Element abundances in PG1159 stars Klaus Werner, Thomas Rauch, Elke Reiff University of Tübingen, Germany and Jeffrey W. Kruk Johns Hopkins University, U.S.A

Outline Basic properties: T eff, log g, mass (spectroscopic vs. asteroseismic results) Evolution: The different final thermal pulse (FTP) variants Abundances: main constituents (He,C,O); compare with [WC] stars and FTP model predictions Hydrogen and nitrogen abundances: which PG1159 star went through which FTP variant Trace element abundances; compare with [WC]s and AGB stellar model intershell abundances Footnote: the (almost) pure C/O atmosphere of H Summary and conclusions

PG1159 stars 40 objects known T eff =75,000 – 200,000 K log g=5.5 – 8 Evolutionary tracks: Wood & Faulkner (1986), Schönberner (1983), Blöcker (1995) (all with H-rich surface) Red track: H-deficient VLTP (Herwig 2003) Mean mass from (H-rich surface) tracks: 0.62 M . More modern tracks are systematically hotter (not a consequence of H-deficiency, but of more realistic AGB progenitor evolution), giving 0.57 M  (Miller Bertolami & Althaus 2006)

Comparison with asteroseismic mass determinations Blue dots: pulsating PG1159 stars Dashed lines: blue and red edges of GW Vir instability strip (Gautschy et al. 2005, and Quirion et al. 2004, respectively) Spectroscopic masses from T eff /log g and Miller Bertolami & Althaus (2006) tracks StarM spec M puls reference PG Corsico & Althaus 2006 PG Corsico et al. 2007b RXJ Corsico et al. 2007a PG Corsico & Althaus 2006 PG Corsico & Althaus 2006

Problems with mass determination Spectroscopic mass determination: Observational errors for T eff /log g. Gravity determination most problematic (line broadening theory). Accuracy in log g is 0.3 dex, at best, corresponding to ΔM  0.05 M  Run of evolutionary tracks depends on intershell opacities (Miller Bertolami & Althaus 2007). Probably most problematic: uncertain Fe abundance (see below); might even differ from star to star Pulsational mass determination: More precise, in principle, but: results depend on utilized method (asymptotic vs. averaged period spacing, Corsico et al. 2007a,b ). At present M puls as uncertain as M spec Conclusion: both methods give consistent results within error limits, with one significant exception: RXJ (M puls =0.56, M spec =0.72 M  ) Why should we care at all about an error of 0.1 M  in mass determination of a PG1159 star?

The uncertainty in the respective MS progenitor mass becomes large, e.g., M final =0.60 → M initial =2.0 M  and M final =0.68 → M initial =3.0 M  (Weidemann 2000). Nucleosynthesis in intershell of AGB stars is strongly mass-dependent, e.g., fluorine production : F intershell abundance as function of stellar mass and metallicity (Lugaro et al. 2004) Problems with mass determination

Conclusion: if we want to use abundance patterns of PG1159 stars as a tool to investigate AGB star nucleosynthesis, then we need a good mass determination. Or, in turn, if we believe that AGB star models describe nucleosynthesis quantitatively correct, then the observed abundance pattern of a PG1159 star constrains its mass. Recent discovery of the first PG1159 star in a close binary system (Nagel et al. 2006) might lead to a dynamical mass determination. Nice independent check for spectroscopic mass! Unfortunately, the star is not a pulsator… See poster by Schuh et al. for this interesting object (a PG1159 star found in the SDSS) Problems with mass determination

Evolution: the different final thermal pulse variants (For details see tomorrow’s talks by Herwig and Miller Bertolami) Idea that re-ignition of He-shell burning during post-AGB evolution (=“late thermal pulse”) causes H-deficiency, is quite old (Iben et al. 1983) Details have been worked out during last two decades (Blöcker & Schönberner, Iben, Herwig, Althaus et al.) We distinguish 3 different scenarios leading to H-deficiency, depending on the moment when the last thermal pulse (TP) occurs: Late TP, very late TP, AGB final TP

1. Very late thermal pulse (VLTP): He-shell burning starts on WD cooling track. Envelope convection above He-shell causes ingestion and burning of H. No H left on surface. 2. Late thermal pulse (LTP): He-shell burning starts on horizontal part of post-AGB track (i.e. H-shell burning still “on”). Envelope convection causes ingestion and dilution of H. Very few H left on surface (below 1%), spectroscopically undetectable in PG1159 and [WC] stars. 3. “AGB final” thermal pulse (AFTP): He-shell burning starts just at the moment when the star is leaving the AGB. Like at LTP, H is diluted but still detectable: H  20%. +CO core material (dredged up) from Lattanzio (2003)

