The Galaxy Zoo survey for giant AGN-ionized clouds: past and present black hole accretion events

Abstract

Some active galactic nuclei (AGN) are surrounded by extended emission-line regions (EELRs), which trace both the illumination pattern of escaping radiation and its history over the light travel time from the AGN to the gas. From a new set of such EELRs, we present evidence that the AGN in many Seyfert galaxies undergo luminous episodes 0.2–2  years in duration. Motivated by the discovery of the spectacular nebula known as Hanny’s Voorwerp, ionized by a powerful AGN which has apparently faded dramatically within  years, Galaxy Zoo volunteers have carried out both targeted and serendipitous searches for similar emission-line clouds around low-redshift galaxies. We present the resulting list of candidates and describe spectroscopy identifying 19 galaxies with AGN-ionized regions at projected radii  kpc. This search recovered known EELRs (such as Mrk 78, Mrk 266 and NGC 5252) and identified additional previously unknown cases, one with detected emission to  kpc. One new Sy 2 was identified. At least 14/19 are in interacting or merging systems, suggesting that tidal tails are a prime source of distant gas out of the galaxy plane to be ionized by an AGN. We see a mix of one- and two-sided structures, with observed cone angles from 23 to 112⁠. We consider the energy balance in the ionized clouds, with lower and upper bounds on ionizing luminosity from recombination and ionization-parameter arguments, and estimate the luminosity of the core from the far-infrared data. The implied ratio of ionizing radiation seen by the clouds to that emitted by the nucleus, on the assumption of a non-variable nuclear source, ranges from 0.02 to ⁠; 7/19 exceed unity. Small values fit well with a heavily obscured AGN in which only a small fraction of the ionizing output escapes to be traced by surrounding gas. However, large values may require that the AGN has faded over tens of thousands of years, giving us several examples of systems in which such dramatic long-period variation has occurred; this is the only current technique for addressing these time-scales in AGN history. The relative numbers of faded and non-faded objects we infer, and the projected extents of the ionized regions, give our estimate (0.2–2  years) for the length of individual bright phases.

galaxies: active, galaxies: ISM, galaxies: Seyfert

1 INTRODUCTION

The compact sizes of the central engines of active galactic nuclei (AGN) have long driven study of their distant surroundings for clues to their geometry and interaction with the surrounding galaxy. Observations of gas seen many kiloparsecs from the AGN itself have proven fruitful in offering views of the core from different angles and implicitly at different times.

Narrow-band images revealed extended emission-line regions (EELRs) around some luminous AGN, particularly radio-loud quasi-stellar objects (QSOs) as well as radio galaxies, as reviewed by Stockton, Fu & Canalizo (2006). Similar structures in lower luminosity Seyfert galaxies often appear as single or double triangles in projection (Unger et al. 1987; Tadhunter & Tsvetanov 1989), generally interpreted as ionization cones. When small-scale radio jets are present, they lie within the ionization cones. However, in many cases, the gas must be ionized by radiation from the nucleus rather than direct interaction with a jet or outflow, as seen from narrow line widths and (particularly diagnostic) modest electron temperatures, both of which would be much larger in the presence of shocks fast enough to match the observed ionization levels. This is particularly true for very large EELRs, where interaction with the radio jet or an origin in outflows alone becomes less and less likely. In fact, the best defined ionization cones are seen in radio-quiet objects (Wilson 1996).

This is one line of evidence linking large-scale structures to the small-scale obscuring regions (‘tori’) implied by other arguments for a unification scheme (Antonucci 1993), in which Seyferts of types 1 and 2 are part of a single parent population, appearing different based on how our line of sight passes this torus. The emission-line structures can be large and well resolved, offering a way to measure the opening angle over which ionizing radiation escapes. Some previous studies have also noted that these emission-line clouds provide a view to the immediate past of the AGN, via light travel time to the cloud and then towards us (Dadina et al. 2010).

Using extended emission-line clouds as probes of AGN history came of age with the discovery of Hanny’s Voorwerp, a high-ionization region extending 45 kpc in projection from the low-ionization nuclear emission region (LINER) galaxy IC 2497 at  (Lintott et al. 2009). Line widths and electron temperature indicate that the gas is photoionized rather than shock excited, while a combination of ionization-parameter and recombination arguments bound the required nuclear ionizing luminosity to be  erg s⁠. However, X-ray spectroscopy shows the nucleus of IC 2497 to be only modestly absorbed, with ionizing luminosity only  erg s (Schawinski et al. 2010b). It is difficult to avoid the conclusion that the nucleus of IC 2497 was in fact a QSO (the nearest known luminous QSO) until roughly  years before our current view, and has faded dramatically in the interim; radio and Hubble Space Telescope (HST) observations offer hints that some of its energy output may have switched to kinetic forms over this time span (Josza et al. 2009; Rampadarath et al. 2010; Schawinski et al. 2010b; Keel et al., in preparation). The unlikeliness of the nearest QSO showing highly unusual behaviour suggests that such variations maybe common among AGN, prompting us to re-examine the incidence and properties of extended ionized clouds around nearby AGN. Such an examination should not be confined to catalogued AGN, since the most interesting objects – those which have faded dramatically – may no longer appear as spectroscopically classified AGN.

Hanny’s Voorwerp was first noted by Dutch teacher Hanny van Arkel in the course of the Galaxy Zoo project (Lintott et al. 2008), on the basis of its unusual structure and colour. In view of the interest of similar ionized clouds for study of both the history and obscuration of AGN, participants in the Galaxy Zoo project carried out a wide search for such clouds using data from the Sloan Digital Sky Survey (SDSS). They examined both known AGN hosts and galaxies not known to have AGN, using the distinctive colour of highly ionized regions across the SDSS  filters as a first selection criterion. We present the results of further analysis of the SDSS images, narrow-band imaging and spectroscopy, yielding a list of 19 galaxies with AGN-photoionized clouds detected to beyond 10 kpc from the nuclei (many of which are newly identified). We consider constraints on changes in ionizing luminosity for these and identify several as the most likely candidates for the kind of long-term fading seen in IC 2497 and Hanny’s Voorwerp.

2 SEARCHES FOR EMISSION-LINE CLOUDS

The Galaxy Zoo search for giant AGN-ionized clouds combined both targeted and serendipitous approaches, to combine a complete examination of known AGN hosts with the possibility of finding ionized clouds around AGN which are yet unknown or in fact optically unseen. In the targeted search, we formed a sample of potential AGN at ⁠. This combined all galaxies whose SDSS pipeline emission-line ratios put them in either the AGN or composite regions of the Baldwin–Phillips–Terlevich (BPT) diagram (Baldwin, Phillips & Terlevich 1981, as revised by Kewley et al. 2001 and Kauffmann et al. 2003) using [O iii]/H and [N ii]/H⁠, and all additional objects listed in the Veron-Cetty & Veron (2010) catalogue at  falling within the SDSS data release 7 (DR7) area. This addition accounted for AGN with no SDSS nuclear spectrum, either because they are relatively bright or, more often, because fibre collisions or sampling rules prevented their selection for spectra, and type 1 AGN where the pipeline spectroscopic classification is less reliable than for narrow-line objects. The merged AGN sample, designed to err on the side of inclusion in borderline cases, included 18 116 objects. With a web interface designed by RP, 199 participants examined all of these within a six-week period in 2009, marking each as certain, possible or lacking an extended emission region. These emission regions have distinctive signatures in both morphology and colour from the SDSS data. They do not follow the usual spiral or annular distributions of star formation in disc galaxies. Such regions show unusual colours in the SDSS composite images, which map  bands to blue, green and red (Lupton et al. 2004). Hence strong [O iii] at low redshift is rendered as a pure blue, as in the discovery of Hanny’s Voorwerp. A combination of strong [O iii] and significant H+[N ii] appears purple; beyond about ⁠, [O iii] falls in the gap between  and  filters, so our search technique loses utility until [O iii] is well within the  band, when the galaxies have much smaller angular sizes. This subproject was known as the ‘voorwerpje hunt’, using the Dutch diminutive form of Voorwerp.

Each galaxy was examined by at least 10 participants; 199 Zoo volunteers participated in this programme, seven of whom examined the entire sample. The final average number of votes was 11.2 per object. After this screening process, a straightforward ranking was by relative numbers of ‘yes’ (weight = 1), ‘maybe’ (weight = 0.5) and ‘no’ (weighted zero) votes.

The most interesting results of such a search would be galaxies with prominent AGN-ionized clouds in which we do not see the AGN, either because of strong obscuration or dramatic variability during the light travel time from the nucleus to the clouds. These would not be found by targeting known AGN, and neither would clouds around AGN which do not have catalogued spectral information. Accordingly, we also posted a request on the Galaxy Zoo discussion forum, with examples of confirmed AGN clouds and various kinds of similar-appearing image artefacts. Participants were invited to post possible cases from among the galaxies they saw in the ordinary course of the Galaxy Zoo classification programmes (Lintott et al. 2008), and some active users reposted examples from other discussion threads. The resulting follow-ups1 provided an additional sample for investigation; W. C. Keel also checked all the threads with image discussion for more such objects, early examples of which instigated this search on the first place. To reduce the number of false positives caused by extended star-forming regions or starburst winds, objects were removed from consideration if an SDSS spectrum shows emission lines characteristic of a starburst. Remaining candidates were examined first on the SDSS composite images for appropriate colour and geometry, and the most promising ones were carried forward for further analysis.

Both targeted and serendipitous lists overlap for many objects with bright emission-line structures and recover such well-studied cases from the literature as Mrk 266, NGC 5252 and Mrk 78; we observed these so as to have a consistent set of spectra for comparison. The entire list of candidates is given in Table 1. In the Survey column, S or T denotes whether the object was found in the serendipitous survey, the targeted survey of known AGN or both. The type of nuclear optical spectrum is listed as Sy 1/1.5/1.9/2, LINER, SB for starburst or non-AGN for an ordinary stellar population. The final column indicates which Galaxy Zoo participant (by user name) first posted objects in the serendipitous survey.

Table 1

Candidate AGN with extended emission-line clouds.

