Imaging the elusive H-poor gas in the high ADF planetary nebula NGC 6778
Jorge García-Rojas, Romano L. M. Corradi, Hektor Monteiro, David Jones, Pablo Rodríguez-Gil, Antonio Cabrera-Lavers
aa r X i v : . [ a s t r o - ph . S R ] J un Draft version April 24, 2018
Preprint typeset using L A TEX style emulateapj v. 5/2/11
IMAGING THE ELUSIVE H-POOR GAS IN THE HIGH adf
PLANETARY NEBULA NGC 6778
Jorge Garc´ıa-Rojas and Romano L. M. Corradi and Hektor Monteiro and David Jones and PabloRodr´ıguez-Gil and Antonio Cabrera-Lavers Draft version April 24, 2018
ABSTRACTWe present the first direct image of the high-metallicity gas component in a planetary nebula(NGC 6778), taken with the OSIRIS Blue Tunable Filter centered on the O ii λ ii ORLs is concentrated in the central parts of the planetary nebula and is notspatially coincident either with emission coming from the bright [O iii ] λ α recombination line. From monochromatic emission line maps takenwith VIMOS at the 8.2m Very Large Telescope, we find that the spatial distribution of the emissionfrom the auroral [O iii ] λ ii ORLs but differs from nebular [O iii ] λ ii emission and the differences with the [O iii ] and H i emission profiles areconsistent with the presence of an H-poor gas whose origin may be linked to the binarity of the centralstar. However, determination of the spatial distribution of the ORLs and CELs in other PNe, and acomparison of their dynamics is needed to further constrain the geometry and ejection mechanism ofthe metal-rich (H-poor) component and hence, understand the origin of the abundance discrepancyproblem in PNe. Subject headings: planetary nebulae: individual — ISM: abundances — binaries: close INTRODUCTION
In photoionized nebulae –both H ii regions and plane-tary nebulae (PNe)– optical recombination lines (ORLs)provide abundance values that are systematically largerthan those obtained using collisionally excited lines(CELs) (see e. g. Peimbert & Peimbert 2006; Liu 2012,, and references therein). This is known as the abun-dance discrepancy problem . It is one of the major un-resolved problems in nebular astrophysics, being knownof for more than seventy years (Wyse 1942), and hasfar-reaching consequences on the measurement of abun-dances throughout the Universe, most often measuredusing CELs from ionized gas.The abundance discrepancy factor ( adf ) is defined asthe ratio between the abundances derived from ORLsand CELs and is usually between 1.5 and 3 (see e. g.Garc´ıa–Rojas & Esteban 2007; Liu 2012), but in PNeit has a significant tail extending to much larger val-ues. It has recently been shown that the largest abun-dance discrepancy factors (up to ∼ ∼ ∼ Instituto de Astrof´ısica de Canarias, E-38200, La Laguna,Tenerife, Spain Universidad de La Laguna. Depart. de Astrof´ısica, E-38206,La Laguna, Tenerife, Spain GRANTECAN, Cuesta de San Jos´e s/n, E-38712 , Bre˜naBaja, La Palma, Spain Instituto de F´ısica e Qu´ımica, Universidade Federal de Ita-jub´a, Av. BPS 1303-Pinheirinho, 37500-903, Itajub´a, Brazil highly enhanced content of heavy elements (which is thecause of the cooling) where only ORLs form. As thisdual nature is not predicted by mass loss theories, theseresults have added a new, unexpected ingredient to theabundance discrepancy problem: very high discrepancyfactors should be explained in the framework of (interact-ing) binary evolution. Some of the proposed explanationsare naturally linked to binarity, such as for instance high-metallicity nova-like ejecta (Wesson et al. 2008). Otherappealing explanations involve the presence of planetarydebris that survived the whole evolution of the centralstar, or the tidal destruction, accretion and ejection ofJupiter-like planets (see the discussion in Corradi et al.2015), but it is difficult at this stage to favor any scenariobecause observational constraints are still very scarce.The existing long-slit studies point out that the plasmaemitting in O ii ORLs (the brightest metal ORLs in theoptical spectrum) is generally more concentrated in thecentral parts of the nebulae than the CEL emitting zonesof the same ion (e. g. Liu 2012; Corradi et al. 2015).However, no direct imaging of the gas component wherethe bulk of the metal ORLs is emitted has ever beenattempted, mainly because O ii ORLs are faint and suit-able filters are generally not available.NGC 6778 ( α = 19 h m s , δ = 01 ◦ ′ ′′ ,PN G034.506.7) was found to show unusually strongORLs reflecting a high adf value (average value ∼ ii ORL and [O iii ] Garc´ıa-Rojas et al.CELs for the first time. OBSERVATIONS AND DATA REDUCTION
