Planetary nebulae with Wolf–Rayet-type central stars – III. A detailed view of NGC 6905 and its central star Free

ABSTRACT

1 INTRODUCTION

By the end of their lives, low- and intermediate-mass stars (0.8 M_⊙ ≲ M_ZAMS ≲ 8 M_⊙) produce copious ejections of material when evolving through the asymptotic giant branch (AGB) phase before the final white dwarf (WD) stage. Up to half of the initial mass of the star is deposited into the interstellar medium through a slow and dense wind (v_AGB = 10−30 km s⁻¹, |$\dot{M}\approx 10^{-6}\!-\!10^{-5}$| M_⊙yr⁻¹; Vassiliadis & Wood 1993). Owing to the low effective temperature in this phase (T_eff < 10⁴ K), a dusty envelope is formed (see Mauron, Huggins & Cheung 2013). After ejecting its outer layers, the star evolves into the post-AGB phase increasing its T_eff and subsequently developing a strong ionizing UV flux and a fast stellar wind (v_∞ = 500−4000 km s⁻¹; e.g. Guerrero & De Marco 2013). The combination of these two effects sweeps, compresses, heats, and ionizes the previously ejected AGB material, forming a planetary nebula (PN; Kwok 2000).

As a result of the vast number of works in the past decades, increasing evidence suggest that binaries could play a key role in PN shaping: producing apparent spherical morphologies (see Guerrero et al. 2020) as well as extremely collimated structures (see Frank et al. 2018). Changes in the orbital parameters caused by the evolution of the primary star affect the mass-loss (Iaconi et al. 2017). Moreover, binarity can produce changes in stellar structure and surface abundance, creating H-poor stars (Lau, De Marco & Liu 2011) and reflecting in the PN chemical composition (Wesson et al. 2018).

About 10 per cent of the central stars of planetary nebulae (CSPNe) are H-poor with strong and broad emission lines from He, C, N, and O in their spectra. These spectral features resemble those typical in massive stars in their Wolf–Rayet (WR) phase (see Crowther 2007) and can be classified following the same scheme (see Crowther, De Marco & Barlow 1998; Acker & Neiner 2003) as C-rich ([WC]), N-rich ([WN]), and with strong O emission lines ([WO]),¹ with sub-classifications assigned depending on the ratio of the line strengths in their spectra. The most numerous [WR]-type stars are those of the carbon sequence, followed by the [WO]-types (see table 2 in Weidmann et al. 2020, and references therein), with a few cases of [WN]-types and transitional [WN/C]-types (e.g. Todt et al. 2010, 2013; Miszalski, Crowther & De Marco 2012).

Statistical analyses of PNe with [WR]-type CSPNe (hereinafter WRPN) have unveiled differences when compared to PNe with H-rich CSPNe. High-resolution spectroscopic observations have shown that WRPNe exhibit larger expansion velocities and present a higher degree of turbulence than non-WRPNe, very likely due to their higher wind mechanical luminosities or the presence of highly-collimated outflows (Gesicki et al. 2003; Medina et al. 2006; Rechy-García et al. 2017). WRPNe have also been found to have higher N and C abundances, suggesting that they are formed as a result of ∼4 M_⊙ stars (García-Rojas et al. 2013). It has also been suggested that there is a relation between the decrease in electron density with decreasing spectral subtype as a sign of evolutionary sequence from late to early [WC]-types (Acker, Górny & Stenholm 1996; Górny & Tylenda 2000). However, the N/O abundance ratio versus spectral type has shown that stars with different masses can evolve through the same [WC] state (Peña et al. 1998; Peña, Stasińska & Medina 2001).

In Gómez-González et al. (2020a, hereinafter Paper I), we started a series of works dedicated to a comprehensive characterization of WRPNe. In Paper I, we noted that statistical studies of WRPNe are helpful in unveiling their properties as a population, but these do not provide the details that particular studies can offer us when combining multiwavelength observations with theoretical predictions. Our analysis of the high-excitation WRPN NGC 2371 and its [WO1] CSPN WD 0722 + 295 presented in Paper I helped us to unveil its physical properties, origin and kinematical structure.