Coordinate name	SDSS ObjID		Nucleus	Name/note	Search	Posted by
SDSS J005607.66+254804.7	587740589487030353	0.1530	Sy 1	Purple haze	S	ElisabethB
SDSS J003507.44+004502.1	587731187281494175	0.1205	LINER	Blue arc	S	scott L
SDSS J004527.06+004237.6	587731187282608299	0.1096	Sy 1.5	–	S	davidjamesjones
SDSS J013037.75+131251.9	587724197207212176	0.0721	Sy 2	CGCG436-065	T	–
SDSS J014238.47+000514.7	588015509280587804	0.1459	Sy 1	–	S	Tsering
SDSS J014644.82004043.1	588015508207304746	0.0827	Sy 1	–	S	davidjamesjones
SDSS J030526.96+005144.9	588015510363373793	0.1181	Sy 1	–	S	Mukund Vedapudi
SDSS J030639.58+000343.2	588015509289762862	0.1074	Sy 1	–	S	mitch
SDSS J033013.26053235.9	587724242842026028	0.0131	Sy 1	NGC 1346	S	Half65
SDSS J040548.78061925.7	587727178476093634	0.0556	Sy 1	–	S	echo-lily-mai
SDSS J074241.70+651037.8	758878270293868614	0.0371	Sy 2	Mrk 78	T	–
SDSS J075910.44+115156.7	588023046395527377	0.0503	Sy 1	–	S	silverhaze
SDSS J080452.73+212050.2	588016878287650850	0.1242	Sy 1	Purple haze	S	davidjamesjones
SDSS J082034.78+153111.3	587741532229337219	0.1435	Sy 1	–	S	Half65
SDSS J082342.37+482754.4	587725470667833620	0.0935	Sy 2	–	S	spiralmania
SDSS J082642.63+111555.5	587745244691628370	0.0884	trans	–	T	–
SDSS J083525.51+104925.7	587744873714679862	0.1172	Sy 1	–	S	spiralmania
SDSS J083818.43+333441.3	587732470387703859	0.0621	LINER	KUG	T	–
SDSS J084002.36+294902.6	587735240637284507	0.0648	Sy 2	4C 29.30	T	–
SDSS J084344.98+354942.0	587732484342415393	0.0539	Sy 2	–	S T	laihro
SDSS J084518.51+142034.1	587742062124269645	0.0606	Sy 1	Purple haze	S	mitch
SDSS J084810.10+351534.0	587732470388818038	0.0573	Sy 2	KUG0845+354	T	–
SDSS J084809.59+351530.3	587732470388818042	0.0567	Sy 2	KUG0845+354	T	–
SDSS J084917.31+531755.7	587725470670585957	0.1112	Sy 1	Purple haze	S	Tsering
SDSS J085625.93+021310.5	587727944563884115	0.1251	RG	–	S	Bruno
SDSS J085729.84+064210.6	587734691423453446	–	–	–	S	lovethetropics
SDSS J085837.52+182221.5	587741708866289722	0.0588	Sy 2	–	T	–
SDSS J085813.75+385631.8	587732053243855000	0.0884	Sy 2	–	T	–
SDSS J090547.33+374738.2	587732152565432366	0.0475	trans	–	T	–
SDSS J090958.07+621450.4	587737826756198431	0.0261	Sy 2	NGC 2742A	S	Ioannab
SDSS J091011.34+230717.9	587741421104136355	0.0362	–	–	S	Citizen_Kirk
SDSS J091708.26+292215.6	587738946130018402	0.0353	Sy 2	KUG 0914+295	S	Half65
SDSS J093033.05+034443.6	587728880335257753	0.0911	Sy 1	–	S	ElisabethB
SDSS J094529.64002154.7	587725074458345485	0.0515	LINER	–	T	–
SDSS J095559.88+395446.9	588016528244670522	0.0483	–	Violet plume	S	StephanieC
SDSS J100507.88+283038.6	587741392112451744	0.0517	Sy 2	–	S T	laihro
SDSS J100529.60+275844.2	587741391575580675	0.0555	Sy 2	–	S	mitch
SDSS J101128.26+260655.4	587741490365792357	0.1164	Sy 2	–	S	mitch
SDSS J101645.11+421025.5	588297863112294442	0.0553	NLSy1	–	S	mitch
SDSS J102016.20+524756.9	587731499185602761	0.0689	Sy 2	–	S	–
SDSS J102108.59+024058.5	587726033311498259	–	–	–	S	mitch
SDSS J103734.22+140120.5	587735349101199404	0.2061	Sy 2	–	S	davidjamesjones
SDSS J104232.05+050241.9	587728880879992930	0.0271	Sy 2	NGC 3341	S	mitch
SDSS J104326.47+110524.2	587734948595499096	0.0475	Sy 1	Purple haze	S	lovethetropics
SDSS J104515.28+421331.7	588017626147782828	0.0990	LINER	Bubble?	S	RandyC
SDSS J110157.90+101739.2	587732772658806856	0.0340	Sy 1	–	S	davidjamesjones
SDSS J110335.42+032014.7	587726033853022296	0.0531	Sy 2	–	S	skwalker
SDSS J110445.46+041755.2	588010358543351857	0.0252	Sy 2	–	S	ElisabethB
SDSS J110756.53+474434.8	588295840708755475	0.0727	Sy 1	–	S	davidjamesjones
SDSS J111100.60005334.9	588848898833580220	0.0904	Sy 2	–	S T	ElisabethB
SDSS J111113.00+284242.7	587741532784361479	0.0294	SB	NGC 3561A	T	–
SDSS J111113.18+284147.0	587741532784361477	0.0295	Sy 2	NGC 3561	T	–
SDSS J111349.74+093510.7	587734892748144649	0.0292	Sy 1.5	IC 2637	S T	stellar190
SDSS J111653.96+593146.8	587729387686461462	0.0815	trans	VII Zw 384	T	–
SDSS J112534.58+523247.0	587732136456487055	0.0270	SB	–	S	errattan
SDSS J112753.87+302138.6	587741491447070913	0.0736	non-AGN	–	S	paulrogers
SDSS J112942.51+235014.1	587742189363331247	0.1277	LINER	–	S	ElisabethB
SDSS J113323.97+550415.8	587733081347063838	0.0085	Sy 1	Mrk 177 compn	S	stellar190
SDSS J113629.36+213551.7	587742013279502427	0.0297	Sy 1	Mrk 739	S T	Budgieye
SDSS J113849.61+574243.4	587735696978215000	0.1162	Sy 1	–	S	ElisabethB
SDSS J114155.61+010516.7	588848901521277093	0.1365	Sy 2	–	S	lovethetropics
SDSS J114454.85+194635.3	588023669168537695	0.0274	Sy 2	–	S T	stellar190
SDSS J114517.10+200121.8	588023669705474229	0.04953	–	(possible)	S	Half65
SDSS J115140.70+675041.9	587725552285122567	0.0629	Sy 2	–	S	StephanieC
SDSS J115739.07023908.3	587724649256779921	0.1308	Sy 1	Purple haze	S	c_cld
SDSS J115906.89+101001.7	587732771591225359	0.1165	Sy 1.8	Purple haze	S	c_cld
SDSS J120114.35034041.0	587725039018311737	0.0196	Sy 1	Mrk 1310	S	Milk_n_cookies
SDSS J120150.80+143323.9	587735348036370587	0.0677	non-AGN	–	S	DuffBeer
SDSS J120719.81+241155.8	587742189367066665	0.0505	NLSy1	Purple haze, Mrk 648	S	davidjamesjones
SDSS J120939.43+643107.6	587729154134966352	0.1042	Sy 2	–	S	mitch
SDSS J121418.25+293146.7	587741532253519916	0.0632	Sy 2	Was 49ab	S T	stellar190
SDSS J121431.32+402902.6	588017979429486656	0.1211	Sy 2	–	S	mitch
SDSS J121452.41+591953.2	587729386079059975	0.0607	Sy 2	VII Zw 444	S	mitch
SDSS J121553.08+051447.8	588010359624827047	0.0803	LINER	–	S	codexluminati
SDSS J121819.30+291513.0	587739719750058064	0.0477	Sy 2	UGC 7342	S T	stellar190
SDSS J122402.57+435814.0	588017603610935505	0.1040	non-AGN		S	fluffyporcupine
SDSS J122546.72+123942.7	588017566564155399	0.0086	Sy 2	NGC 4388	S T	RandyC
SDSS J122802.10+094347.9	587732771057434918	0.1534	LINER	–	S	Bruno
SDSS J123034.25+033800.7	587726016682917948	0.01285	LINER	–	S	StephanieC
SDSS J123038.98+401614.4	587738947758456849	0.1322	Sy 1.5	Purple haze	S	Tsering
SDSS J123046.11+103317.3	587732772131504164	0.01540	SB	VPC 0764	S	lovethetropics
SDSS J123113.12+120307.2	588017702933823557	0.1161	Sy 1	–	S	lovethetropics
SDSS J124036.73+365004.3	587739096991334439	0.0404	Sy 2	–	S	stellar190
SDSS J124046.40+273353.5	587741602034090027	0.0565	Sy 1.5	–	S	Tsering
SDSS J124103.66+273526.0	587741602034155555	0.2007	Sy 1	–	S	Tsering
SDSS J124325.65+365525.3	587739096991596570	0.0839	Sy 2	–	S	Bruno
SDSS J124450.84-042604.5	587745544806727722	0.0147	LINER	IC 0812	S	Milk_n_Cookies
SDSS J124505.56+102433.2	587732772133011652	0.0976	Sy 2	–	S	davidjamesjones
SDSS J124511.84+230210.0	587742014897127434	0.02326	–	IC 0813	S	elizabeth
SDSS J125741.04+202347.7	588023670249750583	0.0807	Sy 2	IC 3929	S T	c_cld
SDSS J130007.06+183914.3	587742575372730410	0.1130	Sy 2	–	S	mitch
SDSS J130234.89+184122.3	588023668102856809	0.0656	Sy 2	–	S	Mukund Vedapudi
SDSS J130422.19+361543.1	587738950954385445	0.0443	Sy 2	WR 470	S	mitch
SDSS J130258.82+162427.7	587742773491531836	0.0673	Sy 1	Mrk 783	S	stellar190
SDSS J130509.98033209.2	587725039025258588	0.0835	Sy 1	Purple haze	S	lovethetropics
SDSS J131555.15+212521.5	587742013289660465	0.0884	Sy 1	–	S T	davidjamesjones
SDSS J131639.74+445235.0	588017605762482225	0.0909	Sy 1.9	–	S	c_cld
SDSS J131913.93+132030.8	587736802936684556	0.0960	Sy 2	–	S	mitch
SDSS J132340.31012749.1	587725041711644785	0.0767	Sy 1	–	S	IC 1101
SDSS J132540.23+275146.1	587739719219675227	0.0377	–	–	S	ElisabethB
SDSS J133227.20+112910.4	588017570316615795	0.0778	Sy 2	–	S	veggy2
SDSS J133416.49+311709.1	587739609171230755	0.0570	SB	Was 75	T	–
SDSS J133718.72+242303.3	587742190986657795	0.1076	NLSy1	–	S	c_cld
SDSS J133815.86+043233.3	587729158970736727	0.0228	Sy 1.5	NGC 5252	S T	laihro
SDSS J133817.11+481636.1	587732483292266549	0.02786	Sy 2	NGC 5256, Mrk 266	S T	Gumbosea
SDSS J134442.16+555313.5	587735666377949228	0.0373	Sy 2	Mrk 273	S T	stellar190
SDSS J134608.10+293810.4	587739504478060626	0.0776	Sy 1	–	S	lovethetropics
SDSS J134630.29+283646.3	587739707943092392	0.0518	Sy 2	–	T	–
SDSS J135255.67+252859.6	587739810484650051	0.06387	Sy 1	KUG1350+257	T	–
SDSS J135602.62+182217.7	587742550676275314	0.0506	Sy 2	Mrk 463	T	–
SDSS J135635.73+232135.9	587739845379883044	0.0668	LINER	–	T	–
SDSS J135712.06-070433.0	587746236298231865	–	–	–	S	stellar190
SDSS J140037.11+622132.7	587728918446407773	0.0752	Sy 1	–	S	mitch
SDSS J141051.82+410412.5	588017604156457121	0.0812	LINER	–	S	Song
SDSS J140506.26+024618.2	587726033335943373	0.0766	Sy 2	–	S	mitch
SDSS J141405.01+263336.8	587739720298201117	0.0357	Sy 1	–	S	davidamesjones
SDSS J142522.28+141126.5	587742609727684701	0.0601	Sy 2	–	T	–
SDSS J142925.07+451831.8	587735490282848380	0.0748	Sy 1.5	Purple haze	S T	Aroel
SDSS J143029.88+133912.0	587736809916399664	0.0852	Sy 2	Teacup	S T	Half65
SDSS J143239.83+361808.0	587736583892238376	0.0132	Sy 2/SB	NGC 5675	T	–
SDSS J144038.10+533015.8	587733427086426161	0.0376	Sy 2	Mrk 477	T	–
SDSS J144240.79+262332.5	587739457225097282	0.1071	Sy 1	–	S	spiralmania
SDSS J144331.19+191121.0	587742062161428638	0.0598	Sy 2	–	S	Bruno
SDSS J145724.63+105937.3	587736807771930760	0.1227	Sy 1	–	S	davidjamesjones
SDSS J150756.88+032037.3	587726100952449048	0.1369	Sy 1	–	S	mitch
SDSS J151004.01+074037.1	588017991773520114	0.0458	Sy 2	–	S T	whitefluffydogs
SDSS J150408.46+143123.3	587742575922708553	0.1181	Sy 1	Mrk 840	S	mitch
SDSS J151141.26+051809.2	587736546312323142	0.0845	Sy 1	–	S	Half65
SDSS J151915.98+104847.8	587736813131989104	0.0988	Sy 1	–	S T	spiralmania
SDSS J152412.58+083241.2	588017703489372418	0.0371	Sy 2	–	S T	Alice
SDSS J152549.54+052248.7	587730022796755031	0.048	Sy 2	–	T	–
SDSS J152907.45+561606.6	587742882456731737	0.0998	Sy 1	–	S	spiralmania
SDSS J153355.15+585756.4	587725818571063416	–	–	–	S	lovethetropics
SDSS J153432.52+151133.2	587742013841145937	0.0066	Sy 2	NGC 5953	S	Half65
SDSS J153508.93+221452.8	587739814240190581	0.0858	trans	Purple haze	T	–
SDSS J153703.36+135944.1	587742590401904799	0.0737	LINER	–	S	RandyC
SDSS J153854.16+170134.2	587739845390761994	0.02974	Sy 2	NGC 5972	S T	NeilGibson
SDSS J155007.62+272814.5	587736941990969374	0.1468	Sy 1	Purple haze	S	ElisabethB
SDSS J160536.79+174807.5	587739720846934175	0.0339	Sy 2	IC 1182	S	stellar190
SDSS J162538.08+162718.1	587739814246023211	0.0343	LINER	Akn 502	T	stellar190
SDSS J162804.06+514631.4	587736980102643827	0.0547	Sy 1.9	Mrk 1498	S	Budgieye
SDSS J162930.01+420703.2	587729653421441105	0.0717	Sy 1	Purple haze	S	Tsering
SDSS J162952.88+242638.4	587736898503639075	0.0368	Sy 1	Mrk 883	S T	Rick Nowell
SDSS J164800.81+295657.4	587733399186898947	0.1059	Sy 1	–	S	mitch
SDSS J172335.75+342133.4	587739849686843709	–	–	–	S	Mukund Vedapudi
SDSS J172747.17+265121.4	587729409160183880	0.0291	–	VV 389	S	elizabeth
SDSS J172935.81+542939.9	587725505559855518	0.0820	Sy 2	–	S	Bruno
SDSS J181611.61+423937.3	758879745074397535	0.04120	Sy 2	UGC 11185	S	stellar190
SDSS J210918.38060754.7	587726879412256901	0.0286	Sy 2	–	S	mitch
SDSS J214150.10+002209.4	587731186725683280	0.1068	LINER	–	S	echo-lily-mai
SDSS J220141.64+115124.3	587727221400862869	0.0296	Sy 2	–	S T	stellar190
SDSS J233254.46+151305.4	587730774959652922	0.2148	Sy 1	–	S	ElisabethB
SDSS J234413.61+004813.9	587731187275923676	0.0497	LINER	–	S	davidjamesjones

