹ or $v=-1670$ km s⁻¹ with respect to the main Lyα emission.

3.1.1. Metal Ion Emission Lines

In addition to the strong Lyα emission, there are a number of weak emission features in the M87 spectrum. As discussed above, the emission feature near a potential Si iii in M87 was fitted concurrently with Lyα but is interpreted as blushifted Lyα, not Si iii. Peaks at ∼1244 Å and ∼1340 Å are consistent with N v doublet and C ii emission, respectively. Best-fit parameters for these lines are given in Table 2.

The C ii 1334.5 Å emission line is modeled as a single Gaussian along with a ${v}_{\mathrm{LSR}}\approx 0$ Galactic C ii absorption line. The Galactic C ii line is fitted with the parameters ${v}_{\mathrm{LSR}}=+26\pm 11$ km s⁻¹, $b=80\pm 16$ km s⁻¹, and $\mathrm{log}\,N({\mathrm{cm}}^{-2})=15.0\pm 0.1$.

The N v doublet ($\lambda =1238.8$, 1242.8) was fitted assuming an identical velocity centroid and FWHM parameters for the two lines, but letting the relative line strength vary from the nominal 2:1, optically thin doublet ratio to an optically thick 1:1 ratio.

A low-contrast peak near 1400 Å is consistent with a blended complex of the Si iv $\lambda =1393.8$, 1402.7 doublet and O iv] $\lambda \approx 1403$ Å emission. However, given the low data quality near the edge of the COS/G130M detector and the complicated nature of the required model, no fit was attempted.

Anderson & Sunyaev (2016) recently reported the tentative detection of [Fe xxi] $\lambda =1354.08$ emission from M87. A careful examination of the data around $\lambda =1360$ Å reanalyzed using the techniques of Danforth et al. (2016) shows a weak ($\sim 5\sigma$) feature consistent with [Fe xxi] emission (Figure 2). We measure ${F}_{[\mathrm{Fe}{\rm{XXI}}]}=(1.8\pm 0.8)\times {10}^{-16}$ erg cm⁻² s⁻¹, ${v}_{\mathrm{lsr}}=1316\pm 29$ km s⁻¹, and $\mathrm{FWHM}=130\pm 70$ km s⁻¹. The line-flux value quoted here is consistent with the upper limit of (4 × 10⁻¹⁶ erg cm⁻² s⁻¹) quoted by Anderson & Sunyaev (2016). While their estimate of the line width of $\approx 290$ km s⁻¹ is much higher than what we obtain, it is based on a marginal detection. A value of $\mathrm{FWHM}\,=130\pm 70$ km s⁻¹ is both larger than a thermal width of ∼50 km s⁻¹ for this species and provides a first estimate of the turbulence of the hot gas close to the SMBH of $\approx 100$ km s⁻¹(Anderson & Sunyaev 2016). New exposures of comparable integration time to the values in Table 1 will make a definitive measurement of turbulence in the accretion flow. This will be reported at a later time.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

3.1.2. Line and Continuum Variability

NGC 4696 shows an asymmetric Lyα emission profile at 1227.6 Å (z = 0.0090, cz = 2710 km s⁻¹) with what appears to be a narrow absorption line superimposed at 1227.3 Å. There is a very low level of continuum flux ($F\lt {10}^{-16}$ $\mathrm{erg}\,{\mathrm{cm}}^{-2}\,{{\rm{s}}}^{-1}\,{\mathring{\rm A} }^{-1}$); the continuum level listed in Table 3 was obtained by averaging the flux in continuum pixels between $1218\lt \lambda \lt 1240\,\mathring{\rm A}$. The component structure of this emission line is unclear. We interpret the narrow absorption as Lyα at the same redshift as NGC 4696. There are no Galactic, z = 0 absorption lines expected at this wavelength and the path length to NGC 4696 is short enough that no intervening, intergalactic absorption lines are expected. The simplest model—a single Gaussian emission component and a single Voigt absorption profile—does not give satisfactory results. Adding a second, weaker emission component helps account for the red wing of the emission, but the peak is still poorly fit. The steep blue edge of the emission suggests the presence of additional absorbing gas slightly blueshifted from the systemic Lyα of NGC 4696, possibly outflowing from the nucleus or associated with the host galaxy itself. If a single, broad absorption line is included in the fit, the profile model fits the data well. However, we caution against overinterpretation of this fit, especially the inferred blue-side absorber. The emission may be intrinsically non-Gaussian or the absorption profile may be more complicated than a single, simple Gaussian component. Fit parameters are given in Table 3.

