Hunting for Planetary Nebulae toward the Galactic Center
Jihye Hong, Janet P. Simpson, Deokkeun An, Angela S. Cotera, Solange V. Ramírez
aa r X i v : . [ a s t r o - ph . GA ] F e b Draft version February 22, 2021
Typeset using L A TEX twocolumn style in AASTeX63
Hunting for Planetary Nebulae toward the Galactic Center
Jihye Hong, Janet P. Simpson, Deokkeun An, Angela S. Cotera, and Solange V. Ram´ırez Department of Science Education, Ewha Womans University, 52 Ewhayeodae-gil, Seodaemun-gu, Seoul 03760, Korea;[email protected] SETI Institute, 189 Bernardo Avenue, Mountain View, CA 94043, USA Carnegie observatory, 813 Santa Barbara Street, Pasadena, CA 91101, USA
Submitted to AJABSTRACTWe present near-infrared (IR) spectra of two planetary nebula (PN) candidates in close lines of sighttoward the Galactic center (GC) using GNIRS at Gemini North. High-resolution images from radiocontinuum and narrow-band IR observations reveal ring-like morphologies of these objects, and theirmid-IR spectra from the
Spitzer
Space Telescope exhibit rich emission lines from highly excited speciessuch as [S IV ], [Ne III ], [Ne V ], and [O IV ]. We also derive elemental abundances using the Cloudysynthetic models, and find an excess amount of the s -process element Krypton in both targets, whichsupports their nature as PNe. We estimate foreground extinction toward each object using near-IRhydrogen recombination lines, and find significant visual extinctions ( A V > . ± . . ± . ∗ (projected distances .
20 pc) make it highly probablethat these objects are the first confirmed PNe objects in the nuclear stellar disk. The apparent scarcityof such objects resembles the extremely low rate of PN formation in old stellar systems, but is in linewith the current rate of the sustained star formation activity in the Central Molecular Zone.
Keywords:
Planetary nebulae(1249), Milky Way Galaxy(1054), Interstellar Line Emission (844), Galac-tic center (565) INTRODUCTIONMost stars with initial masses less than ∼ M ⊙ evolveinto planetary nebulae (PNe) at the end of their life-times (see Balick & Frank 2002, and references therein),as long as their progenitor masses are large enough(see Jacoby et al. 1997). In our Galaxy, a large num-ber of PNe ( ∼ , Corresponding author: Deokkeun [email protected] is an order of magnitude lower than expected from pop-ulation synthesis models (e.g., Moe & De Marco 2006),and the majority of PN populations in the Milky Way re-main to be discovered. The heavy dust obscuration nearthe Galactic plane is likely the main cause of such dis-crepancy (e.g., Jacoby & Steene 2004; Miszalski et al.2008; Parker et al. 2012).In this context, the absence of PNe in the nu-clear bulge of the Milky Way (Serabyn & Morris 1996;Launhardt et al. 2002) can be understood to be the re-sult of the large amount of foreground dust towardsthe Galactic center (GC; A V ∼ ∼ . × M ⊙ (Launhardt et al. 2002), which is approxi-mately 10 times less massive than the kpc-scale classicalbulge. The stellar populations in the nuclear bulge are Hong et al. predominantly old (e.g., Nogueras-Lara et al. 2020), butmay have distinct chemical properties from the classicalbulge (e.g., Schultheis et al. 2020). Most of these starsare confined to the nuclear stellar disk, a rotating disk ofstars around Sgr A ∗ . The nuclear stellar disk spatiallyoverlaps with the Central Molecular Zone (CMZ; seeMorris & Serabyn 1996, and references therein), a mas-sive reservoir of molecular gas clouds with a diameter of ∼
500 pc and a total cloud mass of ∼ × M ⊙ (Dahmen et al. 1998; Tsuboi et al. 1999). Sustainedstar formation activities are observed throughout the re-gion (e.g., An et al. 2011; Longmore et al. 2013). Giventhat both the Galactic bulge and thin disk harbor anoticeable number of PNe, the lack of PNe in the nu-clear bulge poses a challenge to our understanding ofPN formation, stellar populations and evolution in theinner-most region of the Milky Way.Recently, we have identified two objects,SSTGC 580183 (G359.963-0.120 or G-0.037-0.120)and SSTGC 588220 (G0.098-0.051), as candidate PNe(Simpson 2018) while analyzing mid-IR spectra of com-pact IR sources toward the CMZ (An et al. 2009, 2011).They show high excitation lines such as [Na III ] 7 . µ m,[Ne V ] 14 . µ m and 24 . µ m, and [O IV ] 25 . µ m.These lines are not observed in typical H II regions,but are often seen in PNe with an extremely hot(3 × . T eff . × K) central object (e.g.,Osterbrock & Ferland 2006; Peimbert et al. 2017). Thetwo objects have also been observed in high-resolutionradio observations and with near-IR narrow-band fil-ters, revealing a ring-like nebulosity (Wang et al. 2010;Zhao et al. 2020).In this paper, we report near-IR spectroscopic follow-up observations of these two objects using the Gem-ini Near-Infrared Spectrograph (GNIRS; Elias et al.2006a,b) at the Frederick C. Gillett Gemini North Tele-scope. Owing to a paucity of PNe toward the GC, bothobjects are unique, providing an opportunity to explorestellar evolution and chemical characteristics of stellarpopulations in the inner region of the Galaxy. Theirprobable location in the CMZ is an interesting aspect ofthis study, since the massive reservoir of molecular gasin the CMZ is known to maintain the most active starforming activities in the Milky Way ( ∼ . M ⊙ yr − ;An et al. 2011; Longmore et al. 2013); they can be usedto study chemical and kinematical properties of starsthat are distinct from those of the other Galactic com-ponents.The primary goal of this study is to confirm the natureof these nebular sources as PNe and to characterize theirchemical properties based on near-IR and mid-IR spec-tra. The near-IR spectra are particularly useful for this purpose, because hydrogen recombination lines can beused to constrain foreground extinction, which is neces-sary to derive elemental abundances of ionic species frommid-IR spectra. This paper is organized as follows. In §
2, we describe observing tactics and data reduction.In §
3, we present GNIRS spectra, derive foregroundextinction from hydrogen recombination lines, and puta constraint on their distances from a comparison withsize versus surface brightness relations of PNe. We con-duct a joint analysis of near-IR and mid-IR spectra in § § OBSERVATIONS AND DATA REDUCTION2.1.
Candidates
Both SSTGC 580183 ( α = 17h 46m 0 . δ = − ◦ ′ . ′′ ) and SSTGC 588220 ( α =17h 46m 2 . δ = − ◦ ′ . ′′ ) were originally identified as com-pact (within a 2 ′′ beam) sources in the Spitzer
SpaceTelescope (Werner et al. 2004) Infrared Array Camera(IRAC; Fazio et al. 2004) images (Ram´ırez et al. 2008).They were targeted for follow-up observation using theInfrared Spectrograph (IRS; Houck et al. 2004) onboard
Spitzer (An et al. 2009, 2011), as part of a search formassive young stellar objects in the CMZ. There areno parallax measurements, but their proximity to theGC (with a projected angular distance of ∼ ′ ) wasdeemed as a supporting piece of evidence for their po-tential membership in the CMZ, which covers a regionof ∼ ◦ × . ◦ centered at the GC.Figure 1 shows mid-IR spectra of each target, takenusing the high-resolution modules of IRS (see An et al.2009, 2011, for more details). The background emissionfrom surrounding clouds in the CMZ was subtractedfrom the spectrum of each target using a set of back-ground spectra (see below). The continuum of the back-ground spectra is characterized by warm-dust emissionand strong emission from polycyclic aromatic hydrocar-bons at 6 .
2, 7 .
7, 11 .
3, 12 .