2.1 SDSS image analysis and new images

For both subsamples, further winnowing had the same steps. Most importantly, we re-analyzed the SDSS images, to verify that the features do not have continuum counterparts and eliminate artefacts caused by imperfect registration of the images when forming the colour composites. This effect is of particular concern for Seyfert 1 nuclei, where the point spread function of the bright nucleus can produce a decentred colour signature if one of the constituent images is slightly misregistered; Sy 1 galaxy image is more vulnerable to this artefact than normal galactic nuclei. Since many candidates (including some with spectroscopic confirmation) have ‘purple haze’ on the SDSS images, which could either be genuinely extended and somewhat amorphous [O iii] and H or an artefact, this was a helpful step. We adopted a tomographic approach, taking one of the SDSS bands free of strong emission lines (⁠ or ⁠, depending on redshift) as an estimate of the structure of starlight in the galaxy. This was scaled to match the largest part of the  structure, iteratively when necessary. This is illustrated in Fig. 1, isolating the emission-line loop in SDSS 1430+13 (nicknamed the Teacup AGN because of this structure). Chojnowski and Keel inspected the best subtraction among various scalings (often a compromise, due to colour gradients within the galaxy) to assess the reality of extended emission-line features not associated with clear spiral arms or stellar rings. These results let us rank the candidate lists from both targeted and serendipitous searches in order of significance of the emission-line structures based on the SDSS images themselves. We used these results to limit the number of candidates from the targeted search to the top 50; below this there were no convincing candidates based on more detailed analysis of the SDSS images.

Figure 1

Linear combination of SDSS images to isolate candidate [O iii] emission regions, shown with SDSS J143029.88+133912.0 (the Teacup AGN). The region shown spans  SDSS pixels, or  arcsec, with north at the top. Each image is displayed with a logarithmic intensity mapping, with a small zero-point offset to reduce the effects of noise around zero. At the redshift  of this galaxy, H and [N ii] emission fall redward of the  filter band (response 0.2 per cent of peak) so the  image is used as a continuum estimate.

Where appropriate filters were available for [O iii] or H at a galaxy’s redshift, some candidates were imaged at the remote SARA 1-m (Kitt Peak) and 0.6-m (Cerro Tololo) telescopes. For [O iii], we used a filter centered at 5100 Å with half-transmission width 100 Å, useable for the redshift range ⁠. At H⁠, both telescopes have stepped sets of filters 75 Å apart with full width at half-maximum (FWHM)  Å. Continuum was taken from  or ⁠, appropriately scaled for subtraction to show net emission-line structures. These data are particularly helpful in tracing the emission-line structures of UGC 7342 (Fig. 2) and SDSS 2201+11 (Fig. 3).

Figure 2

The extended clouds in UGC 7342. Left: the starlight continuum in a band at 6450 Å from the SARA 1-m telescope. Centre: an estimated [O iii] image from the SDSS data as in Fig. 1. Right: continuum-subtracted H image from the SARA 1-m telescope. North is at the top and east to the left; the field shown spans  arcsec.

Figure 3

The extended clouds in SDSS 2201+11. Left: a  image from the SARA-S 0.6-m telescope, showing the dusty disc. Right: continuum-subtracted [O iii] image from the SARA-N 1-m telescope, smoothed by a Gaussian of 2.0 arcsec FWHM. North is at the top and east to the left; the field shown spans  arcsec.

2.2 Spectroscopy

To confirm that regions are in fact ionized by AGN, and derive diagnostic emission-line properties, we carried out long-slit spectroscopy for the highest priority candidates. Observations used the GoldCam spectrograph at the 2.1-m telescope of Kitt Peak National Observatory (KPNO) and the Kast double spectrograph at the 3-m Shane telescope of Lick Observatory. Table 2 compares the setups used for each session. The slit width was set at 2 arcsec for all these observations, and the spectrographs were rotated to sample the most extended known structures of each galaxy. Scheduling allowed us to reduce the Kitt Peak data before the first Lick observing run, so that the 3 m spectra could be concentrated on the most interesting galaxies. Total exposures ranged from 30 m, for initial reconnaissance to see whether an object might host AGN clouds, to 2 h, for weaker lines in confirmed targets. Either night-sky line or interspersed lamp observations were used to track flexure, as needed. Reduction used the ‘longslit’ package in iraf2 (Tody 1986), and included rebinning to a linear wavelength scale, sky subtraction and flux calibration. Spectrophotometric standard stars were observed to set the flux scale; in a few cases where passing clouds were an issue, the spectra were scaled so that the nucleus within a  arcsec region matched the flux of the SDSS spectra.

Table 2

Spectroscopic data.

Telescope	ut dates	Range (Å)	Resolution (Å)	Slit scale (arcsec pixel⁠)	Galaxies observed
KPNO 2.1 m	2010 June 15–21	3630–5700	3.2	0.78	33
Lick 3 m	2010 July 12–15	5450–8260	4.5	0.78	11
		3495–5605	1.5	0.43	11
Lick 3 m	2010 December 1–3	4630–7410	4.3	0.78	13
		3280–4595	2.7	0.43	13
Lick 3 m	2009 December 17	5250–9940	13.5	0.78	2
		3650–5710	4.8	0.43	2

Our identification of these extended regions as being photoionized by AGN rests on three results – location in the strong-line BPT diagram, strength of the high-ionization species He ii and [Ne v], and electron temperature consistent with photoionization but not with shock ionization. We classify emission regions based on the ‘BPT’ line-ratio diagrams pioneered for galactic nuclei by Baldwin et al. (1981) and refined by Veilleux & Osterbrock (1987), with caution based on the possibility that some of the external gas could have much lower metal abundances than found in galactic nuclei (as seen in Hanny’s Voorwerp; Lintott et al. 2009). Abundance effects in gas photoionized by AGN, as manifested in the BPT diagrams, have been considered in calculations by Bennert et al. (2006a). The largest effect is higher equilibrium temperature at lower O abundance, since it is an important coolant, which drives stronger forbidden lines and higher ionization levels until very low levels (0.1 solar) are reached. In any case, the abundance changes are not large enough to move these clouds across the empirical AGN/starburst ionization boundary. Furthermore, in the galaxies where we have data covering the red emission lines, the clouds’ locations in the (essentially abundance-independent) auxiliary BPT diagram of [O iii]/H versus [O i]/H also indicated photoionization by an AGN continuum. The various BPT diagrams are compared for points along the slit in each of the clouds we classify as AGN-ionized in Fig. 4. This classification is examined more closely in the context of its radial behaviour in the next section.

Figure 4

Summary BPT diagrams for the Lick spectra, where we measured the requisite red emission lines. Circles indicate points along the slit for extended clouds classified as AGN-ionized and the host nuclei. Grey-scale background shows the density of points from all low-redshift SDSS galactic nuclei, as in Schawinski et al. (2010a). The dividing lines between regions photoionized by AGN and by hot stars are shown as given by Kewley et al. (2001; labelled Ke01) and Kauffmann et al. (2003; labelled Ka03).

Independent of these line ratios, strong He ii  or [Ne v]  indicate photoionization by a harder continuum than provided by young stars, and resolved emission from these species is immediately diagnostic of AGN photoionization in this context. For some objects, we do not have red data; in these, we classify the cloud as AGN-ionized based on the presence of the high-ionization lines or continuity of line ratios with the nucleus. Line ratios in the extreme blue may be affected in subtle ways by atmospheric dispersion (Filippenko 1982); the scheduling of our observations forced us to observe most targets at hour angles which did not allow us to put the slit simultaneously along the structures of interest and close to the parallactic angle. The extended regions we observe are generally wider than the slit; to first order line intensities are not affected by atmospheric refraction, since we calibrate with standard stars at low airmass. Some of the Lick blue spectra have atmospheric dispersion contributing as much as 3 arcsec of offset along the slit from red to blue ends of the spectrum, important only for the nuclei and corrected in extracting their spectra.

