Lithium abundances and isotope ratios, and troublesome stellar atmospheres Sean G. Ryan School of Physics, Astronomy and Mathematics University of Hertfordshire Principal collaborators Ana García Pérez(UH/Virginia), Adam Hosford(UH), Andy Gallagher(UH) Wako Aoki (NAOJ), Keith Olive (Minnesota), John Norris (ANU) Structure of this talk Two halo-star lithium problems (7Li and 6Li) Effective temperature scales Line profiles in 1D LTE, 1D NLTE, and 3D NGC 4414: Hubble Heritage Team (AURA/STScI/NASA) + Hack/Ryan (OU) s.g.ryan@herts.ac.uk

Lithium problem #1: 7Li Measurements of CMBR by WMAP give baryon density fraction Bh2 = 0.0224±0.0009 (Spergel et al. 2003). BBN depends on B. WMAP in excellent agreement with B derived from 2H/1H. Uncomfortable discrepancy for 7Li: the “Lithium problem” Coc & Vangioni (2005)

Lithium problem #1: 7Li Several explanations offered to explain WMAP discrepancy intriguing particle physics possibilities (failure of SBBN model): survival of metastable particles for a few ×103 s, i.e. during BBN Bird, Koopmans & Pospelov 2007,hep-ph/0703096: X- + 7Be → 7BeX- ; 7BeX- (p,γ) 8BX- → 8BeX- + β+ + νe Pospelov, M. 2007, hep-ph/0712.0647: X- + 4He → 4HeX- ; +4He → 8BeX- ; 8BeX- + n → 9BeX-* → 9Be + X- decay or annihilation of massive supersymmetric particle, modifying 7Li and 6Li production Jedamzik 2004

Lithium problem #1: 7Li Several explanations offered to explain WMAP discrepancy stellar destruction possibilities: Have some stars partially destroyed 7Li? mundane possibilities: Did we get the abundances wrong? Large uncertainty in low-Z colour-effective-temperature scales E.g. comparison between “cool” Ryan et al (2001) and “hot” Melendez & Ramirez (2004) Teff scales shows difference of up to 400K for [Fe/H] < -3 ΔTeff ≈ +400 K → ΔA(Li) ≈ +0.3 dex; close to discrepancy

Lithium problem #1: 7Li PhD: Adam Hosford Effective temperature scale for metal-poor stars: Use T-dependence of Fe I LTE level populations: Boltzmann factor exp-(χ/kT) Attention to error propagation Uncertainty in χ vs A(Fe) slope being nulled ~ 60-80 K evolutionary state weakly constrained ~ 12-24 K uncertainty in ξ ~ 30-90 K (wrong physics anyway → 3D)

Lithium problem #1: 7Li Hosford: Fe I LTE level populations Asplund et al. 2006: Hα Balmer profile fits Melendez & Ramirez: IRFM T,LTE similar to R01, A05 Hosford, Ryan, Garcia Perez, Norris & Olive 2009, A&A, 493, 601 Asplund’06 analysis: A05 in good agreement with b-y and V-K IRFM of Nissen et al. (‘02,’04): ΔTeff = -34 ± 95 K cooler than “hot” MR04 scale: ΔTeff = 182 ± 72 K @ [Fe/H] < -2.6). T(Ryan) T(Asplund) T(Hosford) T(Hosford) +200 K T(MR05) T(Hosford)

Lithium problem #1: 7Li Tχ,LTE assumes LTE Fe I level populations LTE holds at τcontinuum > 1, but lines form at τcontinuum < 1 NLTE difficult to calculate reliably Collisional excitation very uncertain Collisions with hydrogen parametrized via SH (= 0.001? 1?) Model atom incomplete Ideally calculate populations and radiative & collisional transition rates (need all oscillator strengths) for all levels (populations coupled by radiative and collisional transitions), but ... ... our/Collett model atom contains just 524 levels for Fe I, II and III; cf. NIST lists 493+578+567 levels for Fe I+II+III Confucius say: “Stay away from NLTE, and you can have a nice life.” F. Thevenin, c.2000

Lithium problem #1: 7Li Previous calculations at low Z point to overionisation as major effect: underpopulates Fe I levels relative to LTE e.g. Asplund et al. (1999, A&A, 346, L17; 2005 ARAA, 43, 481, §3.7) transparent layers with τcontinuum < 1 see photons from deep/hot atmosphere, so photon intensity Jν > local Bν. UV photons photoionise excited Fe I states. Additional factors: lack of collisions at τcontinuum < 1 reduces collisional excitation of excited levels compared to what local T suggests via Boltzmann (i.e. populations not in thermal equilibrium with local temperature) Net result: excited level populations lower than in LTE; Assess -dependence using MULTI calculations ...