7, and 16 . µ m; nonetheless,they are a factor of ∼ µ m . λ . µ m), and ∼ µ m . λ . µ m). There are also anumber of strong emission lines seen in the backgroundspectra, which originate from diffuse ionized gas andphoto-dissociation regions (PDRs) in molecular cloudsadjacent to each target.The mean background spectrum of the high-resolutionmodules was constructed using IRS observations at fourcarefully chosen locations ( ∼ ′ away in each direction)with the same instrument setup. Figure 2 shows emis-sion lines observed in the IRS spectra after the back- lanetary Nebulae toward the Galactic Center Figure 1.
Background-subtracted
Spitzer /IRS spectra of PN candidates. Spectra from high-resolution modules are shown byblue lines. For SSTGC 580183, low-resolution spectra are displayed by orange lines. ground subtraction. Strong PDR emission lines fromH S(2) 12 . µ m and H S(1) 17 . µ m are negligibleafter background subtraction. On the other hand, whilesome forbidden emission lines from highly excited ionspecies such as [S IV ] 10 . µ m and [O IV ] 25 . µ mare not visible or very weak in the background spec-tra, they are clearly seen in the background-subtractedspectra. SSTGC 588220 additionally exhibits emissionfrom [Ar V ] 13.10 µ m, [Mg V ] 13.52 µ m, and [Fe VI ]14.77 µ m. Such lines are commonly observed in PNefrom gas ionized by a source of temperature ∼ K,rather than from typical H II regions ionized by massiveOB stars (e.g., Osterbrock & Ferland 2006).2.2. Observations and Data Reduction
Medium-resolution ( R ∼ . µ m to2 . µ m) spectra of the two PN candidates were takenusing Gemini/GNIRS. A 0 . ′′ × ′′ slit with the 110 . [O IV ] 25 . µ m is saturated in the high-resolution IRS spec-trum of SSTGC 588220, for which we used the low-resolutionIRS spectrum in the modeling (see below). mm − grating was used in cross-dispersed (XD) modewith the short-blue camera (0 . ′′ per pixel); the slitwidth was set to approximately match the seeing duringthe observation. Both targets were observed in queueobserving runs; SSTGC 580183 was observed twice onMarch 11 and April 16 in 2016, respectively, with theformer observation taken with a non-zero parallactic an-gle (see below). SSTGC 588220 was observed on March19, 2016. The sky was mostly clear and the seeing wasstable (0 . ′′ –0 . ′′ ) during the observations.Figure 3 displays locations of GNIRS slits forSSTGC 580183, overlaid on top of the UKIDSS(Lawrence et al. 2007) K -band image (top panel) and aJansky Very Large Array (JVLA) 5 . ∼ ′′ along the major axis. The GNIRS slit wasput nearly at the center of the ring.As displayed in Figure 4, SSTGC 588220 was observedin the high-resolution ( ∼ . ′′ ), Paschen- α imaging Hong et al.
Figure 2.
Emission lines in the
Spitzer /IRS spectra of the PN candidates. Original line emissions from each target are shownby the orange-shaded red lines, while the blue lines show emission lines extracted at four nearby locations that do not containbright point sources in the slit. Line centers are marked by a vertical grey stripe. Each panel spans a wavelength range of0 . µ m (each tick mark covers ∆ λ = 0 . µ m). survey using the Hubble Space Telescope (Wang et al.2010). It reveals a bipolar structure of strong emissionnear the center, and a fainter, elongated ring-like struc-ture, which is orthogonal to the axis of the inner lobes.The observed morphology is reminiscent of a quadrupo-lar PN (e.g., Manchado et al. 1996). The center of theslit for SSTGC 588220 was positioned on the bright-est part of the nebula, which is the eastern lobe in the
HST /NICMOS image. The position angle at the time ofobservation was set to the parallactic angle; fortuitously,the slit included not only the eastern rim of the centralregion but also a part of the northern (fainter) ring.Because of the high source density in the GC, theoff-source slit position was carefully chosen near eachtarget, and a background spectrum was obtained in anon-off sequence. A total on-source exposure time was lanetary Nebulae toward the Galactic Center Figure 3.
The UKIDSS K -band image (Lawrence et al.2007, top) and the JVLA 5 .
300 sec each night. We began reducing GNIRS data byremoving pattern noise and flat-field correction using the
Figure 4.
HST /NICMOS Paschen α (F187) image(Wang et al. 2010) of SSTGC 588220. North is to the top,and east to the left. The long rectangle indicates the locationof the slit on 19 March 2016. Gemini
IRAF packages . For each set of data frames, weadjusted bias levels by matching median values of pixelsbetween spectral orders. We then followed the standarddata reduction procedure for GNIRS to correct for theorder distortion, and perform wavelength calibration.Figure 5 shows source profiles along the slit. ForSSTGC 580183 (top panel), a striking bimodal struc-ture is seen in the line emission profiles; we attributethis to the ring-like morphology observed in the radioimage (see the bottom panel of Figure 3). The bimodalstructures in strong lines, such as Br γ and H I . µ m), are similar, and are also seen in the weakerlines, although to a lesser degree. To collect light fromthe entire line emission structure, we used a 4 . ′′ -wideaperture for spectral extraction.On the other hand, SSTGC 588220 (middle and bot-tom panels of Figure 5) shows a single peak, althoughthe slit contained extended emission and its dispersionis larger than the seeing (FWHM ∼ . ′′ –0 . ′′ ). Mostnotably, [Fe II ] 1.644 µ m from SSTGC 588220 is signifi-cantly brighter in the northern ring, and is displaced by ∼ . ′′ from the central emission (bottom panel). Otherlines also show emission at the location of the northern IRAF is distributed by the National Optical Astronomy Obser-vatory, which is operated by the Association of Universities forResearch in Astronomy (AURA) under a cooperative agreementwith the National Science Foundation. Hong et al.
Figure 5.
A flux distribution of individual lines along theslit for SSTGC 580183 (top) and SSTGC 588220 (middleand bottom). Line fluxes summed over 6 ˚A are normalizedwith respect to the maximum value of Br γ emission. Thedot-dashed and solid lines in the top panel show line fluxesfrom the first and the second run on this target, respectively.North is to the left. ring, but its strength is significantly weaker than the oneobserved in the slit center. We used a 3 . ′′ aperture toextract the spectra, collecting both the central emissionand the emission from the northern ring. For [Fe II ],the emission line spectrum was extracted at the placewhere the line emission is strongest. We corrected fortelluric absorptions using the Spectral Extraction Tool Table 1.
Line Flux Measurements from GNIRS Spectra
Wavelength Flux (10 − erg s − cm − )Line ( µ m) SSTGC 580183 SSTGC 588220He I . < .
256 5 . ± . β . < .
256 6 . ± . . < .
256 1 . ± . II ] 1 . < .
256 1 . ± . . < .
256 0 . ± . I . . ± .
389 10 . ± . I . . ± .
020 1 . ± . I . < .
256 0 . ± . . . ± .
059 0 . ± . I . < .
256 0 . ± . I . . ± .
041 2 . ± . γ . . ± .
395 31 . ± . II . . ± .
022 3 . ± . III ] 2 . . ± .
112 0 . ± . h v helio r i +62 ± − − ± − h v LSR r i +72 ± − − ± − Note —Upper limits are shown as a 3 σ detection limit. (Spextool; Cushing et al. 2004) v4.1 , based on nightlyobservation of the telluric standard star (HD 157918). RESULTS3.1.
