A non-LTE spectral analysis of the 3He and 4He isotopes in the HgMn star kappa Cancri
aa r X i v : . [ a s t r o - ph . S R ] D ec Astronomy & Astrophysicsmanuscript no. aa25037-14 c (cid:13)
ESO 2018January 6, 2018
A non-LTE spectral analysis of the He and He isotopesin the HgMn star κ Cancri ⋆ Natalia L. Maza ,⋆⋆ , María-Fernanda Nieva , , and Norbert Przybilla Instituto de Ciencias Astronómicas, de la Tierra y del Espacio (ICATE), Av. España 1512 sur, 5400 San Juan, Argentina Dr. Karl Remeis-Observatory & ECAP, University of Erlangen-Nuremberg, Sternwartstr. 7, 96049 Bamberg, Germany Institute for Astro- and Particle Physics, University of Innsbruck, Technikerstr. 25 /
8, 6020 Innsbruck, Austriae-mail: [email protected]
ABSTRACT
Aims.
We present a pilot study on NLTE line-formation computations for the isotopes He and He in the mercury-manganese star κ Cancri. The impact of NLTE e ff ects on the determination of isotopic abundances and the vertical stratification of helium in theatmosphere is investigated. Methods.
Modern NLTE line-formation computations were employed to analyse a high-resolution and high-S / N ESO-VLT / UVESspectrum of κ Cnc. The atmospheric parameters were determined from fitting the hydrogen Balmer lines and the spectral energydistribution. Multiple He i lines were investigated, including He i λ λ Results.
Half of the observed He i lines in the spectrum of κ Cnc show significant NLTE strengthening, the e ff ects are strongest inthe red lines He i λ i λ i lines are up to a factor of ∼ κ Cnc. While the LTE analysis indicates a step-like profile of the heliumabundance, a gradual decrease with height is indicated by the NLTE analysis. A He / He ratio of ∼ Conclusions.
This work implies that NLTE e ff ects may be ubiquitous in the atmospheres of HgMn stars and may have a significantimpact on abundance determinations and the interpretation of the vertical abundance stratification of elements. Key words. stars: abundances – stars: atmospheres – stars: chemically peculiar – stars: early-type – stars: individual: κ Cancri
1. Introduction
Mercury-manganese stars are chemically peculiar (CP) stars ofthe upper main sequence, with the line spectra of Hg ii and Mn ii indicating large overabundances of the two elements. They con-stitute Preston’s subgroup CP3 (Preston 1974). Their e ff ectivetemperatures range between 10 500 and 16 000 K, correspond-ing to spectral types A0 to B6 (Smith 1996). Some other dis-tinctive characteristics of these stars are low rotational veloc-ities ( v sin i ≤
29 km s − , Abt et al. 1972) and overabundancesof other elements such as P, Ga, Y, and Pt, with abundancepatterns changing from one star to the other. They can showan inhomogeneous distribution of some chemical elements overtheir surface (first discussed by Hubrig & Mathys 1995), whichcauses the observed line-profile variability in the spectra (e.g.Hubrig et al. 2011). HgMn stars are members of binaries in morethan 50% of the cases, with periods ranging between 3 to 20 days(Gerbaldi et al. 1985).The chemical peculiarities in the atmospheres of HgMn starsarise because of their extremely stable atmospheres, exposingthe atoms and ions to the competitive actions of gravitationalsettling and radiative levitation. The observed abundance pecu-liarities can in principle be explained within the framework ofatomic di ff usion (Michaud et al. 1979). Di ff usion theory predicts ⋆ Based on data products from observations made with ESO Tele-scopes at the La Silla Paranal Observatory under programme ID 076.B-0055(A). ⋆⋆ Visiting scientist at the Institute for Astro- and Particle Physics,University of Innsbruck. that the photospheres of all HgMn stars should be deficient in he-lium, which was confirmed for instance in a dedicated study byDworetsky (2004). The reason is that helium has only few andweak lines at wavelengths where the stellar flux is high (long-ward of the Lyman edge), therefore gravitational settling