In this paper, we present a comprehensive multiwavelength characterization of NGC 6905 (PN G 061.4-09.5), a.k.a. the ‘Blue Flash Nebula’ for its characteristic colours. NGC 6905 is a high-excitation WRPN with a clearly clumpy morphology (see Fig. 1), located at a distance of 2.7 kpc (Gaia Collaboration et al. 2020). It is composed of a central roundish cavity with angular radius ∼19 arcsec (0.25 pc) and a pair of extended V-shaped structures extending towards the NW and SE directions with a position angle (PA) of ∼160° and with a total extension ∼80 arcsec (1 pc). These bipolar structures harbour a pair of low-ionization knots at their tips. These low-ionization features expand faster than the central regions of NGC 6905 with an inclination of 60° with respect to the plane of the sky (Sabbadin & Hamzaoglu 1982) and have been suggested to be responsible for giving this WRPN its bipolar morphology (Cuesta, Phillips & Mampaso 1993).

Figure 1.

(left) Nordic Optical Telescope (NOT) Alhambra Faint Object Spectrograph and Camera (ALFOSC) colour-composite image of NGC 6905 ([O iii] – blue, H α – green, [N ii] – red). The white dashed-line region represents the NOT ALFOSC slit at PA of 155°. The solid-line rectangles (0.75 × 5 arcsec), labeled from A1 to A8, indicate our spectra extraction regions. The position of the CSPN (HD 193949) and a background star are indicated with arrows. (right) Individual H α, [N ii], and [O iii] intensity images and line intensity [O iii]/H α ratio map. Each panel have the same field of view (FoV).

Open in new tabDownload slide

The CSPN of NGC 6905 has been classified as [WO]-type star. Due to the significant O vi emission lines, it has been classified as a [WO2]-subtype (Acker & Neiner 2003). Spectroscopic studies in UV of NGC 6905 such as those presented by Keller et al. (2011) and Keller, Bianchi & Maciel (2014) have allowed stellar atmosphere modeling of HD 193949 implying a T_eff in the range of 150–165 kK. These authors also found Ne to be overabundant with a mass fraction of 0.02.

This paper is organized as follows. In Section 2, we describe the observations used in this work. In Section 3, we present the analysis of the spectra and images. A discussion is presented in Section 4. Finally, our summary and conclusions are listed in Section 5.

2 OBSERVATIONS AND DATA PREPARATION

2.1 Nordic Optical Telescope

Optical images and spectra of NGC 6905 were obtained with the ALFOSC² at the 2.5 m NOT of the Observatorio del Roque de los Muchachos (ORM) in La Palma, Spain.

The images were obtained on 2009 July 21, using the H α (λ_c = 6563 Å and Δλ = 33 Å), [N ii] (λ_c = 6584 Å and Δλ = 36 Å), and [O iii] (λ_c = 5010 Å and Δλ = 43 Å) narrow-band filters. Two images of 450 s were obtained in each filter. The images were processed using standard iraf routines (Tody 1993) and are presented in Fig. 1, together with a colour-composite picture and a [O iii]/H α ratio map.

NOT spectroscopic observations were carried out on 2020 July 26. The EEV 231-42 2K×2K CCD was used, providing a pixel scale of 0.211 arcsec per pixel and a FoV of 7 arcmin. A spectrum at a position angle (PA) of 155° was obtained using the grism #7, which covered the spectral range of 3650–7110 Å and a dispersion of ∼1.7 Å per pixel. The total observing time was of 2400 s, which was split into two exposures of 1200 s. We adopted a slitwidth of 0.75 arcsec, with spectral resolution of ∼5 Å. The seeing during the observations was ∼1 arcsec. The details of the spectroscopic observations are presented in Table 1.

The spectra were also reduced and analyzed using standard iraf routines, which include bias subtraction and flat-fielding. The wavelength calibration was performed using HeNe arc lamps and the flux calibration by using the standard star Wolf 1346.