Table 3.  NGC 4696 Line Profile Fit Parameters

Quantity	Comp. 1	Comp. 2	Unit
Lyα Emission Components
Centroid	1227.59 ± 0.30	1230 ± 7	Å
	2940 ± 70	∼3500	km s−1
FWHM	3.2 ± 0.8	∼5	Å
	790 ± 200	∼1200	km s−1
${I}_{\mathrm{Ly}\alpha }$	14 ± 9	∼2	${10}^{-15}\,\mathrm{erg}\,{\mathrm{cm}}^{-2}\,{{\rm{s}}}^{-1}$
${L}_{\mathrm{Ly}\alpha }$	2.3 ± 1.6	∼0.3	${10}^{39}\,\mathrm{erg}\,{{\rm{s}}}^{-1}$ a
continuum	3 ± 1		${10}^{-17}\,\mathrm{erg}\,{\mathrm{cm}}^{-2}\,{{\rm{s}}}^{-1}\,{\mathring{\rm A} }^{-1}$
Lyα Absorption Components
Centroid	2893 ± 10	∼2280	km s−1
$\mathrm{log}\,{N}_{{\rm{H}}{\rm{I}}}$	14.10 ± 0.08	14.7 ± 0.6
b	87 ± 11	270 ± 140	km s−1

Note.

^aAssuming D = 37.6 Mpc.

Download table as:  ASCIITypeset image

Table 4.  Hydra A Lyα Emission Line Fits

Quantity	Comp. 1	Comp. 2	Unit
centroid	1281.85 ± 0.02	1281.80 ± 0.02	Å
	−130 ± 10	−140 ± 10	km s−1 a
FWHM	1.44 ± 0.09	2.61 ± 0.08	Å
	336 ± 21	610 ± 19	km s−1
${I}_{\mathrm{Ly}\alpha }$	16.3 ± 2.9	29.5 ± 2.9	${10}^{-15}\,\mathrm{erg}\,{\mathrm{cm}}^{-2}\,{{\rm{s}}}^{-1}$
${L}_{\mathrm{Ly}\alpha }$	112 ± 20	203 ± 20	${10}^{39}\,\mathrm{erg}\,{{\rm{s}}}^{-1}$ b
continuum	0.58 ± 0.02	⋯	${10}^{-15}\,\mathrm{erg}\,{\mathrm{cm}}^{-2}\,{{\rm{s}}}^{-1}\,{\mathring{\rm A} }^{-1}$
slope	11 ± 2	⋯	${10}^{-18}\,\mathrm{erg}\,{\mathrm{cm}}^{-2}\,{{\rm{s}}}^{-1}\,{\mathring{\rm A} }^{-2}$

Notes.

^aWith respect to a systemic redshift of z = 0.054878. ^bAssuming D = 240 Mpc.

Download table as:  ASCIITypeset image

Anderson & Sunyaev (2016) found a very low-significance emission feature associated with NGC 4696 which they interpreted as [Fe xxi]. We confirm this possible detection in our more sophisticated reduction of the data (Figure 4) at a significance level of $\sim 2\sigma$. The possible emission line is fitted by a Gaussian with $\mathrm{FWHM}=135\pm 83$ km s⁻¹ and ${v}_{\mathrm{lsr}}=2931\pm 38$ km s⁻¹. Our fitted flux of ${F}_{\mathrm{Fe}{\rm{XXI}}}=(8\pm 5)\times {10}^{-17}$ erg cm⁻² s⁻¹ is consistent with the Anderson & Sunyaev 90% upper limit of ${F}_{\mathrm{Fe}{\rm{XXI}}}\leqslant 2.2\times {10}^{-16}$ erg cm⁻² s⁻¹.