Element abundances in PG1159 stars from spectroscopic analyses Abundances of main constituents, He, C, (O) usually derived from optical spectra (He II, C IV, O VI lines), in few cases: H Balmer lines detectable Trace elements: UV spectra (HST, FUSE) Model atmospheres: Plane-parallel, hydrostatic, radiative equilibrium, non-LTE Luminous PG1159 central stars exhibit P Cygni profiles; resonance lines of C IV, N V, O VI; and Ne VII, F VI Expanding NLTE models required for line profile analysis (see next talk, by Elke Reiff) Now: Results for abundances of He, C, O

Main atmospheric constituents: He, C, O Strong variation of He/C/O ratio from star to star: He=0.30 – 0.85 C=0.13 – 0.60 O=0.02 – 0.20 (mass fractions) Preference for helium abundance in range 0.3 – 0.5, no trend with stellar mass Tendency: high O abundance only found in stars with high C He vs. stellar mass O vs. C

He, C, O: comparison to [WC] stars and VLTP models Comparison with evolutionary models: Former TP-AGB star models (Schönberner 1979 ff.) show only low O abundance in intershell region: He  0.75 C  0.25 O=0.01—0.02 New models (Herwig et al ff, Althaus et al ff) assuming strong overshoot at bottom of pulse-driven convection zone, give high O abundance, e.g.: C=0.41 O=0.18 (3 M  model, Herwig 2000 ) C=0.50 O=0.23 (2.7 M  model, Althaus et al ) Compares well with observed PG abundances: C=0.50 O=0.17 (Werner et al. 1991) A series of VLTP models with post-AGB remnant masses M  (Miller Bertolami & Althaus 2006) can be used to compare their intershell abundances with PG1159 stars and [WC] stars:

HeCO PG1159 stars Werner & Herwig 2006 [WCL] stars Koesterke 2001 [WCE] stars Todt et al VLTP models’ intershell Miller Bertolami & Althaus 2006 [WCL] and [WCE] taken together, their abundances agree with PG1159s; clear hint for evolutionary link. Problem: [WCL] and [WCE] seem to form distinct groups. Different evolutionary paths? (see previous talks by Crowther and Todt) (De Marco, Crowther, Marcolino…)

Problem: VLTP models can explain some He/C/O abundance patterns but not all, namely, stars with lowest O abundances and highest He abundances, e.g., HS : He=0.85 C=0.13 O=0.02 (Dreizler & Heber 1998); could be result of different evolutionary path: low-mass star leaving the AGB without preceding thermally pulsing phase (so-called post-early AGB stars), then suffering late TP, i.e. late TP = 1st TP; possibly also related to He-dominant hot O(He) stars (Miller Bertolami & Althaus 2006) Details on this idea in talk by T. Rauch on O(He) stars later today HeCO PG1159 stars Werner & Herwig 2006 [WCL] stars Koesterke 2001 [WCE] stars Todt et al VLTP models’ intershell Miller Bertolami & Althaus 2006

Hydrogen and nitrogen Hydrogen discovered in four PG1159 stars, so-called “hybrid PG1159s”, Balmer lines, H=0.35, see talk by Marc Ziegler Can be explained by AFTP evolution models H found in three [WCL], H= (Hamann group) Nitrogen: Discovered in some PG1159 stars, N= , strict upper limits for some stars: N<3 ∙10 -5 (Dreizler, Heber, Rauch, Werner et al.) Similar result for [WC]s; N= (Hamann et al.) Nitrogen is a reliable indicator of a LTP or VLTP event: N<0.001  LTP, N  0.01  VLTP (nitrogen produced by H ingestion & burning) Hence: From H and N abundances we can conclude which late TP scenario the star underwent

Neon Usually the most abundant element after He, C, O in PG1159 stars Synthesized in He-burning shell starting from 14 N (from previous CNO cycling) via 14 N(α,n) 18 F(e + ) 18 O(α,  ) 22 Ne LTP and VLTP models predict Ne  0.02 Confirmed by spectroscopic analyses of several NeVII lines All NeVII lines were identified for the first time in stellar atmospheres, e.g.:  NeVII 3644Å first identified 1994 (Werner & Rauch) NeVII 973.3Å, one of strongest lines in FUSE spectra, first identified 2004 (Werner et al.)