In order to study the ionization structure of NGC 6905, we have extracted spectra from different regions defined along the slit using the iraf task apall. We first extracted the 1D spectrum of the CSPN tracing the stellar continuum along the 2D spectrum. This information was then used as a reference to trace the nebular spectra along the 2D spectra, as these regions do not show continuum emission. The location of the extracted regions, labelled A1–A8, are shown in Fig. 1, all of them with equal sizes of 0.75 × 5 arcsec. The corresponding extraction region for HD 193619 is A5. Finally, we note that all extracted spectra were corrected for extinction by using the c(H β) value estimated from the Balmer decrement method with an intrinsic H α/H β, H γ/H β, and Hδ/H β ratios of 2.86, 0.45, and 0.25, respectively, which corresponds to mean values for T_e in the range 2000 K and 12 000 K and n_e between 100 cm⁻³ and 1200 cm⁻³ (see Osterbrock & Ferland 2006) and using the reddening curve of Cardelli, Clayton & Mathis (1989) with R_V = 3.1. The spectrum of HD 193949 is presented in Fig. 2 and the spectra from different regions will be discussed in Section 3.3.

Figure 2.

NOT ALFOSC spectrum of HD 193949, the CSPN of NGC 6905, extracted from region A5 defined in Fig. 1 (left-hand panel). The most prominent broad WR features like the violet bump (VB) at ∼λ3820 Å, the blue bump (BB) at ∼λ4686 Å, and the red bump (RB) at ∼λ5806 Å are marked with their respective colours, whilst other WR features are indicated in black. Broad Ne lines of high ionization potential are indicated with orange labels, and the narrow nebular lines are labeled in green. The spectrum is shown normalized to the best-fitting continuum spectrum. See table Table 4 for details.

Open in new tabDownload slide

2.2 Infrared data

In order to produce a consistent model of the nebular and dust properties of NGC 6905, we also need the infrared (IR) photometry of this WRPN. The necessary IR data were retrieved from the NASA/IPAC IR archive.³ Specifically, we retrieved archival IR images obtained from Spitzer, WISE, IRAS, and Akari. With these, we can create a spectral energy distribution (SED) that covers the 5.8–160 |$\mu$|m spectral range. Details of the observations used here are listed in Table 2.

IR photometric measurements were extracted from each image. We defined an extraction region that encompasses the nebular emission for each individual IR image. The corresponding photometric measurements were calculated adding up all the flux from the pixels within this region. Several background regions were selected from the vicinity of NGC 6905 with no contribution from the nebula; additionally, we excised the contribution of background stars like the south-east star at the edge of the nebula. In Table 3 we list the photometric values and their errors. The errors account for the calibration uncertainty from each instrument as well as the standard deviation from the selected background regions. Details of this process can be found in Jiménez-Hernández, Arthur & Toalá (2020), Jiménez-Hernández et al. (2021), and Rubio et al. (2020), and will not be discussed here. We do not include the shorter wavelengths from Spitzer because of the presence of numerous background stars that hinders a proper photometry extraction.

Finally, photometric measurements obtained at radio frequencies were also included. Radio measurements at 1.4 and 4.85 GHz, that correspond to 20 and 6 cm, respectively, were also taken from the NASA/IPAC infrared archive, corresponding to the National Radio Astronomy Observatory (NRAO) Very Large Array (VLA) survey and the NRAO Green Bank telescope (GBT), respectively (see also Hajduk et al. 2018). The radio measurements are also listed in Table 3.

