Discovery of a new PG1159 (GW Vir) Pulsator
S. O. Kepler, Luciano Fraga, Don Earl Winget, Keaton Bell, Alejandro H. Corsico, Klaus Werner
aa r X i v : . [ a s t r o - ph . S R ] M a y Mon. Not. R. Astron. Soc. , 1–5 (2014) Printed 9 June 2018 (MN L A TEX style file v2.2)
Discovery of a new PG1159 (GW Vir) Pulsator
S. O. Kepler ⋆ , Luciano Fraga , Don Earl Winget , Keaton Bell ,Alejandro H. C´orsico and Klaus Werner Instituto de F´ısica, Universidade Federal do Rio Grande do Sul, 91501-900 Porto-Alegre, RS, Brazil Laborat´orio Nacional de Astrof´ısica, Itajub´a, MG, Brazil Department of Astronomy and McDonald Observatory, The University of Texas at Austin, Austin TX 78712-1083, USA, Facultad de Ciencias Astron´omicas y Geof´ısicas, Paseo del Bosque S/N, (1900) La Plata, andConsejo Nacional de Investigaciones Cient´ıficas y T´enicas (CONICET), Argentina Institute for Astronomy and Astrophysics, Kepler Center for Astro and Particle Physics, Eberhard Karls University, Samd I,72076 T¨ubingen, Germany, e-mail: [email protected]
Accepted 2014 May 20. Received 2014 May 19; in original form 2014 April 9
ABSTRACT
We report the discovery of pulsations in the spectroscopic PG 1159 typepre-white dwarf SDSS J075415.12+085232.18. Analysis of the spectrum byWerner, Rauch & Kepler (2014) indicated T eff = 120 000 ±
10 000 K, log g = 7 . ± . M = 0 . ± . M ⊙ , C/He=0.33 by number.We obtained time-series images with the SOAR 4.1 m telescope and 2.1 m OttoStruve telescope at McDonald Observatory and show the star is also a variable PG 1159type star, with dominant period of 525 s. Key words: stars – white dwarf, individual: SDSS J075415.12+085232.18
White dwarf stars are the end product of evolution of allstars with initial masses up to around 8 − M ⊙ , dependingon the metalicity of the progenitor and its effect on massloss and the real value of the C( α, γ )O reaction rate. Theirspatial and mass distributions contain information aboutstar formation history and subsequent evolution in ourGalaxy. As the most common endpoints of stellar evolution,white dwarf stars account for around 95% of all evolvedstars. The GW Vir stars, also called DOVs, are the pulsat-ing variables in the spectroscopic PG 1159 class that linksthe (post-AGB) central stars of planetary nebulae and theH-deficient white dwarf cooling sequence. These stars arenon-radial pulsators and lie in an instability strip boundedby effective temperatures 200 000 T eff
75 000 K, excitedby the κ -mechanism working through partial ionization ofcarbon and oxygen. Asteroseismological analysis of thesestars has provided significant knowledge on the interiorsof the late stages of stellar evolution (Winget & Kepler2008; Althaus et al. 2010). There are 20 known GWVir stars Quirion (2009a); Quirion, Fontaine, & Brassard(2009b); Woudt, Warner, & Zietsman (2012). Findingnew pulsators of this class can improve our knowledgeof the asymptotic giant branch (AGB) and Very Late ⋆ [email protected] Thermal Pulse (VLT) phases, as well as angular mo-mentum loss throughout the extensive mass loss phases(Charpinet, Fontaine, & Brassard 2009; C´orsico et al.2011).In our search for new spectroscopically con-firmed white dwarf stars in the Sloan Digital Sur-vey (SDSS) (Kleinman et al. 2013), we identifiedSDSS J075415.12+085232.18 as a hot pre-white dwarffrom the presence of He II and carbon lines in spec-trum Plate=2945 MJD=54505 Fiber=183 of thisg=18.79 star. It shows no detectable planetary neb-ula. Werner, Rauch & Kepler (2014) fitted its SDSSspectrum with non local thermodynamic equilibrium(NLTE) models and obtained T eff = 120 000 ±
10 000 K,log g = 7 . ± .
3, mass M = 0 . ± . M ⊙ , and C/He=0.33by number, indicating the star is a spectroscopic PG 1159type star, i.e., hotter and with a more complex spectrumthan a normal DO white dwarf, similar to the prototype(Liebert et al. 1989; Werner, Heber, & Hunger 1989),which is also a pulsating star (e.g McGraw et al. 1979;Winget et al. 1991; Kawaler & Bradley 1994; Costa et al.2008a). The observed g = 18 . ± .
01 apparent magnitude,compared to an M g = 5 .
68 for such effective tempera-ture and gravity, implies a distance of 4 . ± .
03 kpc.Such a large distance merits the full extinction correc-tion in that direction, 0.076 mag in g, which bringsthe distance to 4.04 kpc. C´orsico & Althaus (2006a);C´orsico, Althaus, & Miller Bertolami (2006b) computed c (cid:13) Kepler et al.
