THE ORIGIN AND EVOLUTION OF THE HALO PN BoBn 1: FROM A VIEWPOINT OF CHEMICAL ABUNDANCES BASED ON MULTIWAVELENGTH SPECTRA

The spectra of BoBn 1 were taken using the HDS (Noguchi et al. 2002) attached to one of the two Nasmyth foci of the 8.2 m Subaru telescope atop Mauna Kea, Hawaii on 2008 October 6 (program ID: S08B-110, PI: M. Otsuka). In Figure 1, we present the optical image of BoBn 1 taken by the HDS silt viewer camera (∼012 pixel⁻¹, no filters) during the HDS observation. The sky condition was clear and stable, and the seeing was between 04 and 06. The full width at half-maximum (FWHM) of the image is ∼1''. BoBn 1 shows a small protrusion toward the southeast.

Zoom In Zoom Out Reset image size

Spectra were taken for two wavelength ranges, 3600–5400 Å (hereafter, the blue region spectra) and 4600–7500 Å (the red region spectra). An atmospheric dispersion corrector (ADC) was used to minimize the differential atmospheric dispersion through the broad wavelength region. In these spectra, there are many recombination lines of hydrogen, helium, and metals and CELs. These numerous spectral lines allowed us to derive reliable chemical compositions. We used a slit width of 12 (0.6 mm) and a 2 × 2 on-chip binning, which enabled us to achieve a nominal spectral resolving power of R = 30,000 with a 4.3 binned pixel sampling. The slit length was set to avoid overlap of the echelle diffraction orders at the shortest wavelength portion of the observing wavelength range in each setup. This corresponds to 8'' (4.0 mm), in which the nebula fits well and can allow us to directly subtract sky background from the object frames. The CCD sampling pitch along the slit length projected on the sky is ∼0276 per binned pixel. The achieved signal-to-noise ratio (S/N) is >40 at the nebular continuum level even in both ends of each echelle order. The resulting resolving power is around R > 33,000, which derived from the mean of the FWHM of narrow Th–Ar and night sky lines. All the data were taken as a series of 1800 s exposure for weak emission lines and 300 s exposures for strong emission lines. The total exposure times were 16,200 s for red region spectra and 7200 s for blue region spectra. During the observation, we took several bias, instrumental flat lamp, and Th–Ar comparison lamp frames, which were necessary for data reduction. For the flux calibration, blaze function correction, and airmass correction, we observed a standard star HR9087 at three different airmass.

We also used archival high-dispersion spectra of BoBn 1, which are available from the European Southern Observatory (ESO) archive. These spectra were observed on 2002 August (program ID: 069.D-0413, PI: M. Perinotto) and 2007 June (program ID: 079.D-0788, PI: A. Zijlstra), using the Ultraviolet Visual Echelle Spectrograph (UVES; Dekker et al. 2000) at the Nasmyth B focus of KUEYEN, the second of the four 8.2 m telescopes of the ESO Very Large Telescope (VLT) at Paranal, Chile. We call the 2002 August data "UVES1" and the 2007 June data "UVES2" hereafter. We used these data to compensate for unobserved spectral regions and order gaps in the HDS spectra. We normalized these data to the HDS spectra using the intensities of detected lines in the overlapped regions between HDS and UVES1 and 2.

These archive spectra covered the wavelength range of 3300–6600 Å in UVES1 and 3300–9500 Å in UVES2. The entrance slit size in both observations was 11'' in length and 15 in width, giving R > 30,000 derived from Th–Ar and sky lines. The CCDs used in UVES have 15 μm pixel sizes. For UVES1 a 1 × 1 binning CCD pattern was chosen. For UVES2 a 2 × 2 on-chip binning pattern was chosen. The sampling pitch along the wavelength dispersion was ∼0.015–0.02 Å pixel⁻¹ for UVES1 and ∼0.03–0.04 Å pixel⁻¹ for UVES2. The exposure time for UVES1 was 2700 s × 4 frames, 10,800 s in total. The exposure time for UVES2 was 1500 s × 2 frames, 3000 s in total. The standard star Feige 110 was observed for flux calibration.

In Figure 2, we present the combined HDS and UVES spectrum of BoBn 1 normalized to the Hβ flux. The spectrum for the wavelength region of 3650–7500 Å is from the HDS data and that of 3450–3650 Å and >7500 Å is from the UVES data.

Zoom In Zoom Out Reset image size

The observation logs are summarized in Table 2. The detected lines in the Subaru/HDS and VLT/UVES spectra are listed in the Appendix.

We complemented optical spectra with UV spectra obtained by the International Ultraviolet Explorer (IUE) to derive C⁺, C²⁺, N²⁺, and N³⁺ abundances from semi-forbidden lines C ii], C iii], N iii], and N iv], since these emission lines cannot be observed in the optical region. These IUE spectra were retrieved from the Multi-mission Archive at the STScI (MAST). We collected the high- and low-resolution IUE spectra taken by the Short Wavelength Prime (SWP) and Long Wavelength Prime/Long Wavelength Redundant (LWP/LWR) cameras. Our used data set is listed in Table 3. All of the IUE observations were made using the large aperture (10.3 × 23 arcsec²). SWP and LWP/LWR spectra cover the wavelength range of 1150–1980 Å and 1850–3350 Å, respectively. For each SWP and LWP/LWR spectra, we did median combine to improve the S/N. The combined short wavelength spectrum was used to measure fluxes of emission lines in ≲1910 Å because this allowed us to separate C iii] λ1906/08 and C iv λ1548/51 lines. C iii]λ1906/08 are important as a density diagnostic. The combined long wavelength spectrum was for measurements of emission-line fluxes in ≳2000 Å. The measured line fluxes were normalized to the Hβ flux using theoretical ratios of He ii I(λ1640)/(λ4686) for the short wavelength spectrum and I(λ2512)/(λ4686) for the long wavelength spectrum, respectively, adopting an electron temperature T = 8840 K and density n = 10⁴ cm⁻³ as given by Storey & Hummer (1995), then normalized to the Hβ flux. The interstellar extinction correction was made using Equation (1) (see Section 3.1). The observed and normalized fluxes of detected lines are listed in the Columns 4 and 5 of Table 4, respectively.

We used two data sets (program IDs: P30333, PI: A. Zijlstra; P30652, PI: J. Bernard-Salas) taken by the Spitzer Space Telescope in 2006 December. The data were taken by the Infrared Spectrograph (IRS; Houck et al. 2004) with the SH (9.5–19.5 μm), LH (5.4–37 μm), SL (5.2–14.5 μm), and LL (14–38 μm) modules. In Figure 3, we present the Spitzer spectra of BoBn 1. We downloaded these data using Leopard provided by the Spitzer Science Center. The one-dimensional spectra were extracted using Spice version c15.0A. We extracted a region within ±1'' from the center of each spectral order summed up along the spatial direction. For SH and LH spectra, we subtracted sky background using offset spectra. We normalized the SL and LL data to the SH and LH using the measured fluxes of [S iv] λ10.5 μm, H i λ12.4 μm, [Ne ii] λ12.8 μm, [Ne iii] λ15.6 μm, and [Ne iii] λ36.0 μm. Finally, the measured line fluxes were normalized to the Hβ flux. The observed line ratio H i I(11.2 μm)/I(λ4861) (3.1 × 10⁻³) is consistent with the theoretical value (3.15 × 10⁻³) for T = 8840 K and n = 10⁴ cm⁻³ as given by Storey & Hummer (1995). We did not therefore perform interstellar extinction correction.

Zoom In Zoom Out Reset image size

The observed and normalized fluxes of detected lines are listed in Table 5. In addition to the ionized gas emissions, the amorphous carbon dust continuum and the polycyclic aromatic hydrocarbons (PAHs) feature around 6.2, 7.7, 8.7, and 11.2 μm are found for the first time. In Figure 4, we present these PAH features. The 11.2 μm emission line is a complex of the narrow width H i 11.2 μm and the broad PAH 11.2 μm. The 6.2, 7.7, and 8.7 μm bands emit strongly in ionized PAHs, while the 11.2 μm does in neutral PAHs (Bernard-Salas et al. 2009). According to the PAH line-profile classifications by Peeters et al. (2002) and van Diedenhoven et al. (2004), BoBn 1's PAH line profiles belong to class B. Bernard-Salas et al. (2009) classified 10 of 14 Magellanic Clouds (MCs) PNe into class B based on Spitzer spectra. In measuring PAH band fluxes, we used local continuum subtracted spectrum by a spline function fitting. We followed Bernard-Salas et al. (2009) and measured integrated fluxes between 6.1 and 6.6 μm for the 6.2 μm PAH band, 7.2–8.3 μm for the 7.7 μm PAH band, 8.3–8.9 μm for the 8.6 μm PAH band, and 11.1–11.7 μm for the 11.2 μm PAH band. The observed PAH flux ratios I(6.2 μm)/I(11.2 μm) and I(7.7 μm)/I(11.2 μm) follow a correlation among MCs PNe, shown in Figure 2 of Bernard-Salas et al. (2009). BoBn 1 has a hot central star (>10⁵ K), so that ionized PAH might be dominant. However, these line ratios of BoBn 1 are somewhat lower than those of excited MCs PNe. One must take a look at a part of the neutral PAH emissions in a photodissociation region (PDR), too.

Zoom In Zoom Out Reset image size

We found a plateau between 10 and 14 μm, which are believed to be related to PAH clusters (Bernard-Salas et al. 2009). Meanwhile, MgS feature around 30 μm sometimes observed in C-rich PNe, was unseen in BoBn 1.

Data reduction and emission-line analysis were performed mainly with a long-slit reduction package noao.twodspec in IRAF.⁵ Data reduction was performed in a standard manner.

First, we made a zero-intensity level correction to all frames including flat lamp, object, and Th–Ar comparison frames using the overscan region of each frame and the mean bias frames. We also removed cosmic ray events and hot pixels from the object frames. Second, we trimmed the overscan region and removed scattered light from the flat lamp and object frames. Third, we made a CCD sensitivity correction to the object frames using the median flat frames. Fourth, we extracted a two-dimensional spectrum from each echelle diffraction order of each object frame and made a wavelength calibration using at least two Th–Ar frames taken before and after the object frame. We referred to the Subaru/HDS comparison atlas⁶ and a Th–Ar atlas. For the wavelength calibration, we fitted the wavelength dispersion against the pixel number with a fourth- or fifth-order polynomial function. With this order, any systematic trend did not show up in the residuals and the fitting appears to be satisfactory. We also made a distortion correction along the slit length direction using the mean Th–Ar spectrum as a reference. We fitted the slit image in the Th–Ar spectrum with a two-dimensional function. For the HDS spectra, we adopted fourth- and third-order polynomial functions for the wavelength and space directions, respectively. The fitting residual was of the order of 10⁻⁴ Å. For the UVES spectra, we adopted third- and second-order polynomial functions for the wavelength and space directions, respectively. The fitting residual was of the order of 10⁻³ Å. Fifth, we determined a sensitivity function using sky-subtracted standard star frames and obtained sky-subtracted and flux-calibrated two-dimensional PN spectra. The probable error in the flux calibration was estimated to be less than 5%. Finally, we made a spatially integrated one-dimensional spectrum, and we combined all the observed echelle orders using IRAF task scombine.

In measuring emission-line fluxes, we assumed that the line profiles were all Gaussian and we applied multiple Gaussian fitting techniques.