Zoom In Zoom Out Reset image size

The FUV spectrum of Hydra A shows a relatively narrow Lyα emission peak at close to the systemic velocity. It is well fit with a broad-plus-narrow pair of Gaussian emission components and a first-order linear continuum (see Figure 5 and Table 4). Some structure in the residual suggests that there may be additional components, but nothing obvious is seen.

Zoom In Zoom Out Reset image size

Recent HST images of Hydra A in the ultraviolet by Tremblay et al. (2015) show that the nucleus is obscured by a dust disk viewed edge on. Thus one would expect the nuclear FUV continuum and line emission to be suppressed or absent entirely. It is therefore surprising that an FUV continuum is present, the Lyα line emission is many times more luminous than either M87 or NGC 4696 (Tables 2, 3), and that no narrow absorption line is seen superimposed on the Lyα emission as is seen in the other two cases (Figures 1, 3). We conclude that the nuclear region must therefore not be obscured by the dust disk in Hydra A.

In addition to the usual Galactic, z = 0 absorption lines, the Hydra A spectrum shows hints of C ii 1334.5, O i+Si ii 1304, and Si iii 1206.5 absorption at the redshift of Hydra A. It is possible that this absorption arises in the dust disk bisecting the galaxy (Tremblay et al. 2015). However, no H i absorption is seen at the systemic redshift, so modeling of these absorption lines would require data of much higher quality.

No metal ion emission is seen including [Fe xxi].

Although HST/COS was not designed to provide high angular resolution imaging spectroscopy, it has been demonstrated that limited spatial information can be obtained from an analysis of a target's cross-dispersion profile (France et al. 2011). The angular resolution of COS in the G130M mode is $0\buildrel{\prime\prime}\over{.} 8\mbox{--}1\buildrel{\prime\prime}\over{.} 1$. In order to evaluate the spatial extent of the Lyα emission-line region at $\approx 1^{\prime\prime}$ resolution, we compared the angular profiles of the three FR 1 galaxies with (1) a known extended emission source and (2) a known point source. HST orbits in a diffuse cloud of neutral and ionized atoms; the "airglow" from the atomic recombination and resonant scattering of solar photons in this cloud produces a uniform background of hydrogen and oxygen emission that fills the COS aperture and approximates a filled-slit observation of an astronomical target. For point source comparison, we downloaded spectra of a well-studied COS calibration target, WD 0308−565. We analyzed observations from the longest central wavelength settings available from each target (cenwave $=1318$ and 1327) as these grating settings have the smallest intrinsic cross-dispersion heights (Roman-Duval et al. 2013).

Figure 6 shows the cross-dispersion profiles of the Lyα emission lines of the three FR 1 galaxies (top three panels), the FUV continuum of WD 0308 (bottom panel), and the Lyα airglow spectra measured simultaneously. The galactic/stellar profiles were extracted over a 400 dispersion-direction pixel region ($\approx 4$ Å) centered on the source Lyα emission line (offset into the continuum at 1247.8 Å for WD 0308). The geocoronal Lyα airglow profile was fitted with a Gaussian (red) and shown offset from the central cross-dispersion profile of the targets by −50 pixels. Gaussian fits to the galaxy/star cross-dispersion profiles are shown in green. The Gaussian FWHMs (in pixels, recall 1 pixel $\approx 0\buildrel{\prime\prime}\over{.} 11$ for COS/G130M) are shown in the legends. The y-axis label shows the total counts per cross-dispersion extraction region in each of the targets, with the geocoronal profile scaled to the peak of the target flux. For the FR 1 galaxies, the geocoronal Lyα line height is always between 21 and 22 pixels FWHM ($2\buildrel{\prime\prime}\over{.} 3\mbox{--}2\buildrel{\prime\prime}\over{.} 4$), consistent with the expected instrumental profile for a filled aperture and the non-uniform primary science aperture response function of COS. The WD 0308 continuum is bright enough that it has comparable flux to the airglow line, thus its geocoronal line width must be fitted with a pair of Gaussians. The FWHM of the white dwarf cross-dispersion profile is 8.4 ± 0.2 pixels (092), consistent with the expected cross-dispersion profile for a point source. The broader component has FWHM $=\,21\pm 1$ pixels (23), consistent with a filled aperture (bottom panel, Figure 6).