OSIRIS-GTC data
We obtained narrow band imagery of NGC 6778 usingthe Optical System for Imaging and Low Resolution In-tegrated Spectroscopy (OSIRIS, Cepa et al. 2000, 2003)on the 10.4m Gran Telescopio Canarias (GTC), at theObservatorio del Roque de los Muchachos on La Palma.The OSIRIS detector consists of a mosaic of two MarconiCCDs, each with 2048 × . × . , giving a platescale of 0.127 arcsec/pix. We selected the 2 × ii emission at 4649+50 ˚A, the BTF tuning wasdisplaced to 465.4 ± ii emission when the target islocated 1 arcmin from the TF optical center. There wasno need to scan the TF as the nebula is sufficiently small( <
25 arcsec diameter excluding the faintest collimatedoutflows) that the wavelength variation from one side tothe other of the nebula is < × . Sky emission was removed by producing an im-age containing only the sky emission from the series ofdithered images. This sky image was then subtractedfrom the original images of the source, resulting in im-ages cleaned from any background and sky emission. Tosave telescope time, owing to the large exposure timesneeded to obtain the desired S/N ratio in the O ii lines,no attempt was made to subtract the negligible nebularcontinuum.Additionally, we show [O iii ] λ α and[N ii ] λ iii ] image pixel scale(0.189 arcsec pix − ) and astrometrized using the 2MASScatalog and the script designed by Sanabria (2015). VIMOS-VLT data
NGC 6778 was observed with VIMOS (Le F`evre et al.2015), on the ESO-VLT UT3 Melipal on the night of2007 September 14. VIMOS is equipped with an inte-gral field unit containing 6400 fibers and has a change-able scale on the sky that was set to 0.33 arcsec per fiber.This provided a sky coverage of 13.2 × with IRAF is distributed by the National Optical Astronomy Ob-servatory, which is operated by the Association of Universities forResearch in Astronomy (AURA) under a cooperative agreementwith the National Science Foundation. an image formed by an array of 40 ×
40 fibers. We ob-tained observations in the VIMOS high-resolution mode,providing a reciprocal dispersion of 0.6 ˚A pix − and ausable range from 3900 to 7400 ˚A adding the blue andred grisms of the spectrograph. The object was observedwith an exposure time of 300 s in both the red and blueconfigurations. The reduction was performed with theVIMOS pipelines available at the instrument website .For sky subtraction we observed fields offset by 40 arc-sec. The sky level was estimated as an average for eachpixel along the dispersion axis and then subtracted fromthe object data cube. We also applied a correction forthe differential atmospheric extinction using the centralstar of NGC 6778 in the data cube and performing thenecessary shifts along the dispersion axis. To extract thefinal VIMOS emission line maps we used a set of IDLroutines which fit Gaussian profiles to the lines of inter-est in each spaxel in the data cube. In Fig. 1 we showthe area covered by the VIMOS IFU by overlaying theVIMOS map taken in the light of H α on an ALFOSC-NOT narrow-band image centered on the same line fromGuerrero & Miranda (2012). The IFU was positioned tocover the central 13.2 × of the nebula. Fig. 1.—
The VIMOS H α map (color image) overlaid onALFOSC-NOT narrow band image showing the area covered bythe VIMOS IFU (13.2 × ). The ALFOSC-NOT imagewas taken from Guerrero & Miranda (2012). RESULTS AND DISCUSSION
OSIRIS-GTC and ALFOSC-NOT images
In the upper-left panel of Fig. 2 we show the OSIRIS-GTC BTF image centered on the O ii λλ from theselines. In the upper-right panel, we show the narrow-band [O iii ] λ α (left) and [N ii ] λ ii ORLs are contaminated by fluorescent N iii lines and/or maging of NGC 6778 3 Fig. 2.—