We have detected over 600 emission lines in total. Before proceeding to the chemical abundance analysis, it is necessary to correct the spectra for the effects of absorption due to Earth's atmosphere and interstellar reddening. The former was performed using experimental functions measured at the Keck observatories and the ESO/VLT. The interstellar reddening correction was made by determining the reddening coefficient at Hβ, c(Hβ). We fitted the observed intensity ratio of Hα to Hβ with the theoretical ratios computed by Storey & Hummer (1995). Two different situations are assumed for some lines: Case A assumes that the nebula is transparent to the lines of all series of hydrogen; Case B assumes the nebula is partially opaque to the lines of the Lyman series but is transparent for the Balmer series of hydrogen. Initially, we assumed that T = 10⁴ K and n = 10⁴ cm⁻³ in Case B, and we estimated c(Hβ) = 0.087 ± 0.004 from the HDS spectra. That is an intermediate value between Cahn et al. (1992) and Kwitter et al. (2003; see Table 1). For the UVES spectra, we estimated c(Hβ) = 0.066 by the same manner. From Seaton's (1979) relation c(Hβ) = 1.47E(B − V) one obtains E(B − V) = 0.06 for the HDS spectra and 0.04 for UVES spectra, which are comparable to the Galactic value (0.02) to the direction to BoBn 1 measured by the Galactic extinction model of Schlegel et al. (1998).

All of the line intensities were then de-reddened using the formula

$$\begin{equation} \log _{10}\left[\frac{I(\lambda)}{I{\rm (H\beta)}}\right] = \log _{10}\left[\frac{F(\lambda)}{F{\rm (H\beta)}}\right] + c({\rm H\beta })f(\lambda), \end{equation} \tag{ 1 }$$

where I(λ) is the de-reddened line flux, F(λ) is the observed line flux, and f(λ) is the interstellar extinction at λ. We adopted the reddening law of Cardelli et al. (1989) with the standard value of R_V = 3.1 for f(λ). We observed F(Hβ) = 2.56 × 10⁻¹³ ± 2.13 × 10⁻¹⁶ erg s⁻¹ cm⁻² (X(−Y) stands for X × 10^−Y, hereafter) within the 12 slit in the HDS observation. We estimated captured light from BoBn 1 using the image presented in Figure 1, to be about 86.8% of the light from BoBn 1 in the HDS observation. The intrinsic observed Hβ flux is 2.95(−13) ± 3.45(−16) erg s⁻¹ cm⁻² and the de-reddened Hβ flux is 3.63(−13) ± 6.47(−14) erg s⁻¹ cm⁻² including the error of c(Hβ).

We present the line profiles of selected ions in Figure 5. The observed wavelength at the time of observation was corrected to the averaged line-of-sight heliocentric radial velocity of +191.60 ± 1.25 km s⁻¹ among over 300 lines detected in the HDS spectra. The line profiles can be represented by a single Gaussian for weak forbidden lines such as [Ne v] λ3426 and metal recombination lines such as O ii λ4642. For the others, the profiles can be represented by the sum of two or three Gaussian components.

Zoom In Zoom Out Reset image size

Most of the detected lines are asymmetric profile, in particular the profiles of low ionization potential (IP) ions show strong asymmetry. The asymmetric line profiles are sometimes observed in bipolar PNe having an equatorial disk structure. The similar line profiles are also observed in the halo PN H4-1 (Otsuka et al. 2006). H4-1 has an equatorial disk structure and multi-polar nebulae. The elongated nebular shape of BoBn 1 (Figure 1) might indicate the presence of such an equatorial disk. The receding ionized gas (especially, low IP ions) from the observers around the central star would be strongly weakened by the equatorial disk. In contrast, the relatively large extent bipolar flows perpendicular to the equatorial disk might be unaffected by the disk. Due to such a geometry, we observe asymmetric line profiles.

In Table 6, we present twice the expansion velocity 2V_exp measured from selected lines. When we fit the line profile with two or three Gaussian components, we define that 2V_exp corresponds to the difference between the positions of the red- and blueshifted Gaussian peak components. We call the expansion velocity measured by this method "2V_exp(a)."

When we can fit line profile with single Gaussian, we determine 2V_exp from following equation:

$$\begin{equation} 2V_{\rm exp} = \big(V_{\rm FWHM}^2 - V_{\rm therm}^2 - V_{\rm instr}^2\big)^{1/2}, \end{equation} \tag{ 2 }$$

where V_FWHM is the velocity FWHM of each Gaussian component. V_therm and V_instr (∼9 km s⁻¹) are the thermal broadening and the instrumental broadening, respectively (Robinson et al. 1982). V_therm is represented by 21.4(T₄/A)^1/2, where T₄ is the electron temperature (in units of 10⁴ K) and A is the atomic weight of the target ion. For CELs, we adopted T₄ listed in Table 9. For ORLs, we adopted T₄ = 0.88. We call twice the expansion velocity measured by Equation (2) "2V_exp(b)." We converted 2V_exp(a) into 2V_exp(b) using the relation 2V_exp(b) = (1.74 ± 0.12) × 2V_exp(a) for 2V_exp(b) > 16 km s⁻¹, which can be applied only to BoBn 1. We adopted 2V_exp(b) as twice the expansion velocities for BoBn 1. The averaged 2V_exp(b) is 40.5 ± 3.28 km s⁻¹ among selected lines listed in Table 6 and 33.04 ± 2.61 km s⁻¹ among over 300 detected lines in the HDS spectra.

In Figure 6, we present relation between 2V_exp(b) and IP. When we assume that BoBn 1 has a standard ionized structure (i.e., high IP lines are emitted from close regions to the central star and low IP lines are from far regions), expansion velocity of BoBn 1 seems to be proportional to the distance from the central star. BoBn 1 might have Hubble type flows. We found that 2V_exp(b) values from ORLs are slightly smaller than CELs with the same IP, for example, O ii and [O iii] (35.5 eV) and Ne ii and [Ne iii] (41.0 eV). O ii and Ne ii might be emitted from colder regions than [O iii] and [Ne iii] do.

Zoom In Zoom Out Reset image size

3.3.1. CEL Diagnostics

We have detected a large number of CELs, useful for estimations of the temperatures (T) and densities (n). The electron temperature and density diagnostic lines analyzed here arise from various ions, which have a wide variety of IPs ranging from 0 ([N i] and [O i]) to 63.5 eV ([Ne iv]). We have examined the electron temperature and density structure within the nebula of BoBn 1 using 16 diagnostic line ratios. The [O i], [Ne iii], [Ne iv], [S ii], and [S iii] zone electron temperatures and [N i], C iii], and [Ne iii] zone electron densities are estimated for the first time. Electron temperatures and densities were derived from each diagnostic ratio for each line by solving level populations for a multi-level (≧5 for almost all the ions) atomic model using the collision strengths Ω_ij (j > i) and spontaneous transition probabilities A_ji for each ion from the references given in Table 7.

Table 7. Atomic Data References for CELs

Line	Transition Probabilities Aji	Collisional Strength Ωij
[C i]	Froese-Fischer & Saha (1985)	Johnson et al. (1987); Péquignot & Aldrovandi (1976)
[C ii]	Nussbaumer & Storey (1981); Froese-Fischer (1994)	Blum & Pradhan (1992)
C iii]	Wiese et al. (1996)	Berrington et al. (1985)
C iv	Wiese et al. (1996)	Badnell & Pindzola (2000); Martin et al. (1993)
[N i]	Wiese et al. (1996)	Péquignot & Aldrovandi (1976); Dopita et al. (1976)
[N ii]	Wiese et al. (1996)	Lennon & Burke (1994)
N iii]	Brage et al. (1995); Froese-Fischer (1983)	Blum & Pradhan (1992)
N iv]	Wiese et al. (1996)	Ramsbottom et al. (1994)
[O i]	Wiese et al. (1996)	Bhatia & Kastner (1995)
[O ii]	Wiese et al. (1996)	McLaughlin & Bell (1993); Pradhan (1976)
[O iii]	Wiese et al. (1996)	Lennon & Burke (1994)
[O iv]	Wiese et al. (1996)	Blum & Pradhan (1992)
[F ii]	Storey & Zeippen (2000); Baluja & Zeippen (1988)	Butler & Zeippen (1994)
[F iii]	Naqvi (1951)	See the text
[F iv]	Garstang (1951); Storey & Zeippen (2000)	Lennon & Burke (1994)
[Ne ii]	Saraph & Tully (1994)	Saraph & Tully (1994)
[Ne iii]	Mendoza (1983); Kaufman & Sugar (1986)	McLaughlin & Bell (2000)
[Ne iv]	Becker et al. (1989); Bhatia & Kastner (1988)	Ramsbottom et al. (1998)
[Ne v]	Kaufman & Sugar (1986); Bhatia & Doschek (1993)	Lennon & Burke (1994)
[S ii]	Verner et al. (1996); Keenan et al. (1993)	Ramsbottom et al. (1996)
[S iii]	Tayal & Gupta (1999)	Froese-Fischer et al. (2006)
[S iv]	Johnson et al. (1986); Dufton et al. (1982); Verner et al. (1996)	Dufton et al. (1982)
[Cl iii]	Mendoza & Zeippen (1982a); Kaufman & Sugar (1986)	Ramsbottom et al. (2001)
[Cl iv]	Mendoza & Zeippen (1982b); Ellis & Martinson (1984)	Galavis et al. (1995)
	Kaufman & Sugar (1986)
[Ar iii]	Mendoza (1983); Kaufman & Sugar (1986)	Galavis et al. (1995)
[Ar iv]	Mendoza & Zeippen (1982a); Kaufman & Sugar (1986)	Zeippen et al. (1987)
[Fe iii]	Garstang (1957); Nahar & Pradhan (1996)	Zhang (1996)
[Fe iv]	Froese-Fischer & Rubin (1998); Garstang (1958)	Zhang & Pradhan (1997)
[Kr iv]	Biémont & Hansen (1986)	Schöning (1997)
[Kr v]	Biémont & Hansen (1986)	Schöning (1997)
[Rb v]	Persson & Petterson (1984)	...
[Xe iii]	Biémont et al. (1995)	Schöning & Butler (1998)
Ba ii	Klose et al. (2002)	Schöning & Butler (1998)

Download table as:  ASCIITypeset image

The derived electron temperatures and densities are listed in Table 8. Figure 7 is the diagnostic diagram that plots the loci of the observed diagnostic line ratios on the n–T plane. This diagram shows that most CELs in BoBn 1 are emitted from T ∼ 12, 000–16,000 K and log₁₀ n ∼ 3.5 cm⁻³ ionized gas.