domi-nates. Dworetsky also found evidence for a vertical helium abun-dance stratification in two HgMn stars, which would also be aconsequence of atomic di ff usion in the atmosphere.Moreover, helium is a particular case for the occurrenceof di ff erential e ff ects of di ff usion on the isotopes of an ele-ment. The mass di ff erence between its two isotopes, He and He, is largest among the light elements, with gravitational set-tling favouring the heavier isotope. Some CP stars can there-fore show enhanced He well above the protosolar He / He ratioof (1.66 ± × − (Maha ff y et al. 1998). This is most pro-nounced for the (He-weak) He stars, which occupy a narrowstrip in the log T e ff –log g plane between the He-strong B starsand a group of He-weak B stars that show no evidence of He.However, the presence of a significant amount of He was alsosuggested for some HgMn stars (Hartoog & Cowley 1979).Analyses of helium abundances in CP stars (and also forother elements) have so far mostly been performed under theassumption of local thermodynamic equilibrium (LTE). On onehand, this was possibly motivated by the early non-LTE (NLTE)study by Auer & Mihalas (1973), who found that the blue-violetHe i lines are described well under the assumption of LTE inthe temperature range below 15 000 K. On the other hand, thishas also practical reasons because LTE line-formation can bemuch easier implemented than NLTE modelling. However, the Article number, page 1 of 7 & Aproofs: manuscript no. aa25037-14
Fig. 1.
Comparison of our global best-fit NLTE synthetic spectrum (red line) with observed spectral lines of H β to H ǫ (black line) in κ Cnc. quality of the observations has improved tremendously over thepast decades, and we consider the topic worth to merit a re-visitin an era where quantitative spectroscopy can be performed athigh precision and accuracy. A more consistent NLTE modellingof CP stars may therefore help in developing a quantitativelyrefined view of di ff usion theory, which is required to explainthe variety of observed abundance peculiarities even within oneclass of CP stars.We chose the prototype HgMn star κ Cancri (HD 78316,HR 3623) for a pilot study that aims at re-investigating the topicbased on our latest NLTE models. This well-known sharp-linedstar was reported to be a possible He star by Hartoog & Cowley(1979), who derived an upper limit on the He / He ratio of 0.35,more than a thousand times higher than the protosolar value. Thepresence of He in the photosphere of this star was confirmed byDobrichev et al. (1989) and Zakharova & Ryabchikova (1996),concluding on a value of the He / He ratio of 0.35. These previ-ous works were all based on LTE modelling and spectra at lowerresolution than available today. Here, we investigate NLTE ef-fects and their impact on the determination of (isotopic) abun-dances and the vertical stratification of helium, considering var-ious He i lines.
2. Observational data
The spectrum employed in the present work was observed withU ves (UV-Visual Echelle Spectrograph, Dekker et al. 2000) onthe ESO VLT / UT2 at Cerro Paranal inChile under the pro-gram 076.B-0055(A). We extracted the pipeline-reduced spec-trum from the ESO Science Archive Facility. The spectrum wasobtained in dichroic mode, covering the spectral range λλ R = λ/ ∆ λ ≈
110 000. The peak S / Nratio is ∼ κ Cnc.
U BV photometry was adopted from Mermilliod (1991)and
JHK data from the Two Micron All Sky Survey (2MASS,Skrutskie et al. 2006). Two low-dispersion spectra that were ob-served with the International Ultraviolet Explorer (IUE) using alarge aperture were extracted from the MAST archive . The ex-posures SWP06911 and LWR05873 cover the range from 1150to 1980 Å and from 1850 to 3290 Å.
3. Model calculations
We employed a hybrid NLTE approach for our line-formationcalculations, which has been successfully used before for quan-titative analyses of main-sequence OB stars (Nieva & Przybilla http://archive.stsci.edu/ Table 1.