OSIRIS-GTC Tunable-filter image of NGC 6778 in the O ii iii ] 5007 CEL (upper right), H α (lower left) and [N ii ] 6548 (lower right). All the images wereregistered to the [O iii ] image pixel scale (0.189 arcsec/pixel) and astrometrized using the 2MASS catalog. It is worth mentioning that O ii and [O iii ] emission comes from the same ion: O . C iii ORLs. In their deep VLT-FORS2 spectrum ofNGC 6778, Jones et al. (2016) did not report any con-tamination in the wavelength range covered by our tun-able filter. In Fig. 3, we show a section of the observedspectrum of NGC 6778 around the multiplet 1 of O ii . Itis clear that there is no emission from the C iii λ iii .Therefore, all the emission covered by the tunable filtercomes purely from O ii recombination lines.From inspection of Fig. 2, it is clear that the [O iii ] λ α emissions share a very similar spatialdistribution in the central parts of NGC 6778 (see alsothe discussion in Guerrero & Miranda 2012). On theother hand, the [N ii ] λ ii ]-bright equatorialring aligned along a line close to the east-west direction.This [N ii ] CEL innermost region is characterized by thepresence of filamentary low-ionization structures (LIS).Their origin, however, remains unknown but has beenlinked to post-CE nebulae through photoionization ofneutral material deposited in the orbital plane duringthe CE phase (see Miszalski et al. 2011; Manick et al.2015, and references therein). Fig. 3.—
Section of the VLT-FORS2 observed spectrum (blackline) of NGC 6778 presented in Jones et al. (2016) where the O ii ORLs, as well as some N iii lines, have been highlighted. The TFwidth is shown as a red line.
The most important finding from inspection of Fig. 2 isthat in NGC 6778 the location of the O ions producingthe O ii ORLs at 4649+50 ˚A does not match the spatialdistribution of the same ions emitting in the [O iii ] λ ii ORL emission is concentrated inside the[O iii ] CEL emission or H i emission. This is the first timethat the different distribution of O ii ORL and nebular Garc´ıa-Rojas et al.[O iii ] CEL emission has been shown using direct imag-ing. This finding clearly supports the hypothesis of theexistence of two separate plasmas, with the additionalindication that they are not well mixed, perhaps becausethey were produced in distinct ejection events. Previousworks based on long-slit and 2D spectrophotometry inpartial regions of PNe found similar results. Hence, thisbehaviour seems to be a standard characteristic of theORL emission in PNe with high adf s (see e. g. Liu et al.2000; Tsamis et al. 2008; Corradi et al. 2015).
VIMOS monochromatic emission line maps
We have analyzed the morphology of the emission com-ing from various other lines on monochromatic emissionline maps taken with VIMOS-VLT. In the left columnof Fig. 4, we show the C ii λ ii λλ and O recombination emission. Although theC ii emission has a rather low contrast relative to thebackground, its global morphology resembles the O ii λλ iii ] λ iii ] λ ii λ iii ] λ iii ] λ iii ] λ iii ] λ ii ORL emission. This causes a radially increasing[O iii ]5007/4363 flux ratio, which may be interpretedas being due to a markedly negative electron temper-ature gradient (Jones et al. 2016, estimated a negativegradient of more than 1000 K, see their Fig. 5). Thiseffect has been found in other PNe (see Liu et al. 2000;Garnett & Dinerstein 2001, for the PNe NGC 6153 andNGC 6720, respectively). However, the presence of ex-tremely high-density clumps ( n e > cm − ) in the in-ner regions would cause the quenching of nebular [O iii ]lines in the central part of the PN owing to collisional de-excitation effects. The effect of high-density ionized gasin the determination of chemical abundances from ORLsand CELs has been studied in detail in i. e. the Orionnebula, by Tsamis et al. (2011) and Mesa-Delgado et al.(2012). This would affect the determination of physicalconditions ( n e and T e ) and abundances from CELs, cre-ating a spurious temperature gradient. Moreover, if theseclumps are H-poor and only contribute a few percent ofthe total H mass of the PN, as it has been proposed forPN, they would not be seen in H i images. Therefore, ourobservations support the presence of a metal rich, densecomponent in the inner regions of NGC 6778, where ORLemission mainly form, in addition to the ”standard”, lowdensity and H-rich gas component where CELs are emit-ted. Other than its location in the inner nebular regions,which may indicate a younger origin, the morphology ofthe ORL emitting gas does not provide any other cluesto its origin. Detailed 3D photoionization models aretherefore needed to fully constrain the nebula and shedfurther light on its formation.All these results are in very good agreement with whatwas found by Jones et al. (2016), although they were lim-ited to a 1-D portion of the nebulae as selected in their long-slit FORS2 observations. C II 4267 − . . . . . . . . .