Zoom In Zoom Out Reset image size

Table 8. Plasma Diagnostics

Parameter	ID	Diagnostic	Ratio	Result
T	(1)	[Ne iv] (λ2422+λ2425)/(λ4715/16/25/26)	100.78 ± 10.84	14,920 ± 810
(K)	(2)	[O iii] (λ4959+λ5007)/(λ4363)	85.01 ± 4.32	13,650 ± 290
	(3)	[Ar iii] (λ7135)/(λ5192)	85.70 ± 35.63	13,330 ± 3310
	(4)	[Ne iii] (λ15.5 μm)/(λ3869+λ3967)	0.20 ± 0.01	13,050 ± 140
	(5)	[Ne iii] (λ3869+λ3967)/(λ3344)	333.67 ± 14.63	12,870 ± 170
	(6)	[N ii] (λ6548+λ6583)/(λ5755)	57.55 ± 1.77	12,000 ± 190a
	(7)	[S iii] (λ9069)/(λ6312)	7.42 ± 0.53	12,460 ± 490
	(8)	[O i] (λ6300+λ6363)/(λ5577)	68.65 ± 10.76	9520 ± 550
	(9)	[O ii] (λ3726+λ3729)/(λ7320+λ7330)	10.47 ± 0.21	12,100 ± 180b
	(10)	[S ii] (λ6716+λ6731)/(λ4069+λ4076)	13.84 ± 5.17	12,420−3590
		Averagec		13,050
		He i (λ7281)/(λ6678)	0.21 ± 0.01	9430 ± 310
		He i (λ7281)/(λ5876)	0.05 ± 0.01	7340 ± 110
		He i (λ6678)/(λ4471)	0.83 ± 0.02	7400+1070
		He i (λ6678)/(λ5876)	0.27 ± 0.01	9920 ± 310
		Average		8520
		(Balmer jump)/(H 11)		8840 ± 210
n	(11)	[N i] (λ5198)/(λ5200)	1.43 ± 0.03	1030 ± 130
(cm−3)	(12)	[O ii] (λ3726)/(λ3729)	1.65 ± 0.03	1510 ± 60
	(13)	C iii] (λ1906)/(λ1909)	1.39 ± 0.03	3590 ± 1000
	(14)	[Ar iv] (λ4711)/(λ4740)	1.05 ± 0.07	3960 ± 1090
	(15)	[Ne iii] (λ15.5 μm)/(λ36.0 μm)	12.11 ± 1.69	4400+9010
	(16)	[S ii] (λ6716)/(λ6731)	8.57 ± 0.03	5740 ± 1310
		Averagec		3370
		Balmer decrement		5000–10,000

Notes. ^aCorrected for recombination contribution to [N ii] λ5755 (see the text). ^bCorrected for recombination contribution to [O ii] λλ7320/30 (see the text). ^cFrom ions with IP > 13.6 eV.

Download table as:  ASCIITypeset image

First, we calculated electron densities assuming a constant electron temperature of 12,800 K. Estimated electron densities range from 1030 ([N i]) to 5740 cm⁻³ ([S ii]). Although [S ii] and [O ii] have similar IPs, a large discrepancy between their electron densities is found (see Table 8). Kniazev et al. (2008) and Kwitter et al. (2003) estimated [S ii] electron densities as large as 9600 and 7100 cm⁻³, respectively. Stanghellini & Kaler (1989), Copetti & Writzl (2002), and Wang et al. (2004) found that the [S ii] density is systematically larger than the [O ii] density in a large number of samples. The curve yielded by the [S ii] λ6716/31 ratio in the n versus T plane indicates higher electron density than critical density of these lines, 1600 and 4100 cm⁻³ at T = 12,800 K for λ6716 and λ6731, respectively (cf. 5740 cm⁻³ in Figure 7). This density discrepancy is not due to the errors in the [O ii] atomic data. Wang et al. (2004) also found the density discrepancy between [S ii] and [O ii] that might be likely caused by errors in the transition probabilities of [O ii] given by Wiese et al. (1996). In the case of BoBn 1, this possibility can be ruled out because we obtained similar [O ii] electron densities even when with the other transition probabilities. The high [S ii] density might be due to high-density blobs in the outer nebula. This could have contributed to producing the small [S ii] λ6717/λ6731 ratio and give rise to an apparently high density. Zhang et al. (2005b) pointed out the possibility that a dynamical plow by the ionization front effects yields large density of [S ii] because the IP of S⁺ is close to the H⁺ edge. Since the estimated upper limit to the [S ii] density is close to the H⁺ density derived from the Balmer decrement (see below), this explanation might be plausible. In BoBn 1, caution is necessary when using the [S ii] electron density.

Next, we calculated the electron temperature. An average electron density of 3370 cm⁻³, which excluded the n([N i]) and n([S ii]), was adopted when estimating electron temperatures except the T([O i]). [N i] and [O i] are representative of the very outer part of the nebula and probably do not coexist with most of the other ions. T([O i]) was, therefore, estimated using n([N i]).

To obtain the [N ii], [O ii], and [O iii] temperatures, it is necessary to subtract the recombination contamination to the [N ii] λ5755, [O ii] λλ7320/30, and [O iii] λ4363 lines, respectively. For [N ii] λ5755, Liu et al. (2000) estimated the contamination to [N ii] λ5755, I_R([N ii] λ5755) in the range 5000 K ≦ T ≦ 20,000 K as

$$\begin{equation} \frac{I_{R}( [{\rm N}\,\mathsc {ii}]\,\lambda 5755)}{I(\rm H\beta)} = 3.19\left(\frac{T_{\epsilon }}{10^4}\right)^{0.33}\times \frac{\rm N^{2+}}{\rm H^{+}}, \end{equation} \tag{ 3 }$$

where N²⁺/H⁺ is the doubly ionized nitrogen abundance. Adopting the value derived from the ORL analysis (see Section 3.6) and using Equation (3), we estimated I_R([N ii] λ5755)  ∼ 0.1, which is approximately 7% of the observed value. Given the corrected [N ii] λ5755 intensity, the [N ii] temperature is 12,000 K, which is 400 K lower than that obtained without taking into account the recombination effect.

The same effect also exists for the [O ii] λλ7320/30 lines. We estimated the recombination contribution using the doubly ionized oxygen abundance derived from O ii lines and the equation of Liu et al. (2000) for these lines in the range 5000 K ≦ T ≦ 10,000 K,

$$\begin{equation} \frac{I_{R}(\rm [O\,\mathsc {ii}]\,\lambda \lambda 7320/30)}{I(\rm H\beta)} = 9.36\left(\frac{T_{\epsilon }}{10^4}\right)^{0.44}\times \frac{\rm O^{2+}}{\rm H^{+}}. \end{equation} \tag{ 4 }$$

Using Equation (4), we estimated a contribution of ∼7 % of the observed value and obtained 12,100 K, lower by 700 K than that without the recombination contribution. For [O iii] λ4363, we estimated the recombination contribution using the O³⁺ abundance derived from the fine-structure line [O iv] λ25.9 μm adopting T([Ne iv]) and n([Ar iv]) and the equation of Liu et al. (2000). We chose the value from this line because O iii lines could be affected by star light excitation and the abundance derived from them could be erroneous. Assuming the ratio of O³⁺(ORLs)/O³⁺(CELs) = 10, we estimated the recombination contribution to [O iii] λ4363 less than 1% of the observed value, which has a negligible effect on the T([O iii]) derivation.

The electron temperature in BoBn 1 ranges from 9520 ([O i]) to 14,920 K ([Ne iv]). Our estimated electron temperatures except for [O ii] are comparable to those of Kwitter et al. (2003) and Kniazev et al. (2008). Their estimated temperatures are 12,400–13,720 K for [O iii], 11,320–11,700 K for [N ii], and 13,250 K for [Ar iii]. Note that the [O ii] electron temperature of 8000 K of Kwitter et al. (2003) was estimated adopting the [S ii] density of 7100 cm⁻³. Our n–T plane predicts that the [O ii] temperature is ∼10,000 K when adopting [S ii] density.

The ionic abundances derived from the CELs depend strongly on the electron temperature. In the case of [O iii] λ5007, for example, only 500 K change makes a difference of over 10% for O²⁺ abundance. It is therefore essential to find the proper electron temperature for each ionized stage of each ion. To that end, we examined the behavior of the electron temperature and density as a function of IP The upper panel of Figure 8 shows that T is increasing proportional to IP The observed behavior of T is consistent with the schematic picture of stratified physical conditions in ionized nebula, where the electron temperature of ions in the inner part should be hotter than that in the outer part. n is simply monotonically increasing up to ∼40 eV as IP, except for [S ii]. The [S ii] might be emitted in high-density blobs in the outer nebula as we mentioned above.

Zoom In Zoom Out Reset image size

To minimize the estimated error for ionic abundances due to electron temperature, we have assumed a seven-zone model for BoBn 1 by reference to Figure 8. Adopted T and n for each ion are presented in Table 9. T([O i]) and n([N i]) are adopted for ions in zone 0, which have <10 eV. T([N ii]) and n([S ii]) are adopted for ions in zone 1 (IP < 11.3 eV). T([N ii]) and n([O ii]) are for zone 2 (11.3–20 eV). T([S iii]) and n(C iii]) are for zone 3 (20–25 eV). For zone 4 (25–41 eV), 5 (41–63.5 eV), and zone 6 (>63.5 eV), we adopted n([Ar iv]) and T([O iii]), the averaged value from T([O iii]) and T([Ne iv]), and T([Ne iv]), respectively.

Table 9. Adopted T and n for CEL Ionic Abundance Calculations

Zone	Ions	T (K)	n (cm−3)
0	C0, N0, O0, Ba+	9520	1030
1	C+, S+	12,000	5740
2	O+, N+, F+, Fe2+	12,000	1510
3	C2+, Ne+, S2+, Cl2+, Xe2+	12,460	3590
4	N2+, O2+, Ne2+, Ar2+, Ar3+, S3+	13,650	3960
	Cl3+, F2+, Fe3+, Kr3+, Kr4+
5	C3+, N3+, O3+	14,290a	3960
6	Ne3+, Ne4+, F3+	14,920	3960

Note. ^aThe averaged value from T([O iii]) and T([Ne iv]).

Download table as:  ASCIITypeset image

3.3.2. ORL Diagnostics

We detected a large number of ORLs. C iii, iv, O ii, iii, iv, N ii, iii, and Ne ii are for the first time detected. To calculate ORL abundances, the electron temperature and density derived from ORLs are needed. We estimated the electron temperature using the Balmer discontinuity and He i line ratios, and the electron density using the Balmer decrement. The results are listed in Table 8.

The ratio of the jump of continuum emission at the Balmer limit at 3646 Å (BJ) to a given hydrogen emission line depends on the electron temperature. Following Liu et al. (2001), we use this ratio to determine the electron temperature. This temperature, T(BJ), is used to deduce ionic abundances from ORLs. Defining BJ as I(3646 Å) − I(3681 Å), and taking the emissivities of H i Balmer lines and H i, He i, and He ii continuum emissivities, Liu et al. (2001) gave the following equation:

$$\begin{eqnarray} T_{\epsilon }({\rm BJ}) &=& 368\left(1 + 0.259{\rm \frac{{\it N}(He^{+})}{{\it N}(H^{+})}} + 3.409{\rm \frac{{\it N}(He^{2+})}{{\it N}(H^{+})}}\right) \nonumber \\ &&\times \left(\frac{I(3646\,\AA) - I(3681\,\AA)}{I{\rm (H\,11)}}\right)^{-1.5} \end{eqnarray} \tag{ 5 }$$

[$I(\rm 3646\,\AA)-I(\rm 3681\,\AA)]/I({\rm H\,11})$ is in units of Å⁻¹. T(BJ) is valid over a range from 4000 to 20,000 K. The process was repeated until self-consistent values for the N(He⁺)/N(H⁺), N(He²⁺)/N(H⁺) and T(BJ) were reached, we estimated T(BJ) of 8840 K.