Zoom In Zoom Out Reset image size

Contrasting the filled-aperture geocoronal observations and the point source white dwarf observations with the cross-dispersion heights of the FR 1 galaxies, we see that all three galaxies are in an intermediate category: they are clearly extended emission sources, but do not fully fill the COS aperture. For M87, we find a cross-dispersion line height of 11.5 ± 0.1 pixels (127), with 18.7 ± 0.3 pixels (206) and 15.7 ± 0.5 pixels (173) for Hydra A and NGC 4696, respectively. We conclude that the Lyα-emitting regions for these FR 1 galaxies are indeed extended, although they do not fill the aperture.

A similar analysis of a region of the FUV continuum away from the galactic Lyα emission signal shows an interesting result: the FUV continua in M87 and Hydra A are also extended (that is, not consistent with a point-like AGN alone) and the FUV continuum in M87 shows an angular extent 20%–40% larger than the Lyα emission region (FWHM $=\,14\mbox{--}16$ pixels over a range of continuum wavelengths and extraction region sizes). NGC 4696 does not have sufficient FUV continuum flux for a cross-dispersion profile to be reliably measured.

Furthermore, in M87 the Lyα core emission, blue wing emission, and red wing emission are centered at slightly different locations in the aperture (see Figure 7). Although slight, these offsets are consistent with Lyα-emitting material outflowing along the direction of the jet (the blue wing is in the direction of the jet; the red wing is in the direction of the counter jet).

Zoom In Zoom Out Reset image size

The inferences on the spatial extent and location of the line and continuum emission in M87 can be tested by obtaining long-slit FUV spectra of the nucleus at position angles along the jet and perpendicular to the jet using HST/STIS. The results of these observations (approved for HST cycle 24) will be reported in a future publication.

Several BL Lac objects in the literature show weak Lyα emission: Mrk 421, Mrk 501, PKS 2005−489 (Paper 1) and, more recently, H2356−309 (Fang et al. 2014). To these we add new detections of weak Lyα emission in PMN J1103−2329 and 1ES 1028+511. As with Paper 1, we fit power-law continua to the spectra and model the Lyα emission profiles as single, Gaussian components.

Neither PMN J1103−2329 nor 1ES 1028+511 show high-significance Lyα emission features like the FR 1s, nor even as large an equivalent width as the previously published BL Lac objects with Lyα emission (Paper 1). Nevertheless, a Gaussian fit in the expected location of the systemic Lyα shows weak, broad emission with significance levels of $\sim 6\sigma$ and $\sim 3\sigma$ for PMN J1103−2329 and 1ES 1028+511, respectively (Figure 8).

Zoom In Zoom Out Reset image size

Of the seven BL Lac objects with well-known redshifts and high S/N FUV spectra, six show intrinsic Lyα emission. Weak, relatively narrow Lyα emission appears to be a generic feature of both BL Lac objects and their parent population FR 1 radio galaxies.

While all three FR 1 objects studied here show intriguing complexity, more constraints on source size and geometry are available for M87 due to its proximity to us and its previous observations in the FUV by HST. In particular, M87 is the only one of these objects where the jet viewing angle ($\approx 15^\circ$) is well constrained (Perlman et al. 2011; Meyer et al. 2013). Variability information is particularly important and puzzling. The Lyα-emitting gas cannot possibly be as close as 10¹⁶ cm (0.003 pc) to the SMBH as claimed by Sankrit et al. (1999) from their variability constraints. The Schwarzschild radius for the SMBH in M87 is ∼10¹⁵ cm. Even at 10¹⁷ cm ($100\,{R}_{S}$) the Keplerian speed is $\gt {10}^{4}$ km s⁻¹, so the lines would be an order of magnitude broader than we observe if the Lyα-emitting gas were even this close to the SMBH. A more likely location for the Lyα clouds is $\sim {10}^{4}\,{R}_{S}$ (∼10¹⁹ cm; ∼3 pc), the inner reaches of the narrow-line region and of the observed gaseous disk. So, there is little chance that the FOS-observed variability could be due to a very small source size. The larger distance of ∼3 pc from the SMBH is consistent with the observed angular extent of the Lyα components in the COS aperture as well as the observed line width, but leaves the large variability amplitude over short time periods as a puzzle. It seems most likely that the reported five-day Lyα variability is spurious due to a mis-placement of the FOS aperture as suggested by Sankrit et al. (1999). Currently there is no unambiguous indication for emission-line gas closer than 3 pc, at much larger radii than where the BLR of Seyferts exists.