He IIC IV1.01.5 4000 4500 5000 5500 6000 6500 7000 wavelength / A o r e l a t i v e f l u x Figure 1.
Spectrum of SDSS J075415.12+085232.1 and the model (overplotted in red) with T eff = 120 000 K, log g = 7 and C/He=0.33by number. fully evolutionary models and non-adiabatic pulsationmodels for stars in the GW Vir instability strip and foundthat they agree with the observed strip. We first obtained time-series photometry ofSDSS J075415.12+085232.18 with the 4.1-m SOAR telescope and using the Soar Optical Imager[SOI, Schwarz et al. (2004)] during the night of28 Jan 2014 (Barycentric Julian Terrestrial TimeBJTT=245 6685.7200646). SOI is a mini-mosaic of two E2V2k ×
4k CCDs covering a 5.26 arcmin square field of view at aplate scale of 0.077 arcsec/pixel. We obtained a total of 263SOI frames with a Bessel-B filter, exposure time of 30 s and4 × daophot (Stetson 1991) routines inIRAF. From the Fourier analysis, we achieved a mean noiselevel of h A i =1.4 mma, and detected for the first time aperiodicity, with a period of 525 s at 6.8 mma, therefore at4.8 h A i , well above the 1/1000 false alarm probability limit.On the three consecutive nights of 3, 4, and 5 Feb2014 (BJTT=245 6691.699600605) we obtained follow-upobservations of the star with the Cassegrain-mountedProEM camera and the PuokoNui data acquisition software(Chote et al. 2014) at McDonald Observatory’s 2.1-m OttoStruve telescope. From 3127 images with 10 to 30s expo-sures, we confirmed the 525 s periodicity at 5.9 mma, com-pared to the average noise level h A i =1.03 mma. The frameswere binned at 4 ×
4, giving a 0.36 arcsec/pixel plate scaleacross the 2.3 × Based on observations obtained at the Southern AstrophysicalResearch (SOAR) telescope, which is a joint project of the Min-ist´erio da Ciˆencia, Tecnologia, e Inova¸c˜ao (MCTI) da Rep´ublicaFederativa do Brasil, the U.S. National Optical Astronomy Ob-servatory (NOAO), the University of North Carolina at ChapelHill (UNC), and Michigan State University (MSU). package ccd hsp (Kanaan, Kepler, & Winget 2002) and cal-culated barycentric corrections with the
WQED software(Thompson & Mullally 2009).We obtained additional time-series observa-tions with the SOAR Goodman Spectrograph(Clemens, Crain, & Anderson 2004) in imaging modeduring the night of 27 Feb 2014 (BJTT=245 6715.5192530).Goodman is mounted at the SOAR Optical Nasmyth andits detector is a 4k ×
4k Fairchild 486 back-illuminated CCD,with a un-binned plate scale of 0.15 arcsec/pixel. We carriedout the photometric observations with a S8612 red blockfilter, a region of interest (ROI) of 800 ×
800 pixel square anda 2 × h A i =0.94 mma, with which we were able to detectthree periodicities, 523.5 s at 7.0 mma, 457.2 s at 3.8 mma,and 439.2 at 3.5 mma, all above the 1/1000 false alarmprobability.Figure 2 shows the Fourier transform of all datasets. Analyzing the whole data set at once, we obtained523.480 ± ± ± ± δν ≃ . µ Hz is similar to that for ℓ = 1 modes ofPG 1159 −
035 (4.1 µ Hz). If the spacing is real, it indicatesa rotation period of 28 h, similar to those derived for othervariable PG 1159 stars.
The pulsation modelling and seismological analysis pre-sented in this section rely on a set of stellar models thattake into account the complete evolution of PG 1159 pro-genitor stars. The models were extracted from the evo-lutionary calculations presented by Althaus et al. (2005)and Miller Bertolami & Althaus (2006), who computed thecomplete evolution of model star sequences with initialmasses on the zero-age main sequence (ZAMS) rangingfrom 1 to 3 . M ⊙ . All of the post-AGB evolutionary se-quences were computed using the LPCODE evolutionary code(Althaus et al. 2005) and were followed through the very c (cid:13) , 1–5 iscovery of a new PG1159 (GW Vir) Pulsator Figure 2.