The BPT diagrams are designed to separate common sources of photoionization in galaxies; temperature and kinematic data are also important to understand whether shocks pay a significant role. In a few cases, the [O iii]  line was measured in the extended clouds with sufficient signal-to-noise ratio for a measurement of the electron temperature via its ratio to the strong  lines. Using the iraf task ‘temden’, which implements the algorithm of Shaw & Dufour (2007), and considering  cm⁠, we find  values of  in the SDSS 2201+11,  for Mrk 266 and  for the Teacup system. These confirm that the gas is photoionized rather than shocked; for comparison, temperatures in the [O iii] zone of supernova remnants (including some with lower ionization levels than in these clouds) range from 20 000 to 69 000 K (e.g., Fesen, Blair & Kirshner 1982; Wallerstein & Balick 1990; Morse, Winkler & Kirshner 1995). In addition, very high shock velocities  km s are needed to produce significant [Ne v] emission (Dopita & Sutherland 1996). This is far in excess of the local velocity ranges we observe (Section 6); even though we would not necessarily observe material on both sides of a shock in the same ion, it is difficult to envision a situation with large-scale shocks of this velocity without observable velocity widths or structures exceeding 100 km s⁠.

While not the main thrust of our survey, it is worth noting that we find a few instances of either double AGN in interacting systems or AGN in the fainter member of a close pair (Mrk 177, Was 49, possibly SDSS J111100.60005334.9 and SDSS J142522.28+141126.5). These maybe worth deeper spectroscopy in the context of mapping AGN obscuration; if a high-ionization component can be isolated in the gas of the other galaxy, its distribution could show where ionizing radiation escapes any circumnuclear absorbing structure. This offers a distinct way of tracing the ionizing radiation even in the absence of extensive gaseous tidal features, in an approach that has been discussed for Was 49ab by Moran et al. (1992).

Table 3 lists the results of our spectroscopy. Confirmed, resolved clouds ionized by the AGN are separated from other results (unresolved AGN emission, extended star-forming regions denoted as H ii, and so on). The instruments used are denoted by GCam (Kitt Peak GoldCam) and Lick (Lick 3 m with Kast spectrograph). New redshifts and spectral classifications are marked with asterisks. We separate the AGN clouds of most interest based on the detected extent of [O iii] ⁠; our spectra have a lower detection threshold than our images for this, roughly  erg cm s arcsec for emission regions a few arcseconds in size. Spectra of the nuclei and representative cloud regions are shown in Figs 5 and 6. Table 4 lists emission-line ratios and selected fluxes for the same regions plotted in these figures. Fluxes are given both for [O iii]  and H⁠, since these were usually measured with different gratings and detectors. For some of the nuclei, correction of the H flux for underlying absorption in the stellar population was significant; we have applied an approximate correction based on typical values for synthetic stellar populations from Keel (1983).

Table 3

Results of long-slit spectroscopy.

Coordinate name	SDSS ObjID		Nucleus	Name	PA(⁠⁠)	Source	Region type
Confirmed AGN ionized:
SDSS J074241.70+651037.8	758878270293868614	0.0371	Sy 2	Mrk 78	90	Lick	AGN
SDSS J095559.88+395446.9	588016528244670522	0.0483	Sy 2	–	148	Lick	AGN
SDSS J100507.88+283038.5	587741392112451744	0.0517	Sy 2	–	62	Lick	AGN
SDSS J111349.74+093510.7	587734892748144649	0.0292	Sy 1.5	IC 2637	47	GCam	AGN
SDSS J113629.36+213551.7	587742013279502427	0.0297	Sy 1	Mrk 739	168	GCam	AGN
SDSS J121819.30+291513.0	587739719750058064	0.0477	Sy 2	UGC 7342	133	GCam	AGN
SDSS J122546.72+123942.7	588017566564155399	0.0086	Sy 2	NGC 4388	26	GCam	AGN
SDSS J133815.86+043233.3	587729158970736727	0.0228	Sy 1.5	NGC 5252	175	GCam	AGN
SDSS J133817.11+481636.1	587732483292266549	0.0279	Sy 2	Mrk 266	176	GCam, Lick	AGN
SDSS J134442.16+555313.5	587735666377949228	0.0373	Sy 2	Mrk 273	57	GCam	AGN
SDSS J135602.62+182217.8	587742550676275314	0.0504	Sy 2	Mrk 463	8	GCam	AGN
SDSS J143029.88+133912.0	587736809916399664	0.0852	Sy 2	Teacup	37	GCam, Lick	AGN
SDSS J151004.01+074037.1	588017991773520114	0.0458	Sy 2	–	175	GCam	AGN
SDSS J152412.58+083241.2	588017703489372418	0.0371	Sy 2	CGCG 077117	150	GCam	AGN
SDSS J153854.16+170134.2	587739845390761994	0.0297	Sy 2	NGC 5972	167	GCam, Lick	AGN
SDSS J162804.06+514631.4	587736980102643827	0.0547	Sy 1.9	Mrk 1498	150	GCam, Lick	AGN
SDSS J162952.88+242638.4	587736898503639075	0.0368	Sy 1	Mrk 883	91	Lick	AGN
SDSS J181611.61+423937.3	758879745074397535	0.0412	Sy 2	UGC 11185	90	GCam, Lick	AGN
SDSS J220141.64+115124.3	587727221400862869	0.0296	Sy 2	–	19	GCam, Lick	AGN
Other cloud types:
SDSS J003507.44+004502.1	587731187281494175	0.1205	LINER	–	35	Lick	H ii regions
SDSS J005607.66+254804.7	587740589487030353	0.1530	Sy 1	–	200	Lick	No
SDSS J012839.87+144553.8	587724233179660360	0.0452	Sy 2	CGCG436060	45	Lick	No
SDSS J014238.47+000514.7	588015509280587804	0.1459	Sy 1	–	80	Lick	AGN
SDSS J030639.58+000343.2	588015509289762862	0.1074	Sy 1	–	40	Lick	AGN (small)
SDSS J033013.26053235.9	587724242842026028	0.0131	SB	NGC 1346	76	Lick	H ii
SDSS J040548.78061925.7	587727178476093634	0.0556	Sy 1	–	76	Lick	H ii
SDSS J080452.73+212050.2	588016878287650850	0.1242	Sy 1	–	150	Lick	AGN (small)
SDSS J082642.63+111555.5	587745244691628370	0.0884	trans	–	72	Lick	Resolved?
SDSS J083525.51+104925.7	587744873714679862	0.1172	Sy 1	–	150	Lick	Small?
SDSS J084344.98+354942.0	587732484342415393	0.0539	Sy 2	–	141	Lick	AGN
SDSS J111100.60005334.9	588848898833580220	0.0904	Sy 2	–	101	GCam	Unresolved cloud at 4 arcsec
SDSS J113323.97+550415.8	587733081347063838	0.0085	Sy 1	Mrk 177	136	GCam	AGN in small compn
SDSS J114454.85+194635.3	588023669168537695	0.0274	Sy 2	–	131	GCam	Unresolved  arcsec
SDSS J121418.25+293146.7	587741532253519916	0.0632	Sy 2	Was 49ab	63	GCam	Off-nuc AGN or cloud
SDSS J123046.11+103317.3	587732772131504164	0.01540	SB	VPC 0764	30	GCam	Unresolved
SDSS J124450.84042604.5	587745544806727722	0.0147	LINER	IC 0812	62	GCam	arcsec
SDSS J125741.04+202347.7	588023670249750583	0.0807	Sy 2	IC 3929	47	GCam	H ii to 9 arcsec
SDSS J134630.29+283646.3	587739707943092392	0.0518	Sy 2	–	164	GCam	34 arcsec dim AGN cloud?
SDSS J135255.67+252859.6	587739810484650051	0.06387	Sy 1	KUG1350+257	163	GCam	H ii
SDSS J142522.28+141126.5	587742609727684701	0.0601	Sy 2	–	110	GCam	Two AGN?
SDSS J150408.46+143123.3	587742575922708553	0.1181	Sy 1	Mrk 840	70	GCam	7 arcsec
SDSS J151915.98+104847.8	587736813131989104	0.0099	Sy 1	–	89	GCam	Unresolved  arcsec
SDSS J153508.93+221452.8	587739814240190581	0.0858	trans	–	70	GCam	arcsec
SDSS J153703.36+135944.1	587742590401904799	0.0737	LINER	–	21	GCam	Unresolved
SDSS J160536.79+174807.5	587739720846934175	0.0339	Sy 2	IC 1182	96	GCam	H ii
SDSS J210918.38060754.7	587726879412256901	0.0286	Sy 2	–	70	GCam	Unresolved  arcsec
SDSS J214150.10+002209.4	587731186725683280	0.1068	LINER	–	70	GCam Lick	Unresolved  arcsec
SDSS J233254.46+151305.4	587730774959652922	0.2148	Sy 1	–	138	Lick	Unresolved  arcsec
SDSS J234413.61+004813.9	587731187275923676	0.0497	LINER	–	71	Lick	H ii

Figure 5

Lick spectra for nuclei and associated AGN-ionized clouds. Small insets at left show the H + [O iii] region for nuclei and clouds, scaled to the peak of  emission. Panels on the right show the [O i][S ii] region at the same flux scale as the blue spectra. He ii and [Ne v] emission, especially important as indicators of a hard ionizing radiation field, are indicated by red dotted lines when clearly detected in clouds. Nuclear spectra represent  arcsec areas, and cloud spectra are summed over  arcsec areas. Distances and directions of cloud relative to nuclei are indicated as shown. Three spectra have gaps in the blue region, since they were taken with the dichroic splitting red and blue optical trains near 4600 Å.

Figure 6

KPNO GoldCam spectra sampling nuclei and associated AGN-ionized clouds. Spectra are scaled to show [O iii] ⁠, with nuclear spectra representing  arcsec areas and cloud spectra representing  arcsec areas. He ii and [Ne v] emission, especially important as indicators of a hard ionizing radiation field, are indicated by red dotted lines when clearly detected in clouds. Distances and directions of cloud relative to nuclei are indicated as shown.

Table 4

Emission-line ratios and selected fluxes.