Neon Newly discovered NeVII multiplet in VLT spectra (Werner et al. 2004): Allows to improve atomic data of highly excited NeVII lines (line positions, energy levels). Was taken over into NIST database (Kramida et al. 2006)

Neon The NeVII 973Å line has an impressive P Cygni profile in the most luminous PG1159 stars (first realized by Herald & Bianchi 2005): Also seen in [WCE] stars, allowing for abundance determination: Ne  0.02 (Bianchi & Herald 2006) In [WCL] stars, Ne I lines were identified in four objects, Ne= (Hamann and co-workers) In conclusion: Neon abundance in PG1159 stars and [WC] stars agree with predictions from (V)LTP models

Recent identification of NeVIII (!) lines in FUSE spectra of PG1159 and [WCE] stars (Werner et al. 2007) has important consequences NeVIII lines allow more precise T eff determination for hottest stars, e.g., the PG1159 star Longmore 4 (see Howard Bond’s talk later today) is not T eff =120,000, but 170,000 K (affects spectroscopic mass determination) Neon PG1159 [WCE] PG1159

Neon Another example: the hottest known DO white dwarf KPD is even much hotter than hitherto assumed: not T eff =120,000, but 200,000 K! (see poster by Werner et al.) We also find that optical emission lines in the hottest PG1159 and [WCE] stars are from NeVIII and NeVII, and not as hitherto assumed, from O VIII and O VII (as proposed by Barlow, Blades, & Hummer 1980) This solves a longstanding problem, because O ions have high ionization potentials, cannot be thermally excited Note that this affects the [WCE] classification scheme which is based on the ``O VII and O VIII´´ lines (e.g., Acker et al. 2003)

Fluorine ( 19 F) Interesting element, its origin is unclear: formed by nucleosynthesis in AGB stars or Wolf-Rayet stars? Or by neutrino spallation on 20 Ne in type II SNe? Up to now F only observed as HF molecule in AGB stars, F overabundant (Jorissen et al. 1992), i.e. AGB stars are F producers Discovery of high F overabundance in a C-enhanced metal-poor star (CEMP) is interpreted as pollution by AGB companion (Schuler et al. 2007) Would be interesting to known the AGB intershell abundance of F, use PG1159 stars as “probes”! Discovery of F V and F VI lines in a number of PG1159 stars (Werner et al. 2005) is the first identification of fluorine in hot stars at all! log F  = -6.3, i.e., fluorine overabundant by factor 200!

Fluorine ( 19 F) Wide spread of F abundances in PG1159 stars, solar Qualitatively explained by evolutionary models of Lugaro et al. (2004), large F overabundances in intershell, strongly depending on stellar mass: Range of fluorine intershell abundance coincides amazingly well with observations !!! But: we see no consistent trend of F abundance with stellar mass (our sample has M initial =0.8-4 M  ) Conclusion: fluorine abundances in PG1159 stars are (well) understood

Fluorine ( 19 F) F abundances in PG1159 stars and in AGB stars show a common trend with 12 C abundance Strong evidence that path for 19 F nucleosynthesis is strongly tied to 12 C production during TP-AGB evolution From Schuler et al. (2007) PG1159 stars (CEMP) 14 N(α,  ) 18 F(  + ) 18 O(p,α) 15 N(α,  ) 19 F

Fluorine ( 19 F) F VI λ Å line in [WCE] stars? No systematic search performed, yet (to my knowledge), but: P Cyg profile in a [WC]-PG1159 transition object (Abell 78) identified and modeled, see following talk by Reiff F in PNe? Slight overabundances (2 times solar) found for many PN, strong overabundance in PN around [WCL] NGC 40 (10 times solar) (Zhang & Liu 2006)

Argon Up to now, never identified in any hot star First identification of an Ar VII line (λ Å) in several hot white dwarfs and one PG1159 star (Werner et al. 2007); problem with other objects: blending H 2 interstellar line Argon abundance solar, ( ) in agreement with AGB star models, intershell abundance becomes only slightly reduced (Gallino priv. comm.)

Silicon Si abundance in AGB star intershell remains almost unchanged; solar Si abundances expected in PG1159 stars and [WC]s Practical problem: Si IV lines rather weak in hottest objects. But: Si V and Si VI lines discovered in HST spectra (first identification of these ions in any star) (Jahn et al. 2007) Results for 5 PG1159s show wide range, from solar down to <0.05 solar, i.e., Si=<4∙10 -5 … 70∙10 -5 (Jahn et al. 2007, Reiff et al. 2007) Surprising results for 6 [WCL]: Si= , i.e., 8-45 times oversolar. Two [WCL] have upper limits at  solar (Hamann group), and two have solar abundance (Marcolino et al. 2007) Large Si scatter (3 dex!) around solar value cannot be explained by final TP models

Phosphorus Discovered in PG1159 stars by identification of P V resonance doublet λλ 1118,1128 Å Two PG1159 stars have about solar P abundance (within 0.5 dex), three have upper limit solar abundance (Reiff et al. 2007) P in [WCE]? Only one rough estimate, “oversolar” in NGC 5315 (Marcolino et al. 2007) Strong enrichment predicted in the intershell of a M initial =3 M  model, 4-25 times solar, dependent on assumption of convective extramixing (Lugaro priv. comm.). Systematic investigations for different stellar masses lacking; consequences of uncertainties in n-capture reaction rates unknown Conclusion: Observed (roughly solar) P abundance is not understood. Systematic nucleosynthesis study in AGB stars badly needed