The suppression of BLR emission in BL Lacs was first explored by Guilbert et al. (1983) who pointed out that the steep spectrum ionizing continua seen in BL Lacs will inhibit the creation of cool, 10⁴ K clouds that are typical of Seyferts and QSOs. For this mechanism to be operable in FR 1s, this steep ionizing continuum should be illuminating much of the potential BLR of these AGNs. Indeed, the steep X-ray spectra of BL Lacs must be much less beamed (half illumination angle ∼1 radian) than the radio and optical continua based on their observed properties and source counts (Urry & Padovani 1995) when selection is made in the X-ray versus radio or optical bands. Therefore, the presence of broad-line-emitting gas is not expected in this class of AGNs, as we observe.

Unlike the narrow core emission in M87, the red and blue wings to the Lyα emission are much broader and appear to be offset spatially from the main Lyα emission along the jet axis (see Section 3.4). The red wing emission also varies over the same observing epochs and is very likely intrinsic variability since there is no observed absorption in M87 at these wavelengths. Given their oppositely trending spatial extents, the red and blue emission wings are likely to arise in gas illuminated or shocked by the jet (blue wing) and counter jet (red wing). Spatially resolved STIS spectra can confirm this suggestion.

In the three FR 1s observed by HST/COS, the Lyα luminosities are comparable to or somewhat less than what is expected in "Case B" recombination theory assuming unity covering factor. In M87 previous epochs saw lower Lyα luminosities while the continuum luminosities remained the same. Therefore, at the highest Lyα luminosities observed, an ionization-bounded scenario is possible in these AGNs but, even then, the amount of gas must be close to the amount needed to absorb all the available ionizing photons given the OPF $\approx 1$ values found here. If the line-emitting clouds are in a density-bounded regime then, as illustrated in Paper 1 using the Lyα line luminosities seen in BL Lac objects, there is very little warm gas in this region, ${10}^{-4}\mbox{--}{10}^{-5}\,{M}_{\odot }$. This amount is inferred by assuming that there is one hydrogen atom for every Lyα photon emitted. A reasonable energy conversion rate of this mass (e.g., ∼1%) finds a very low estimated accretion rate, which is consistent with earlier estimates using a variety of methods (Kuo et al. 2014; Russell et al. 2015; Di Matteo et al. 2016).

It is also possible that the observed Lyα gas is outflowing rather than infalling, which decreases the estimated accretion rate even further. This possibility is consistent with the limited red/blue wing spatial information from the COS aperture (Section 3.4). It is important to verify this assertion using the full HST spatial resolution with STIS.

There could be a significant amount of much hotter, "coronal" gas in the line-emitting region surrounding the cooler, Lyα-emitting clouds. Using the results from Anderson & Sunyaev (2016) on the detection of [Fe xxi], which we confirm here, an estimate of the particle density of $0.2\,{\mathrm{cm}}^{-3}$ can be derived for a temperature of T ≈ 10⁷ K based on the presence of the [Fe xxi] feature. Assuming this hot coronal gas fills the region around the SMBH out to ∼3 pc, this phase contains $\sim 0.3\,{M}_{\odot }$ of gas. These values are consistent with the recent re-analysis of Chandra X-ray images of M87 conducted by Russell et al. (2015). Although we have no good, direct measurement of the density of the Lyα-emitting clouds, if the coronal gas is in pressure equilibrium with these clouds, their densities are a few $\times {10}^{-4}\,{\mathrm{cm}}^{-3}$. It is reasonable to suspect that the Lyα-emitting cloud ensemble is gas that has condensed out of this hotter phase in either an infalling or outflowing wind.