Atmospheric parameters
Parameter κ CncSp. Type B8 III V (mag) 5.233 B − V (mag) − E ( B − V ) (mag) 0.038 T e ff (K) 12800 ± g (cgs) 3.70 ± ξ (km s − ) 0 + v sin i (km s − ) 6 ± ζ (km s − ) 4 ± tlas , chemical homogeneity, and hy-drostatic, radiative, and local thermodynamic equilibrium (LTE).A value of triple solar metallicity was adopted for the model at-mosphere calculations for κ Cnc, which is found to reproduce thelow-resolution UV spectrum overall well.Then, NLTE line-formation computations were performedwith the codes D etail and S urface (Giddings 1981; Butler &Giddings 1985). D etail calculates atomic-level populations bysolving the coupled radiative transfer and statistical equilibriumequations, and S urface computes the formal solution using real-istic line-broadening functions. The following model atoms wereemployed: H (Przybilla & Butler 2004), He i (Przybilla 2005),and, for some supplementary calculations, O i/ii (Przybilla et al.2000; Becker & Butler 1988, updated). To facilitate NLTE cal-culations for the He isotope in addition to the usual He, a He i model atom in analogy to that of He by Przybilla (2005) wasrealized, taking into account isotopic line shifts as measured byFred et al. (1951). Both isotopes were treated simultaneously toaccount for the overlap of lines and continua.Our approach allows LTE spectrum synthesis to be per-formed by running S urface only on top of the A tlas ff ects.
4. Atmospheric parameter determination κ Cnc is a known single-lined spectroscopic binary (SB1) star.The system has been resolved among others by Schöller et al. The ratio of atmospheric thickness (geometrical height from the outerrim to the optical continuum-forming layers from the A tlas ∼ ff ects are negligible here,despite the star’s classification as giant.Article number, page 2 of 7aza et al.: NLTE spectral analysis of He and He in κ Cnc
Fig. 2.
Comparison of the A tlas (2010), who found a ∼ ff erence in the K band betweenthe primary and the (later-type) secondary. Because the bright-ness ratio increases towards shorter wavelengths, we do not ex-pect the available spectrum to be contaminated by any significantsecond light.Therefore, the analysis of the observed spectrum followed asimilar philosophy as previously employed by us for studies ofnormal B-type stars in Nieva & Przybilla (2007, 2012). We usedthe hydrogen Balmer lines and the SED as principal indicatorsto determine the e ff ective temperature T e ff and the surface grav-ity log g . Supplementary indicators were two hydrogen Paschenlines and the ionization balance of O i/ii . A fine-tuning of the at-mospheric parameters was achieved by multiple iterations aim-ing at reproducing all indicators simultaneously. At this stage ofthe analysis, an approximate average value for the helium abun-dance in the line-formation region was determined, which hadto be assumed to be homogeneous throughout the entire atmo-sphere because of the basic assumptions made in A tlas
9. Notethat this introduces no restrictions to the further discussion, ashelium behaves as a trace element in our case. A zero microtur-bulent velocity ξ was adopted, which facilitates reproducing theweak and strong oxygen lines for one abundance value simul-taneously. The (radial-tangential) macroturbulent velocity ζ andprojected rotational velocity v sin i were also determined basedon fits to the oxygen lines. The finally adopted atmospheric pa-rameters together with a few supplementary data are summarisedin Table 1. Our T e ff is significantly lower (by up to ∼ g slightly lower than in earlier LTEanalyses (e.g. Roby & Lambert 1990; Dworetsky & Budaj 2000;Adelman et al. 2004; Bailey & Landstreet 2013; Takeda et al.2014). The higher temperatures are not compatible with theobserved SED. On the other hand, our results agree ex-cellently well with the atmospheric parameters adopted byZakharova & Ryabchikova (1996) and Maza et al. (2011).Figures 1 and 2 show a comparison of a global (by-eye)best-fit NLTE spectrum with H β to H ǫ and the fit of the A tlas E ( B − V ) as indicated in Ta-ble 1 and a ratio of total-to-selective extinction R V = A V / E ( B − V ) = Fig. 5.
Comparison of our NLTE spectrum synthesis (red lines) withobservation for the two lines with the widest isotopic separation (blacklines). Synthetic LTE spectra computed for the same isotopic heliumnumber fractions are also shown (dotted red line). The positions of theline centres of the isotopic components are indicated.
Fig. 6.