3 ANALYSIS

The NOT ALFOSC spectrum of HD 193949 presented in Fig. 2 corroborates its [WR] nature with the clear presence of the broad features at ∼4686 Å and ∼5806 Å, the so-called blue and red bumps (hereinafter BB and RB), together with other WR features. The CSPN of NGC 6905 also exhibits the broad O vi feature at ∼3820 Å (hereinafter the violet bump or simply VB), which has been used to classify this object as part of the oxygen sequence (Acker & Neiner 2003). Additionally, 21 WR emission lines can be spotted in the ALFOSC spectrum at different wavelengths between 3680 Å and 7060 Å, plus several high-excitation Ne lines, which have never been discussed before. Several contributing narrow emission lines of nebular origin (non-stellar) are also labeled on the spectrum. All the ions responsible for the broad emission lines displayed in the spectrum of HD 193949 are listed in Table 4. Most of the observed wavelengths of the ions analysed here were consulted in the National Institute for Standards and Technology (NIST) Atomic Spectra Database.⁴

Previous authors have developed a methodology to sub-classify [WR]-type CSPNe using the line fluxes and/or the equivalent widths (EW) of the broad spectral emission features (e.g. Acker & Neiner 2003; Crowther et al. 1998). For this, we need to calculate the line fluxes for all the WR features in the spectrum; we note that some broad features are usually the result of blends of stellar broad features with some contribution from nebular lines. Although minor, these need to be taken into account. To help us disentangle the presence from the contributing nebular lines to the WR bumps, we applied the multi-Gaussian fitting technique discussed in length in Paper I. This method allows us to assess the contribution from nebular lines (e.g. [Ar iv] λ4711 and He ii λ4686 in the BB). We give a brief description in the following.

3.1 Line fluxes of the CSPN HD 193949

The WR features in the spectrum of the CSPN HD 193949 (Fig. 2) were decomposed into their individual emission lines by following the multi-Gaussian approach (discussed in length in Gómez-González et al. 2020b). In brief, the method consists of fitting the individual WR features with multi-Gaussian components using a tailor-made code that uses the idl routine lmfit that performs a non-linear least squares fit of a function with an arbitrary number of parameters by using the Levenberg-Marquardt algorithm (Press et al. 1992). The resultant fits to each broad spectral feature give us the line fluxes (F), central wavelength (λ_c), FWHM, and EW from the contributing lines from stellar and nebular origin. Table 4 resumes our findings.

Examples of multi-Gaussian fitting to several WR features in HD 193949 are illustrated in Fig. 3. In order of increasing ionization potential, the WR features detected are: He ii λλ4200, 4541, 4686, 5412, 6406, 6560; C iv λλ3689, 3933, 4227, 4440, 4658, 4789, 5470, 5806, 7060; O v λλ4933, 5100, 5600, 6740, and O vi λλ3811, 3834, 4500, 4604, 5166, 5290, 6200. We highlight the presence of three Ne broad emission lines: Ne vii λλ4555, 5666 and Ne viii λ6068, of stellar origin according to Werner et al. (2007). These lines are discussed in the following sections.

Figure 3.

Multicomponent Gaussian fits to the bumps and other WR features of the CSPN on NGC 6905: VB (top left); BB (top centre), and RB (top right). The rest of the panels show other WR features and high ionization Ne emission lines (see Fig. 2). The blue dashed lines represents the fits to the broad WR features and the red dotted lines the narrow nebular emission. The sum is shown in black. The fitted continuum is shown by the dashed straight line. The residuals are shown at the bottom of each panel (in per cent). The results are listed in Table 4.

Open in new tabDownload slide

We found that the VB is composed of two broad emission lines of O vi λλ3811, 3834, with negligible contribution of [Ne iii] and Ne vii at its red wing (see Fig. 3 top left-hand panel). The BB, composed by the broad He ii λ4686 and C iv λ4658 features, has some contribution from nebular lines of He ii and [Ar iv] (Fig. 3 top middle panel). The RB is fitted by a single broad line at ∼λ5806 Å, from the blended components of C iv λλ5801, 5812 (Fig. 3 top right-hand panel). Notice that they all correspond to broad emission lines (FWHM>5.4 Å; see Table 4, column 8), according to their WR nature.

In Table 4 we also list what we call ‘special lines’, corresponding to high ionization potential Ne lines (see column 5) from presumably stellar emission origin. Before Werner et al. (2007), the Ne viii emission line at ∼6068 Å, with a ionization potential of 239.1 eV, was previously identified as a non-stellar ultra-high ionization O feature, O viii, with a much higher ionization potential of 871.4 eV. This ion supposedly formed in shock wind regions.