If the ionizing continuum radiation extrapolated from the HST/COS spectra of M87 and NGC 4696 is typical of FR 1 radio galaxies, then the contribution of this class of AGNs to the extragalactic UV background (UVB) in the local universe may be substantial since this AGN class is so numerous (∼3% of all bright cluster ellipticals have $\mathrm{log}P({\rm{W}}\,{\mathrm{Hz}}^{-1})\geqslant 22.0;$ Ledlow & Owen 1996).

The beaming geometry in FR 1s must also be taken into account in estimating the amount of UVB contributed by these objects. For example, there is some evidence from BL Lac population studies (e.g., Perlman & Stocke 1993; Urry & Padovani 1995; Nieppola et al. 2006; Padovani et al. 2007; Meyer et al. 2011; Giommi et al. 2013) that the amount of Doppler boosting (i.e., the relativistic Γ) varies with the frequency of the emitted radiation. This is also known as the re-collimating or accelerating jet model (Ghisellini & Maraschi 1989; Urry & Padovani 1995; Böttcher & Dermer 2002) for blazars. X-ray-selected BL Lacs (most of which are now termed High-energy peaked BL Lacs or HBLs) often lack the optical nonthermal BL Lac spectrum, instead showing either a pure stellar spectrum or a mixture of an old stellar population and a nonthermal continuum. This property and their large space density means that the X-ray beam must be much broader than the optical- and radio-emitting beams, presumably coming from particles lower in Γ. This hypothesis is supported by the optical polarization studies conducted by Jannuzi et al. (1994), which showed that HBLs possess only modestly variable polarization amounts and relatively constant position angles of polarization as if we are viewing HBL optical beams from well off their axes. A gradient of broader to narrower emission beams (lower to higher Γs) with increasing wavelength predicts that the UV continuum beam is somewhat narrower than the $\sim 25^\circ \mbox{--}30$° for the X-ray emission (e.g., Perlman & Stocke 1993). Due to this somewhat rudimentary understanding of the beaming geometry, here we adopt a simplistic model with a beamed ionizing flux emanating from two cones with half angles of 20° and an unbeamed flux given by our FR 1 observations emitted over the rest of the sphere.

Since the extrapolated Lyman continuum flux of Hydra A is almost two orders of magnitude more luminous than that of M87, we assume that some significant beaming is present in our direction toward this source and we do not use its ionizing flux as typical of unbeamed FR 1s. Using just M87 and NGC 4696 as typical for the unbeamed FR 1 population, the ionizing continuum luminosity in these objects is ∼10⁴⁰ erg s⁻¹. If only a small fraction of luminous early-type galaxies are FR 1s, then the ionizing radiation contributed by this class is also quite small. But the evidence says otherwise; for example, Ledlow & Owen (1996) find that 3% of all bright cluster ellipticals have radio luminosities with $\mathrm{log}{P}_{1.4\mathrm{GHz}}({\rm{W}}\,{\mathrm{Hz}}^{-1})\geqslant 22$ while Lin & Mohr (2007) find a larger fraction of 5% with $\mathrm{log}{P}_{1.4\mathrm{GHz}}({\rm{W}}\,{\mathrm{Hz}}^{-1})\geqslant 23$ in X-ray-emitting clusters. At X-ray wavelengths, Martini et al. (2006) found that ∼2% of all bright cluster ellipticals emit X-rays at ${P}_{x}\geqslant {10}^{42}$ erg s⁻¹. In a combined X-ray/radio study of cluster AGNs, Hart et al. (2009) found 6% of bright cluster ellipticals are radio sources at $\mathrm{log}{P}_{1.4\mathrm{GHz}}({\rm{W}}\,{\mathrm{Hz}}^{-1})\geqslant 23.5$ and 1% are X-ray sources at ${P}_{x}\geqslant {10}^{42}$ erg s⁻¹. Extrapolation of these results to lower radio and X-ray power levels is consistent with all bright cluster ellipticals being radio sources at $\mathrm{log}{P}_{1.4\mathrm{GHz}}({\rm{W}}\,{\mathrm{Hz}}^{-1})\geqslant 21.4$ and X-ray sources at ${P}_{x}\geqslant {10}^{40}$ erg s⁻¹. Indeed, a deep Chandra observation of the central region of the Perseus Cluster (Santra et al. 2007) has detected weak (10^40–41 erg s⁻¹) nonthermal X-ray point sources from all 13 early-type galaxies brighter than $0.2\,{L}^{* }$ in this region. In the Santra et al. (2007) X-ray imaging, there is no obvious trend in X-ray luminosity with optical luminosity, suggesting that the AGN emission in these X-ray point sources is similar in luminosity for all bright cluster ellipticals. Together, all these studies suggest that virtually all bright early-type cluster galaxies are weak FR 1s and could be emitting Lyman continuum radiation at levels similar to M87. Despite consistent results from the several studies cited above, this important assumption must be confirmed in order to validate our conclusions concerning the FR 1 contribution to the UVB.