We estimated the He i electron temperature T(He i) from the ratios of He i λ7281/λ6678, λ7281/λ5876, λ6678/λ4471, and λ6678/λ5876 assuming a constant electron density = 10⁴ cm⁻³, estimated from the Balmer decrement as described below. All the He i line ratios we chose here are insensitive to the electron density. We adopted the emissivities of He i from Benjamin et al. (1999). We estimated T(He i) values as 7340–9920 K. The T(He i) from λ7281/λ6678 ratio appears to be the most reliable value because (1) He i λ6678 and λ7281 levels have the same spin as the ground state and the Case B recombination coefficients for these lines by Benjamin et al. (1999) are more reliable than the other He i λ4471 and λ5876 and (2) the effect of interstellar extinction is less due to the close wavelengths. We adopted 9430 K as T(He i). Note that in Figure 7 the electron temperatures and densities derived from the H i and He i are not presented.

The intensity ratios of the high-order Balmer lines Hn (n > 10, n: the principal quantum number of the upper level) to a lower Balmer line, e.g., Hβ, are also sensitive to the electron density. In Figure 9, we plot the ratio of higher-order Balmer lines to Hβ with the theoretical values by Storey & Hummer (1995) for the cases of electron temperature of 8840 K (=T(BJ)) and electron densities of 1000, 5000, 10⁴, and 10⁵ cm⁻³. This diagram indicates that the electron density in the ORL emitting region is between 5000 and 10⁴ cm⁻³, which is fairly compatible with the CEL electron densities. Zhang et al. (2005a) estimated T(He i) from the ratio of He i λ7281/λ6678 for 48 PNe and found that high-density blobs (10⁵–10⁶ cm⁻³) might be present in nebula if T(He i) ≃ T(BJ). Figure 9 indicates that such components do not coexist in BoBn 1.

Zoom In Zoom Out Reset image size

The derived CEL ionic abundances X^{m +}/H⁺ are listed in Table 10. X^m+ and H⁺ are the number densities of an m times ionized ion and ionized hydrogen, respectively. To estimate ionic abundances, we solved level populations for a multi-level atomic model. In the last one of the line series of each ion, we present the adopted ionic abundances in bold face characters. These values are estimated from the line intensity-weighted mean or average if there are two or more available lines. Over 10 ionic abundances of some elements are estimated for the first time. These newly estimated ionic abundance would reduce the uncertainty of estimation of each elemental abundance, in particular, N, O, F, Ne, S, Fe, and some s-process elements, which are key elements to the nucleosynthesis in low-mass stars and chemical evolution in galaxies.

Table 10. Ionic Abundances from CELs

Xm+	λlaba	I(λlab)	T	n	Xm+/H+
	(Å/μm)	[I(Hβ) = 100]	(K)	(cm−3)
C0	8727.12	7.93(−2) ± 4.73(−3)	9520	1030	4.74(−7) ± 9.34(−8)
C+	2324	3.51(+1) ± 4.46(0)	12000	5740	2.30(−5) ± 3.48(−6)
C2+	1906	8.39(+2) ± 1.27(+1)	12460	3590	7.71(−4) ± 1.59(−4)
	1908	6.03(+2) ± 1.03(+1)			7.70(−4) ± 1.59(−4)
					7.71(−4) ± 1.59(−4)
C3+	1548	1.05(+3) ± 2.02(+1)	14290	3960	2.42(−4) ± 4.47(−5)
	1551	5.19(+2) ± 1.50(+1)			2.36(−4) ± 4.38(−5)
					2.40(−4) ± 4.44(−5)
N0	5197.90	2.73(−1) ± 7.90(−3)	9520	1030	4.81(−7) ± 1.01(−7)
	5200.26	1.91(−1) ± 5.47(−3)			4.82(−7) ± 9.67(−8)
					4.82(−7) ± 9.90(−8)
N+	5754.64	1.23(0) ± 1.40(−2)	12000	1510	7.20(−6) ± 4.74(−7)
	6548.04	1.56(+1) ± 8.34(−1)			5.84(−6) ± 3.75(−7)
	6583.46	5.04(+1) ± 1.28(0)			6.40(−6) ± 2.79(−7)
					6.27(−6) ± 3.02(−7)
N2+	1750	4.81(+1) ± 1.32(+1)	13650	3960	6.24(−5) ± 1.88(−5)
N3+	1485	4.58(+1) ± 1.78(+1)	14290	3960	3.83(−5) ± 1.65(−5)
O0	5577.34	1.70(−2) ± 2.64(−3)	9520	1030	2.06(−6) ± 7.06(−7)
	6300.30	8.72(−1) ± 1.55(−2)			2.04(−6) ± 3.97(−7)
	6363.78	2.91(−1) ± 9.91(−3)			2.13(−6) ± 4.20(−7)
					2.06(−6) ± 4.03(−7)
O+	3726.03	1.09(+1) ± 6.85(−2)	12000	1510	4.00(−6) ± 2.23(−7)
	3728.81	6.61(0) ± 9.99(−2)			4.03(−6) ± 2.42(−7)
	7319	1.02(0) ± 2.00(−2)			7.13(−6) ± 5.67(−7)
	7330	7.81(−1) ± 1.38(−2)			6.74(−6) ± 5.30(−7)
					4.01(−6) ± 2.30(−7)
O2+	4363.21	5.57(0) ± 9.73(−2)	13650	3960	4.76(−5) ± 4.87(−6)
	4931.23	4.36(−2) ± 3.88(−3)			4.37(−5) ± 4.58(−6)
	4958.91	1.22(+2) ± 5.96(0)			4.80(−5) ± 3.52(−6)
	5006.84	3.51(+2) ± 2.18(+1)			4.76(−5) ± 3.94(−6)
					4.77(−5) ± 3.83(−6)
O3+	25.9	1.25(+1) ± 1.36(−1)	14290	3960	3.41(−6) ± 8.21(−8)
F+	4789.45	5.58(−2) ± 5.20(−3)	12000	1510	2.16(−8) ± 2.23(−9)
	4868.99	1.34(−2) ± 3.00(−3)			1.66(−8) ± 3.79(−9)
					1.98(−8) ± 2.74(−9)
F2+	5721.20	2.70(−2) ± 2.93(−3)	13650	3960	6.59(−7) ± 1.03(−7)
	5733.05	2.68(−2) ± 5.43(−3)			6.70(−7) ± 1.55(−7)
					6.65(−7) ± 1.29(−7)
F3+	3996.92	4.09(−2) ± 2.53(−3)	14920	3960	1.47(−8) ± 2.21(−9)
	4059.90	1.19(−1) ± 3.72(−3)			1.51(−8) ± 2.12(−9)
					1.50(−8) ± 2.14(−9)
Ne+	12.8	2.49(0) ± 8.17(−2)	12460	3590	2.97(−6) ± 1.14(−7)
Ne2+	3342.42	8.47(−1) ± 2.75(−2)	13650	3960	6.40(−5) ± 8.10(−6)
	3868.77	2.17(+2) ± 1.05(+1)			8.41(−5) ± 6.75(−6)
	3967.46	6.39(+1) ± 4.13(−1)			5.94(−5) ± 3.82(−6)
	4011.60	1.48(−2) ± 3.92(−3)			9.71(−5) ± 2.65(−5)
	15.6	1.61(+2) ± 1.30(0)			9.15(−5) ± 1.27(−6)
	36	1.33(+1) ± 1.84(0)			9.07(−5) ± 1.26(−5)
					8.32(−5) ± 4.33(−6)
Ne3+	2423.50	1.71(+1) ± 1.78(0)	14920	3960	3.97(−6) ± 8.99(−7)
	4714.25	5.52(−2) ± 2.76(−3)			4.35(−6) ± 1.23(−7)
	4715.80	1.87(−2) ± 2.12(−3)			5.05(−6) ± 1.52(−6)
	4724.15	5.08(−2) ± 1.88(−3)			3.60(−6) ± 1.01(−6)
	4725.62	4.52(−2) ± 2.58(−3)			3.43(−6) ± 9.78(−7)
					3.97(−6) ± 9.01(−7)
Ne4+	3345.83	3.22(−1) ± 1.98(−2)	14920	3960	1.99(−7) ± 3.40(−8)
	3425.87	8.71(−1) ± 8.29(−3)			1.97(−7) ± 3.15(−8)
					1.98(−7) ± 3.22(−8)
S+	4068.60	3.95(−1) ± 8.37(−3)	12000	5740	4.05(−8) ± 1.96(−9)
	4076.35	2.45(−2) ± 9.13(−3)			7.44(−9) ± 2.79(−9)
	6716.44	1.23(−1) ± 4.55(−3)			1.03(−8) ± 5.12(−10)
	6730.81	2.16(−1) ± 4.88(−3)			1.03(−8) ± 4.01(−10)
					1.03(−8) ± 4.41(−10)
S2+	6312.10	4.77(−2) ± 4.75(−3)	12460	3590	6.87(−8) ± 1.07(−8)
	9068.60	3.78(−1) ± 9.97(−3)			7.34(−8) ± 4.79(−9)
	18.7	6.92(−1) ± 4.67(−2)			6.81(−8) ± 4.82(−9)
					6.99(−8) ± 5.06(−9)
S3+	10.5	1.92(0) ± 5.03(−2)	13650	3960	5.28(−8) ± 1.51(−9)
Cl2+	5517.66	1.81(−2) ± 2.82(−3)	12460	3590	1.36(−9) ± 2.42(−10)
	8500.20	<1.81(−3)			<2.79(−9)
Cl3+	8046.30	2.05(−2) ± 2.60(−3)	13650	3960	7.82(−10) ± 1.04(−10)
Ar2+	5191.82	3.90(−3) ± 1.61(−3)	13650	3960	1.23(−8) ± 5.17(−9)
	7135.80	2.73(−1) ± 1.14(−2)			1.30(−8) ± 7.41(−10)
	7751.10	6.11(−2) ± 2.51(−3)			1.22(−8) ± 6.83(−10)
					1.29(−8) ± 7.30(−10)
Ar3+	4711.37	9.40(−2) ± 5.23(−3)	13650	3960	7.53(−9) ± 5.46(−10)
	4740.17	8.99(−2) ± 3.71(−3)			7.58(−9) ± 4.60(−10)
	7170.50	6.53(−3) ± 7.75(−4)			4.32(−8) ± 6.13(−9)
	7262.70	4.58(−3) ± 3.99(−3)			3.52(−8) ± 3.08(−8)
					7.56(−9) ± 5.04(−10)
Fe2+	4881.00	2.14(−2) ± 4.84(−3)	12000	1510	1.16(−8) ± 2.80(−9)
	5270.40	2.19(−2) ± 3.69(−3)			1.04(−8) ± 1.79(−9)
					1.10(−8) ± 2.29(−9)
Fe3+	6740.63	1.58(−2) ± 4.81(−3)	13650	3960	1.02(−7) ± 3.26(−8)
Kr3+	5346.02	4.56(−3) ± 2.02(−3)	13650	3960	1.41(−10) ± 6.31(−11)
	5867.70	<8.20(−3)			<1.89(−10)
Kr4+	6256.06	<6.27(−3)	13650	3960	<4.09(−10)
	8243.39	<5.12(−3)			<3.45(−10)
					<3.77(−10)
Xe2+	5846.66	<1.56(−3)	12460	3590	<2.30(−11)
Ba+	4934.08	6.42(−3) ± 1.80(−3)	9550	1030	1.90(−10) ± 6.45(−11)
	6141.70	3.43(−3) ± 7.66(−4)			2.12(−10) ± 6.42(−11)
					1.98(−10) ± 6.44(−11)

Note. ^aFor emission lines in 1 μm or longer wavelengths, the wavelength unit is μm; for the others, the unit is Å.