The same applied to other spectral features like O viii λ4340 now identified as Ne viii; C vi λ4500 now identified as O vi; O vii λ4555 now identified as Ne vii; C v λ4945 now identified as N v; C vi λ5290 now identified as O vi; and O vii λ5665 now identified as Ne vii. We note that although the presence of X-ray-emitting gas is invoked in low T_eff CSPN to explain the ultrahigh-ionization O features (see e.g. the case of IC 4593; Herald & Bianchi 2011; Toalá et al. 2020), this might not be the case for hot WRPN. Furthermore, Werner et al. (2007) showed that the mere presence of Ne vii and Ne viii puts a strict lower limit of T_eff ≈ 150 kK.

In Table 4 we present the flux, FWHM, EW, as well as the parameters F_λ/F_RB adopting F_RB = 100 and EW_λ/EW_RB adopting EW_RB = 100, that are used as a classification criteria for CSPNe, according with Acker & Neiner (2003) and Crowther et al. (1998). The mere presence of the strong O vi broad emission line ∼λ3820 Å, the VB in the CSPN spectrum of NGC 6905 (see Fig. 2), suggests a [WO]-type classification. Moreover, the presence of high-ionized Ne lines constrains it to a very early-type (e.g Werner et al. 2007). A classification of CSPNe later than [WO2]-subtypes can be safely discarded since they do not display neither Ne vii nor Ne viii in their spectra. Besides, later [WO]-subtypes, like [WO4], present weaker O vi and C iv intensities, in general. Also, any [WC] fingerprint (early or later) is also discarded: C iii lines, expected in [WCL]-subtypes, are absent in our spectrum, and in the classification schemes by Acker & Neiner (2003) and Crowther et al. (1998), [WCE]-subtypes do not display a strong VB by definition. It is worth to explicitly mention that we also did not detect any broad N lines, which also rules out any [WN]-subtype or transitional stages [WN/C].

In order to quantitatively classify the CSPN of NGC 6905 as [WO]-type, Table 5 summarizes the values of the main parameters among the [WO]-subtypes in which our [WR] star is expected to be ([WO1], [WO2], and [WO3]), according to the lines present in our spectrum. The spectrum of HD 193949 suggest a [WO2]-subtype based on the Acker & Neiner (2003) criteria, whilst the Crowther et al. (1998) scheme suggest a [WO1]-subtype. In conclusion, we classified the CSPN of NGC 6905 as a [WO]-type, specifically an early one, between [WO1] and [WO2]-subtype.

3.2 Non-local thermodynamic equilibrium analysis of HD 193949

We have analysed UV and optical spectra by means of the most updated version (2021 January 24) of the non-local thermodynamic equilibrium (NLTE) stellar atmosphere code Potsdam Wolf–Rayet (PoWR; Gräfener, Koesterke & Hamann 2002; Hamann & Gräfener 2004)⁵ in order to infer the stellar parameters of HD 193949. Details of the computing scheme can be found in Todt et al. (2015).

Available UV Far Ultraviolet Spectroscopic Explorer (FUSE) and Hubble Space Telescope (HST) Space Telescope Imaging Spectrograph (STIS) observations of HD 193949 were retrieved from the Mikulski Archive for Space Telescopes (MAST).⁶ The UV observations were analysed in conjunction with the optical NOT ALFOSC spectra with the PoWR stellar atmosphere code. The FUSE observation corresponds to Obs. ID. a1490202000 (PI: F. Bruhweiler) and were obtained on 2000 August 11 with the LWRS aperture for a total exposure time of 9487 s. The HST STIS data were taken on 1999 June 29 and correspond to Obs. ID. o52r01010 (NUV-MAMA) and o52r01020 (FUV-MAMA), each with an exposure time of 802 s (PI: A. Boggess) and taken with the time-tag mode with the 52 × 0.5 arcsec aperture. The FUV-MAMA observation was performed with the G140L grating at a central wavelength of 1425 Å, and the NUV-MAMA with the G230L grating at a central wavelength of 2376 Å. We notice that there might be a problem with the background subtraction for the FUV-MAMA observation, as the flux at the centre of the interstellar Lyman-α absorption line as well as the absorption trough of the C iv resonance doublet does not reach zero.

With T_⋆ and R_t determined from the continuum-normalized spectra, we can fit the synthetic SED to Gaia eDR3 photometry (Gaia Collaboration et al. 2020) and the flux-calibrated UV spectra to estimate L_⋆ and E(B − V). Taking into account the loss of flux from the CSPN due to the long-slit spectroscopic observations, the stellar luminosity is L_⋆ = 7600 L_⊙ for the photogeometric distance of 2.7 kpc (Bailer-Jones et al. 2021). Our fit implies a reddening of E(B − V) = 0.21 mag using the extinction law by Cardelli et al. (1989), consistent with value derived by previous authors (see Keller et al. 2014). We note, however, that such model requires a higher value of R_V= 3.8 that suggests that the extinction towards HD 193949 is higher due to the presence of different dust compositions along the line of sight (see e.g. Todt et al. 2010).

The spectral fit to the blue edges of the P-Cygni line profiles resulted in a terminal wind velocity of v_∞ = 2000 km s⁻¹ by adopting a β-law with β = 1. Additional line broadening by microturbulence is also included in our models. From the shape of the line profiles we estimate a value of about |$100\, \mathrm{km}\, \mathrm{s}^{-1}$|⁠. From equation (3) we obtained a mass-loss rate of |$\dot{M}=1.1\times 10^{-7}$| M_⊙ yr⁻¹ adopting a density contrast of D = 10, which is a typical value also found for other [WC]-CSPNe (e.g. Marcolino et al. 2007; Todt, Hamann & Gräfener 2008). We achieve a similar fit quality also with a lower value of D = 4 and log (R_t/R_⊙) = 0.8, while even smaller values of D, e.g. a smooth wind with D = 1, yield too strong electron scattering line wings. For larger values of D the fit quality is worse.

To determine the C and He abundances we mainly focused on the ‘diagnostic line pair’ He ii λ5411 and C iv λ5471 (Todt, Gräfener & Hamann 2006). Both features are formed at the same region in the wind, hence their relative strength is less sensitive to temperature and |$\dot{M}$|⁠. For the hot [WC] and [WO] stars an abundance ratio of about He:C ≈ 60:30 yields approximately equal strength for both emission features, as it is observed. For the O abundance we used primarily the O vi multiplet at ≈5290 Å, as this line is less sensitive to T_⋆ and R_t and more sensitive to the O abundance. The spectrum of HD 193949 displays many Ne vi and Ne vii lines in the UV and optical. With the inferred Ne abundance of about 2 per cent by mass we got a reasonable fit to most of these lines. However, we also note the presence of Ne viii features around 1162 Å and 1165 Å (see Fig. 4, top panel), as described in Werner et al. (2007). While these lines are in absorption in the model, it appears that they are in emission in the FUSE observation. There is also the emission feature at 6068 Å in the NOT spectrum, which might be also attributed to Ne viii. However, our model is not hot enough to produce this potential Ne viii emission line. For our object, this would require a stellar temperature above 170 kK, which is excluded by the presence of the observed O v lines. This emission feature has been reported to appear in the spectrum of other hot CSPNe such as NGC 2371 (see Paper I) and PC 22 (Sabin et al. in preparation). It seems that there is no N line present in the UV and optical spectra: the N v resonance doublet is not detectable and the emission features around 4604 Å and 4933 Å, which are also seen in hot WN stars, originate from O vi and O v, respectively. Therefore, we estimate the N fraction to be at most 1/10 of the solar value. For the H abundance we can only give an upper limit too, as stellar Balmer lines would be blended with the corresponding He ii lines from the Pickering series and nebular H emission lines. We find that a H mass fraction below 5 per cent would escape detection. In the absence of strong spectral lines from the iron group elements and with regard to the limited quality of the UV spectra we could not determine an iron abundance and adopted instead the solar value. Fig. 4 shows the comparison between the best-fitting PoWR model for HD 193949 and its FUSE, HST STIS, and NOT optical spectra, respectively.