Line Flux Measurements
Our GNIRS spectra of SSTGC 580183 andSSTGC 588220 are shown in Figures 6 and 7, respec-tively. The observed spectra contain several hydro-gen and helium recombination lines, as well as [Fe II ]1.644 µ m, [Kr III ] 2.199 µ m, H µ m,and S(2) 2.0338 µ m. Among these, Br γ (2 . µ m)and Pa β (1 . µ m) are of particular importance inthis study, since they provide a strong constraint onthe amount of foreground extinction toward each ob-ject. For SSTGC 580183, only the K -band spectrumfrom the highest order is shown, because most lines areessentially undetected (see individual panels at the bot-tom). Even [Fe II ] shows a possible absorption feature,which might reflect the widespread [Fe II ] emission fromthe interstellar medium (ISM) around the source (seeAn et al. 2013; Simpson 2018). Pa β was detected at2 . σ in March, but it was not seen in the April data dueto a slightly lower signal-to-noise ratio (SNR). The weakPa β is indicative of its large foreground extinction (seebelow).Table 1 contains the line fluxes measured from theGNIRS spectra. They were measured by fitting a Gaus-sian function after subtracting the local continuum usinga straight line. For blended lines such as He I (2 . µ m)and Br γ , two Gaussians were used to simultaneously fit lanetary Nebulae toward the Galactic Center Figure 6.
GNIRS spectra of SSTGC 580183. The other spectral orders are omitted, because none of the emission lines aredetected at shorter wavelengths. both emission lines. The uncertainties represent the dif-ference between the best-fitting Gaussian and a directintegration of the line. A 3 σ upper limit is shown ifthe lines are not detected, as is the case for most of theshort wavelength lines in SSTGC 580183. Heliocentricradial velocities ( v helio r ) are flux-weighted, averaged val-ues from the observed lines, and the uncertainties rep-resent the standard deviation of these measurements.Radial velocities with respect to the Local Standard ofRest (LSR) are also given ( v LSR r ).3.2. Foreground Extinction Estimates
As seen from Br γ emission, which is stronger thanPa β , both objects suffer strong attenuation by a largeamount of foreground dust in the Galactic disk. Sucha high extinction is indicative of a large distance fromthe Sun, but it also means that modeling emission lines is sensitively affected by the adopted foreground extinc-tion. The impacts of extinction corrections on IR linesare not as significant as those required at optical wave-lengths, but systematic differences in the IR extinctioncurves, as demonstrated below, and patchy extinctiontoward the CMZ generally make extinction correctionsdifficult.Table 2 lists individual extinction estimates from theobserved Br γ and Pa β lines for each object, based onthree near-IR extinction curves (Chiar & Tielens 2006;Boogert et al. 2011; Fritz et al. 2011). For the estimatederived from Chiar & Tielens (2006), we employed theirextinction curve in the line of sight to the GC. The 3 σ upper limit on Pa β was used to constrain the fore-ground extinction toward SSTGC 580183. We assumedthe Case B emissivity ratios from Storey & Hummer(1995), j (Pa β ) /j (Br γ ) = 5 . Hong et al.
Figure 7.
GNIRS spectra of SSTGC 588220. lanetary Nebulae toward the Galactic Center Table 2.
Foreground Extinction Estimates
Extinction SSTGC 580183 a SSTGC 588220 A K (Fritz et al. 2011) > .
72 1 . ± . A K (Chiar & Tielens 2006) > .
64 2 . ± . A K (Boogert et al. 2011) > .
95 2 . ± . h A K i > .
26 2 . ± . h A V i b > . . ± . h τ . i c > .
06 3 . ± . h A K i (Schultheis et al. 2009) d . ± .
49 2 . ± . h τ . i (An et al. 2013) e . ± .
37 3 . ± . a σ upper limits. b Assuming A K /A V = 0 .
11 (Figer et al. 1999). c Assuming 1 . ≤ τ . /A K ≤ . d Extinction within 2 ′ from the source. e Extinction within 0 . ′ –1 . ′ from the source. ular electron temperature ( T e = 10 K) and density( n e = 10 cm − ) (e.g., Zhang et al. 2004) and comparedthem to observed line ratios to derive the extinction at2 . µ m, A K . The uncertainties are the quadratic sum ofpropagated values from flux measurement uncertaintiesand systematic differences from other Case B conditions(2 × ≤ T e ≤ × K and 10 ≤ n e ≤ cm − ).As shown in Table 2, all of the three extinction curvesproduce large A K for both objects, as expected from theobserved Br γ and Pa β line ratios. However, the exactvalues of the foreground extinction strongly depend onthe adopted extinction curve. In particular, the curveof Fritz et al. (2011) has the largest slope in the near-IRamong the three curves, which results in a systematicallysmaller A K almost by ∼ mean extinctiontoward sources in the CMZ. The average foreground ex-tinction toward the CMZ is h A V i ≈
30 mag, or h A K i ≈ . A K /A V = 0 .
11 is adopted (Figer et al. 1999).The A K from Fritz et al. (2011) is almost a factor oftwo lower than this average. In case of SSTGC 580183,where we used a 3 σ upper limit on Pa β , the foregroundextinction estimates are slightly larger than the GC av-erage, except from the Fritz et al. curve.In Table 2, local values of the mean foreground ex-tinction are also included. Schultheis et al. (2009) pro-vided an extinction map across the CMZ, based on IRcolors of red giant branch stars in a grid of 2 ′ × ′ .The A K represents the mean and standard deviation oftheir measurements within 2 ′ from each source. Taking A K /A V = 0 .
11, their estimates correspond to A V ∼ . µ m ( τ . ) from the flux ratio between 10 µ mand 14 µ m, measuring the strength of the silicate ab-sorption band centered at 9 . µ m (see also Simpson2018). The h τ . i and its uncertainty are the average andstandard deviation from four nearby (within 0 . ′ –1 . ′ ) Spitzer /IRS slits that were originally designed to mea-sure background CMZ emissions (An et al. 2009, 2011).In contrast, Simpson (2018) estimated values of τ . =2 .
725 for SSTGC 580183 and 2 .
803 for SSTGC 588220,using a combination of the shapes of the 10 µ m silicatefeature and the [S III ] 18 . µ m/33 . µ m line ratios. Weemphasize that these extinction values, like those fromAn et al. (2013), apply to the diffuse ISM in the GC andnot to objects much smaller than the Spitzer
IRS aper-tures that may be substantially in front of (or behind)the CMZ of the GC.The conversion between A K and τ . also depends onthe shape of the extinction curve over the wavelength in-terval. If A V /τ . = 9 (Roche & Aitken 1985) is taken,along with A K /A V = 0 . τ . /A K becomes unity. Onthe other hand, the two extinction curves employed inthis work predict τ . /A K = 1 . . A K . In Table 2, h τ . i is the mean from the two cases with Chiar & Tielens(2006) and Fritz et al. (2011) laws, while the uncertaintyindicates half of the difference. The τ . estimates fromGNIRS spectra are 15%–20% smaller than the ISM ex-tinction measurements from An et al. (2013), implyingthat the source is likely located in front of gas and dust inthe CMZ. Taking these at face value, the larger A K frombackground giants (Schultheis et al. 2009) also impliesa shorter distance to SSTGC 588220 than the GC. Incase of SSTGC 580183, the 3 σ upper limits are compa-rable to the foreground extinction from Schultheis et al.(2009), while they are higher than the An et al. (2013)estimates.3.3. Comparison to the Size versus Surface BrightnessRelations of PNe
Galactic and extra-Galactic PNe exhibit a tight cor-relation between the radius and surface brightness (e.g.,Frew et al. 2016b; Stanghellini et al. 2020, see Figures 8and 9), according to which a smaller PN tends to have0
Hong et al. a higher mean surface brightness. Because other as-trophysical nebulae, such as classical nova shells, shownoticeable offsets from this relation (e.g., Frew et al.2016a), its direct comparison can be used not only toconfirm the nature of our sources as PNe, but also toconstrain a distance range assuming that our sourcesdirectly follow the mean PN relation.For this comparison, we measured an angular size ofSSTGC 588220 from the
HST /NICMOS Pa α image(Wang et al. 2010). Since the bright inner rim showsa mild ellipticity (Figure 4), we took an average of theangular diameter measured along the major and minoraxes, 2 . ± . ′′ . Reassuringly, this size is comparableto the spatial extent of the observed emission line pro-files (Figure 5). On the other hand, emission lines fromSSTGC 580183 show double peaks along the slit, and theobserved full extent is approximately equal to the meandiameter from high-resolution radio images (Zhao et al.2020), 3 . ± . ′′ .For our targets, the H α and H β fluxes were com-puted from the extinction-corrected Br γ line flux, as-suming the same Case B recombination coefficients asin the previous section: j (H α ) /j (Br γ ) = 103 . j (H β ) /j (Br γ ) = 36 . × ≤ T e ( K ) ≤ × and10 ≤ log N e ≤ (92 . . .