Phosphorus P V

Sulfur Discovered in a number of PG1159 stars by identification of S VI resonance doublet λλ 933, 945 Å One PG1159 star shows S solar (Miksa et al. 2002) while five others have 0.1 solar (Jahn et al. 2007, Reiff et al. 2007) Only upper limits (solar) derived for some [WC] (Marcolino et al. 2007) In contrast, only mild depletion occurs in the Lugaro models mentioned above: S=0.6 – 0.9 solar. Conclusion: Strong S deficiencies not understood. Like for P, systematic nucleosynthesis study in AGB stars urgently required

Iron and nickel A slight reduction of Fe, down to  90% solar (below spectroscopic sensitivity), in the AGB star intershell is expected because of n-captures on Fe nuclei (s-process) In Miksa et al. (2002) and subsequent work, however, significantly stronger Fe deficiencies were claimed in PG1159 stars and [WC] stars. Remarkable: in no single case, a  solar Fe abundance was found

Iron and nickel Detailed results on Fe: <10% solar in 2 PG1159 stars and a [WC]-PG1159 object (Miksa et al. 2002, Werner et al. 2003). One of it is a H-rich (i.e. AFTP) PG1159 star. Very strange! Watch talk by Ziegler. <20% solar in the prototype PG (Jahn et al. 2007)  25% solar, a [WCL] and a [WCE] (Marcolino et al. 2007)  33% solar, a [WCL] (Crowther et al. 1998) <10% solar, a [WCE] (Stasinska, Gräfener et al. 2004) Latest results on Fe in other PG1159 stars see talk by Reiff, she also will make a first estimate on nickel abundance. Was Fe transformed into Ni? Is Ni overabundant? If not, then Fe- deficiency is even harder to explain!

[WC]-PG1159 transition object

Dream: discovery of trans-iron group elements in PG1159 stars Ge overabundance (3-10 solar) detected in PNe (Sterling et al ), also for other s-process elements, Xe, Kr, overabundances claimed Indicative s-process elements being dredged up (TDU) Strongest Ge enrichment in PNe of [WC] stars Interpreted as consequence of late TP, but in contrast, Xe, Kr, should also show strongest enrichment in [WC] PNs, which is not the case (Sterling & Dinerstein 2006, Zhang et al. 2006) It would be highly interesting to discover n-capture elements in PG1159 stars in order to conclude on s-process nucleosynthesis in AGB stars. But: atomic data is the problem (almost no UV/optical line data available for high ionisation stages) Also, systematic intershell abundance predictions for such elements required, as, e.g., presented by Karakas et al. (2007) for Ge.

Composition profile of intershell abundances before last computed TP. Ge abundance near 10 -6, could be detectable spectroscopically (we found Ar at that abundance level in a H-rich central star). Search for these species (Ge, Ga, As, Xe, Kr ….) is not completely hopeless. Future HST/COS spectroscopy might play key role.

Remark: the engimatic case of H H is the hottest known PG1159 star (T eff =200,000 K, log g=8.0) H and He-deficient (He<0.01), dominated by C and O: C=0.48 O=0.48 Ne=0.02 Mg=0.02 Most massive PG1159 star: 0.74 – 0.97 M , but uncertain (which evolutionary track to compare with?) Speculation: H is a bare CO or even ONeMg white dwarf. Evolutionary scenario: completely unknown; may be a former C-burning super-AGB star; CO surface chemistry result of late carbon shell flash? You, too, are free to speculate! This star is a real challenge. No other star with this surface composition discovered, yet. Where are progenitors or descendants? Since its discovery in 1983, H is a “singularity”, but: watch Patrick Dufour’s talk on Thursday: announces discovery of possible descendants: cool WDs with pure C atmospheres!

Summary: PG1159 stars Main atmospheric constituents (He,C,O,Ne) are in agreement with stellar models Stars showing H or N are explained by an AFTP or VLTP event, respectively. All others experienced a LTP Theory can also explain the abundances of F, Ar Interpretation of  solar P abundances unclear (strong overabundance expected), could be explained by systematic stellar model calculations Strong depletion of S and Si in some objects (and strong Si enrichment in some [WC]) is probably a serious problem The extent of the observed iron deficiency is most surprising and lacks an explanation. Efficiently destroyed by n-captures? Is nickel overabundant? Element abundances in [WC] stars very similar. It seems clear that they are the progenitors of PG1159 stars Progeny of PG1159 stars: AFTP and LTP make DA WDs (H diluted, floats up); VLTP (H burnt) makes non-DA WDs.