Download table as:  ASCIITypeset images: 1 2

Ne²⁺ (zone 4 ion) and S²⁺ (zone 3) abundances are derived from CELs seen in both the UV–optical and mid-infrared regions. CELs in the mid-infrared, namely, fine-structure lines, have an advantage in derivations of ionic abundances. Since the excitation energy of the fine-structure lines is much lower than that of the other transition lines, ionic abundances derived from these lines are nearly independent of the electron temperature or temperature fluctuation in the nebula. Note that these ionic abundances derived from fine-structure lines are almost consistent with those from other transition lines. This means that the adopted electron temperature and density for the ions in zones 3 and 4 are appropriate at least.

Followings are short comments on derivations of ionic abundances. We subtract the [O iii] λ2322 contamination from C ii] λ2324 using the theoretical intensity ratio [O iii] I(λ2322)/(λ4363) of 0.24 and then estimate the C⁺ abundance. The N⁺ and O⁺ abundances are derived only from [N ii] λλ6548/83 and [O ii] λλ3726/29 to avoid recombination contamination, respectively, while The Ne⁺ abundance is calculated from [Ne ii] λ12.8 μm by solving a two level atomic model.

We have detected two fluorine lines [F iv] λλ3996, 4060 and estimate an F³⁺ abundance of 1.50(−8). Otsuka et al. (2008a) detected these fluorine lines and estimated an F³⁺ abundance of 5.32(−8). The F³⁺ abundance discrepancy between Otsuka et al. (2008a) and the present work is due to different adopted electron temperature and c(Hβ) values. We have detected candidates of [F ii] λλ4790, 4868 and [F iii] λλ5721/33. In the previous section, we confirmed that BoBn 1 has no high-density components larger than the critical density of these [F iii] lines. The critical density of [F ii] λλ4790/4868 is ∼2 × 10⁶ cm⁻³, and that of [F iii] λλ5721/33 is ∼8 × 10⁶ cm⁻³. Therefore, the effect of collisional de-excitation is negligibly small. Accordingly, the ratios of [F ii] I(λ4790)/(λ4868) and [F iii] I(λ5733)/(λ5721) depend on their transition probabilities. When adopting the transition probabilities by Baluja & Zeippen (1988) for [F ii] and Naqvi (1951) for [F iii], the theoretical intensity ratios of [F ii] I(λ4790)/(λ4868) and [F iii] I(λ5721)/(λ5733) are ∼3.2 and ∼1, which are in agreement with our measurements (4.2 ± 1.0 for [F ii] and 1.0 ± 0.3 for [F iii]). Hence, these four emission lines can be identified as [F ii] λλ4790, 4868 and [F iii] λλ5721/33. The F²⁺ and F³⁺ abundances are estimated from the each detected line by solving the statistical equilibrium equations for the lowest five energy levels. For [F iii] lines the relevant collision strength has not been calculated. However, since F²⁺ is isoelectronic with Ne³⁺, and collision strengths for the same levels along an isoelectronic sequence tend to vary with effective nuclear charge (Seaton 1958). We therefore assume that the collision strengths for [F iii] are ∼22% smaller than those for [Ne iv] and estimate F³⁺ abundances. Otsuka et al. (2008a) showed a correlation between [Ne/Ar] and [F/Ar] in PNe, suggesting that Ne and F were synthesized in the same layer and carried to the surface by the TDU. If this is the case, the ionic abundance ratios of F²⁺ (IP = 35 eV) and F³⁺ (62.7 eV) to F⁺ (17.4 eV) should be comparable to those of Ne²⁺ (41 eV) and Ne³⁺ (63.5 eV) to Ne⁺ (21.6 eV). Indeed, these ionic abundance ratios follow our prediction; F⁺:F²⁺:F³⁺ ∼1:34:1 and Ne⁺:Ne²⁺:Ne³⁺ ∼1:28:1. This means that our identifications of two [F ii] and [F iii] lines and the estimated ionic abundances are reliable. So far, fluorine has found in only a handful of PNe (see Zhang & Liu 2005; Otsuka et al. 2008a). Among them, BoBn 1 appears to be the most F-rich PN.

We subtract the contribution to [Cl iii] λ8500.2 due to C iii λ8500.32 using the C³⁺ ORL abundance and give upper limit of Cl²⁺ abundance from this line. The adopted Cl²⁺ abundance is from [Cl iii] λ5517 only. The Ar³⁺ abundance is from [Ar iv] λλ4717/40. We adopt the S⁺ abundance based on [S ii] lines except [S ii] λ4068, because [S ii] λ4068 could be partially blended with C iii λ4068. To estimate Fe²⁺ and Fe³⁺ abundances, we solved a 33 level model (from ⁵D₃ to b³P₂) for [Fe iii] and a 18 level model (from ⁶S_5/2 to ²F_5/2) for [Fe iv]. We adopt the transition probabilities of [Fe iv] recommended by Froese-Fischer & Rubin (1998). For those not considered by Froese-Fischer & Rubin (1998), the values by Garstang (1958) were adopted.

We have detected 10 emission-line candidates of krypton (Kr), rubidium (Rb), xenon (Xe), and barium (Ba). Kr and Rb are light s-process elements (30 ≦ Z ≦ 40, Z: atomic number), Xe and Ba are heavy s-process (Z ≧ 41). Kr has been detected in nearly 100 PNe (Sterling & Dinerstein 2008), while the latter three s-process elements have been detected in only a handful of PNe (Sharpee et al. 2007). Selected line profiles of these candidates are presented in Figure 10. The Kr³⁺, Kr⁴⁺, Xe²⁺, and Ba⁺ abundances in this object are estimated for the first time.

Zoom In Zoom Out Reset image size

We have detected two nebular lines of [Kr iv] λλ5346.7, 5867.7 (⁴S^o_3/2 − ²D^o_5/2 and ⁴S^o_3/2 − ²D^o_3/2, respectively). For [Kr iv] λ5346.7, the possibility of blending with C iii λ5345.85 (multiplet V13.01) is low, and the contribution to this [Kr iv] line is probably negligible because other V13.01 C iii lines are not detected. For [Kr iv] λ5867.7, we estimated the contamination from He ii λ5867.7 using theoretical ratios of He ii I(λ5867.7) to I(λ5828.4), I(λ5836.5), I(λ5857.3), I(λ5882.12), and I(λ5896.8) given by Storey & Hummer (1995) assuming T = 8840 K and n = 10⁴ cm⁻³, from which the contribution from He ii λ5867.7 is estimated to be ∼64 %. In Figure 10, we present the isolated [Kr iv] λ5867.7 profile. The observed ratio of [Kr iv] I(λ5867.7)/I(λ5346.7) (<1.8) is comparable to the theoretical value (∼1.5) if the population at each energy level does not exceed the critical densities of ∼1.5(+6) cm⁻³ (²D^o_5/2) and ∼2(+7) cm⁻³ (²D^o_3/2). We therefore identified these two lines as [Kr iv] λλ5346.7, 5867.7.

[Kr v] λ8243.4 (³P₂ − ¹D₂) is likely to be blended with the blue wing of the Paschen series H i λ8243.7 (n = 3–43). Using theoretical ratios of H i I(λ8243.7) to I(λ8247.7) and I(λ8245.6) given by Storey & Hummer (1995), we subtracted the contribution of H i λ8243.7, then estimated the intensity of [Kr v] λ8243.4. We found another nebular line [Kr v] λ6256.1 (³P₁ − ¹D₂). The theoretical intensity ratio of I(λ6256.1)/I(λ8243.7) (∼1.1) is in good agreement with ours (1.2).

[Xe iii] λ5846.8 (³P₂ − ¹D₂) appears to be blended with He ii λ5846.7. We subtracted the He ii λ5846.7 contribution from it using the theoretical ratios of I(λ5846.7) to I(λ5828.4), I(λ5836.5), I(λ5857.3), I(λ5882.12), and I(λ5896.8) given by Storey & Hummer (1995), then obtained an upper limit to the intensity of [Xe iii] λ5846.8. In Figure 10, we present the isolated [Xe iii] λ5846.8 profile.

Two Ba ii recombination lines λλ4934, 6141.7 (6s²S_1/2 − 6p²P^o_1/2 and 5d²D_5/2 − 6p²P^o_3/2) are detected. Following Sharpee et al. (2007), we estimated the Ba⁺ abundances adopting transition electron temperature and density (zone 0). The Ba⁺ abundances from these lines are in good agreement each other.

We have detected a candidate [Rb v] λ5363.6 (auroral line; ⁴S^o_3/2 − ²D^o_3/2). Rb is one of the important elements as tracers of the neutron density. In the case of NGC 7027, Sharpee et al. (2007) argued the possibility that this line is O ii λ5363.8 (4fF²[4]^o_7/2 − 4d'²F_7/2). They also suggested that the intensity of O ii λ5363.8 is comparable to O ii λ4609.4 (3d²D_5/2 − 4fF²[4]^o_7/2), arising from the lower level of O ii λ5363.8. In BoBn 1, based on the ORL O²⁺ abundance of 1.45(−4) from the 3d–4f O ii lines (see the next section) the expected intensity of O ii λ4609.4 is ∼6.8(−5)I(Hβ), which is lower than the observed intensity of the [Rb v] λ5363.6 candidate. No 4fF²[4]^o − 4d'²F O ii lines are detected in BoBn 1. Therefore, we consider that the detected line is [Rb v] λ5363.6. Since there are no available collision strengths for this line at present, we do not estimate an Rb⁴⁺ abundance.

We have detected many ORLs of helium, carbon, nitrogen, oxygen, and neon. To our knowledge, the nitrogen, oxygen, and neon ORLs are detected for the first time from this PN. These lines provide us with a new independent method to derive chemical abundances for BoBn 1. The recombination coefficient depends weakly on the electron temperature (∝ T^−1/2). The ionic abundances are, therefore, insignificantly affected by small-scale fluctuations of electron temperature. This is the most important advantage of this determination method. The ionic abundances from recombination lines are robust against uncertainty in electron temperature estimation.

The ORL ionic abundances X^m+/H⁺ are derived from

$$\begin{equation} \frac{{\rm X^{\it m+}}}{{\rm H^{+}}} = \frac{\alpha ({\rm H\beta })}{\alpha ({\rm X^{\it m+}})} \frac{\lambda ({\rm X^{\it m+}})}{\lambda ({\rm H\beta })}\frac{I({\rm X^{\it m+}})}{I({\rm H\beta })}, \end{equation} \tag{ 6 }$$

where α(X^m+) is the recombination coefficient for the ion X^m+. For calculating ORL ionic abundances, we adopted T of 8800 K and n of 10⁴ cm⁻³ from the hydrogen recombination spectrum.

Effective recombination coefficients for the lines' parent multiplets were taken from the references listed in Table 11. The recombination coefficients for each multiplet at a given electron density were calculated by fitting the polynomial functions of T. The recombination coefficient of each line was obtained by a branching ratio, B(λ_i), which is the ratio of the recombination coefficient of the target line, α(λ_i) to the total recombination coefficient, ∑_iα(λ_i) in a multiplet line. To calculate the branching ratio, we referred to Wiese et al. (1996) except for O ii 3p–3d and 3d–4f transitions and Ne ii. For O ii 3p–3d and 3d–4f transition lines, the branching ratios were provided by Liu et al. (1995) based on intermediate coupling. For Ne ii, Kisielius et al. (1998) provided the branching ratios based on LS-coupling.