0, respec-tively). Because the foreground extinction was com-puted from the line ratio between Br γ and Pa β , the H α and H β line fluxes depend on which of the three extinc-tion curves included in this study is taken; therefore, weexplicitly included it as a source of the systematic un-certainties. Since the GNIRS targets are more extendedthan the slit width, we estimated the amount of lightlost by simulating our observations with the continuum-subtracted HST Pa α images (Wang et al. 2010). Fromthis, we found that the 0 . ′′ -wide slit collects approxi-mately 41% ±
3% of the total light from SSTGC 588220.In case of SSTGC 580183, we used the 5 . ± β surface bright-ness relation of the Galactic PNe in Stanghellini et al.(2020), which is based on accurate distances to PNecentral stars from Gaia
DR2 (Gaia Collaboration et al.2018). A comparison to our GNIRS observa-tion for SSTGC 580183 is shown in the top, andSSTGC 588220 in the bottom panel, assuming thatthey are located in the GC at a distance of 8 . β surface brightness esti- Figure 8.
Comparisons of SSTGC 580183 (top) andSSTGC 588220 (bottom) to the radius versus H β surfacebrightness relation of PNe in Stanghellini et al. (2020). Thesurface brightness estimates from GNIRS observations areindicated by the blue open diamonds. Other measurementsfrom JVLA 5.5 GHz (Zhao et al. 2020, red triangle in thetop), VLA 20 cm (Yusef-Zadeh et al. 2004, green boxes), and HST Pa α observations (Wang et al. 2010, red triangle inthe bottom) are also shown assuming that the sources arelocated at the GC distance (8 . ± σ lower limit. Thegrey crosses are Galactic PNe in Stanghellini et al. (2020),and the open circles indicate a subset of these with low ion-ized mass (see their Figure 3); their best-fitting power lawis shown by a solid line, and its extrapolation beyond thesample limit is indicated by a dotted line. lanetary Nebulae toward the Galactic Center Figure 9.
Same as in Figure 8, but showing compar-isons to the radius versus H α surface brightness relationin Frew et al. (2016b). The relations derived from Galac-tic bulge and disk populations are indicated by the dottedand dashed lines, respectively. The best-fitting power law oftheir full calibrator sample is shown by a solid line. mated from the GNIRS Br γ flux is shown by blue opendiamonds, assuming an average extinction toward eachobject (Table 2). Since the lower limit on the extinctioncorrection is available for SSTGC 580183, its 3 σ lowerlimit is marked. The vertical error bars indicate thequadrature sum of the uncertainties from the flux mea-surements, extinction corrections, Case B emissivities,and a 20% uncertainty in the absolute flux calibration.The horizontal error bars indicate a range of a physi-cal radius, assuming ± < . M ⊙ ) of these objects among other possiblecauses (see discussions in Stanghellini et al. 2020). Ac-cordingly, if our candidate PNe do not follow the meansize versus surface brightness relation, but are outlierswith systematically smaller radii, the inferred distanceswould become significantly larger by an order of mag-nitude. However, the ionized masses of SSTGC 580183and SSTGC 588220 based on radio fluxes ( ∼ . M ⊙ assuming 4–8 kpc distances) are larger than the upperlimit for their outliers, rejecting such hypothesis. Thesolid line shows the best-fitting relation to the data inStanghellini et al. (2020) without the low-ionized masssample.In addition, Figure 9 shows comparisons to the sizeversus H α surface brightness relation of the Galacticand extra-Galactic PNe of Frew et al. (2016b). The greycrosses indicate the Galactic and extra-Galactic ‘calibra-tor’ sample in their study, which do not show the samelarge scatter at small radii as in Figure 8. The blacksolid line indicates their best-fitting relation to all sam-ple PNe. The dotted and dashed lines show the observedrelations for the bulge and disk PNe in the Milky Way,respectively, which are not significantly different fromeach other and from the mean relation.In all cases, the GNIRS observations are in good agree-ment with the PN relations if the sources are put at theGC distance. The same conclusion can be drawn us-ing the Pa α flux measurement from HST /NICMOS forSSTGC 588220 (red open triangles in the bottom pan-els of Figures 8 and 9). As an independent check onthese results, we also employed radio continuum fluxesfrom VLA 20 cm (Yusef-Zadeh et al. 2004) and JVLA5 . α and H β surface brightnesses, which are essentiallyfree of extinction errors. Moreover, the radio imagesencompass the whole structure of each target and there-fore do not need slit-loss corrections. We employed therelation between the hydrogen recombination line fluxand the free-free emission in Scoville et al. (2003) andassumed the same set of recombination line Case B emis-sivities as in the previous section. The surface bright-ness estimates derived from radio fluxes are markedby green boxes (VLA 20 cm) and red triangles (JVLA5 . Hong et al.
Table 3.
Heliocentric Distance Estimates
Observations/Relation a SSTGC 580183 SSTGC 588220Br γ /Frew et al. < . . ± . γ /Stanghellini et al. < . . ± . α /Frew et al. · · · . ± . α /Stanghellini et al. · · · . ± . . ± . . ± . . ± . . ± . . ± . · · · . ± . · · · Average 9 . ± . b . ± . a Observations: wavelengths at which fluxes were measured to de-rive surface brightness – Br γ (this study), Pa α (Wang et al.2010), 20 cm (Yusef-Zadeh et al. 2004), and 5 . b An weighted average distance from radio observations. ies, we proceed to constrain the range of distance to eachobject by directly comparing its measured angular sizeto the inferred physical size for the estimated averagesurface brightness. The heliocentric distances computedin this way are summarized in Table 3. The first tworows show distance estimates using the Br γ fluxes mea-sured with GNIRS in this study, based on Frew et al.(2016b) and Stanghellini et al. (2020), respectively. ForSSTGC 580183, upper distance limits are shown fromthe lower limit in the foreground extinction. The un-certainties include those propagated from the surfacebrightness measurements and the fitting coefficients ineach of the relations. Within the uncertainties, both PNrelations give consistent distance estimates with eachother for each PN candidate.As shown in Table 3, the distances to SSTGC 588220derived from the near-IR line fluxes are slightly smallerthan those estimated using radio flux measurements, al-though they are in mutual agreement within the esti-mated uncertainties. On the other hand, the 3 σ up-per distance limits for SSTGC 580183 seem too smallcompared to those derived from radio fluxes. This im-plies that our estimated foreground extinction towardSSTGC 580183 may be too large, or that there maybe systematic errors in the adopted extinction curves.The average distance to SSTGC 580183 in Table 3 indi-cates a mean distance from radio observations; an aver-age distance to SSTGC 588220 is derived from near-IRand radio observations. These average distances, as wellas high foreground extinction, indicate that both targetsare likely located in the central region of the Milky Way. NEBULAR ABUNDANCES
Table 4.
Physical Parameters used in the Cloudy Models
SSTGC 580183 SSTGC 588220 T eff (K) a ,
555 133 , − ) 46 .