Figure 4.

PoWR model (red dashed line) compared with the UV and optical observations of the CSPN HD 193949 (blue): FUSE (first panel), HST STIS (second panel), and optical ALFOSC (third and fourth panel). The model spectra are convolved with a Gaussian to match the instrument’s spectral resolution. The black lines below the continuum in the third and fourth panel mark the nebular Balmer lines. See text for details.

Open in new tabDownload slide

Our analysis yields similar values for the abundances and stellar parameters reported in previous works (Marcolino et al. 2007; Keller et al. 2014), although this work is based on a measured distance, hence we can give now a reliable estimate for L and |$\dot{M}$|⁠. We note that our C abundance is lower than that derived by Keller et al. (2014), who inferred the abundances from the UV spectra only, while we fitted also the optical spectrum.

3.3 Physical properties of NGC 6905

The seven ID spectra obtained from regions A1–A4 and A6–A8 (see Fig. 1) corresponding to the nebular emission of NGC 6905 are shown in Fig. 5. Fluxes of the nebular lines were measured using the Gaussian line fitting option of the splot task in iraf. The resultant line fluxes and statistical errors are listed in Table 7. Following Sabin et al. (in preparation), we quadratically add 5 per cent to the statistical errors described in Section 3.1 to take into account plausible additional uncertainties from the various calibrations, the reddening correction and the uncertainties on the atomic data used to derive the abundances.

Figure 5.

NOT ALFOSC spectra from the extracted regions A1–A8 in NGC 6905. The main emission lines are labeled. For the sake of clarity, H-Balmer lines are indicated in green, He lines in orange, and the rest of the lines in black. The spectrum of the CSPN HD 19394 (A5) is shown in red dotted line in the middle for reference. The colours of the spectra were coded as the extraction apertures indicated in Fig. 1. To avoid overlap, the spectra are shifted upwards (A6–A8) and downwards (A4–A1) with respect to the spectrum of the CSPN by appropriate amounts (its continuum level corresponds to its aperture number). Thereby, the ‘onion-like’ ionization structure of the object can be clearly appreciated. See Table 7 for details of the present lines and their fluxes.

Open in new tabDownload slide

The spectra have been analyzed by means of the pyneb code (version 1.1.15; Luridiana, Morisset & Shaw 2015). First, the correction from extinction was performed using the Cardelli et al. (1989) law with R_V = 3.1. The logarithmic extinction c(H β) was found to be in the range 0–0.22, consistent with that estimated from UV analysis towards the CSPN (see Table 6). Our NOT ALFOSC spectra are the highest-quality spectra of NGC 6905 presented so far (see e.g. Kingsburgh & Barlow 1994; Peña et al. 1998), which will allow us to assess possible physical and/or abundance differences within this WRPN.

We applied a Monte Carlo (MC) procedure to determine the propagation of the uncertainties from the emission lines into the subsequent determination of physical parameters and abundances. The details are described in length in Sabin et al. (in preparation) for the case of the WRPN PC 22. For all the parameters derived in the following we present the mean and standard deviation obtained from the MC analysis. We did not consider in our analysis the lines showing an error on their measurements greater than 50 per cent. The electron temperature (T_e) and density (n_e) are calculated for all the available ions and we distinguish the low and high ionization potentials with the [N ii] and [S ii] in one hand, and the [O iii], [Cl iii] and [Ar iv] on the other hand. The estimated T_e and n_e values for the different regions in NGC 6905 are listed in Table 8. We note for example that three different estimates for the T_e([O iii]) were attempted which correspond to three n_e estimations using the [S ii], [Cl iii], and the [Ar iv] doublets. Conversely, the corresponding n_e([S ii]), n_e([Cl iii]) and n_e([Ar iv]) were computed using the T_e([O iii]).