10 5 10 15 20 25 30 3505101520253035
O II 4649+50 . . . . . . . . .
35 0 5 10 15 20 25 30 3505101520253035 [O III] 4363 . . . . . . . . .
350 5 10 15 20 25 30 3505101520253035 [O III] 5008 spaxel s pa x e l Fig. 4.—
VIMOS-VLT emission line maps in several emissionlines. Left column: C ii λ ii λλ iii ] λ λ × and the orientation is north up and eastleft. Position of the central star is marked with a white cross. Finally, some kinematical studies point to the pres-ence of a colder plasma where ORLs are preferentiallyemitted. From very high spectral resolution observa-tions (R=150,000) of two moderate to high adf
PNe(NGC 7009 with adf ∼ adf ∼ ii ORLs were significantlysmaller than that shown by [O iii ] CELs. They alsofound that auroral and nebular [O iii ] lines in NGC 6153showed different velocity profiles, which implies largetemperature fluctuations. Such large temperature fluc-tuations can also contribute to the observed adf (seeGarc´ıa–Rojas & Esteban 2007, and references therein).Richer et al. (2013) made a detailed kinematical studyof ORLs and CELs in a limited region of PN NGC 7009.They found that ORLs exhibit different kinematics toCELs but no clear conclusion could be drawn because ofthe limited spatial coverage. Otsuka et al. (2010) ana-lyzed high resolution (R ∼ adf ( ∼ ii and Ne ii ORLs expansion velocities lower than that of[O iii ] and [Ne iii ] CELs, which they attributed to differ-ent thermal motions. It is worth mentioning that in thecase of NGC 6778, where the O ii ORL emission comesfrom the innermost zones of the PN, the kinematics canbe affected by the expansion velocity field combined withthe ionization stratification of the gas in the PN. Thiswould result in narrower O ii ORLs compared to nebu-lar [O iii ] CELs. CONCLUSIONS
We present direct imaging in the light of the O ii λλ emission comingfrom these lines in NGC 6778 is more centrally concen-trated than the emission coming from either the strongest[O iii ] λ α RL. Also, from 2D recon-structed VIMOS monochromatic emission line maps, wemaging of NGC 6778 5find that the auroral [O iii ] λ ii ORLs emission but not the nebular [O iii ]emission. This may be due to presence of a high-density,H-poor gas component in the inner regions of the neb-ula, which could be responsible for the bulk of the O ii emission and cause the abundance discrepancy. How-ever, we do not have enough information to generalizethe scenario of metal-rich inclusions to explain the bulkof low-to-moderate ADF PNe or H ii regions. Moreover,the effect of high-velocity flows, such as in HH objects,and/or high-density inclusions, such as proplyds, has def-initely something to say in the abundance discrepancyproblem in H ii regions.The physical origin of the enhancement of the O ii emission in the central parts of this PN is still elusive,but should almost certainly be linked to the binary evo-lution of its central star. Additional observations areneeded to put this important result on solid ground andfurther progress in our understanding of the origin ofthe phenomenon. Determining the spatial distribution of the ORLs and CELs, and confronting their dynamicsin several PNe with high adf values would help constrainthe geometry and ejection mechanism of the metal rich(H-poor) component.This work is based on observations obtained with the10.4m GTC, operating on the island of La Palma at theSpanish Observatories of the Roque de Los Muchachosof the Instituto de Astrof´ısica de Canarias. We thankthe anonymous referee for his/her comments. This re-search has been supported by the Spanish Ministry ofEconomy and Competitiveness (MINECO) under grantsAYA2012-35330, AYA2011-22614, and AYA2012-38700.JGR acknowledges support from Severo Ochoa excellenceprogram (SEV-2011-0187) postdoctoral fellowship. PRGwas supported by a Ram´on y Cajal fellowship (RYC2010-05762). JGR acknowledges fruitful discussions with C.Esteban. Facilities: