First discovery of trans-iron elements in a DAO-type white dwarf (BD−22°3467) Free

ABSTRACT

We have identified 484 lines of the trans-iron elements (TIEs) Zn, Ga, Ge, Se, Br, Kr, Sr, Zr, Mo, In, Te, I, Xe, and Ba, for the first time in the ultraviolet spectrum of a DAO-type white dwarf (WD), namely BD−22°3467, surrounded by the ionized nebula Abell 35. Our TIE abundance determination shows extremely high overabundances of up to 5 dex – a similar effect is already known from hot, H-deficient (DO-type) WDs. In contrast to these where a pulse-driven convection zone has enriched the photosphere with TIEs during a final thermal pulse and radiative levitation has established the extreme TIE overabundances, here the extreme TIE overabundances are exclusively driven by radiative levitation on the initial stellar metallicity. The very low mass (⁠|$0.533^{+0.040}_{-0.025}\, \mathrm{M}_\odot$|⁠) of BD−22°3467 implies that a third dredge-up with enrichment of s-process elements in the photosphere did not occur in the asymptotic giant branch (AGB) precursor.

1 INTRODUCTION

Trans-iron elements (TIEs) are synthesized during the asymptotic giant branch (AGB) phase of a star by the slow neutron-capture (s-)process. Depending on the initial stellar mass, its yields vary strongly (Karakas & Lugaro 2016). To become detectable, TIEs have to be transported from the helium-rich intershell region to the stellar surface. This happens, if the star experiences a third dredge-up (TDU; cf. Herwig 2000). A scenario in which the envelope becomes mixed with the intershell region is known as the late helium-shell flash. Such a late thermal pulse (LTP) was predicted e.g. by Iben et al. (1983). When it occurs after the star’s descent from the AGB at already declining luminosity, i.e. close to the end of nuclear burning, the H-burning shell is ‘off’ and a pulse-driven convection zone (PDCZ) establishes between the He-burning shell and the photosphere. The remaining H is mixed into the stellar interior, becomes diluted or even burned, making the star H-deficient (cf. Fujimoto 1977; Schönberner 1979; Iben et al. 1983; Blöcker 1995). Thus, it was – although surprising – well understandable that lines of 10 TIEs were identified (Werner et al. 2012a) in the ultraviolet (UV) spectrum of the DO-type white dwarf (WD) RE 0503−289 (effective temperature |${T_\rm{eff}} = 70\, 000 \pm 2000 \, \rm{K}$|⁠, surface gravity |$\log (g/\mathrm{cm\, s^{-2}}) = 7.5 \pm 0.1$|⁠; Rauch et al. 2016b), which became an archetype for TIE search in WDs. Presently, 18 of these species are identified in RE 0503−289 (Rauch et al., submitted). Chayer et al. (2005) first succeeded in the detection of TIEs in DO WDs, namely six species in two other objects (HD 149499 B, HZ 21).

In a subsequent investigation, TIE line identification was successfully performed in three related H-deficient objects (two DO-type WDs and one PG 1159-type WD, namely WD 0111+002, PG 0109+111, and PG 1707+427; Hoyer et al. 2018). The commonality of these stars is that they are located close to the so-called PG 1159 wind limit (Unglaub & Bues 2000) that approximately separates the regions of PG 1159-type stars and DO-type WDs in the Hertzsprung–Russell diagram (HRD). Here, the stellar wind is already weak enough and diffusion can establish strong TIE overabundances of up to 5 dex in the photosphere (Rauch et al. 2016a).

The search for TIE lines has not been restricted to He-rich WDs. Vennes, Chayer & Dupuis (2005) discovered the first TIE in WDs at all, namely Ge in three DA WDs. One of them is G191−B2B, an object that is employed as spectrophotometric flux standard for the Hubble Space Telescope (e.g. Bohlin 2007; Rauch et al. 2013). Recently, the TIEs Cu, Zn, Ga, Ge, As, Sn, and Ba were identified (Rauch et al. 2012, 2013, 2014a,b, 2015, 2016a; Rauch et al., submitted). The TIE abundance pattern is similar to RE 0503−289, but at a lower absolute level probably because of the lower T_eff of G191−B2B (⁠|${T_\rm{eff}} = 60\, 000 \, \rm{K}$|⁠; Rauch et al. 2013). TIE line search and abundance analyses are also successfully performed in the field of He-rich, hot subdwarf stars. The first one was LS IV−14°116, for which extreme overabundances of Fe, Sr, Y, and Zr were detected (Naslim et al. 2011). The most recent members of the group of ‘heavy metal’ subdwarfs are HZ 44, HD 127493, and Feige 46 (e.g. Dorsch, Latour & Heber 2019; Latour, Dorsch & Heber 2019), with TIE enrichment patterns similar to those in Fig. 1.

Abell 35 was discovered by Abell (1955) and characterized as homogeneous disc planetary nebula (PN; Abell 1966). Jacoby (1981) classified the visible nucleus as G8 III−IV. Later, Grewing & Bianchi (1988) classified the hot, ionizing central star as DAO-type white dwarf (WD). Shortwards of 2800 Å, the WD dominates the flux, whereas the cool companion outshines it in the optical. Herald & Bianchi (2002) analysed the binary and found |${T_\rm{eff}} = 80\, 000 \, \rm{K}$| and |$\log (g/\mathrm{cm}\,\mathrm{s}^{-2}) = 7.7$| for the hot and |${T_\rm{eff}} = 5000 \, \rm{K}$| and |$\log g = 3.5$| for the cool star. Ziegler et al. (2012) corrected the surface gravity of the WD to |$\log g = 7.2$| and Frew & Parker (2010) classified the nebula as ‘bow shock nebula in a photoionized Strömgren sphere’. Recently, a close reinspection of the UV spectrum of the exciting star of the ionized nebula Abell 35, BD−22°3467 (WD 1250−226; McCook & Sion 1999), led us to the identification of TIE absorption lines. In this work, a systematic TIE line search was performed in order to constrain abundances analogously to Hoyer et al. (2017, for RE 0503−289). It is based on the BD−22°3467 model of Ziegler et al. (2012, atmospheric parameters given in Table A1) that was calculated using the non-local thermodynamic equilibrium (NLTE) model atmosphere code of the Tübingen NLTE Model Atmosphere Package (tmap;¹ Werner et al. 2003; Werner, Dreizler & Rauch 2012b). In Sections 2 and 3, we briefly describe the available observations and the model atmospheres used for the spectral analysis, respectively. In Section 4, the process of line identification and subsequent abundance measurement is explained. Lastly, we summarize our results and conclude in Section 5.

2 OBSERVATIONS

Our analysis is based on high-resolution Far Ultraviolet Spectroscopic Explorer (FUSE) and Hubble Space Telescope/Space Telescope Imaging Spectrograph (HST/STIS) observations. These were obtained from the MAST² archive. The FUSE spectrum taken with the LWRS aperture has a resolving power of |$R = \lambda \, /\, \Delta \lambda \approx 20\, 000$|⁠. Four STIS observations with grating E140M and |$R = 45\, 800$| are available. The observation log is shown in Table B1. We convolved the synthetic spectra with Gaussians to model the respective instrument’s resolution. The signal-to-noise ratio of the STIS observations was improved by co-adding the observations. The combined spectra are the same as used by Ziegler et al. (2012). No optical observation of the DAO WD is available, since the G-star companion dominates this spectral range.

3 MODEL ATMOSPHERES AND ATOMIC DATA

The analysis was carried out using tmap. This code assumes plane-parallel geometry and calculates chemically homogeneous NLTE atmospheres in radiative and hydrostatic equilibrium.

For the TIEs Cu, Zn, Ga, Ge, Se, Br, Kr, Sr, Zr, Mo, In, Te, I, Xe, and Ba, we used the recently calculated data that is available via the Tübingen Oscillator Strengths Service (TOSS). For the elements with Z ≥ 20, it is necessary to create model atoms using a statistical approach that calculates super levels and super lines (Rauch & Deetjen 2003) to take their complex atomic structure into account for the calculation. The statistics of all elements considered in our model-atmosphere calculations are summarized in Table B2.

We constructed a new classical model ion for Ba viii from the level and line data of Churilov, Joshi & Gayasov (2001) available via the National Standards and Technology Institute (NIST) Atomic Spectra Database (ASD;³ Kramida et al. 2018), which was incorporated into the Tübingen Model Atom Database (TMAD).

For all considered elements with an atomic number Z ≤ 28, we used the same model atoms like Ziegler et al. (2012). For Z < 20 these were obtained from the TMAD (Rauch & Deetjen 2003) that was constructed as part of the German Astrophysical Virtual Observatory (GAVO). For the iron-group elements (IGEs, atomic number 20 ≤ Z ≤ 28), Kurucz’s line lists⁴ (Kurucz 2009, 2011, 2017) were utilized.

For this analysis, we adopted the photospheric parameters of Ziegler et al. (2012) and used their final model (⁠|${T_\rm{eff}} = 80\, 000 \, \rm{K}$|⁠, |$\log g = 7.2$|⁠, see Table A1 for the element abundances) to start our TIE analysis. To identify lines and determine abundances of the 15 TIEs (Cu, Zn, Ga, Ge, Se, Br, Kr, Sr, Zr, Mo, In, Te, I, Xe, and Ba), we performed line-formation calculations by adding each of them individually to the start model from Ziegler et al. (2012), while temperature and density structure of the atmosphere were kept fixed. To verify our method, a final model including all TIEs with their previously determined abundances was then calculated with temperature and density corrections. The deviations in the abundances were marginal.

The observed spectra are affected by reddening due to interstellar material within the line of sight. By comparing the slope of the flux calibrated observations and the Galaxy Evolution Explorer (GALEX), Hipparcos, and Two Micron All-Sky Survey (2MASS) magnitudes (Cutri et al. 2003; Perryman et al. 2009; Bianchi et al. 2011) with the synthetic spectra of the central star, Ziegler et al. (2012) found a colour excess |${E({B-V})} = 0.02 \pm 0.02$|⁠. This value was used to apply interstellar reddening following the law of Fitzpatrick (1999, with the standard R_V = 3.1) to the model spectra to reproduce the observation. Absorption due to neutral interstellar hydrogen, assuming a column density of |${N_{\rm H \thinspace\small {I}}} = 5.0 \times 10^{20}\, \rm{cm}^{-2}$| (Ziegler et al. 2012), was applied to the synthetic spectra. Furthermore, we applied the interstellar line-absorption model of Ziegler et al. (2012) that was calculated using the program owens (Hébrard et al. 2002; Hébrard & Moos 2003) to unambiguously identify lines of stellar and interstellar origin.

4 LINE IDENTIFICATION AND ABUNDANCE DETERMINATION

We calculate synthetic spectra from our line-formation models with each of the 15 elements added individually to the best model of Ziegler et al. (2012). The spectrum is crowded with a multitude of blended metal lines that hampers their unambiguous identification. To clearly see the contribution of the individual TIE elements, we divided the synthetic spectrum including this species by another model spectrum without it. The individual abundances were varied by small steps of 0.2 dex or smaller to derive the final values from evaluation of line-profile fits by eye. To estimate the influence of the uncertainty in T_eff of |${\pm} 10\, 000$| K and in log g of ±0.3 for the error propagation, we redid the abundance determination for models with |${T_\rm{eff}} = 90\, 000 \, \rm{K}$| and |$\log g = 6.9$| and for |${T_\rm{eff}} = 70\, 000 \, \rm{K}$| and |$\log g = 7.5$|⁠. The abundances are affected by typical errors below 0.3 dex.

For Cu, a line identification was not possible with appreciable certainty. Instead, upper limits were determined by reducing the abundance until the strongest computed lines become undetectable. An equivalent width of W_λ = 5 mÅ was set as a detection limit. Tables B3–B17 list all lines of TIEs that appear with an equivalent width above the threshold in the model spectrum. These tables include also those lines that could not be identified in the spectrum of BD−22°3467 due to e.g. blending with other photospheric or interstellar lines to make them a useful tool for the identification of TIE lines in the spectra of other DAO-type WDs. The abundances are given in Table A1 and are illustrated in Fig. 1. The complete FUSE and STIS observations compared to our best model are shown online⁵ within the Tübingen VISualization (TVIS) tool. The ionization fractions and the temperature structure and electron density in the final atmosphere model are shown in Fig. B1.

The number of identified lines per TIE ion is shown in Table 1. The observation is well reproduced by our final model with the abundances shown in Table A1 as it is illustrated in Figs B2–B16 for prominent lines of each of the TIEs.

5 RESULTS AND CONCLUSIONS

To identify TIE lines, the UV spectrum of BD−22°3467 was closely inspected that led to the discovery of Zn, Ga, Ge, Se, Br, Kr, Sr, Zr, Mo, In, Te, I, Xe, and Ba (Table 1). In total, 484 TIE lines were discovered.

Our spectral analysis has shown that the enrichment of TIEs in BD−22°3467 (⁠|${T_\rm{eff}} = 80\, 000 \pm 10\, 000 \, \rm{K}$|⁠, |$\log g = 7.2 \pm 0.3$|⁠) and RE 0503−289 (⁠|${T_\rm{eff}} = 70\, 000 \pm 2000 \, \rm{K}$|⁠, |$\log g = 7.5 \pm 0.1$|⁠) is comparably high (⁠|${\approx} 1.5 \rm {-} 5\, \mathrm{dex}$|⁠, Fig. 1). The origin of the high enrichment of TIEs is diffusion, i.e. efficient radiative levitation. This was shown already for RE 0503−289 by detailed diffusion calculations (Rauch et al. 2016a). While it was possible to determine abundances for several TIEs with consecutive atomic number in RE 0503−289 and find that the odd–even shape of the solar abundance pattern seems to be reflected also by the enriched TIEs (Fig. 2, cf. Rauch et al., submitted), there is not enough information to confirm this finding based on the results for BD−22°3467.

The evolutionary difference between BD−22°3467 and RE 0503−289 is that the latter most likely experienced an LTP in which it became hydrogen deficient. As a result, a pulse-driven convection zone, established during the flash, enriched the TIEs in the atmosphere. Their abundances were later on amplified to the observed values by radiative levitation. In contrast, the high abundances of TIEs in BD−22°3467 are possibly the result of radiative levitation on the initial stellar metallicity without previous enrichment by s-processed matter. This is because of the very low mass of BD−22°3467 (Fig. 1), which corresponds to an initial mass of below |$1.0\, \mathrm{M}_\odot$| (Cummings et al. 2018), implying that no TDU occurred on the AGB (Karakas & Lugaro 2016). From the position in the T_eff–log g diagram (Fig. 3) only, a possible evolution without an AGB phase, directly from the extended horizontal branch (EHB) to the WD cooling sequence, cannot be excluded. Conversely, it is then possible that the high amount of TIEs in RE 0503−289 is solely due to diffusion, independent of the occurrence of a previous LTP. This is an interesting conclusion, because a large fraction of DOs is not initiated by an LTP but by a merger event with the so-called O(He) stars as merger products and DO precursors (Reindl et al. 2014). DOs from this evolutionary channel should therefore also go through a phase with an extreme TIE enrichment by diffusion.

We have to keep in mind that diffusion-established abundance patterns do not contain any more information about the previous stellar evolution, i.e. wherever the TIEs (or other elements) stem from, the exhibited surface abundance may be the same. Investigations on yields of the AGB s-process nucleosynthesis elements have to be performed before diffusion dominates the stellar evolution. This is the phase of just declining luminosity when the strength of the stellar wind decreases but is still high enough to maintain the original abundance ratios produced by the s-process. Spectral analyses of stars in that evolutionary phase might help to directly constrain AGB nucleosynthesis.

The formation of Abell 35-like central stars of planetary nebulae (CSPNe), i.e. binary CSPNe with a rapidly rotating late-type (sub)giant and an extremely hot companion (Bond, Ciardullo & Meakes 1993), is discussed controversially in the literature. Thevenin & Jasniewicz (1997) found the companion of BD−22°3467 to be enriched in Ba indicating that this still unevolved star experienced mass transfer from a (post-) AGB star. However, this formation channel is debated since Abell 35 was found to only mimic a PN (Frew & Parker 2010) and the mass of the ionizing star was considered to be too low for a post-AGB star (Ziegler et al. 2012). As explained above, an evolution directly from the EHB to the WD cooling track is possible (Fig. 3). The peculiar ionized nebula around BD−22°3467 is not a real PN but, nevertheless, it also cannot be excluded that the nebula material is in some way connected to the evolution of the central star. Assuming that it ejected a PN as an AGB star, the original PN might have already dispersed. The star has a high proper motion (μ_α = −54.566 ± 0.226 mas yr⁻¹ and μ_δ = −10.097 ± 0.187 mas yr⁻¹; Gaia Collaboration et al. 2018) and, thus, might now be passing through the edges of the ejected former nebula material or another dense interstellar medium (ISM) region while ionizing the surrounding material. The classification of Abell 35 as a bow shock nebula in a photoionized Strömgren sphere in the ambient ISM (Frew & Parker 2010) does not necessarily include a PN but also, does not rule out the post-AGB nature of BD−22°3467. Further detailed abundance analyses of a sample of Abell 35-like CSPNe and their ambient nebulae and companion stars should give us a better handle on their evolution. The nebulae, if ejected from an AGB star, contain signatures of s-process elements (Madonna et al. 2017) and the unevolved companions, if they accreted a fraction of the ejected material. Therefore, a precise knowledge of the companion is mandatory because any accreted material would become diluted in this star’s convective envelope.

To better understand the late evolutionary phases of low-mass stars, it is highly desirable to improve the determination of T_eff and log g with much narrower error ranges. For this purpose, the analysis of high-resolution optical spectra may be helpful. Although the cool companion dominates this wavelength regime, broad lines of H and He of BD−22°3467 should be detectable like demonstrated by Aller et al. (2015) for the binary central star of NGC 1514. These lines may reduce the error and, thus, allow to better constrain the stellar mass (Fig. 3).

As a remark, we would like to mention that the discrepancy found by Ziegler et al. (2012) between the spectroscopic distance of |$361^{+195}_{-137}$| pc and the distance based on the Hipparcos parallax is still present in the era of Gaia with a distance of |$124.84^{+2.21}_{-2.13}$| pc (Bailer-Jones et al. 2018). Ziegler et al. (2012) demonstrated that the H i lines in the FUSE range are poorly reproduced with models with log g > 7.7. This phenomenon of too large spectroscopic distances has already been reported in the literature for CSPNe (Schönberner, Balick & Jacob 2018; Schönberner & Steffen 2019). The argument that missing metal-line blanketing and back warming may result in too high temperatures does not hold in our analysis, because all elements in the model calculation were considered in full NLTE computations. The objects of the mentioned studies are all located before the knee at highest temperatures in the HRD. Thus, the spectroscopic mass derived from fitting of the spectral energy distribution using the parallax distance is systematically too high. In our case, the spectroscopic mass derived from the dereddened GALEX FUV flux (Bianchi et al. 2011) using the Gaia distance is unreasonably low. With the given T_eff, at least a |${\log g} \gt 8.0$| would be required to reach masses above 0.4 M_⊙. In conclusion, the discrepancy remains unexplained and needs further investigation.

Following the discovery of TIE lines in BD−22°3467, we have initiated an analogous search in other DAO-type WDs. Since the FUSE and HST archives provide quite a number of high-quality UV spectra of such stars, which have not been inspected in focus of TIEs, we expect to identify TIE enrichment as a common phenomenon in many hot WDs.

ACKNOWLEDGEMENTS

We thank the referee for the very useful comments that improved this paper. LL has been supported by the German Research Foundation (DFG) under grant WE 1312/49 − 1. MAM had been supported by the DAAD RISE Germany program. The GAVO project at Tübingen had been supported by the Federal Ministry of Education and Research (BMBF) at Tübingen (05 AC 6 VTB, 05 AC 11 VTB). Financial support from the Belgian FRS-FNRS is also acknowledged. PQ is research director of this organization. Some of the data presented in this paper were obtained from the Mikulski Archive for Space Telescopes (MAST). STScI is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS5-26555. Support for MAST for non-HST data is provided by the NASA Office of Space Science via grant NNX09AF08G and by other grants and contracts. The TIRO (http://astro-uni-tuebingen.de/ TIRO), TMAD (http://astro-uni-tuebingen.de/ TMAD), TOSS (http://astro-uni-tuebingen.de/ TOSS), and TVIS (http://astro-uni-tuebingen.de/ TVIS) tools and services used for this paper were constructed as part of the Tübingen project (https://uni-tuebingen.de/de/122430) of the German Astrophysical Virtual Observatory (GAVO, http://www.g-vo.org). This research has made use of NASA’s Astrophysics Data System and of the SIMBAD data base, operated at CDS, Strasbourg, France. This work has made use of data from the European Space Agency (ESA) mission Gaia (https://www.cosmos.esa.int/gaia), processed by the Gaia Data Processing and Analysis Consortium (DPAC, https://www.cosmos.esa.int/web/gaia/dpac/consortium). Funding for the DPAC has been provided by national institutions, in particular the institutions participating in the Gaia Multilateral Agreement.

Footnotes

REFERENCES

APPENDIX A: PHOTOSPHERIC PARAMETERS OF BD−22°3467

APPENDIX B: ADDITIONAL FIGURES AND TABLES