892 47 . . . N e ; cm − ) 5550 ± ± N H ; cm − ) 3945 3023Filling Factor ( f ) 0 . . a Effective temperature of an ionizing star from stellar atmosphere modelsat log g = 6 (Rauch 2003). Cloudy Models
Estimates were made of the abundances of thetwo candidate PNe using Cloudy (Ferland et al.2017). The input parameters to the Cloudy models aresummarized in Table 4. These include the effective tem-perature ( T eff ) of the exciting star (here, log g = 6 whitedwarf models at solar abundance taken from Rauch(2003)), the hydrogen nucleus density N H (the electrondensity N e being variable with depth in the models de-pending on the local ionization of the multi-electron ele-ments), the filling factor f (see Simpson (2018) for equa-tions relating N H and f ), and the abundances of heliumand the heavy elements (see below). The total ionizingluminosities Q (H) (number of photons s − ) were esti-mated from the radio fluxes measured by Zhao et al.(2020) for SSTGC 580183 and by Yusef-Zadeh et al.(2004) for SSTGC 588220, with the assumption of arounded GC distance of 8 kpc, T e = 10 K, and the useof Equation (1) of Simpson & Rubin (1990). The in-ner radii were measured from Figures 3 and 4, and werescaled at 8 kpc. We note that, since all model compar-isons are made using line flux ratios, the model resultswill not change if the estimated distances are differentfrom 8 kpc.In this modeling effort, we included various mid-IRlines measured with the
Spitzer
IRS, in addition to thehydrogen and helium lines observed with GNIRS. Weutilized the line intensity measurements from Simpson(2018), which were based on the spectral extraction tool
CUBISM (Smith et al. 2007) with point-source flux cali-bration. All of the lines were measured by fitting Gaus-sians to the spectra. Some lines were re-measured inthis study using the IRS analysis program
SMART-IDEA (Higdon et al. 2004) for better accuracy. These include[S IV ] 10.51 µ m, [Ar V ] 13.10 µ m, [Mg V ] 13.52 µ m,[Fe VI ] 14.77 µ m, [P III ] 17.88 µ m, [Fe II ] 17.94 µ m,[Ar III ] 21.83 µ m, [Ne V ] 24.32 µ m, and [Fe III ]22.93 µ m. As described in §
2, background subtractionwas performed on a line-by-line basis by taking the aver- lanetary Nebulae toward the Galactic Center SH and LH modules) fluxes atfour nearby positions, while for the low-resolution mod-ules ( SL and LL ), background positions were measuredfrom slightly distant positions along the same slits.In Table 5, the extinction-corrected line fluxes are ex-pressed as line ratios with respect to the hydrogen re-combination lines, where H I γ ) was used forthose observed with GNIRS, and H I µ m) forthe IRS observations. The division into two wavelengthintervals was necessary to minimize systematic errorsfrom different slit sizes – the GNIRS observations weremade with a relatively narrow slit (‘slit’ models), whilethe longer-wavelength IRS slits were large enough to in-clude almost the whole source (‘whole nebular’ models).We corrected the observed near-IR line fluxes for ex-tinction using A K estimates derived from the Fritz et al.(2011) extinction curve (Table 2). The same extinctioncurve was used for mid-IR emission lines up to the peakof the 9 . µ m silicate feature. At longer wavelengths,however, we employed the Chiar & Tielens (2006) ex-tinction curve because it extends to a longer wavelength( ∼ µ m instead of the 27 µ m of Fritz et al. (2011)).Also, it is based on a source that should have absorp-tion only (the Quintuplet Cluster star GCS 3) and nota source for which one must compensate for dust emis-sion (Sgr A) in the extinction curve estimate at thelongest wavelengths (Fritz et al. 2011). The uncertain-ties in the background-subtracted line intensities includeuncertainties from the flux measurements and the back-ground subtraction, which are added in quadrature.For the reasons described above, we performed sep-arate Cloudy runs to account for the ‘slit’ models andthe ‘whole-nebula’ models. In the former, the near-IRlines Br γ , He I 2.059 µ m, and He II 2.189 µ m lineswere used to produce estimates of T eff and the He/Hratio. In the whole-nebula integrated models, the mid-IR lines were used to produce estimates of N H and f forSSTGC 588220.We used Cloudy with its ‘optimize’ commands, where,in each call, a given parameter or a set of parameters isallowed to vary with the best solution based on the de-viation of the model predictions for a set of lines withthe observed fluxes. Because of the large number of in-put parameters, relative to limited observational infor-mation, a uniform density and a constant filling factorwere assumed in these models. The elemental abun-dances were fit by hand by comparing the model output The other K -band He I lines are listed here by their most im-portant components, but are actually blends with lines that aredistinguished in Cloudy. We are unable to separate them owingto the moderate spectral resolution of our data. with the observed line flux ratios. The filling factor wascomputed by ‘optimize’ command for SSTGC 588220,but this was not successful for SSTGC 580183. Instead,models with a variety of filling factors were computedand the median model with f = 0 .
10 was selected.Elemental abundances in the models, other than thosederived in the modeling, were taken from Cloudy’s stan-dard set for an H II region, nominally that of the OrionNebula. The abundances of carbon, nitrogen, and sili-con were set to the GC abundance ratios with respect tooxygen in Simpson (2018): C/O=0.75, N/O=0.17, andSi/O=0.035. This allows for some depletion into grains,whose opacities are included in the calculation.Criteria for an acceptable fit and steps in the modelingprocess are:(i) The He II 2.189 µ m/He I 2.059 µ m line ratio, whichis used to estimate T eff of the PN central star, should beequal to the observed ratio. Iteration was needed, sincethe exact T eff is somewhat dependent on N H and f asthey both affect the ionization parameter U (Simpson2018, see their Equation 2).(ii) The density-sensitive line pairs (mainly [S III]18.7 µ m and 33.5 µ m, but also [Ne V] 14.3 µ m and24.3 µ m for SSTGC 588220) should agree with the ob-served line ratios; the [Ar III] 8.99 µ m and 21.8 µ m linepair in SSTGC 580183 was not used because of theirlarge uncertainties. As shown in the high-resolution im-ages (bottom panel of Figures 3 and 4), the PNe maynot have a constant density. Indeed, it was not possibleto find constant density models that produce line ratiosthat agree with all line pairs. For this reason, less weightwas given to the [S III] line pair in SSTGC 588220, whichhas larger uncertainties than the [Ne V] line pair.(iii) The line ratios with respect to hydrogen shouldagree with the observed ratios. To assure this, the rela-tive abundances of helium and the heavy elements wereadjusted in each iteration. In the models, the abun-dances of carbon, nitrogen, and silicon, all unobservedin our candidate PNe, were scaled to the abundance ofoxygen according to the H II region abundances used bySimpson (2018). This resulted in additional iterationsas changing the abundances modifies the cooling func-tion and hence the electron temperature, which in turnaffects all computed fluxes and flux ratios.4.2. Results
The elemental abundances in the best-fitting Cloudymodels are given in Table 6. Although we ran mod-els assuming constant N H and f , the above modelingexercise provides useful and reliable information on thecentral star T eff and, in particular, the abundances of he-lium, neon, and sulfur, all of which were observed in lines4 Hong et al.
Table 5.
Normalized Line Fluxes
SSTGC 580183 (G359.963 − − µ m) Flux a Uncertainty b Flux c Ion. d Flux a Uncertainty b Flux c Ion. d Gemini GNIRS (slit models)Paschen beta 1 . · · · · · · · · · · · · .
877 0 .
015 5 . − . . · · · · · · · · · · · · .
168 0 .
031 0 . − . . .
000 0 .
047 1 . − .
053 1 .
000 0 .
014 1 . − . P –2 S . · · · · · · · · · · · · .
445 0 .
020 8 . − . P –2 S . . .
079 0 . − .
050 0 . .
015 0 . − . P –3 S . . .
070 0 . − .
050 0 .
041 0 .
046 0 . − . P –3 S . · · · · · · · · · · · · .
023 0 .
336 0 . − . F –4 F . · · · · · · · · · · · · .
023 0 .
070 0 . − . G –4 F . · · · · · · · · · · · · .
075 0 .
087 0 . − . . . .
101 0 . − .
247 0 . .
020 0 . − . II ] 1 . · · · · · · · · · · · · .
225 0 .
203 11 . − . Spitzer
IRS (whole nebula models)H 7–6 12 . .
000 0 .
055 1 . − .
066 1 .
000 0 .
099 1 . − . IV ] 25 . .
68 0 .
152 63 . − .
765 330 .
95 0 .
102 330 . − . II ] 12 . .
79 0 .
057 44 . − .
861 15 .
57 0 .
111 28 . − . III ] 15 . .
64 0 .
056 379 . − .
075 413 .
37 0 .
099 470 . − . V ] 14 . · · · · · · · · · − .
897 160 .
13 0 .
099 140 . − . V ] 24 . · · · · · · · · · − .
897 113 .
35 0 .
099 94 . − . III ] 7 . .
36 0 .
077 5 . − . · · · · · · · · · − . V ] 13 . · · · · · · · · · − .
697 0 .
654 0 .
283 0 . − . III ] 17 . .
916 0 .
469 0 . − .
460 1 .
716 0 .
307 1 . − . III ] 18 . .
28 0 .
057 107 . − .
271 81 .
82 0 .
102 91 . − . III ] 33 . .
74 0 .
232 32 . − .
271 40 .
02 0 .
316 34 . − . IV ] 10 . .
39 0 .
056 104 . − .
069 245 .
80 0 .
099 219 . − . II ] 14 . .
246 0 .
067 1 . − . · · · · · · · · · − . II ] 6 . .
46 0 .
088 10 . − . · · · · · · · · · − . III ] 8 . .
60 0 .
068 83 . − . · · · · · · · · · − . III ] 21 . .
816 0 .
121 5 . − .
094 4 .
626 0 .
121 2 . − . V ] 13 . · · · · · · · · · − .
604 6 .
031 0 .
101 11 . − . III ] 22 . .
348 0 .
145 4 . − .
530 17 .
67 0 .
110 17 . − . VI ] 14 . · · · · · · · · · − .
908 0 .
446 0 .
139 14 . − . a Extinction-corrected observed line fluxes with respect to Br γ (GNIRS) from Table 1 or H I 7–6 12.37 µ m (IRS, Simpson2018). b Uncertainties in the observed line flux ratios, consisting of the statistical uncertainties from the line flux measurementscombined quadratically with the uncertainty of the associated hydrogen recombination line. c Modeled line fluxes with respect to Br γ or H I 7–6 12.37 µ m. d Logarithm of the fractional ionization of each ion. from more than one ionization stage. Of lesser reliabilityare the elements observed from only a single ionizationstage, although some have, at least, the dominant ion-ization stage observed. Even so, a detailed modeling ofas many ionization stages as is done here can producemore reliable results than the use of estimated ionizationcorrection factors (
ICF s) that is otherwise common indetermining nebular abundances.The uncertainties in the background subtraction prop-agate into significant uncertainties in the final abun-dances for most of heavy elements. This is becauseabundances are tabulated with respect to hydrogen, where the strongest hydrogen line in the IRS wavelengthis the H 7–6 line at 12.37 µ m, which is sensitive to thebackground subtraction. Whereas the highest excitationlines had little or no presence in the background spec-tra, the low excitation lines have substantial backgroundemission that needs to be removed. In particular, back-ground subtraction removes 27% of the measured targetflux of the H 7–6 line for SSTGC 580183, and 40% ofthe measured target flux for SSTGC 588220 (see Fig-ure 2); the final SNRs are 18 and 10 for SSTGC 580183and SSTGC 588220, respectively (Table 5). The othersingly ionized lines ([Ne II ], [Ar II ], and [Fe II ]) have even lanetary Nebulae toward the Galactic Center Table 6.
Elemental Abundances
Log Abundances a Elements SSTGC 580183 SSTGC 588220 Solar b Hydrogen 12 .
00 12 .
00 12 . . ± .
04 10 . ± .
03 [10 . . ± .
27 8 . ± .
80 8 . . ± .
04 8 . ± .
04 [7 . . ± . · · · . · · · . ± .
25 7 . . ± .
16 5 . ± .
16 5 . . ± .
18 7 . ± .
22 7 . . ± . · · · . . ± .
04 6 . ± .
05 [6 . . ± .
19 7 . ± .
15 7 . c . ± .
20 3 . ± .
06 [3 . a log ( N A /N H ) + 12. b Solar abundances from Asplund et al. (2009) with the ex-ceptions of Na, Mg, and Fe, which are from Scott et al.(2015a,b). The brackets around the noble gas abundancesindicate that these Solar abundances are ‘indirect photo-spheric estimates’ (Asplund et al. 2009). c Krypton (Kr) was modeled separately from the Cloudy runs(see text). more flux subtracted (so much so for [Fe II ] that this linewas dropped from the analysis), but these elements havedoubly ionized lines in the IRS spectra, which make ourabundance estimates more secure.The abundance uncertainties in Table 6 are the sum,in quadrature, of the measurement uncertainty, the un-certainty due to a possible error in the extinction of∆ τ . = 0 .
5, and an estimate of the model error ex-pressed as the uncertainty in the
ICF . This
ICF un-certainty is estimated to be the value of 0.05 divided bythe fractional ionization of the observed ion for elementswhere only one ionization state was observed or it is es-timated to equal 0.25 times the fractional ionization ofthe unobserved ionization states when multiple ioniza-tion states were observed. Which of these errors dom-inates depends on the line – the measurement error isimportant for the two blended lines, [P
III ] 17.88 µ m and[Fe II ] 17.94 µ m (the latter having such a large uncer-tainty that it was not further included); the extinctionerror is mainly important for the lines with wavelengthsnear 10 µ m and 18 µ m (this uncertainty is large becausethe silicate extinction curve is not that well known); andthe ICF error dominates for lines arising from ions withvery low fractional ionization, such as O and Mg .Finally, the uncertainty for O/H in SSTGC 588220 wasincreased because the line was saturated in the high-resolution IRS module ( LH ) and the flux had to be takenfrom the low-resolution ( LL1 ) observation. 4.2.1.
SSTGC 588220 (G0.095-0.051)
The fluxes predicted by the Cloudy models providea reasonably good agreement with the observed fluxes.However, in spite of the overall agreement, notablediscrepancies are found in some lines. For instance,the best-fitting model over-predicts fluxes from [Ar V ]13.1 µ m and [Fe VI ] 14.7 µ m, while it under-estimatesthe He I 1.083 µ m flux. The over-estimated high-excitation lines are probably the result of the source notbeing spherically symmetric and having large densityfluctuations. On the other hand, the He I 1.083 µ m lineis more sensitive to the amount of extinction than thelonger-wavelength He lines, and its emissivity is moredependent on details of the electron density distribu-tion than the emissivity of the He I 2.059 µ m line (seePorter et al. 2012, 2013).4.2.2. SSTGC 580183 (G359.963-0.120/G-0.037-0.120)
The agreement with the best-fitting model is less sat-isfactory in case of SSTGC 580183, most likely due tothe PN having components of multiple densities and fill-ing factors, rather than having constant values as as-sumed in the current analysis. In particular, the ‘opti-mize’ command in Cloudy initially produced a very lowfilling factor ( f ≈ .
03) in the models, driven by theabsence of high ionization lines such as [Ne V ], [Ar V ],and [Mg V ] in the Spitzer mid-IR spectra, except [O IV ]26 µ m. The models described in Tables 4 and 5 wereproduced with f = 0 . f ) and inner radius0 .