Object							F(5007)				F(H⁠)
IC 2637	Nuc	0.23	0.28	0.81	0.11	3.94	2.7e14	–	–	–	–
	Cloud		0.12	2.43		1.34	4.6e15	–	–	–	–
Mrk 78	Nuc (1)	0.35	0.32	0.10	0.29	13.5	6.8e14	0.05	1.00	0.23	6.5e13
	Cloud	2.00	0.16	0.43	0.37	14.00	7.8e15	0.094	0.51	0.22	2.0e15
Mrk 266	Nuc	0.69	0.17	0.43	0.19	4.08	3.4e14	0.052	0.58	0.35	4.0e14
	Cloud	0.46	0.40	0.22	0.38	8.20	3.7e14	0.029	0.25	0.23	1.1e14
Mrk 273	Nuc	0.27	0.11	0.90	0.12	3.93	2.4e14	–	–	–	–
	Cloud	0.06	0.26	0.93	0.27	9.3	1.0e14	–	–	–	–
Mrk 463	Nuc	0.26	0.29	0.30	0.08	7.7	3.7e13	–	–	–	–
	Cloud	0.41	0.16	0.51	0.17	11.3	1.8e14	–	–	–	–
Mrk 739	Nuc (2)	1.16	0.49	0.15	0.18	5.88	1.5e14	–	–	–	–
	Cloud	1.31	0.22	0.45	0.24	8.30	6.5e15	–	–	–	–
Mrk 883	Nuc	0.31	0.12	0.81	0.08	2.82	5.3e14	0.12	0.47	0.47	8.4e14
	Cloud	0.30	0.22	0.30	0.19	7.84	6.6e15	0.08	0.31	0.49	4.1e15
Mrk 1498	Nuc (2)	0.73	0.82	0.07	0.10	3.72	1.5e13	0.07	1.00	0.22	4.9e14
	Cloud	1.08	0.62	0.18	0.32	12.3	1.0e14	0.06	0.14	0.20	3.4e15
NGC 4388	Nuc	0.45	0.25	0.27	0.23	11.5	3.6e13	–	–	–	–
	Cloud	0.39	1.15	0.57		4.72	4.3e15	–	–	–	–
NGC 5252	Nuc	0.31	0.27	0.55	0.19	9.22	6.0e14	–	–	–	–
	Cloud	1.31	0.34	0.31	0.34	10.1	1.1e14	–	–	–	–
NGC 5972	Nuc	2.25	0.31	0.29	0.40	2.56	5.1e14	0.07	0.69	0.49	1.9e14
	Cloud	0.15	0.20	0.63	0.26	8.48	2.6e14	0.12	0.79	0.61	9.5e15
SDSS 0955+39	Nuc		0.05	4.60		2.90	6.9e16	0.32	0.84	1.02	1.6e15
	Cloud	2.64	0.47	0.28	0.51	7.00	3.2e15	0.08	0.26	0.37	9.5e16
SDSS1005+28	Nuc	0.55	0.29	0.30	0.24	7.96	1.1e14	0.07	0.57	0.48	5.8e15
	Cloud	1.52	0.60	0.24	0.58	5.46	4.5e15		0.25	0.34	3.3e15
SDSS 1510+07	Nuc	0.04	1.15	1.86		2.11	3.2e15	–	–	–	–
	Cloud	1.76	0.22	0.56	0.39	6.64	4.4e15	–	–	–	–
SDSS 1524+08	Nuc	0.78	0.19	0.91	0.37	8.00	4.8e15	–	–	–	–
	Cloud	1.58	0.15	0.49	0.44	7.84	5.4e15	–	–	–	–
SDSS 2201+11	Nuc	0.17	0.07	1.65	0.28	5.48	1.7e15	0.17	1.21	1.02	7.0e15
	Cloud	0.53	0.24	0.21	0.18	10.1	7.8e15	0.07	0.40	0.35	3.0e15
Teacup	Nuc	0.31	0.23	0.30	0.09	7.95	1.3e13	0.15	0.23	0.15	1.1e14
	Cloud	0.14	0.17	0.23	0.15	8.07	2.1e14	0.11	0.34	0.36	7.6e15
UGC 7342	Nuc	0.75	0.22	0.58	0.22	9.14	2.9e14	–	–	–	–
	Cloud	0.77	0.26	0.41	0.55	9.90	8.9e15	–	–	–	–
UGC 11185	Nuc (1)	0.17	0.19	0.37	0.16	8.89	8.5e14	0.19	1.29	0.88	3.8e14
	Cloud	0.28	0.28	0.25	0.26	8.11	1.4e14	0.07	1.10	0.72	5.1e15

Note. Line fluxes are in units of erg cm s⁠.

(1) Blend of two velocity components.

(2) BLR present; flux is estimated NLR only.

The upper part in Table 3, with AGN-ionized gas detected more than 10 kpc from the nucleus, forms the sample for our subsequent analysis. As a sign of completeness, of these 19, 14 were found in both targeted and serendipitous searches. SDSS J095559.88+395446.9 was newly identified as a type 2 Seyfert in our spectrum, after having been found in the ‘blind’ search of galaxies independent of prior classification as an AGN host (so it was not included in the targeted sample). Of the remainder, Mrk 78 and Mrk 463 were selected in the targeted AGN sample, while Mrk 1498 and UGC 11185 were recognized only in the serendipitous survey. It maybe relevant that both Mrk 78 and Mrk 463 have ionized regions with a relatively small projected extent, easily lost against the galaxy starlight (which in Mrk 463 is morphologically complex).

The [S ii]  lines are particularly important, tracing electron densities and thereby providing one estimate of the intensity of the impinging ionizing radiation. Since the densities in these extended clouds are low, and the ratio is generally near its low-density limit, where the mapping from line ratio to density is highly non-linear, we have examined the errors in measuring the line ratio closely. We generated multiple realizations of pixel-to-pixel noise, and each was scaled to four representative fractions of the stronger line peak. This was added to line pairs, modelled to match the line widths and pixel separation of the red Lick data. Gaussian fitting of the lines with added noise gave a relation between the peak signal-to-noise ratio and error of the fitted ratio which we adopted; we use  error bounds to derive bounds on the density. Density values were calculated using the iraf task ‘temden’.

3 ENERGY BUDGET IN EXTENDED CLOUDS: OBSCURATION VERSUS VARIABILITY

Seeing the effects of radiation from an AGN on gas tens of kpc from the nucleus allows the possibility of tracing dramatic changes in core luminosity. One straightforward way to approach this question is a simple energy balance. The spectra give us upper and lower bounds on the required ionizing luminosity. To probe the most extreme conditions, we analyze galaxies in which we detect ionized gas at projected distances  kpc. For all distances and luminosities, we use the WMAP ‘consensus’ cosmological parameters, with H km s Mpc (Spergel et al. 2007).

The lower bound comes from the highest recombination-line surface brightness we observe; the central source must provide at least enough ionizing photons to sustain this over periods longer than the recombination time-scale (which maybe as long as  years at these low densities). This is a lower limit, since the actual emission-line surface brightness of some regions maybe smeared out by seeing, and we do not know that a given feature is optically thick at the Lyman limit. This limit depends only very weakly on the slope of the ionizing continuum, since helium will generally absorb most of the radiation shortward of its ionization edge leaving only the 13.6–54.4 eV range to consider for hydrogen ionization. We base our bounds on the highest implied luminosity among structures at various projected radii in a given system, with no correction for projection effects. This makes our limits conservative, since a given cloud will always lie farther from the nucleus than our projected measurement. In essence, this argument is based on the surface brightness in a recombination line; we use H since we have these data for the whole sample. In a simple approximation, we take the surface brightness in the brightest portion of a cloud, assuming this to be constant across the slit. We take the region sampled in this way to be circular in cross-section as seen from the nucleus, so its solid angle is derived from the region subtended by the slit. We then see this region occupying a small angle  as seen projected at angular distance  from the AGN; the required ionizing luminosity is given from observed quantities as  for  in degrees. The derived values are listed in Table 5, along with complementary quantities related to the nuclear luminosity (as collected below). The derived ionizing luminosities are lower limits, since there maybe unresolved regions of higher surface brightness, and we do not know whether a given cloud is optically thick in the Lyman continuum. Higher resolution imaging in the emission lines could thus increase these values.

Table 5

Ionizing and FIR luminosity comparison.

SDSS ID	Other name	IRAS m	m	AKARI m	m	L(FIR)	(arcsec)	F(H⁠)	⁠)		Ratio
074241.70+651037.8	Mrk 78	1.11	1.13			2.4e44	10	4.0e16	11	e43	0.25
095559.88+395446.9	–	–	–			5.0e44	7	2.5e16	16	e43	0.06
100507.88+283038.5	–	–	–	–	–	e44	8	4.2e16	14	e44	0.5
111349.74+093510.7	IC 2637	1.75	3.39			3.0e44	10	3.0e16	11	e43	0.10
113629.36+213551.7	Mrk 739	1.3	2.4			2.3e44	18	2.1e16	6.3	e43	0.26
121819.30+291513.0	UGC 7342			–	–	44	35	1.8e16	43.2	e44	3.6
122546.72+123942.7	NGC 4388	10.0	17.1			1.4e44	15	3.8e16	7.6	e42	0.02
133815.86+043233.3	NGC 5252				–	4.0e43	32	3.8e16	3.5	e44	5
133817.11+481636.1	Mrk 266	7.3	10.3			1.0e45	27	6.8e16	4.2	e44	0.4
134442.16+555313.5	Mrk 273	22.5	22.5			5.0e45	12	9.0e16	10	e44	0.03
135602.62+182217.8	Mrk 463	2.33	1.94			9.0e44	16	3.8e16	7	e44	0.3
143029.88+133912.0	Teacup	0.26		–	–	e43	5	1.66e15	22	e44	14
151004.01+074037.1	–			–	–	e44	14	8.2e16	8	e45	3.4
152412.58+083241.2	–	0.71	0.73	0.63		1.6e44	5.6e-16	12	10	1.0e44	0.6
153854.16+170134.2	NGC 5972	0.24		–	–	e43	35	7.2e17	3	e43	1.8
162804.06+514631.4	Mrk 1498	0.34		–	–	e43	10	9.8e17	11	e43	0.6
162952.88+242638.4	Mrk 883	1.01	1.13		–	2.2e44	10	4.0e16	11	e43	0.26
181611.61+423937.3	UGC 11185			–	–	e44	10	1.7e15	11	e44	1.7
220141.64+115124.3	–	0.28		–	–	e43	23	4.6e16	5.0	e44	3.4

Note. FIR fluxes are in Jy. Mean IRAS detection limits are used when no specific value is available. Luminosities are in erg s⁠; H fluxes are in erg cm s⁠. Values of the ionizing/FIR luminosity ratio are all lower limits.

Upper limits to the incident ionizing flux come from a complementary analysis using the ionization parameter (⁠⁠, the ratio of ionizing photons to particles), since these emission-line features all have [S ii] line ratios near the low-density limit. Our density results from the  [S ii] line ratio are given in Table 6. Values are listed only for objects with useful measures far from the core. In each case, we evaluated the density at a typical temperature of  K, and at the higher temperature  K found in Hanny’s Voorwerp (Lintott et al. 2009) and in our data for Mrk 266 and SDSS 2201+11, where the higher temperature is set by thermal equilibrium for substantially subsolar oxygen abundance. We quote the extreme range of density values between these two cases (allowing in the Teacup an upper bound on the electron density as high as 240 cm⁠, and in some cases limits cm⁠), since the temperature-sensitive [O iii]  line ratio is not well-enough measured in most of these objects to use individual  values. We derive  from the [O ii] /[O iii]  ratio using the power-law continuum models from Komossa & Schulz (1997), and the analytic fits from Bennert (2005) as interpolation tools. For fully ionized hydrogen at a distance  from the AGN, the photon flux in the ionizing continuum is ⁠. For objects with red spectra, giving densities from the [S ii] lines, limits to the luminosity are given in Table 6. It is re-assuring that the upper limits to ionizing luminosity derived from  and  always fall above the lower limits from recombination balance.

Table 6

[S iii] density measures and limits.

Object	Distance (arcsec)	(kpc)	[S iii] ratio	(cm⁠)	[O iii]/[O iii]	log	(erg s⁠)
Teacup	5–14	8–22		130–240	0.15	2.16
Mrk 883	12–16	9–12		12–100	2.62	3.34
SDSS 2201	8–16	5–10			0.25	2.42
NGC 5972	26–30	15–18		3–300	0.11	1.99
Mrk 266	20–29	11–16		12–100	0.22	2.36

The lower limits from recombination balance are independent of assumptions about the local density ⁠, making it more robust than ionization-parameter arguments when we have no independent tracer at these low densities. Fig. 7 shows several of our objects in one of the ‘BPT’ diagrams, going beyond their initial use to classify the gas as AGN-ionized to examine changes with projected distance from the nuclei. Some of these, such as Mrk 1498 and the Teacup 1430+13, show a phenomenon remarked earlier in, for example, NGC 5252 (Dadina et al. 2010) – the ionization balance stays roughly constant with radius, which is naturally explained if the characteristic density ⁠. This might occur naturally for gas in the host galaxy; tidal streams of gas would not be likely to match the extrapolated behaviour of gas within the galaxy and indeed we see some cases (Mrk 266, NGC 5972 and SDSS 2201+11) with substantial radial changes in ⁠. However, for Seyfert narrow-line regions, Bennert et al. (2006b) find a shallower density gradient ⁠, which would imply  for gas which is optically thin (or has a small covering fraction). These objects have heterogeneous behaviour. In the ionization cone of NGC 7212, Cracco et al. (2011) find no radial trend of ⁠.