Comparison of our two-component NLTE fits with the strongestobserved di ff use He i lines. The upper curves in each panel visualise thefit to the line wings, the lower ones the fit to the line core. See Fig. 5 forthe description of the line encoding. The two vertical lines indicate thedegree of the isotopic shift in the two cases. tion is achieved. This also includes the confirmation of the atmo-spheric parameter determination by the secondary indicators, thePaschen lines P and P (the only useful lines from the near-IR region, see the online Fig. 3) and the O i/ii NLTE ionizationbalance (see the online Fig. 4). The latter is fulfilled for a ho-mogeneous oxygen abundance of log (O / H) + =
5. Helium abundance analysis
The visual inspection of the helium spectrum of κ Cnc showstwo peculiarities. First, an asymmetry of He i λ i λ He isotope in the stel-
Article number, page 3 of 7 & Aproofs: manuscript no. aa25037-14
Fig. 7.
Comparison of our NLTE spectrum synthesis with all other He i lines that were analysed here. Shown are individual best fits. Lines areencoded in the same way as in Fig. 5. lar atmosphere (see Fig. 5), confirming the detection ofHartoog & Cowley (1979) at much higher spectral resolution(cf. also Zakharova & Ryabchikova 1996). Second, the strongestamong the di ff use He i lines, at λ λ i spectrum uses line-profile fits to the observed (largely) unblended features (He i λλ i λ ii λ i λ ff ects. A helium abun-dance higher by a factor of almost 3 is indicated by the quanti-tative LTE analysis in the latter case. The He and He num-ber fractions ε (He) derived from the NLTE and LTE analysisare summarised in Table 2, together with the Rosseland opticaldepths τ ross at which the line cores are formed. The abundanceratio He / He is about 0.25 to 0.30 at the atmospheric depthscovered by the observations, slightly lower than the value of 0.35derived by Zakharova & Ryabchikova (1996).The isotopic shifts are much smaller in all other helium lines,such that only a combined abundance from both isotopes can bederived (we kept the He / He ratio fixed to 0.3, consistent withthe value range determined above during our modelling). For thestrongest di ff use He i lines we performed a two-component fit(Fig. 6). One model was tuned to reproduce the line core, theother to reproduce the line wings. The derived abundances aresummarised in Table 3 in analogy to Table 2. Data for the wingsof the lines were evaluated 2 Å redwards of the line core. Inthe cores of the two lines, NLTE abundances are smaller thanderived in LTE, by about 15% and 40%. The wings are un-a ff ected by NLTE e ff ects. Individual best fits to all other He i Table 2.
Isotopic helium abundances λ (Å) log τ ross ε (He) NLTE ε (He) LTE4 He i − He i He i − He i − Table 3.
Line-by-line abundances of He i . λ (Å) log τ ross ε (He) NLTE ε (He) LTE − − − ff ects are weakin these mostly weak lines. Strong e ff ects are present only in He i λ NLTE effects.
Departure coe ffi cients b i = n i / n ∗ i (Zwaan defini-tion, the n i and n ∗ i are the NLTE and LTE level population num-bers) for the helium energy levels are displayed in Fig. 8 as afunction of optical depth. He i is the main ion at the temper-atures found in the visible layers of the atmosphere of κ Cnc.The population of the He i ground state shows practically nodeviations from detailed equilibrium. The levels with principalquantum number n = s , S levels are not shown in Fig. 8 for clarity, asthey show qualitatively and quantitatively a very similar be-haviour as the 2 p , P ◦ levels). This is because radiative (the2 s S state is metastable) and collisionally induced decays tothe ground state are ine ffi cient, while the level populations areconstantly fed by the recombination cascade from higher lev-els, which drains the populations of these. In consequence, allhigher levels are underpopulated compared with the LTE val-ues. They closely follow the behaviour of the He ii ground state.Detailed balance is approached for all levels at continuum for-mation depths (at log τ ross ≈ ff er at most by a few percent from LTE valuesin the line-formation region (see Tables 2 and 3). The di ff erencesin the behaviour of the departure coe ffi cients between He and He levels are negligible. All the He i lines that were analysedhere have either the 2 p P ◦ or the 2 p P ◦ level as the lower tran-sition level and upper levels with a higher principal quantumnumber. Therefore, NLTE strengthening of the lines can occuras the consequence of overpopulated lower and underpopulatedupper levels, depending on the formation depth. The (stronger)red He i lines are more sensitive to NLTE e ff ects also because ofthe pronounced response of the line-source function to variationsof the departure coe ffi cients in the Rayleigh-Jeans limit (see e.g.Przybilla & Butler 2004). Even small di ff erences in level popu-lations on the percent-level can therefore result in fundamentallyaltered line strengths. Article number, page 4 of 7aza et al.: NLTE spectral analysis of He and He in κ Cnc
Fig. 8.