02 pc (estimated from the ring size in Figure 3, bot-tom). The uncertainties in these tables do not includeany uncertainty for choice of model, but an additional ∼
10% should be added to the abundance uncertaintiesin the tables to account for the variations in the possiblemodel parameters.Notably, the oxygen abundance from the best-fittingmodel is super-solar (Table 6). With a low f , the frac-tion of elements in the highest ionization states (likeO ) is very small, and the oxygen abundance (O/H) be-comes substantially higher than solar. We also searchedfor models of more compact PNe with lower f andsmaller inner radii that produce a higher ionization frac-tion of oxygen. However, we found that none of thesemodels could produce as strong [O IV] emission as isobserved, unless the total oxygen abundance is substan-tially higher than Solar. In addition, because T eff of thecentral star can be reliably determined from the He II2.189 µ m/He I 2.059 µ m line ratio, and both He ++ andO require >
54 eV photon energies for ionization tothat level, the higher oxygen abundance is less affected6
Hong et al. by the choice of the spectral energy distribution of theexciting star.Along with oxygen, the helium abundance is also sig-nificantly higher than in SSTGC 588220. Because theemission from heavy element ions, especially oxygen, isthe primary coolant for the gas, increasing their abun-dances results in lower electron temperatures, which inturn increases the emissivities of the hydrogen recombi-nation lines much more than they change the emissivitiesof the He I 2.059 µ m line (see Storey & Hummer 1995;Porter et al. 2012, 2013). The consequence is that theabundance of helium with respect to hydrogen must beincreased above Solar to match the observed line fluxes.4.3. [Kr III] 2.199 µ m In both candidate PNe, we detected the s -processnoble gas krypton (Kr) at 2.199 µ m, which was firstidentified in PNe by Dinerstein (2001). From the lineflux measurements, we computed the Kr ++ /H + ra-tios of (3 . ± . × − and (1 . ± . × − for SSTGC 580183 and SSTGC 588220, respectively,based on the Br γ emissivities of Storey & Hummer(1995), the effective collisional strengths for Kr ++ ofSch¨oning (1997), and the transition probabilities forKr ++ of Eser & ¨Ozdemir (2019). Because [Kr III ]2.199 µ m is not modeled in the current version ofCloudy, we adopted the analytical relation for the ICF from Sterling et al. (2015) between the Kr ++ /Kr andthe S ++ /S ratios, which show the tightest fit amongother ionic ratios (such as Ar ++ /Ar). In addition, ourS ++ /S ratio is more reliable than Ar ++ /Ar, because theS ++ and S measurements were made using higher res-olution modules in the IRS with less extinction. TheKr/H abundance ratios are included in Table 6.4.4. Significance
Krypton is significantly enriched in both of our targets([Kr/Fe] = 1 . ± . . ± . , indicating that this s -process element has been over-produced, possibly duringthe late asymptotic giant branch evolution. This is inagreement with other lines of evidence supporting theirstatus as a PN (morphology, high excitation lines, anda match to the PNe size-surface brightness relations)presented in this paper.The abundance patterns of our targets are generallyconsistent with those in the GC region. Stars in this re-gion have a bimodal metallicity distribution with peaks Here, we used a conventional notation, [X/Y] ≡ log ( N X /N Y ) ∗ − log ( N X /N Y ) ⊙ , where log N indicates abundances of specific el-ements. at [Fe/H] ∼ − . ∼ +0 .
3, and exhibit an elevated α -element abundance in the metal-poor component,while the metal-rich counterpart shows near-solar abun-dance patterns (e.g., Schultheis et al. 2020). Our de-rived metallicities are [Fe/H] ≈ − . ≈ +0 . α -element abundances also seem to follow the observedabundance pattern – higher α -element abundances ofSSTGC 580183 ([O/Fe]= 1 .
2, [S/Fe]= 1 . .
2, [Mg/Fe]=0 .
4, [S/Fe]= 0 .
2) – although the former estimates are sig-nificantly higher than the GC sample in Schultheis et al.(2020).Our radial velocity measurements were based on neb-ular emission lines, instead of observations of a centralstar of a PN. Given that the expansion speed of a PNis an order of a few tens of km s − , our measurementsare uncertain by this amount. A large positive radialvelocity of SSTGC 580183 ( v LSR r ∼ +70 km s − ) in thefourth Galactic quadrant and an extreme negative radialvelocity of SSTGC 588220 ( v LSR r = −
150 km s − ) in thefirst quadrant are at odds with the sense of rotation ofthe nuclear stellar disk (a positive correlation between v r and l ), but there is a large scatter in the v r vs. l diagram of stars in the nuclear bulge (Sch¨onrich et al.2015; Schultheis et al. 2020).SSTGC 580183 is extremely helium-rich N (He)/ N (H)= 0 .
12, while SSTGC 588220 has a nearlysolar helium abundance. According to the PN classifica-tion scheme in Peimbert (1978), the helium abundanceof SSTGC 580183 is at the border between the helium-and nitrogen-rich PNe (Type I) and other types withnormal helium abundances. Since Type I PNe areconsidered a descendant of a massive progenitor star( > M ⊙ ) by their super-helium and nitrogen abun-dances (see also Stanghellini & Haywood 2010), it isconceivable that the progenitor of SSTGC 580183 isalso likely as massive as ∼ M ⊙ . This implies thatSSTGC 580183 originated from binary interactions ofold stellar populations in the nuclear bulge, or fromrecent star formation in the CMZ. DISCUSSION AND SUMMARYIn this paper, we have presented medium-resolutionnear-IR spectra taken with Gemini/GNIRS of twocandidate PNe, which were serendipitously found in
Spitzer /IRS spectra in close lines of sight to Sgr A ∗ . Be-sides showing strong emission lines in the mid-IR fromseveral high-excitation ions, their appearance on high-resolution images is consistent with their being PNe.Moreover, our Gemini/GNIRS near-IR spectra reveal lanetary Nebulae toward the Galactic Center s -process elementsin the PN envelope. Their memberships in the nuclearstellar disk seem feasible from the proximity to the GCon the sky and distances based on a comparison to PNsize-surface brightness relations, and are favored by ourjoint analysis of near- and mid-IR spectra, instead of be-longing to other Galactic structural components. OurCloudy modeling assumes uniform density and fillingfactors of these targets, but future observations withhigh spatial-resolution mapping in the various near- andmid-IR lines can potentially reveal high-density, or evenhigh-excitation structures within the overall morphologyof these sources.The expected number of PNe in the nuclear bulge canbe estimated from a comparison with other Galacticstellar components. The total stellar mass of the nu-clear bulge ( ∼ . × M ⊙ ; Launhardt et al. 2002) isapproximately 10 times less massive than the classicalbulge. If we assume that most stars in the nuclear bulgeare as old as those in the bulge ( &
10 Gyr), and simplyscale the result obtained from the populations synthesismodels in Moe & De Marco (2006) by stellar mass, thereshould be ∼ PNe found in this region, although thisestimate is uncertain by a factor of two or more. At theother extreme, if we scale the expected number from thestellar halo in Moe & De Marco (2006), the number ofPNe decreases by a factor of ∼ due to older ages ofthe system (a relatively shorter dissipation timescale ofstellar envelopes), but this estimate is quite uncertain.On the observational ground, however, there are 8 PNeknown in the Galactic globular clusters (Jacoby et al.1997; Minniti et al. 2019), from which one can naivelyexpect ∼ PNe since the nuclear bulge is ∼