Figure 7

BPT diagrams for spatial slices in Lick data. Each row shows the three classical BPT diagrams for each object, highlighting radial ionization behaviours. The nucleus is indicated by crosshairs, with colors changing from white to red with increasing distance from the nucleus. The grey-scale background and dividing lines are the same as in Fig. 4; these show only the region around the AGN loci in each case for discrimination of detail. Nearly all measurements lie firmly in the AGN domain, with possible exceptions in some regions of SDSS 1005 and SDSS 0955. The starburst/AGN ionization boundary from Kewley et al. (2001) is shown as the red full curve, while the boundary from Kauffmann et al. (2003) is the black dashed curve. All these measurements lie firmly in the AGN domain, with the possible exception of two regions in SDSS 1005 and the nucleus of Mrk 883, whose red spectrum shows a broad-line region and strong [O i]. The grey-scale shows the density of points representing low-redshift galactic nuclei in the SDSS, from Schawinski et al. (2010a), which we also follow in adopting the straight line as the LINER/Seyfert boundary.

Similar conclusions come from the more limited blue-line diagram also considered by Baldwin et al. (1981), which we can apply to the objects for which we have only blue-light spectra from KPNO. Some of the objects in this diagram as well as in Fig. 7 show systematic changes in ionization level with radius, manifested as offsets from upper left (higher ionization) to lower right (lower ionization). We show this behaviour in Fig. 8.

Figure 8

An alternate BPT diagram, usable when only blue spectra are available (in this case, from the Kitt Peak GoldCam). As in Fig. 7, points are colour-coded with projected distance from the galaxy nuclei. Since Mrk 739 has a significant broad-line region, its H components were deblended using a broad Gaussian and narrow Lorentzian, with narrow-line ratios plotted here.

We use far-infrared (FIR) data to estimate (or limit, for non-detections) the amount of AGN radiation absorbed (and reradiated) by nearby dense material, whether in an AGN ‘torus’ or in the inner parts of the host interstellar medium (ISM). The FIR luminosity is conservatively high as an estimate of the potential-obscured AGN luminosity, since there maybe a non-trivial contribution from star formation in the host galaxy as well as the AGN, and in some cases companion galaxies might blend with the target in the FIR beam. In a simple picture where a fraction  of the AGN radiation is absorbed by nearby dust and reradiated, the FIR luminosity will be of order ⁠, with an additional scaling factor of a few to account for non-ionizing radiation heating the grains (which we omit at this point for the sake of a conservative calculation). For convenience, we approximate the total FIR output by the FIR parameter introduced for Infrared Astronomical Satellite (IRAS) point-source catalogue data (Fullmer & Lonsdale 1989), a linear combination of flux values in the 60 and 100 m bands which gives a reasonable approximation to the total flux between 42 and 122 m. Numerically,

for IRAS fluxes in the 60 and 100 m bands given in Jy (multiplied by  for a result in erg cm s⁠). IRAS data were supplemented, where possible, by AKARI data (Kawada et al. 2007; Murakami et al. 2007; Yamamura et al. 2010) of higher accuracy. The positions of all these galaxies were covered in the IRAS survey, so we can assign typical upper limits to non-detections depending on ecliptic latitude; AKARI added two additional detections not found in the IRAS data, using only quality 1 (confirmed detection) fluxes. For non-ultraluminous infrared galaxy (non-ULIRG) objects (⁠ erg s⁠), we can reproduce the IRAS FIR parameter from AKARI 90 m fluxes and mean colours via

with 30 per cent accuracy (⁠ dex), and we use this to fill in FIR luminosities for the objects detected only by AKARI. The input values and results of this energy-balance test are shown in Table 5. Within our sample, Mrk 273 and Mrk 266 are classic ULIRGs, with  erg s⁠. Some of the others remain undetected in both the IRAS and AKARI surveys, leading to limits typically  erg s⁠. An index of whether the extended clouds can be ionized by an obscured AGN is provided by the ratio of required ionizing luminosity to FIR luminosity, given in Table 5. These values are all lower limits, since the ionizing luminosity is a lower limit. This ratio ranges from 0.02 to values  (Fig. 9). Low values clearly represent AGN which are strongly obscured along our line of sight but not towards the EELR clouds. Large ratios indicate long-term fading of the AGN, a spectral shape strongly peaked in the ionizing ultraviolet (UV), or very specific geometry for obscuring material, and thus indicate objects worthy of close attention.

Figure 9

Distribution of the ratio  in the log. Small values are consistent with AGN heavily obscured along our line of sight but not towards the emission-line clouds, while large values indicate fading of the AGN. Shaded regions show that there are no obvious differences in the distributions among type 1 Seyferts (black), intermediate type 1.5 and 1.9 objects (grey) and type 2 nuclei (white).

Arguments for long-time-scale variations in the central sources here depend on our having estimates for their total luminosity as seen directly, which could in principle fail either if their ionizing radiation were collimated by something other than obscuration, or the spectral shapes in the deep UV differ from our expectations based on the UV and X-ray behaviour of familiar AGN. Collimation by relativistic beaming would not account for the combinations of opening and flux ratio required (Keel et al., in preparation). A spectral solution to the behaviour would require an extreme-UV ‘bump’ dominating the ionizing flux from 13.6 to 54 eV by more than an order of magnitude. However, known AGN do not provide evident examples of either solution; the most straightforward interpretation of the data suggests that some of these clouds are ionized by AGN which have faded over the differential light travel time between our views of the clouds and nuclei.

4 NUCLEAR AND EXTENDED RADIO EMISSION

To further characterize the AGN in these galaxies, we collected radio fluxes at 1.4 GHz from the NRAO VLA Sky Survey source catalogue (Condon et al. 1998). All but two objects (SDSS 1510+07 and 1005+28) were detected above the 2.5 mJy survey limit; the source luminosity L (1.4 GHz) ranges from  to  W Hz⁠, the latter for the double source associated with NGC 5972 and comprising 94 per cent of the galaxy’s total flux. Eight of the galaxies qualify as radio-loud if one uses a simple, representative division at  W Hz⁠, and only one lies above 10⁠. This one – NGC 5972 – represents an interesting departure from the usual alignment of emission-line and radio structure (Section 7).

5 HOST AND CLOUD MORPHOLOGY

The examples of Hanny’s Voorwerp (Josza et al. 2009) and NGC 5252 (Prieto & Freudlng 1996) suggest that a common source of extraplanar gas at large radii is tidal debris. The host morphologies of the galaxies where we find extended ionized clouds support this notion. Table 7 lists morphological information on these galaxies, including warps, close companions, asymmetries or ongoing mergers. The actual incidence of tidal structures will be higher – for example, the inclined ring of gas with distinct kinematics in NGC 5252 has no optical counterpart. Of the 19 confirmed large-scale clouds, 14 are in systems classified from SDSS data alone as interacting, merging or post-merging (still showing tidal tails). This remarkably high incidence of disturbed systems (at least 73 per cent, even without including NGC 5252) supports the idea that most very extended emission-line clouds around local AGN represent illuminated tidal debris. We illustrate this in Fig. 10, showing the SDSS colour images with the -band [O iii] contribution enhanced to show the clouds’ locations. In this section, we include IC 2497/Hanny’s Voorwerp in the tabulations for comparison. A striking instance of a QSO-ionizing gas in a companion and tidal tail, on similar scales  kpc, has been reported by da Silva et al. (2011).

Table 7

Morphologies of AGN hosts with extended clouds.

SDSS designation		Sy type	Name	(kpc)	Morphology	Cone angle (⁠	Disc/cloud angle (⁠⁠)	Sides
SDSS J074241.70+651037.8	0.0371	2	Mrk 78	16	E	55	–	2
SDSS J095559.88+395446.9	0.0483	2	–	10	Interacting S	88	–	1
SDSS J100507.88+283038.5	0.0517	2	–	13	Sb, disturbed companion	92	–	1
SDSS J111349.74+093510.7	0.0292	1.5	IC 2637	11	Merger remnant	60	–	1
SDSS J113629.36+213551.7	0.0297	1	Mrk 739	17	Ongoing merger	28	–	1
SDSS J121819.30+291513.0	0.0477	2	UGC 7342	38	Ongoing merger; tails	86	–	2
SDSS J122546.72+123942.7	0.0086	2	NGC 4388	13	Edge-on Sc	80	53	1
SDSS J133815.86+043233.3	0.0228	1.5	NGC 5252	21	Edge-on S0, tilted H I ring	59	31	2
SDSS J133817.11+481636.1	0.0279	2	Mrk 266	21	Ongoing merger	112	–	2
SDSS J134442.16+555313.5	0.0373	2	Mrk 273	19	Ongoing merger	75	–	2
SDSS J135602.62+182217.8	0.0504	2	Mrk 463E	16	Ongoing merger	55	–	2
SDSS J143029.88+133912.0	0.0852	2	Teacup	18	Stellar tail and arc	80	–	1
SDSS J151004.01+074037.1	0.0904	2	–	10	Symmetric disc; S0 or Sa	85	–	2
SDSS J152412.58+083241.2	0.0371	2	CGCG 077-117	19	Merger remnant	56	–	1
SDSS J153854.16+170134.2	0.0297	2	NGC 5972	33	Warped disc and tails	35	18	2
SDSS J162804.06+514631.4	0.0547	1.9	Mrk 1498	21	E	42	–	1
SDSS J162952.88+242638.4	0.0368	1	Mrk 883	37	Ongoing merger	73	–	1
SDSS J181611.61+423937.3	0.0412	2	UGC 11185	11	Strong interaction	48	–	1
SDSS J220141.64+115124.3	0.0296	2	–	16	Edge-on warped disc, tails	23	30	2
SDSS J094104.11+344358.4	0.0499	LINER	IC 2497	40	Warped disc, H i tail	46	65	1

Figure 10

Host morphologies and ionized-gas structures in the Galaxy Zoo sample of AGN-ionized clouds. These are based on SDSS colour image products, with  filters mapped to blue–green–red for display. However, to enhance visibility of the ionized clouds, the contrast of the  filter has been increased in this figure, across the intensity ranges needed to show the gas effectively. Each image cutout spans   kpc with north at the top. As in Table 7, IC 2497 is included for comparison.

Several of these galaxies show discs seen nearly edge-on. From these, it is clear that the ionizing radiation can emerge well away from the disc poles. The projected angles from a stellar disc to the axis of the ionized clouds, when it is well defined, range from 30 to 54⁠. This fits with the statistics reported by Schmitt et al. (1997), in which obscured (type 2) objects show a wider range of angle than type 1 objects between the host-galaxy axes and the AGN axes as traced by radio jets.

We see both one- and two-sided emission regions. The two-sided regions are generally highly symmetric in angular location, although not necessarily in radial extent or surface brightness, fitting with biconical illumination patterns. As listed in Table 7, nine of 19 of our confirmed objects have emission detected on both sides of the nucleus. Particularly in very disturbed systems, a strong asymmetry may reflect the location of cold gas rather than the pattern of escaping ionizing radiation, so that we cannot necessarily conclude that the one-sided clouds are in galaxies that do not have two-sided radiation patterns.