Departure coe ffi cients b i of the He (full lines) and the Helevels (dotted lines) as a function of optical depth for a model with ε ( He) = ε ( He) = n of the levels are indicated. The dots mark the runof the He ii ground-state departure coe ffi cient.
6. Discussion
Using abundances from lines of di ff erent strength – which sam-ple the line-formation region to a di ff erent extent –, one can re-construct the abundance profile, that is, the vertical abundancestratification of an element in the atmosphere. The run of the he-lium number fraction ε (He) with optical depth in the atmosphereof κ Cnc is shown in Fig. 9. The LTE analysis indicates a step-like helium stratification, dropping in the region log τ ross ≈ ε (He) ≈ ff erent behaviour. The NLTEabundances are lower by almost a factor 3 at the lowest opticaldepths that can be traced. Note that while the formation regionsof the weaker lines agree well when computed under LTE andNLTE conditions, they can di ff er by up to ∼ τ ross for the stronger lines. The LTE computations indicate a forma-tion of the line cores farther out in the atmosphere.It would be highly interesting to know whether the two he-lium isotopes follow the same stratification profile, or if di ff eren-tial e ff ects prevail. Unfortunately, with data from only two linesthis cannot be decided here with confidence.Finally, it may be appropriate here to comment on NLTE ef-fects in CP stars in the general context, based on the present find-ings. The helium lines in κ Cnc are weak because of the ratherlow T e ff for populating the excited levels and because of the lowhelium abundance that is due to gravitational settling. Neverthe-less, NLTE e ff ects are found for all but the weakest He i lines,leading to changes in the derived abundances relative to an LTEanalysis by up to a factor ∼
3. On the other hand, most of themetals show large overabundances because of radiative levita-tion, yielding spectral lines much stronger than usually found innormal stars. The formation regions for the strong lines extend tomuch lower optical depths than investigated here, possibly giv-ing rise to much stronger NLTE e ff ects than encountered in thepresent work. More NLTE studies for CP stars are encouragedbecause this may have a significant e ff ect on the results of ele-mental abundance determinations and on the deduced abundancestratifications. Acknowledgements.
We thank V. Scha ff ernroth for helping us to improve thenormalization of the observed spectrum. We also thank our anonymous refereefor several suggestions that helped to improve the manuscript. NLM acknowl- Fig. 9.