The estimated ORL ionic abundances are listed in Tables 12 and 13. In general, a Case B assumption applies to the lines from levels having the same spin as the ground state, and a Case A assumption applies to lines of other multiplicities. In the last one of the line series of each ion, we present the adopted ionic abundance and the error, which are estimated from the line intensity-weighted mean.

3.6.1. Helium

3.6.2. Carbon

3.6.3. Nitrogen

3.6.4. Oxygen

3.6.5. Neon

If the ionic abundances in all ionization stages are known, an elemental abundance will be simply the sum of its ionic abundances. Actually, it is, however, impossible to probe all of the ionization stages of an element using UV to mid-infrared spectra. To estimate elemental abundances, we must correct for unobserved ionic abundances. This correction was performed using ionization correction factors, ICF(X). ICFs(X) for each element are listed in Table 14.

Table 14. Adopted Ionization Correction Factors (ICFs)

X	Line	ICF(X)	X/H
He	ORLs	1	He++He2+
C	CELs	1	C++C2++C3+
	ORLs	$\rm \left(\frac{N}{N^{2+} + N^{3+}}\right)$a	ICF(C)$\rm \left(C^{2+}+C^{3+}+C^{4+}\right)$
N	CELs	1	N++N2++N3+
	ORLs	$\rm \left(\frac{N}{N^{2+}+N^{3+}}\right)$a	ICF(N)$\rm \left(N^{2+}+N^{3+}\right)$
O	CELs	1	O++O2++O3+
	ORLs	$\rm \left(\frac{O}{O^{2+}}\right)$a	ICF(O)O2+
F	CELs	$\rm \left(\frac{Ne}{Ne^{+}+Ne^{2+}+Ne^{3+}}\right)$a	ICF(F)$\rm \left(F^{+}+F^{2+}+F^{3+}\right)$
Ne	CELs	1	Ne++Ne2++Ne3++Ne4+
	ORLs	$\rm \left(\frac{Ne}{Ne^{2+}}\right)$a	ICF(Ne)Ne2+
S	CELs	$\rm \left(\frac{N}{N^{+}+N^{2+}}\right)$a	ICF(S)$\rm \left(S^{+}+S^{2+}+S^{3+}\right)$
Cl	CELs	$\rm {\left(\frac{O}{O^{2+}}\right)}$a	ICF(Cl)$\rm \left(Cl^{2+}+Cl^{3+}\right)$
Ar	CELs	$\rm \left(\frac{Ne}{Ne^{+}+Ne^{2+}+Ne^{4+}}\right)$a	ICF(Ar)$\rm \left(Ar^{2+}+Ar^{3+}\right)$
Fe	CELs	$\rm {\left(\frac{O}{O^{+}+O^{2+}}\right)}$a	ICF(Fe)$\rm \left(Fe^{2+}+Fe^{3+}\right)$
Kr	CELs	$\rm \frac{Cl^{2+}+Cl^{3+}}{Cl^{3+}}$	ICF(Kr)Kr3++Kr4+
Xe	CELs	$\rm \frac{S}{S^{2+}}$	ICF(Xe)Xe2+
Ba	CELsb	1	Ba+

Notes. ^aIonic and elemental abundances derived from CELs. ^bThe value is the lower limit (see the text).

Download table as:  ASCIITypeset image

3.7.1. Helium, Carbon, Nitrogen, Oxygen, and Neon

3.7.2. Other Elements

The resultant elemental abundances are presented in Table 15. We recognized that BoBn 1 is a C-, N-, and Ne-rich PN: the [C/O], [N/O], and [Ne/O] abundances from ORLs are +1.23, +1.12, and +0.81. The ratios derived from CELs are +1.58, +1.10, and +1.04, respectively. Comparing the C, N, O, Ne ORL and CEL abundances, ORLs might be emitted from O-, Ne-rich region. The ORL C, N, O, Ne abundances are larger by 0.14–0.49 dex than the CEL abundances. We need to look for reasons for the abundance discrepancy.

We derived ionic and elemental abundances using CELs and ORLs and found somewhat large abundance discrepancies between them. So far, abundance discrepancies have been found in about 90 Galactic disk PNe, three Magellanic PNe (Tsamis et al. 2003, 2004; Liu et al. 2004; Robertson-Tessi & Garnett 2005; Wesson et al. 2005, etc.), and one halo PN (DdDm 1; Otsuka et al. 2009). We define the ionic abundance discrepancy factor (ADF) as the ratio of the ORL to the UV or optical CEL abundances. In BoBn 1, the ADFs are 0.98 ± 0.21 for C²⁺, 2.39 ± 0.45 for C³⁺, 4.21 ± 1.59 for N²⁺, 0.67 ± 0.29 for N³⁺, 3.05 ± 0.54 for O²⁺, and 1.82 ± 0.39 for Ne²⁺, respectively

Up to now, three models have been proposed to explain abundance discrepancies in PNe: temperature fluctuations, high-density components, and hydrogen-deficient cold components. We examine what can cause the abundance discrepancies in BoBn 1 using these models.

4.1.1. Temperature Fluctuations

The emissivities of the CELs increase exponentially as the electron temperature becomes higher. The electron temperature derived from the CELs, T(CELs) will be indicative of the hot region nearby the radiation source. If we adopt T(CELs) for abundance estimations using the CELs, the ionic abundances might be underestimated.

Peimbert (1967) considered the effect of electron temperature fluctuation in a nebula, which sometimes gave high electron temperature, on the determinations of the ionic abundances derived from CELs. For example, Torres-Peimbert et al. (1980) characterized the electron temperature fluctuations in term of t² as the cause of the abundance discrepancy between CELs and ORLs. Assuming the validity of the temperature fluctuation paradigm, the comparison of the ionic abundances derived from CELs and ORLs may provide an estimation of t².

The relation between ADFs and T(CELs) is presented in Figure 11. We recognize that ADFs would approach to 1 if T(CELs) for each zone dropped by >1000 K. Using the formulations for temperature fluctuations given by Peimbert (1967), we have estimated the t² parameter and the mean electron temperatures T₀ for each zone. The resultant t² and T₀ are listed in Table 16. The derived t² = 0.027 ± 0.011 indicates that the temperature fluctuations are ∼16 % inside nebula. The temperature fluctuation in BoBn 1 is low, compared with typical Galactic PNe, i.e., t² < 0.1 (cf. Zhang et al. 2004).

Zoom In Zoom Out Reset image size

Table 16. t² and T₀ for Each Zone

Zone	t2	T0
0	0.027 ± 0.011	8920 ± 840
1, 2	0.027 ± 0.011	11540 ± 400
3	0.027 ± 0.011	12220 ± 630
4	0.027 ± 0.011	12950 ± 610
5	0.027 ± 0.011	12870 ± 730
6	0.027 ± 0.011	12790 ± 840

Download table as:  ASCIITypeset image

When we take the temperature fluctuation effect into account, the derived ADFs become lower than the t² = 0 case; ADFs are 0.87 ± 0.25 for C²⁺, 1.12 ± 0.38 for C³⁺, 2.95 ± 1.32 for N²⁺, 0.32 ± 0.16 for N³⁺, 2.63 ± 0.55 for O²⁺, and 1.69 ± 0.37 for Ne²⁺, respectively. The great improvements of C²⁺ and C³⁺ might have been caused by the large temperature dependency of C iii] λλ1906/09 and C iv λλ1549/50; the energy difference between upper and lower level, Δ E = kΔ T, where Δ T = 75,380 K and 44,820 K, respectively. This also implies that C²⁺ and C³⁺ abundances from ORLs are more reliable than those from CELs. Concerning O²⁺ ADFs, large discrepancy still exists. Since compared with the C iii] λλ1906/09 and C iv λλ1549/50 lines, the [O iii] nebular lines depend more weakly on the electron temperature (Δ T ∼ 14,100 K), so we realize that the temperature fluctuation model alone cannot improve CEL O²⁺ over >1 dex. We need to seek other explanations for the large ADF(O²⁺). Potentially this model can explain the discrepancies of N²⁺ and Ne²⁺ abundances, taking into account the uncertainties of measured fluxes of the observed ORLs N ii and Ne ii.

In Table 17, we present elemental abundances from the CELs and ORLs, taking into account the above temperature fluctuations. The CEL C, N, and Ne abundances become comparable to the ORL abundances. For the O abundance, there still exists a large discrepancy.

Table 17. The Elemental Abundances Derived from CELs and ORLs in the Case of t² ≠ 0

X	X/H			log(X/H)+12a			[X/H]b
	CELs	ORLs		CELs	ORLs		CELs	ORLs
He	...	1.18(−1) ± 2.12(−3)		...	11.07 ± 0.01		...	+0.17 ± 0.01
C	1.44(−3) ± 3.31(−4)	1.44(−3) ± 4.96(−4)		9.16 ± 0.10	9.16 ± 0.16		+0.77 ± 0.11	+0.77 ± 0.16
N	1.77(−4) ± 5.37(−5)	2.99(−4) ± 1.45(−5)		8.25 ± 0.14	8.48 ± 0.23		+0.42 ± 0.18	+0.66 ± 0.26
O	6.35(−5) ± 7.38(−6)	1.67(−4) ± 3.98(−5)		7.80 ± 0.05	8.22 ± 0.11		−0.89 ± 0.07	−0.47 ± 0.12
F	5.43(−7) ± 1.57(−7)	...		5.73 ± 0.13	...		+1.27 ± 0.14	...
Ne	1.01(−4) ± 9.26(−6)	1.71(−4) ± 4.08(−5)		8.00 ± 0.04	8.23 ± 0.11		+0.13 ± 0.11	+0.36 ± 0.15
S	2.48(−7) ± 1.16(−7)	...		5.32 ± 0.22	...		−1.80 ± 0.23	...
Cl	2.47(−9) ± 4.02(−10)	...		3.39 ± 0.07	...		−1.94 ± 0.09	...
Ar	2.36(−8) ± 3.61(−9)	...		4.37 ± 0.07	...		−2.18 ± 0.10	...
Fe	1.53(−7) ± 5.71(−8)	...		5.18 ± 0.17	...		−2.29 ± 0.17	...
Kr	<8.74(−10)	...		<2.94	...		<−0.42	...
Xe	<1.33(−10)	...		<2.12	...		<−0.09	...
Ba	2.54(−10) ± 1.00(−10)	...		2.41 ± 0.18	...		+0.23 ± 0.18	...

Notes. ^aThe number density of hydrogen is 12. ^bSolar abundances are taken from Lodders (2003).

Download table as:  ASCIITypeset image

4.1.2. High-Density Components

4.1.3. Hydrogen-deficient Cold Components

In Table 19, we compiled results for BoBn 1 from the past 30 years. Our estimated CEL elemental abundances except for Fe are in good agreement with previous works. The large discrepancy for Fe between us and Kniazev et al. (2008) could be due to adopted electron temperatures for the Fe²⁺ abundance estimation. The depletion of Fe relative to the Sun is almost consistent with that of Ar (Table 15). Since Ar and Fe are not to be synthesized in low-mass stars, the abundances of these elements must be roughly same, if the large amount of the dust does not coexist in the nebula. In addition, the mid-IR spectra show no astronomical silicate or iron dust such as FeS features. Therefore, our estimated Fe abundance seems more reliable than Kniazev et al. (2008). Sneden et al. (2000) estimated [Fe/H] = −2.37 ± 0.02 as the metallicity of M 15 using >30 giants. The [Fe/H] abundance of BoBn 1 corresponds to that of M 15 within error, implying that the progenitor of BoBn 1 might have formed at ∼10 Gyr ago.