The angular width of regions of escaping radiation may constrain the geometry of obscuring regions, if the ionization is bounded by the availability of radiation rather than gas. We list in Table 7 a cone angle which is the projected angular width of each half of a notional bicone encompassing all the high-ionization regions seen in our images, outside of a usual nuclear emission region (Fig. 11). Projection effects make the observed angle an upper limit to the three-dimensional opening angle of a cone. The sample of large emission clouds exhibits a wide range, from 23 to 112⁠. The narrower ones are challenging to understand from obscuration by a circumnuclear torus, suggesting absorbers that are geometrically quite thick compared to the opening angle for escape of ionizing radiation, to an extent which might better be described as an obscuring shell with small polar holes. However, some of these objects have dual clouds in very symmetric locations, which would be most naturally explained by such a scheme.

Figure 11

Illustration of angles listed in Table 7. The disc angle, defined only for highly inclined and reasonably symmetric host discs, is measured from the projected major axis of the disc to the mid-point of the cloud (bi)cones. The cone angle, as shown, measures the width of the paired triangles about the nucleus that encompass the extranuclear emission detected in available images. The base image is the SDSS  observation of SDSS 2201+11, with the [O iii] clouds shown schematically.

Several of the two-sided clouds show near-symmetry in radial extent on opposite sides of the galaxy. This could reflect episodic activity on the light travel time-scale, although front–back geometric effects would generally break an exact symmetry.

6 KINEMATICS OF IONIZED GAS

Extended ionized regions around AGN may commonly be separated into kinematically quiescent components, such as would be seen for disc gas ionized from the nucleus, and outflow, with additional superimposed radial motion which might be manifested in a well-sampled velocity field as misalignment of the velocity field with the galaxy morphology if the superimposed velocity components are not spectrally resolved (Barbosa et al. 2009). In addition, for disturbed systems, tidal features may show motions decoupled from the disc itself. We consider here the information on gas kinematics provided by our spectra, noting that in most cases we sample only a single position angle through each galaxy.

Redshifts were measured for each pixel along the slit using Gaussian fitting in iraf. We show results for [O iii], H⁠, and when observed, H and [N ii]. Velocity errors are based on the propagation of photon statistics (Keel 1996).

Fig. 12 shows a selection of these velocity slices, relative to the nucleus in each case. Despite the angular offset from the edge-on disc, the gas velocities in SDSS 2201+11 are continuous with the pattern in the inner rotating section, and closely symmetric. Similarly, the emission clouds in NGC 5972 fall along an extrapolation of the inner-disc rotation curve (as traced by [O iii]). Despite its very disturbed morphology, UGC 11185 shows near-symmetry in kinematics, with a very strong velocity gradient crossing the nucleus.

Figure 12

Radial-velocity slices (shown relative to the nuclei) along our slit locations for selected hosts and AGN clouds. Nuclei are indicated by dark red circles. Lines used in calculating radial velocities from GCam and Lick data sets are indicated in the plots for NGC 5972, one of several objects observed at both locations though typically at slightly different position angles. In some cases, the extended emission follows the inner rotation closely even for extraplanar gas, as in SDSS 2201+11 and NGC 5972. Others are more chaotic, as expected for mergers. Particularly for Mrk 266, some structures show significant differences between the lines, indicating that we are seeing blends of multiple components with very different line ratios. Intensity slices in the continuum and lines are plotted across the bottom to show correlations between the location and velocity structure, scaled in flux to match [O iii] ⁠. Velocity errors are calculated as  from photon statistics following Keel (1996); in some cases, larger errors maybe appropriate from such factors as blending of multiple components.

Other systems in our sample show less ordered velocity slices. The gas in UGC 7342 at all radii has a single sense of motion on each side of the nucleus, but local departures have amplitudes up to 120 km s⁠. A central gradient in the Teacup (SDSS 1430+13) maybe reversed where the slit crosses its loop of emission. The kinematics in Mrk 883 and Mrk 739 are very disordered, as expected for a merging system. In Mrk 78, multiple components are seen in the inner few arcseconds, even in [Ne v] (Fischer et al. 2011).

The northern filament in Mrk 266 presents interesting kinematic behaviour, with a large and consistent velocity offset between [O iii], H and H⁠. This difference is seen in spectra from both spectrographs. A likely explanation is superposition of structures with quite different ionization states as well as velocity, so that the weighting of lines in our spectra, even though they are not separately resolved, gives different velocity centroids. The offset is close to 50 km s along the entire filament. Localized instances of similar mismatches between [O iii] and H velocities on one side of SDSS 2201+11 and possibly in NCG 5972. Outflows are typically inferred from blue wings on [O iii], but far from the nuclei where disc extinction is unlikely to be a major effect, velocity offsets of either sign could show outflows. One corollary of this distinction is that there exists a gas component of much higher excitation than implied by the ionization parameter we derived from ratios of total line flux at these locations, suggesting higher ionizing luminosities in these galaxies.

The relatively quiescent kinematics of most of these features may indicate a contrast in origin of the extended gas when compared with the radio-loud QSOs (Fu & Stockton 2009a,b). In their sample, modest star formation rates led them to suggest that much of the ionized gas was expelled from the system by the launch of powerful radio jets. The galaxies in our EELR sample are mostly radio-quiet (or at least radio-weak), as noted above. Also unlike their QSO sample, we see significant metallicity differences between the nuclei and EELRs, most strongly shown in the [N ii]/H ratio, again consistent with the EELR gas having an external tidal origin.

7 NOTEWORTHY SYSTEMS

From our results or previously reported data, several of these galaxies have interesting individual features.

The inner parts of the EELR in Mrk 78 have long been known (Pedlar et al. 1989, who detected much of the [O iii] extent we observe), and observed with HST in both imaging and spectroscopic modes (Capetti et al. 1994; Whittle & Wilson 2004). A detailed fit to the HST radial velocities explains the double-line profiles near the core without requiring a second AGN (Fischer et al. 2011); optical and near-IR line spectra suggest that the gas is photoionized from the nucleus with at most a very localized role for excitation by interaction with small-scale radio jets (Whittle et al. 2005; Ramos Almeida et al. 2006). Our data also show the complex spatial and velocity structure in the inner few arcseconds. The outer emission is spatially smooth and is measured to much larger radii in our spectra than in the initial SDSS imaging detection.

In Mrk 883, the blue line ratios indicate that it lies near the Seyfert/starburst boundary. Only in the red do weak broad H and [O i] definitely indicate an AGN. We do not detect a broad component at H⁠.

For NGC 4388, the SDSS images detect only a few inner knots of the extensive emission region revealed by, for example, Subaru imaging (Yoshida et al. 2002). Our spectra detect more of this structure. Our cone angle is estimated from the Subaru image. Detailed spectroscopy by Yoshida et al. (2004) confirms that this distant gas is photoionized by the AGN.

NGC 5252 has been considered the archetypal Seyfert galaxy with ionization cones. The implied energetics of the nucleus depend critically on the density in the ionized filaments. Our implied limits from photoionization balance via surface brightness in H are significantly greater than the values suggested from pressure balance with the galaxy’s hot ISM (Dadina et al. 2010), while we concur with the X-ray results that the ionization parameter remains roughly constant with radius among the ionized features. In turn, the interpretation of this behaviour depends on fine structure (much still unresolved) in the emission-line filaments, as seen in Hanny’s Voorwerp (Keel et al., in preparation).

Two objects in this sample appear to violate the usual pattern of ionization cones encompassing radio-source axes. NGC 5972 is the most radio-luminous of our galaxies and shows a typical double-lobed structure (Condon & Broderick 1988). The lobes are separated by 9.4 arcmin (330 kpc) in projection and are oriented near PA ⁠, quite different from the optical emission at PA ⁠. In this source, the most radio-powerful in our sample, the very different geometries of the line and radio emission make ionization from interaction with the radio plasma unlikely, and their near-perpendicular orientation is unlike the typical case for Seyfert galaxies (Wilson & Tsvetanov 1994). This could be explained if the ionization cones have extremely broad opening angles, or if the radio structure makes a dramatic and yet-unobserved twist on small scales. Similarly, Mrk 1498 is associated with a giant low-frequency double radio source (Röttgering et al. 1996), with projected separation 1.1 Mpc. In this case as well, the orientations of the emission-line structures and the large radio source differ strongly, by about ⁠.

SDSS 1430+13, the ‘Teacup’ AGN, is distinguished by a 5 kpc loop of ionized gas. The Faint Images of the Radio Sky at Twenty-one centimeters Very Large Array data at 20 cm show extended structure roughly coextensive with this feature, possibly indicating a related origin.

In both SDSS 2201+11 and SDSS 111100 (the latter observed spectroscopically but with emission smaller than our 10 kpc limit), the extranuclear clouds outshine the AGN itself in the [O iii] lines.

UGC 11185 shows a second, weaker set of emission-line components near the nucleus, peaking about 1.8 arcsec to the east along our slit, roughly 600 km s to the red of the main peaks, and including about 1/4 of the nuclear [O iii] flux within a  arcsec aperture.

In both Mrk 463 and Mrk 739, Chandra imaging has furnished evidence for double AGN (Bianchi et al. 2008; Koss et al. 2011). In both cases, the emission regions are much larger than the separation between AGN components, so we do not know whether the ionization is associated mostly with one or the other. More detailed [O iii] images could resolve this. However the flux sources are apportioned between components, one AGN in each system must have an ionizing/FIR ratio at least as high as our tabulated limit. Several earlier studies have noted the extended [O iii] emission around Mrk 463 (Mazzarella et al. 1991; Uomoto et al. 1993; Chatzichristou & Vanderriest 1995).

Wu et al. (2011) summarize polarimetric detections of ‘hidden’ broad-line regions in nearby AGN. Their list includes four of the nuclei in our sample: NGC 4388, NGC 5252, Mrk 78 and Mrk 463. Broad wings to the Balmer lines are seen in Mrk 266 (southwestern nucleus) and Mrk 739 (eastern nucleus), making them clear Sy 1 nuclei with ‘non-hidden’ broad-line regions. Weak wings are seen at H in Mrk 1498, which would then be classified as a type 1.8 object (Osterbrock 1977).

8 CONCLUSIONS

Volunteers in the Galaxy Zoo project have carried out a search for AGN-ionized gas clouds on large scales (10–40 kpc). This paper has documented the search and spectroscopic observations of candidates yielding 19 such features. These clouds were classified as AGN-photoionized based on their locations in the BPT line-ratio diagrams, strength of [Ne v] and He ii emission, and (when measurements are sufficient) modest electron temperatures  K, consistent with photoionization but not with shock heating. Most of the host galaxies show signs of interaction, suggesting that the extended ionized gas in many cases rises from tidal tails.

We consider upper and lower bounds to the luminosity of the AGN as it reached the clouds – lower limits from recombination and upper limits from density and ionization parameter. We compare these with the obscured luminosity estimated from FIR measurements; an excess in ionizing luminosity (or deficit in the FIR) could signal long-term variability of the AGN. The ratio of ionizing to obscured luminosity spans a wide range, from 0.02 to ⁠. Over a third of them (7/19) exceed unity, making this kind of energy deficit a common issue. Small values fit with an origin in obscured AGN, requiring only a small fraction of the extreme UV to escape. In contrast, large values may require a long-term fading of the AGN. An extreme case of this is represented by Hanny’s Voorwerp near IC 2497 (Lintott et al. 2009). In this object, the required ionizing flux indicates that the AGN has faded by a factor  within the last  years, sampling a time-scale on which we otherwise have no information. More detailed observations of this new sample, including pending X-ray measurements, could give statistics adequate to show how common such variations are.