Helium number fraction as a function of Rosseland optical depthin the atmosphere of κ Cnc (NLTE results: dots, LTE: circles). edges a CONICET postdoc stipend and additional financial support by CON-ICET under the ‘Programa de financiamiento parcial de estadías breves en el ex-terior para becarios postdoctorales’, and MFN an equal opportunity FFL stipendfrom the University of Erlangen-Nuremberg. Financial support by the Interna-tional Relations O ffi ce to a research stay of NLM at University of Innsbruck isgratefully acknowledged. References
Abt, H. A., Cha ff ee, F. H., & Su ff olk, G. 1972, ApJ, 175, 779Adelman, S. J., Pro ffi tt, C. R., Wahlgren, G. M., Leckrone, D. S. & Dolk, L.2004, ApJS, 155, 179Auer, L. H. & Mihalas, D. 1973, ApJS, 25, 433Bailey, J. D. & Landstreet, J. D. 2013, A&A, 551, A30Becker, S. R. & Butler, K. 1988, A&AS, 74, 211Bessell, M. S., Castelli, F., & Plez, B. 1998, A&A, 337, 321Butler, W. P. & Giddings, J. R. 1985, in Newsletter of Analysis of AstronomicalSpectra, No. 9 (Univ. London)Cardelli, J. A., Clayton, G. C., & Mathis, J. S. 1989, ApJ, 345, 245Castelli, F. & Hubrig, S. 2007, A&A, 475, 1041Dekker, H., D’Odorico, S., Kaufer, A., Delabre, B., & Kotzlowski, H. 2000,Proc. SPIE, 4008, 534Dobrichev, V. M., Raikova, D. V., Ryabchikova, T. A., & Topil’skaya, G. P. 1989,Afz, 30, 91Dworetsky, M. M. 2004, in IAU Symposium, Vol. 224, The A-Star Puzzle, ed.J. Zverko, J. Ziznovsky, S. J. Adelman, & W. W. Weiss, 727Dworetsky, M. M. & Budaj, J. 2000, MNRAS, 318, 1264Fred, M., Tomkins, F. S., Brody, J. K., & Hamermesh, M. 1951, Phys. Rev., 82,406Gerbaldi, M., Floquet, M., & Hauck, B. 1985, A&A, 146, 341Giddings, J. R. 1981, Ph.D. Thesis (Univ. London)Hartoog, M. R. & Cowley, A. P. 1979, ApJ, 228, 229Heber, U., Moehler, S., Napiwotzki, R., Thejll, P., & Green, E. M. 2002, A&A,383, 938Hubrig, S., González, J. F., Ilyin, I., et al. 2011, Astron. Nachr., 332, 998Hubrig, S. & Mathys, G. 1995, Comments on Astrophysics, 18, 167Kurucz, R. L. 1993, CD-ROM No. 13 (SAO, Cambridge, Mass.)Maha ff y, P. R., Donahue, T. M., Atreya, S. K., Owen, T.C., & Niemann, H.B.1998, Space Sci. Rev., 84, 251Maza, N. L., Levato, H. O., López-García, Z. & Sa ff e, C. 2011, Bol. Asoc. Ar-gentina Astron., 54, 101Mermilliod, J. C. 1991, Catalogue of Homogeneous Means in the UBV System(Institut d’Astronomie, Universite de Lausanne)Michaud, G., Martel, A., Montmerle, T., et al. 1979, ApJ, 234, 206Nieva, M. F. & Przybilla, N. 2007, A&A, 467, 295Nieva, M. F. & Przybilla, N. 2008, A&A, 481, 199Nieva, M.-F. & Przybilla, N. 2012, A&A, 539, A143Preston, G. W. 1974, ARA&A, 12, 257Przybilla, N. 2005, A&A, 443, 293Przybilla, N. & Butler, K. 2004, ApJ, 609, 1181Przybilla, N., Butler, K., Becker, S. R., Kudritzki, R. P., & Venn, K. A. 2000,A&A, 359, 1085Przybilla, N., Butler, K., Becker, S. R., & Kudritzki, R. P. 2006a, A&A, 445,1099Przybilla, N., Nieva, M.F. & Edelmann, H. 2006b, Balt. Astron., 15, 107Roby, S. W. & Lambert, D. L. 1990, ApJS, 73, 67Schöller, M., Correia, S., Hubrig, S. & Ageorges, N. 2010, A&A, 522, A85Skrutskie, M. F., Cutri, R. M., Stiening, R., et al. 2006, AJ, 131, 1163Smith, K. C. 1996, Ap&SS, 237, 77Takeda, Y., Kawanomoto, S. & Ohishi, N. 2014, PASJ, 66, 23Zakharova, L. A. & Ryabchikova, T. A. 1996, Astronomy Letters, 22, 152 Article number, page 5 of 7 &A–aa25037-14,
Online Material p 6
Fig. 3.
Comparison of our global best-fit NLTE synthetic spectrum (redline) with the observed Paschen lines Pa11 and Pa12 (black line) in κ Cnc. &A–aa25037-14,
Online Material p 7
Fig. 4.
Comparison of our global best-fit NLTE synthetic spectrum (for log (O / H) + = i/ii lines (black lines) in κκ