BoBn 1 is a quite Ne-rich PN, and this Ne abundance is comparable with those of bulge and disk PNe; the averaged Ne abundance is 8.09 (CELs) and 9.0 (ORLs) for bulge PNe (Wang & Liu 2007) and 7.99 (CELs) and 9.06 for disk PNe (Tsamis et al. 2004; Liu et al. 2004; Wesson et al. 2005). The Ne isotope, ²⁰Ne is the most abundant, and it is not altered significantly by H- or He-burning (Karakas & Lattanzio 2003). Therefore, the Ne overabundance would be due to an increase of the neon isotope, ²²Ne. During helium burning, ¹⁴N captures two α particles, and ²²Ne are produced. The Ne overabundance also implies that BoBn 1 might have experienced a very late thermal pulse (Otsuka et al. 2008a). If this is the case, Ne abundance might be an indirect evidence of He-rich intershell activity. The Ne overabundance would be concerned with extra mixing during the RGB phase, which would increase N. The Ne and N enhancements of BoBn 1 might be also involved with the chemical environment where the progenitor formed. So far, four objects including BoBn 1 have been regarded as Sagittarius dwarf galaxy PNe, and three objects of them showed C, N, and Ne-rich ([C, N, Ne/O] > 0; cf. Zijlstra et al. 2006).

The leading and trailing streams of the Sagittarius dwarf galaxy trace several globular clusters. Terzan 8 is a member of the Sagittarius dwarf galaxy. Mottini et al. (2008) investigated the chemical abundances in three red giants in Terzan 8. The averaged [Fe/H], [O/Fe], and [Mg, Si, Ca, Ti/Fe] among these objects were −2.37 ± 0.04, +0.71 ± 0.14, and +0.37 ± 0.14, respectively. Note that the metallicity of Terzan 8 is very close to BoBn 1. The amounts of O and other α elements such as Mg, Ne, S, and Ar do not significantly change during RGB phase. Therefore, the pattern of α elements derived from these RGB stars should be close to those in BoBn 1's progenitor. Based on this assumption, the initial [O/H] abundance of BoBn 1 is estimated to be −1.66 ± 0.15, which is log(O/H) + 12 = 7.03 ± 0.15. [O/H] ∼ +0.77 would need to have been synthesized in BoBn 1 during helium burning. Assuming that the ²²Ne(α,n)²⁵Mg reaction is inefficient, the initial [Mg/H] abundance is estimated to be −2.07 ± 0.15, which is comparable to the observed [S, Ar/H] in BoBn 1. The observed S and Ar abundances in BoBn 1 could show the original abundances of the progenitor. [Ne/H] = +2.16 ± 0.18 could have been synthesized in BoBn 1 by ¹⁴N capturing two α particles. Mottini et al. (2008) also estimated the light and heavy s-process enhancements; [Ba/Fe] = −0.09 ± 0.17 and [Y/Fe] = −0.29 ± 0.17. From these values, the initial s-process elemental abundances would be −2.46 ± 0.18 for heavy s-process elements such as Xe and Ba and −2.66 ± 0.18 for light s-process elements such as Kr. The progenitor of BoBn 1 could have synthesized [s/H] ∼+2 during He-burning phase.

To investigate the properties of the ionized gas, dust, and the PN central star in a self-consistent way, we have constructed a theoretical P-I model which aims to match the observed flux of emission lines and the spectral energy distribution (SED) between UV and mid-IR wavelength, using Cloudy c08.00 (Ferland 2004).

First, a rough value of the distance to BoBn 1 is necessary in fitting the observed fluxes. The distance to this object is estimated to be in the range between 16.5 and 29 kpc (see Table 1). Based on the assumption that BoBn 1 is a member of the Sagittarius dwarf galaxy, we fix the distance to be 24.8 kpc (Kunder & Chaboyer 2009). P-I model construction needs information about the incident SED from the central star and the elemental abundances, geometry, density distribution, and size of the nebula. Once one gets a proper prediction of line intensities from a proper modeling procedure, the central stellar physical properties employed in the P-I model can give us a hint of the nebular evolutionary history or that of its progenitor star. Especially, the central star temperature T_⋆ and the SED of the PN central star are an important physical parameter in constructing the correct P-I model.

We estimated T_⋆ of 125,930 ± 6100 K using the energy balance methods proposed by Gurzadyan (1997). Being guided by this T_⋆, we used theoretical model atmosphere for a series of values of T_eff to supply the SED from the central star. We used Thomas Rauch's non-LTE theoretical model atmospheres⁷ for halo stars ([X, Y] = 0 and [Z] = −1) with the surface gravity log  g = 6.0, 6.125, 6.25, 6.375, 6.5, and 6.625. We varied T_eff and the luminosity L_⋆ to match the observations.

For the elemental abundances X/H, we used the values from the case of t² = 0 as a starting point. For the C, N, O, and Ne abundances, we used the CELs abundances. We assumed no high-density cold clumps. Using the HDS slit-viewer image (Figure 1), we measured the radius of the outer nebular shell R_out of ∼06 (=0.072 pc) and fixed this value. We assumed the hydrogen density N_H to be a R⁻² smooth distribution, i.e., N_H = N_H(R_in) × (R_in/R)². In the models, within a small range we varied X/H, N_H(R_in), and R_in to match the observed line fluxes from UV to mid-infrared wavelength, including Two Micron All Sky Survey (2MASS) JHK bands, and our mid-infrared bands, i.e., between 17 and 23 μm (IRS B) and between 27 and 33 μm (IRS C).

Since the IRS spectra show that the dust grains coexist in the nebula of BoBn 1, we need information about the dust composition. Here, we considered amorphous carbon and PAH grains. The optical constants were taken from Rouleau & Martin (1991) for amorphous carbon and from Desert et al. (1990), Schutte et al. (1993), Geballe (1989), and Bregman et al. (1989) for PAHs. The observed emission-line and base line continuum fluxes between 5.9 and 6.9 μm (PAH 6.4 μm) and between 7.4 and 8.4 μm (PAH 7.9 μm) were used to determine the PAH abundance. We assumed that the gas and dust coexist in the same sized ionized nebula. We adopted a standard MRN a^−3.5 distribution (Mathis et al. 1977) with a_min = 0.001 μm and a_max = 0.25 μm for amorphous carbon. For PAHs, we adopted an a⁻⁴ size distribution with a_min = 0.00043 μm and a_max = 0.0011 μm. We adopted an R⁻² smooth dust density distribution. High-density clumped dust grains were not considered.

In Table 20, we compare the predicted with observed relative fluxes where I(Hβ) = 100. For most CELs and He lines and the wide band fluxes, as well, the agreement between the P-I model and the observation is within 30%. The poor fit of [N i], [O i], and [S ii] would be due to the assumed density profile. The relation between n and IP (Figure 8, lower panel) shows that the electron density jumps from ∼2000 to ∼6000 cm⁻³ around [S ii] emitting region. In our model, such a density jump was not considered. For important lines such as He i, ii, and C iii], [N ii], [O ii], [O iii], [Ne iii], fairly good agreements in calculating ionic abundances are achieved. However, most of the C and O ORLs fit to the observations poorly. In most cases, the P-I models underestimate their line fluxes. As we discussed above, the O ORLs, and likely the Ne ORLs too, might be emitted from cold and metal-enhanced clumps.

In Table 21, we list the derived parameters of the PN central star, ionized nebula, and dust, and in Figure 12 we present the predicted SED (black line). The predicted SED matches the UV to mid-infrared region well. Through the P-I modeling, we found the PN central star's parameters: T_eff = 125,260 ± 200 K, L_⋆ = 1180 ± 240 L_☉, log  g = 6.5, and a core mass of ∼0.62 M_☉. The estimated 0.62 core mass is comparable to that of K 648 (0.62 M_☉, Bianchi et al. 2001; 0.57 M_☉ Rauch et al. 2002) and the high-excitation halo PN NGC 4361 (0.59 M_☉; Traulsen et al. 2005). The ionized mass of BoBn 1 is 0.09 M_☉, which is comparable to K 648 (0.07 M_☉; Bianchi et al. 1995). In Figure 13, we plot the locations of BoBn 1 and K 648 (Rauch et al. 2002) and the post-AGB He-burning evolutionary tracks for LMC metallicity (Z ∼ 0.5 Z_☉) by Vassiliadis & Wood (1994). These evolutionary tracks would suggest the possibility that the progenitors of BoBn 1 and K 648 were single 1–1.5 M_☉ stars which would end their lives as white dwarfs with a core mass of ∼0.6 M_☉. Alternatively, these halo PNe might have evolved from binaries composed of a 0.8 M_☉ (=a typical halo star mass) secondary and a more massive primary, and gained mass (∼0.1 M_☉) from the primary through mass transfer or coalescence.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Table 21. The Derived Properties of the PN Central Star, Ionized Nebula, and Dust by the P-I Model

Central Star		Nebula		Dust
Parameter	Value/Assumption	Parameter	Value/Assumption	Parameter	Value/Assumption
d (kpc)	24.8	Composition	He:11.11, C:8.63, N:7.96,O:7.70,	Grains	AmC and PAHs
L⋆ (L☉)	1180	(Conti.)	F:5.85, Ne:7.90, S:5.01, Cl:3.22,	Mdust (M☉)	5.78(−6)
T⋆ (K)	125260	(Conti.)	Ar:4.29, Fe:5.05,others:[X] = −2.13	Tdust (K)	80–180
log  g (cm2 s−1)	6.5	Rin/Rout ('')	0.43/0.60	Mdust/Mgas	5.84(−5)
Composition	[X, Y] = 0, [Z] = −1	NH(Rin) (cm−3)	3890	$\dot{M}_{\rm dust}$ (M☉ yr−1)	∼3(−9)
Mcore (M☉)	∼0.62	Geometry	Spherical
		log F(Hβ)	−12.44
		Mgas/Matom (M☉)	0.09/0.04

Download table as:  ASCIITypeset image

For K 648, Alves et al. (2000) support a binary evolution scenario. From F enhancement and similarity to CEMP stars, Otsuka et al. (2008a) argued that BoBn 1 might have evolved from a binary composed of a 0.8 M_☉ secondary and a > 2 M_☉ primary star. We should consider two possibilities: these halo PNe have evolved from single stars or from binaries.

The P-I model indicated elemental abundances except for C and S are in excellent agreement with those estimated by the empirical method using the ICFs. The discrepancy for C is due to the underprediction of C iv lines. The model predicted C⁺ = 2.0(−5) and C²⁺ = 3.4(−4), which are comparable to the observations. However, it predicted lower line fluxes of C iv λλ1548/51 than the observations, and accordingly an underestimated C³⁺ as 6.95(−5). These might suggest that the origin of C iv lines is not the ionized nebula but the stellar wind zone. The discrepancy for S could be due to the low ionic the S⁺ abundance, which depends strongly on the assumed radial density profile.