An important use of this sample is in addressing the history of AGN luminosity – on what time-scales do episodes of high luminosity persist and fade? Broad arguments suggest that AGN episodes extend over spans comparable to the duration of a galaxy merger (several 10 years), if statistics associating excess AGN with strong interactions and mergers are representative. We note that establishing a link between galaxy interactions and AGN episodes has proven remarkably elusive, with the results depending on details of comparison sample selection and what kind of AGN is studied; as recent examples, Maia, Machado & Willmer (2003), Alonso et al. (2007) and Li et al. (2008) reached different conclusions – a null result, enhancement limited to certain kinds of AGN, or a weak overall enhancement of AGN – from similar analysis of nearby galaxy samples. Therefore, even within such long time spans, we have little information on how episodic the accretion and associated luminous output might be.

The relevant equation for time delay between radiation reaching us directly from the nucleus and that reprocessed in a cloud follows usage for light echos in ordinary reflection, except that here we are constrained by the location of gas so we deal with a constant observed radius  and an unknown angle  between the illumination direction and the plane of the sky; the long recombination times at low density impose a convolution with a non-trivial time span for response. With the geometry defined in Fig. 13, keeping  fixed by the observations means that the geometrical time delay  for observing reprocessed nuclear radiation depends on the viewing angle  (from the observer to the nucleus of the cloud, with a cloud along the line of sight at zero and increasing away from the observer) as given by

1

derived in the approximation of an infinite distance from the observer. Two-sided symmetric sets of clouds have progressively much longer differential delays when seen with their axis near the line of sight, so that a faded source in this regime should eventually be seen ionizing only the farside cloud. Our ability to reconstruct the actual distribution of  is hampered by an inner cut-off in  (10 kpc, so that the cloud detection is not hampered by galaxy starlight) and the lack of knowledge of the distribution of cloud extent from the nuclei. To be conservative, our calculations of ionizing luminosity (above) assume ⁠, the minimum possible distance for the nucleus and thus minimal ionizing luminosity,

Figure 13

Behaviour of the differential time delay  between radiation seen directly from the nucleus and reprocessed through a cloud at distance  from the nucleus, as a function of angle  from the line of sight, for a fixed value of the projected radius ⁠. The inset diagram shows the geometric quantities. Units of  are 

We might expect our sample to be complete at least for the lowest redshifts and highest surface brightnesses, but there are a few objects with selection priority as high as some of our cloud hosts for which we do not yet have confirming optical spectra. Of our 19 confirmations, 14 were found in both the targeted and serendipitous searches. Two were found only in the targeted search, and three were found in the serendipitous search. Of these three, one (SDSS J095559.88+395446.9) had no previous optical spectrum and could not have been included in the targeted sample.

A first hint as to characteristic time-scales comes for the relative numbers of galaxies with and without deficits in ionizing luminosity, since the ones with deficits in the energy budget would be seen during the appropriate delay time after fading of the nucleus. There is no obvious reason for this ratio to be biased in our sample, since the serendipitous survey was independent of the presence of an AGN, and even in the targeted search there are many AGN which are too weak to ionize the extended gas; in essence, given a luminous AGN, our selection is for objects with outlying gas available to be ionized. In a toy model where all objects’ delay times  are equal, the time-scale for the AGN to be at high luminosity before fading would be of order ⁠. From Table 5, our estimate is  when we divide the bright and faded groups at an ionizing/FIR ratio of 1.5. The projected extent of the clouds  from our [O iii] data is listed in Table 7; for the 19 galaxies in our sample, the mean value is 19 kpc with a median of 17. For a typical projected extent of 20 kpc, this becomes a range 25 000–175 000 years, taking the sample to populate values of  at this small sample size. For the luminosity range of Seyfert galaxies we have probed, the fading maybe an order of magnitude in ionizing luminosity, but this sample includes no cases in which we see AGN-ionized clouds around a galaxy with no optical trace of an AGN. IC 2497 (Lintott et al. 2009) must be extreme in this respect, having faded from a QSO to a borderline LINER/Sy 2 nucleus. As noted by Schawinski et al. (2010b), these time-scales are rapid compared to expectations from scaling up the behaviour of accretion discs around stellar-mass black holes, perhaps indicating that disc self-gravity enhances the growth of accretion instabilities.

There are several directions in which we can expand this study. In a ‘Dead Quasar Survey’, we are conducting [O iii] imaging of samples of luminous AGN hosts and galaxies without AGN signatures, to seek fainter (and possibly older) clouds than can be detected from the SDSS  images. H i selection should help pick out objects with tidal tails in suitable positions to be ionized at tens of kiloparsecs from the core; we are beginning with the Kuo et al. (2008) sample of Seyfert galaxies mapped in H i.

For the ‘faded’ galaxies in these samples, we are pursuing XMM–Newton and HST observations to clarify the obscuration towards the nucleus, seek any signs of outflow-induced star formation as seen in Hanny’s Voorwerp and refine estimates of the ionizing luminosity through the highest recombination-line surface brightness in the clouds.

Footnotes

1

See http://www.galaxyzooforum.org/index.php?topic=275014.0.

2

iraf is distributed by the National Optical Astronomy Observatory, which is operated by the Association of Universities for Research in Astronomy (AURA) under cooperative agreement with the National Science Foundation.

This work would not have been possible without the contributions of citizen scientists as part of the Galaxy Zoo project. We particularly thank Hannah Hutchins, Elizabeth Baeten, Massimo Mezzoprete, Elizabeth Siegel, Aida Berges, and users voyager1682002 and Caro, who each examined all of the galaxies in the targeted AGN sample, and in addition Christian Manteuffel, for assistance in compiling the list of SDSS AGN candidates. We are grateful to the following additional Galaxy Zoo participants who contributed to the targeted AGN search: Michael Aarons, Mark Ackland, AdrianusV, Aerial, alexob6, Daniela Alice, Norvan Allen, Anderstp, AndrewM, angst, Anjinsan, ARCHEV, artemiit, aryamwojn, astrobrainiac, astronomicom1, Markku Autio, Michelle Ayers, Elisabeth Baeten, R. Balick, Michael Balzer, Michael Derek Barnett, Kirsten Barr, Barbara Ann Barrett, David Bartlett, Coral Benham, Aida Berges, Mark Bernaldo, Chiheb Boussema, Gwen Brogmus, Dave Browne, brunochi, buddyjesus, David Burt, cadou, caliz83, Capella, Alice Carlsen, Caro, Jiri Cejka, Theodore J. Celaya Sr., chairstar, citisue3, Nick Clarke, Ana Claudia, cloud9, clua, David N. Cook, coral, Gemma Coughlin, Rob Cowhey, Penny Cox, Laurence Cuffe, cyprien, DancesWithWords, Darren, DarthKeribo, Lloyd Daub, daveb, dave3, david_mbe, david_nw, Michael C. Davis Jr., distel, Dobador, Shane Dobkins, drawm, Juliette Dowle, Elizabeth Duff, Graham Dungworth, dxjerlubb, dzd, Michael Easterly, echo, Alan Eggleston, Thomas Erickson, ErroneousBee, Falconet, firejuggler, frisken, Gino, glyphon, GNB080, gordhaddow, Michael Gronceski, grrower1, Michael Hand, Thomas Hardy, Hans Heilman, Steph Hill, Thomas Hobbs, Rick Holtz, Rob Hounsell, hrutter, Mikko Huovinen, IC1101, ixzrtxp, Nina Jansen, Alain Jaureguiberry, jayton, jczoehdo, jhyatt, David E. Johnson, Steve Johnson, David James Jones, John Kelly, khwdfnwit, Pat Kieran, KillerSkaarj, kiske1, knuid, kokdeblade, Anuradha Koratkar, Michi Kovacs, kzhndepnd Marc Laidlaw, laihro01, landersonzych, Lily Lau WW, lawless, Bill Lawrence, Kathleen Littlefield, Liz, Marc Lluell, Michael Lopez, lpspieler, luigimx, Lzsp, Michael MacIsaac, Christine Macmillan, Katie Malik, Steve Malone, mardo, Lelah Marie, Mark, Michael Marling, Stephanie R. Marsala, Mauro Marussi, marxpmp, Mark McCormack, Rob Mellor, Massimo Mezzoprete, mgn, Michaelr1415, MichaelRoberts, MichaelSangerTx, milkncookies, miraculix250, Elspeth Mitchell, Graham Mitchell, mlvgofjedxv, mothic, Mukund, mykyij, NGC3372, Julian Nicol, Rick Nowell, nrbeuw, Richard Oram, orion, oswego9050, pbungaro, Alice Peachey, Thomas Perraudin, Amanda Peters, Erica Pinto, plummerj, Jim Porter, Steven Porter, Richard Proctor, ptkypxdh, randa, RandyC, Kim Reece, Jessica Reeder, RelativisticDog2, Thomas Rickenbach, ripw, rjwarmv, rnjrchd, Michael Roberts, RobinMiller, Jim Robinson, roborali, Rona, Geoff Roynon, Paul Rutten, Rynnfox, S4CCG, Michael Salmon, salteV, Jeroen Sassen, second_try, Matt Sellick, sheba, Alice Sheppard, SianElderxyz, Nanne Sierkstra, Michael Simmons, SJPorter, skepticdetective, Stephen Sliva, Mark Smith, Sophie378, spat, Maria Steinrueck, stella13, stellar390, John H. Stewart, Doug Stork, sumoworm, superhouse, tadaemdg, Auralee Tamison, Chet Thomas, thom_2, Michael Thorpe, timchem, torres, Trixie64, Ramon van der Hilst, Marcel Veillette, Rob H.B. Velthuis, John Venables, Michael Viguet, vkhtmhfigou, Aileen Waite, David Walland, wbybjbpv, weezerd, Mark Westover, Julia Wilkinson, Nat T. Winston III, Windsmurf, wpubphx, xuhtjhc, xzxupfqjd, and Mairi Yates. We also thank the referee, who caught a mistake in calculating light-delay times and helped make the discussion more comprehensive. Jean Tate helped to untangle some issues of participant discovery order.

WCK acknowledges support from a Dean’s Leadership Board faculty fellowship. CJL acknowledges funding from the Leverhulme Trust and the STFC Science in Society Programme. Galaxy Zoo was made possible by funding from a Jim Gray Research Fund from Microsoft and the Leverhulme Trust. SDC participated through the SARA Research Experiences for Undergraduates programme funded by the US National Science Foundation. This research is based on observations with AKARI, a JAXA project with the participation of ESA. We thank the Lick Observatory staff for their assistance in obtaining the data. Support for the work of KS was provided by NASA through Einstein Postdoctoral Fellowship grant number PF9-00069 issued by the Chandra X-ray Observatory Center, which is operated by the Smithsonian Astrophysical Observatory for and on behalf of NASA under contract NAS8-03060.

Funding for the creation and distribution of the SDSS Archive has been provided by the Alfred P. Sloan Foundation, the Participating Institutions, the National Aeronautics and Space Administration, the National Science Foundation, the US Department of Energy, the Japanese Monbukagakusho and the Max Planck Society. The SDSS website is http://www.sdss.org/. The SDSS is managed by the Astrophysical Research Consortium (ARC) for the Participating Institutions. The Participating Institutions are the University of Chicago, Fermilab, the Institute for Advanced Study, the Japan Participation Group, the Johns Hopkins University, Los Alamos National Laboratory, the Max-Planck-Institute for Astronomy (MPIA), the Max-Planck-Institute for Astrophysics (MPA), New Mexico State University, Princeton University, the US Naval Observatory and the University of Washington.

This research has made use of the NASA/IPAC Extragalactic Database (NED), which is operated by the Jet Propulsion Laboratory, Caltech, under contract with the National Aeronautics and Space Administration.