The CfA galaxy luminosity function of elliptical galaxies from Marzke et al. (1994) finds that the number density of $L\gt 0.2L*$ ellipticals is 4 × 10⁻³ Mpc⁻³ (we ignore S0 galaxies in this estimate since they appear not to harbor FR 1s in general). Following the Santra et al. (2007) result as well as the extrapolations from higher power levels made in the other studies quoted above, we assume that all bright ellipticals emit Lyman continuum radiation at a level similar to M87 and NGC 4696. From this assumption we derive an estimate of the unbeamed UV emission from FR 1s of ∼4 × 10³⁷ erg s⁻¹ Mpc⁻³. This amount is ∼7% of the total ionizing background at $z\approx 0$ (assuming the total UVB at $z\sim 0$ from Haardt & Madau 2012).

If the UV luminosity of 10⁴⁰ erg s⁻¹ is the unbeamed continuum, then beaming of these sources can add significantly to this total, perhaps as much as an additional 7% at $z\approx 0$ if the standard beaming model of Urry & Padovani (1995) is assumed (half-opening angle of $\approx 20$°; see above). This standard model assumption is equivalent to assuming that 1% of all FR 1s are seen as beamed sources at any one location and that their mean observed Doppler boosting factor is 20 in the UV. There are a very few (roughly one in 10⁶) FR 1s whose beamed nonthermal luminosities are at the luminosities of ${L}_{x}\gt {10}^{44}$ erg s⁻¹ (Morris et al. 1991), which is typical of HBL-type BL Lac objects, but these contribute negligibly to the UVB total.

Overall, the FR 1 AGN class could contribute to the ionizing background an amount of up to 1/7 (10%–15% of the total UVB) of the amount contributed by QSOs and Seyferts locally. If the above assumptions about beaming and the FR 1 source population are correct, luminous ellipticals contribute non-negligibly to the amount of ionizing background inferred from source population studies at low-z (Haardt & Madau 2001, 2012). A sizable contribution from FR 1s could account for the discrepancy between the most recent UVB model of Haardt & Madau (2012; which has minimal contribution from normal, star-forming galaxies) and the observed absorber statistics in the Lyα forest (Kollmeier et al. 2014; Shull et al. 2015). However, since most bright ellipticals are found in rich clusters, this putative additional UV radiation could be very non-uniform, clumped around regions rich in early-type galaxies. A more detailed look at FR 1 radio galaxy beaming geometries and number density statistics is required to assess these possibilities in more detail.

Support for the idea that a large contribution to the UVB could be made by FR 1s/BL Lacs comes from the all-sky survey of FUV sources made by the Extreme Ultraviolet Explorer. This continuum survey conducted at a wavelength closer to the Lyman limit than any other survey detected more BL Lacs and narrow-line Seyferts than any other AGN class (Marshall et al. 1995). These two AGN classes share an unusual SED with a very steep soft X-ray spectrum as well as the absence of broad permitted lines in their optical and UV spectra. The three FR1 s studied here share the absence of broad emission lines with these more luminous, EUV-bright AGN classes.