For BoBn 1, we have for the first time estimated a dust mass of 5.78(−6) M_☉ and the temperature of 80–180 K. The dust in BoBn 1 is carbon rich. The dust composition suggests that BoBn 1 had experienced the TDU during the latest thermal pulsing AGB phase (TP-AGB). Since the TDU efficiently takes place in >1 M_☉ stars, the progenitor of BoBn 1 might be 1–3.5 M_☉ from the aspect of elemental abundances and dust composition.

The dust-to-gas mass ratio ψ of 5.84(−5) is much lower than the typical value in PNe (<∼10⁻³; Pottasch 1984). For AGB stars, Lagadec et al. (2008) argued that ψ scales linearly with the metallicity. They assume the ratio is

$$\begin{equation} \psi = \psi _{\odot }\times 10^{[{\rm Fe/H}]}, \end{equation} \tag{ 7 }$$

where ψ_☉ is 0.005. When we adopt [Fe/H] = −2.22 for BoBn 1, we obtain ψ = 3.01(−5).

Assuming that the inner and outer shells were expanding with ∼10 km s⁻¹ and most of the dust was formed during the ending period of the thermal pulse AGB phase, we estimated the dust mass-loss rate $\dot{M}_{\rm dust}$ of ∼3(−9) M_☉ yr⁻¹. Lagadec et al. (2010) estimated $\dot{M}_{\rm dust}$ in metal-poor ([Fe/H] ∼ −1) carbon stars IRAS 16339 − 0317 and 18120+4530 in the Galactic halo and IRAS 12560+1656 in the Sgr stream. They estimated 4–18(−9) M_☉ yr⁻¹. For IRAS 12560+1656, Groenewegen et al. (1997) estimated $\dot{M}_{\rm dust}$ of 1.9(−9) M_☉ yr⁻¹. These $\dot{M}_{\rm dust}$ values are comparable to that of BoBn 1.

Through P-I modeling, we found two possibilities: BoBn 1 might have evolved from (1) a 1–1.5 M_☉ single star or (2) a binary composing of ∼0.8 M_☉ secondary and a more massive primary. In this section, we explore these possibilities by comparing the observed and predicted elemental abundances employing theoretical nucleosynthesis models for low- to intermediate-mass stars.

In Table 22, we present observed elemental abundances and predicted values from the theoretical models of Karakas & Lugaro (2010) for 1.0, 1.5, and 2.0 M_☉ stars with Z = 10⁻⁴ ([Fe/H] ∼ −2.3). The abundances from the models are the values at the end of the AGB phase. For these models, Karakas & Lugaro (2010) chose an initial α-enhanced abundance pattern, i.e., [α/Fe] = +0.4. For s-process elements, they chose scaled solar abundances, i.e., [X/Fe] = 0. These [α/Fe] and [X/Fe] ratios are consistent with the values for RGB stars in Terzan 8, therefore the assumption of initial abundances seems very reasonable for BoBn 1. The accuracy of the predicted abundances by the models is within 0.3 dex. For the observed abundances, we adopted the t² = 0 CEL abundances except for C. The adopted C abundance was from ORLs.

On the possibility (1) that BoBn 1 has evolved from a single star and has survived in the Galactic halo, the abundances of BoBn 1 except for N can be properly explained by the 1.5 M_☉ star model including a partial mixing zone of 0.004 M_☉, which produces a ¹³C pocket during the interpulse period and releases free neutrons through ¹³C(α, n)¹⁶O. This model assumes that the stars end as white dwarfs with a core mass of ∼0.7 M_☉, which is comparable to our estimated core mass.

The ¹³C(α, n)¹⁶O reaction proceeds in the upper surface layer of the He-burning shell. The F and s-process elements are synthesized by capturing these neutrons in the He-intershell. The observed F, probably Kr and Xe abundances are systematically larger (∼+0.5 dex) than 1.5 M_☉ star + partial mixing model. This suggests that hydrogen mixing mass is likely to be >0.004 M_☉ or that BoBn 1 had experienced helium-flash-driven deep mixing (He-FDDM; Fujimoto et al. 2000; Suda et al. 2004) and obtained the extra neutrons. This process can occur in stars with [Fe/H] < −2.5 at the bottom of the He-burning shell because the entropy barrier between the H- and He-shell becomes low. The lower limit to [Fe/H] for BoBn 1 is −2.46. This process would also produce ¹⁴N through the ¹³C(p, γ)¹⁴N reaction, by mixing protons into the He-burning shell.

We note that BoBn 1 is similar to K 648, for the latter is also known as an extremely metal-poor, C- and N-rich halo PN. On the assumption that K 648 has been evolved from a single star, the abundances of K 648 except for Ne can be explained by a 1.5 M_☉ model. K 648 would not have experienced He-FDDM for abnormally increasing N.

However, can halo single stars with an initial mass of ∼1.5 M_☉ and Z = 10⁻⁴ survive up to now? Such stars would end as white dwarfs in ∼2–3 Gyr. Their lifetime is much shorter than the age of Terzan 8; Forbes et al. (2004) estimated the age of Terzan 8 to be 13 ± 1.5 Gyr from an age–metallicity relation for the Sagittarius dwarf globular cluster. If the progenitor of BoBn 1 was a ∼0.8 M_☉ single star, then it can have survived up to now, however it cannot evolve into a visible PN and cannot become extremely C-rich. To circumvent the evolutionary time scale problem, we should consider the other evolutionary scenario for BoBn 1.

As mentioned earlier, BoBn 1 is likely to be of a binary origin because the [C, N, F/Fe] abundances of BoBn 1 are comparable to those of the CEMP star HE 1305+0132 (Schuler et al. 2007) and other CEMP stars.

Most CEMP stars show large enhancements of C and N abundances. Some evolutionary models for CEMP stars have demonstrated that the C and N overabundances would be reproduced by binary interactions. Schuler et al. (2007) concluded that HE 1305+0132 might have experienced mass transfer and that s-process elements should be enhanced. Lugaro et al. (2008) concluded that HE 1305+0132 consisted of ∼2 M_☉ (primary) and ∼0.8 M_☉ (secondary) stars with Z = 10⁻⁴ and that the enhanced C and F could be explained by binary mass transfer from the primary star via Roche lobe overflow and/or wind accretion. In Figure 14, we present the diagram of [Xe, Ba/Ar]–[C/Ar]. The [Xe/Ar] for BoBn 1 is an upper limit. The data for Galactic PNe except BoBn 1 are taken from Sharpee et al. (2007). The data for s-process elements enhanced CEMP (CEMP-s) with [Fe/H] > −2.5 and C-rich AGB are from the SAGA database (Suda et al. 2008). For PNe, we use Xe as a heavy s-process element and adopt Ar as a metallicity reference. For CEMP-s and C-rich AGB stars, we use Ba and Fe as a metallicity reference. The diagram indicates that C and s-process elements are certainly synthesized in the same layer and brought up to the stellar surface by the TDU. We note that the enhancement of heavy s-process elements in BoBn 1 is comparable to CEMP-s stars, in particular CS22948-027 (Aoki et al. 2007; [Fe/H] = −2.21, [Ba/Fe] = +2.31, [C/Fe] = +2.12, [N/Fe] = +2.48). The chemical similarities between BoBn 1 and CEMP-s stars suggest that this PN shares a similar origin and evolutionary history.

Zoom In Zoom Out Reset image size

BoBn 1 is similar to K 648 in elemental abundances and nebular shape (see Figure 1 and Table 19). K 648 has been for a long time suspected to have experienced binary evolution. Rauch et al. (2002) and Bianchi et al. (2001) analyzed the spectrum of the central star and estimated a core mass ∼0.6 M_☉. The mass ∼0.6 M_☉ corresponds to the initial mass of 1–1.5 M_☉ from the HR diagram as shown in Figure 13. The initial mass of 1–1.5 M_☉ suggests that K 648 might have evolved from a binary and accreted a part of the ejected mass by a massive primary or coalescence during the evolution. Alves et al. (2000) argued that K 648 has experienced mass augmentation in a close-binary merger and evolved as a higher mass star to become a PN. Such a high-mass star would be a blue straggler. Ferraro et al. (2009) observed stars in the globular cluster M 30 using the HST/WFPC2 and concluded that blue stragglers are results of coalescence or binary mass transfer.

In view of the internal kinematics and chemical abundances, the progenitor of K 648 seems to be a binary. K 648 has a bipolar outflow (Tajitsu & Otsuka 2006) and bipolar nebula (Alves et al. 2000). The statistical study of PN morphology shows that bipolar PNe have evolved from massive stars with initial mass ≳2.4 M_☉ or binaries. Otsuka et al. (2008b) showed that the [C/Fe] and [N/Fe] abundances of K 648 are compatible with CEMP stars. So far, there have been no reports on the detection for any binary signatures in both objects. The contradiction to the evolutionary time scale of this object can be avoided if BoBn 1 has indeed evolved from a binary. Similar to K 648, BoBn 1 could have evolved from a binary and undergone coalescence to become a PN.

We explore the possibility of binary evolution of BoBn 1 using binary nucleosynthesis models by Izzard et al. (2004, 2009). We assume a binary system composed initially of a 0.75 M_☉ secondary and a 1.5/1.8/2.1 M_☉ primary with Z = 10⁻⁴, separation = 219/340/468 R_☉, respectively. We set eccentricity e = 0, common envelope efficiency α = 0.5, and structure parameter λ_CE = 0.5. He-FDDM is not considered. We set a ¹³C pocket mass of 7.4 × 10⁻⁴M_☉. The ¹³C pocket contains 4.1 × 10⁻⁶M_☉ ¹³C and 1.3 × 10⁻⁷ M_☉ ¹⁴N. We set the ¹³C pocket efficiency = 2 and adopt wind mass-loss rates from Reimers formula on the RGB and by Vassiliadis & Wood (1993) on the TP-AGB.

When we choose these initial primary and secondary masses and α, the binary system will experience Roche lobe overflow; and it will merge into a 1.2–1.4 M_☉ single star at the TP-AGB phase and end its life as a white dwarf with a core mass of 0.62–0.68 M_☉, 10.4–12.5 Gyr after the progenitor was born. We present the results in Table 22. The binary models might seem to explain the elemental abundances of BoBn 1 and K 648. Concerning K 648, the 0.75 M_☉ + 1.5 M_☉/1.8 M_☉ models fairly well match the prediction to the observed abundances. Among the models, the 0.75 M_☉ + 1.5 M_☉ model seems to be the best fit to BoBn 1 for the moment because this model can explain not only the abundance patterns but also the observed core mass (the predicted core mass ∼0.64 M_☉). If the progenitor has experienced extra mixing in the RGB phase and increased N, the N overabundance can be accommodated by this model. The issues with the evolutionary time scale and C and N enhancements might be resolved simultaneously if BoBn 1 has evolved from such a binary. At the present, we conclude that binary evolution scenario is more plausible for BoBn 1.

To further discuss the evolution of BoBn 1 and K 648, we need to increase detection cases of s-process elements and to investigate the isotope ratios of ¹²C/¹³C, ¹⁴N/¹⁵N, ¹⁶O/¹⁷O, and ¹⁶O/¹⁸O, which would be useful to investigate nucleosynthesis in the progenitors. It would be also necessary to completely trace mass-loss history to improve mass-loss rate. So far, mass-loss history of evolved stars has been revealing by investigating spatial distribution of dust grains and molecular gas using far-infrared to millimeter wavelength data. The Atacama Large Millimeter Array (ALMA) and the thirty meter telescope (TMT) could open new windows to study the evolution of metal-poor stars such as halo PNe.