Lithium problem #1: 7Li NLTE effects clearly depend on χ b ≡ nNLTE/nLTE (SH = 1) NLTE effects clearly depend on χ χLTE vs A(Fe) affected by NLTE, hence Tχ, LTE affected by NLTE Calculations vary from star to star, but (for six stars): T,NLTE ~ 110-160 K hotter than R01, A05, ~ 190 K cooler than MR04 Hosford, García Pérez, Collet, Ryan, Norris, Olive, 2010, A&A, 511, 47

Lithium problem #2: 6Li 6Li isotope shift = 0.15 Å; same as fine structure splitting Adds a little asymmetry to asymmetric line ... as does convection – but hard to model in 3D Cayrel ,et al. (incl. Ludwig), 2007, A&A, 473, 37 6Li < 0.00001 ppb in standard bbn Serpico et al. 2004 6Li not produced in stars: no stable A = 5 or 8 nuclei 6Li produced via galactic cosmic ray (GCR) spallation In Pop I alongside 9Be and 10,11B; at low Z via 4HeISM + αGCR Steigman & Walker 1992, ApJ, 385, L13; Yoshii et al. 1997, ApJ, 485, 605 (YKR) Boesgaard et al. 1999, AJ, 117, 1549 (BDKRVB) Duncan et al. 1997, ApJ, 488, 338 (DPRBDHKR) 6,7Li destroyed in stars in (p,α) reactions S-factor = 3140 keV barns for 6Li(p,3He)4He Elwyn et al. 79, PhysRevC, 20, 1984 S-factor = 55 keV barns for 7Li(p,4He)4He Pizzone etal. 03, A&A, 398, 423 6Li(p,α)3He ~2.0×106 K 7Li(p,α)4He ~2.6×106 K Survives (if at all) in warmest low-Z stars Brown & Schramm 88, ApJ, 329, L103

Lithium problem #2: 6Li Aoki et al. 2004, A&A, 428, 579 (AIKRSST) S/N = 1000 R = 90000 6Li/7Li = 0.00, 0.04, 0.08

Lithium problem #2: 6Li Two major results from Asplund et al. 2006: Abundance high compared to models that are consistent with spallative 9Be, 10,11B, especially if depletion allowed for. Trend with [Fe/H] looks like plateau, unlike strong [Fe/H] dependence of models.

Lithium problem #2: 6Li Subaru/HRS data on 5 stars. Isotope ratio VERY sensitive to systematic uncertainties: e.g. macroturbulent width, wavelength shifts, continuum errors, flat field errors, 7Li abundance fair choices → uncertainties Δ(6Li/7Li) ~ 3-4% García Pérez, Aoki, Inoue, Ryan, Suzuki, & Chiba, 2009, A&A, 504, 213

Lithium problem #2: 6Li Subaru/HRS data very similar to Asplund et al. VLT/UVES data ... ... but we are not confident of our “detections” Working at margins of significance due to systematic limitations VLT observations 4% 3% 2% 1% Asplund et al. 2006

Troublesome stellar atmospheres Barium isotope ratios Truran (1981) proposed that at low Z, r-process dominates over s-process since s-process seeds have low abundance whereas r-process seeds are made in the SN precursor (based partly on Spite & Spite (1978) Eu/Ba) Truran 1981, A&A, 97, 391 Spite & Spite 1978, A&A, 67, 23 Travaglio et al (1999) numerical GCE simulations confirm moderate-Z onset of s-process Travaglio et al. 1999, ApJ, 521, 691 But ... Magain (1995) found Ba 4554 isotope profile in HD 140283 more like s-process than r-process Magain 1995, A&A, 297, 686 Andy Gallagher (PhD thesis with SGR and AEGP): Use 2 analysis techniques (ex-6Li) to attempt to study 135,137Ba isotopic splitting in low-Z stars

Troublesome stellar atmospheres Sensitivities: macroturbulent broadening key (Lambert & Allende-Prieto, 2002, MNRAS, 335, 325) Fit via ~90 Fe I lines with WFe ~ WBa 4554 Gallagher, Ryan, Garcia Perez & Aoki, 2010, A&A, 523, A24 Gallagher, Ryan, Hosford, Garcia Perez, Aoki & Honda, 2012, A&A, 538, A118 Co-add residuals for all Fe I lines to see if any asymmetry 4/4 dwarfs show asymmetric red wing ~ 130 mÅ from line core Not improved switching ATLAS to MULTI LTE, or LTE to NLTE 2/2 giants are symmetric (though still large residuals)

Troublesome stellar atmospheres Experimented with three formalisms for macroturbulence, again fitting to ~90 Fe I lines Gaussian profile (+ Gaussian instrumental) Radial-tangential profile (+ Gaussian instrumental) vsini (+ Gaussian instrumental) Results: vsini : rarely the best profile (~ 5% of lines) Gaussian macroturbulence: sometimes the best profile (~20% of lines) Radial-tangential macroturbulence: most often the best profile (~80% of lines)

Concluding remarks 7Li: temperature scales from colours, IRFM, T,LTE and T,NLTE suggest 7Li not compatible with BBN/WMAP. 6Li: our Subaru data at best only marginally significant; uncertainties ~3-4% of A(7Li); not significant detections. 6Li, Ba & Fe: asymmetries seen in Fe I line residuals (and Ba II); could also be important for 6Li. Radial-tangential macroturbulence better than Gaussian ... but still artificial ... Need 3D atmospheres and radiative transfer. Observation-based challenge for emerging 3D codes: to reproduce observed shapes of Fe I lines in dwarfs and giants.

Cautionary remark 3D modelling (in NLTE) motivated by: observed asymmetries in Fe I dissatisfaction with  (microturbulence) dissatisfaction with  and/or  (macroturbulence) realisation that 3D radiative transfer in dynamical models may better explain line formation and hence affect interpretation of spectra But it may not deliver! M.Spite, 1997, IAUS, 189, 185

NGC 4414: Hubble Heritage Team (AURA/STScI/NASA) + Hack/Ryan (OU)