100 timesmore massive than the entire globular cluster system inthe Milky Way.In this regard, the scarcity of known PNe in the nu-clear bulge raises a more severe problem than the ob-served paucity of PNe in other Galactic components.This may be due to extreme foreground extinction to-wards the GC. Such an observational bias may exist, be-cause the two PNe presented in this paper have the mostcompact sizes (implying early stages of PN evolution) ascan be seen from the size-surface brightness distributionsof other Galactic and extra-Galactic PNe (see Figures 8and 9). Their compactness has even made them ap-pear as ‘point-like’ sources in the original
Spitzer /IRACcatalog (Ram´ırez et al. 2008). As a consequence, therecould be more PNe in lines of sight toward the CMZ thatare more difficult to detect due to extended envelopeswith lower surface brightness in the high source-density,highly dust-obscured region. On the other hand, if the observed lack of PNe isrooted in astrophysical origins, this may indicate thatthe PN formation channel such as binary interactions(e.g., Minniti et al. 2019) is overpowered by suppres-sion mechanisms that can reduce the envelope mass ofasymptotic giant branch (AGB) stars. Some stars canalso skip the AGB phase and directly evolve into hotsubdwarfs, if the envelope mass is not high enough.These AGB manqu´e stars can be formed from helium-rich progenitor stars by enhanced mass loss during thered giant branch evolution (Bressan et al. 2012).While PNe are strikingly rare in the nuclear bulge,however, the number of the observed PNe is not greatlydifferent from what one would expect from ongoing starformation in the CMZ, with a rate of ∼ . M ⊙ yr − (e.g., An et al. 2011; Longmore et al. 2013). Since thePN lifetime is approximately ∆ t ∼ years, the to-tal mass of stars formed at any time interval ∆ t is ∼ M ⊙ , if the star formation has proceeded con-stantly over the past few billion years. If stars as mas-sive as 2 M ⊙ can evolve into a PN, for reasons describedabove, a total mass of PN progenitors would be ∼ M ⊙ at any time. Therefore, the expected number of PNewould become ∼ M ⊙ in the nuclear bulge. Indeed, Simpson (2018)find half a dozen more candidates with exceptionallyhigh excited lines from neon and/or oxygen, in addi-tion to the two sources included in this study, whichserve as good PN candidates in this region. Future high-resolution narrow-band imaging or mid-IR spectroscopicsurveys will help hunt for additional PNe in the nuclearbulge. ACKNOWLEDGMENTSWe thank Kris Sellgren and Harriet Dinerstein formany helpful and interesting discussions over the years.We thank Thomas Geballe for his technical assistancebefore and during the observing runs. J.H. and D.A.acknowledge support provided by the National Re-search Foundation of Korea (NRF) to the Center forGalaxy Evolution Research (2017R1A5A1070354) andby Basic Science Research Program through the NRFgrant funded by the Ministry of Education (NRF-2018R1D1A1A02085433).The Gemini data (Gemini program GS-2016A-Q-42,NOAO prop. ID 2016A-0431), acquired through theGemini Observatory Archive at NSF’s NOIRLab andprocessed using the Gemini IRAF package, are basedon observations obtained at the international GeminiObservatory, a program of NSF’s NOIRLab, which ismanaged by the Association of Universities for Research8 Hong et al. in Astronomy (AURA) under a cooperative agreementwith the National Science Foundation on behalf of theGemini Observatory partnership: the National ScienceFoundation (United States), National Research Council(Canada), Agencia Nacional de Investigaci´on y Desar-rollo (Chile), Ministerio de Ciencia, Tecnolog´ıa e Inno-vaci´on (Argentina), Minist´erio da Ciˆencia, Tecnologia,Inova¸c˜oes e Comunica¸c˜oes (Brazil), and Korea Astron- omy and Space Science Institute (Republic of Korea).This work was enabled by observations made from theGemini North telescope, located within the MaunakeaScience Reserve and adjacent to the summit of Mau-nakea. We are grateful for the privilege of observing theUniverse from a place that is unique in both its astro-nomical quality and its cultural significance.REFERENCES
An, D., Ram´ırez, S. V., Sellgren, K., et al. 2009, ApJL, 702,L128. doi:10.1088/0004-637X/702/2/L128An, D., Ram´ırez, S. V., Sellgren, K., et al. 2011, ApJ, 736,133. doi:10.1088/0004-637X/736/2/133An, D., Ram´ıirez, S. V., & Sellgren, K. 2013, ApJS, 206,20. doi:10.1088/0067-0049/206/2/20Asplund, M., Grevesse N., Sauval A. J., & Scott P. 2009,ARA&A, 47, 481.doi:10.1146/annurev.astro.46.060407.145222Balick, B. & Frank, A. 2002, ARA&A, 40, 439.doi:10.1146/annurev.astro.40.060401.093849Beaulieu S. F., Freeman K. C., Kalnajs A. J., Saha P., &Zhao H. 2000, AJ, 120, 855. doi:10.1086/301504Boogert, A. C. A., Huard, T. L., Cook, A. M., et al. 2011,ApJ, 729, 92. doi:10.1088/0004-637X/729/2/92Bressan, A., Marigo, P., Girardi, L., et al. 2012, MNRAS,427, 127. doi:10.1111/j.1365-2966.2012.21948.xChiar, J. E., & Tielens, A. G. G. M. 2006, ApJ, 637, 774.doi:10.1086/498406Cushing, M. C., Vacca, W. D., & Rayner, J. T. 2004,PASP, 116, 362. doi:10.1086/382907Dahmen, G., Huttemeister, S., Wilson, T. L., et al. 1998,A&A, 331, 959Dinerstein, H. L. 2001, ApJ, 550, L223. doi:10.1086/319645Durand, S., Acker, A., & Zijlstra, A. 1998, ApJS, 132, 13.doi:10.1051/aas:1998356Elias, J. H., Rodgers, B., Joyce, R. R., et al. 2006a,Proc. SPIE, 6269, 626914. doi:10.1117/12.671765Elias, J. H., Joyce, R. R., Liang, M., et al. 2006b,Proc. SPIE, 6269, 62694C. doi:10.1117/12.671817Eser, S., & ¨Ozdemir, L. 2019, CJP, 97, 529.doi:10.1139/cjp-2018-0385Fazio, G. G., et al. 2004, ApJS, 154, 10. doi:10.1086/422843Ferland, G. J., Chatzikos, M., Guzm´an, F., et al. 2017,RMxAA, 53, 385Figer, D. F., McLean, I. S., & Morris, M. 1999, ApJ, 514,202. doi:10.1086/306931 Frew, D. J., Bojiˇci´c, I. S., & Parker, Q. A. 2016a, Journalof Physics Conference Series, 728, 032015.doi:10.1088/1742-6596/728/3/032015Frew, D. J., Parker, Q. A., & Bojiˇci´c, I. S. 2016b, MNRAS,455, 1459. doi:10.1093/mnras/stv1516Fritz, T. K., Gillessen, S., Lutz, D., et al. 2011, ApJ, 737,73. doi:10.1088/0004-637X/737/2/73Gaia Collaboration, Brown, A. G. A., Vallenari, A., et al.2018, A&A, 616, A1. doi:10.1051/0004-6361/201833051Gravity Collaboration, Abuter, R., Amorim, A., et al. 2019,A&A, 625, L10. doi:10.1051/0004-6361/201935656Higdon, S. J. U., Devost, D., Higdon, J. L. et al. 2004,PASP, 116, 975. doi:10.1086/425083Houck, J. R., Roellig, T. L., van Cleve, J., et al. 2004,ApJS, 154, 18. doi:10.1086/423134Jacoby G. H., Kronberger, M., Patchick, D., et al. 2010,PASA, 27, 156. doi:10.1071/AS09025Jacoby, G. H., Morse, J. A., Fullton, L. K., et al. 1997, AJ,114, 2611. doi:10.1086/118671Jacoby, G. H., & van de Steene, G. 2004, A&A, 419, 563.doi:10.1051/0004-6361:20035809Launhardt, R., Zylka, R., & Mezger, P. G. 2002, A&A, 384,112. doi:10.1051/0004-6361:20020017Lawrence, A., Warren, S. J., Almaini, O., et al. 2007,MNRAS, 379, 1599. doi:10.1111/j.1365-2966.2007.12040.xLongmore, S. N., Bally, J., Testi, L., et al. 2013, MNRAS,429, 987. doi:10.1093/mnras/sts376Manchado, A., Stanghellini, L., & Guerrero, M. A. 1996,ApJL, 466, L95. doi:10.1086/310170Mills, E., Morris, M. R., Lang, C. C., et al. 2011, ApJ, 735,84. doi:10.1088/0004-637X/735/2/84Minniti, D., Dias, B., G´omez, M., et al. 2019, ApJL, 884,L15. doi:10.3847/2041-8213/ab4424Miszalski B., Parker Q. A., Acker A., Birkby J. L., FrewD. J., & Kovacevic A. 2008, MNRAS, 384, 525.doi:10.1111/j.1365-2966.2007.12727.xMoe, M., & De Marco, O. 2006, ApJ, 650, 916.doi:10.1086/506900 lanetary Nebulae toward the Galactic Center19