Fourier transform of the two SOAR data sets (blackand blue lines), and the McDonald data set (cyan shaded). The3 h A i line, corresponding to the false alarm probability of 1/1000,refers only to the equally spaced SOAR data set from 27 Feb2014, the one with lowest noise, shown in black. late thermal pulse (VLTP) and the resulting born-againepisode that gives rise to the H-deficient, and He-, C-, andO-rich composition characteristic of PG 1159 stars. Themasses of the resulting remnants are 0.530, 0.542, 0.556,0.565, 0.589, 0.609, 0.664, and 0 . M ⊙ .With only three periods detected forSDSS J075415.12+085232.18, we cannot estimate themean period spacing, and cannot constrain the stellar massby comparing with the mean period spacing of the models,as done in the case of other pulsating PG 1159 stars (e.g.,C´orsico et al. 2009). The way to infer the stellar mass, alongwith the effective temperature and also details of the internalstructure of SDSS J075415.12+085232.18 is through theirindividual pulsation periods. This has been the approachemployed by C´orsico et al. (2007a,b, 2008, 2009) for thepulsating PG 1159 stars RX J2117.1+3412, PG 0122+200,PG 1159 − ℓ = 1 , g -mode adia-batic pulsation periods used in C´orsico et al. (2007a,b, 2008,2009). For details of the adiabatic pulsation code ( LP-PULcode ) and methods employed to produce the set of pe-riods, see C´orsico & Althaus (2006a). We analyzed morethan about 3000 PG 1159 models covering a wide rangeof effective temperatures [5 . & log( T eff ) & . . log( L ∗ /L ⊙ ) . . . M ∗ /M ⊙ . k asso-ciated to the observed periods ( ∼ −
524 s) is large(as we shall see below), the pulsation g -modes of SDSSJ075415.12+085232.18 are probably not in the asymptoticregime (see, for instance, C´orsico & Althaus (2006a)). Be-cause the models are evolutionary, not started from a poly- Π obs Π theor ℓ k Table 1.
Observed and theoretical periods for the best model fitin Case 1. trope, they cannot achieve any combination of mass, lu-minosity and effective temperature, and do not cross eachother in the Hertzsprung-Russell diagram. The best solu-tions, quoted, are not just samples of possible solutions, butlimited solutions. As there are three independent modes, onecan estimate up to three parameters of the models.We seek pulsation models that best match the individ-ual pulsation periods of SDSS J075415.12+085232.18. Thegoodness of the match between the theoretical pulsation pe-riods (Π k ) and the observed individual periods (Π obs ,i ) ismeasured by means of a quality function defined as: χ ( M ∗ , T eff ) = 1 N N X i =1 min[(Π obs ,i − Π k ) ] (1)where N (= 3) is the number of observed periods. In theabsence of any additional information, we assume that thethree observed periods of SDSS J075415.12+085232.18 cor-respond to eigenmodes with azimuthal order m = 0, butMetcalfe (2003) shows the effect of the assumption is neg-ligible. We evaluate the function χ ( M ∗ , T eff ) for evolution-ary models with stellar masses of 0.530, 0.542, 0.556, 0.565,0.589, 0.609, 0.664, 0 . M ⊙ . The PG 1159 model thatshows the lowest value of χ is adopted as the “best-fitmodel”. Since we do not know at the outset the harmonicdegree ( ℓ ) identification of the observed modes, we have todistinguish three cases. ℓ = 1 modes Here, we consider that all the three measured periods areassociated to modes with ℓ = 1. We obtain a best fit so-lution characterized by: M ∗ = 0 . M ⊙ , T eff = 130 100 K,and L/L ⊙ = 170. A comparison between the observed andtheoretical periods, along with the derived ℓ and k (radialorder) values associated to this solution is shown in Table 1.The quality function for this case is displayed at theupper panel of Figure 3. ℓ = 1 and ℓ = 2 modes In this case we consider that the observed periods are asso-ciated with a mix of ℓ = 1 and ℓ = 2 modes. We perform aperiod fit in which the value of ℓ for the theoretical periodsis not fixed, but instead is obtained as a result of our periodfit procedure, with allowed values of ℓ = 1 and ℓ = 2. Thesolution is displayed in the central panel of Figure 3 and has M ∗ = 0 . M ⊙ , T eff = 128 300 K, L/L ⊙ = 156. The agree-ment between theoretical and observed modes is shown inTable 2. c (cid:13) , 1–5 Kepler et al. Π obs Π theor ℓ k Table 2.
Observed and theoretical periods for the best model fitin Case 2. Π obs Π theor ℓ k Table 3.
Observed and theoretical periods for the best model fitin Case 1. ℓ = 2 modes Finally, we assume that all three identified periods are asso-ciated with ℓ = 2, even though it is improbable that a pul-sating pre-white dwarf star shows only quadrupole modes.The solution is displayed in the bottom panel of Figure 3and has M ∗ = 0 . M ⊙ , T eff = 86 900 K, L/L ⊙ = 21 (Table3).This solution, already unlikely from the point of view ofgeometrical cancellation, gives too low of a temperature,compared with the spectral determination. A secondary so-lution is observed for M ∗ = 0 . M ⊙ and T eff = 121 600K, but it can be discarded because of its very high massvalue, as compared with the spectroscopically inferred mass(0 . ± . M ⊙ ) of SDSS J075415.12+085232.18.The agreement between theoretical and observed pe-riods of the solution in Case (2) is excellent, with two ℓ = 1 modes and one ℓ = 2, and the mean difference of0.17 s is within observational and theoretical uncertain-ties. This solution agrees with the spectral temperature, T eff = 120 000 ±
10 000 K, and is within the real uncertainty(i.e., including systematic uncertainties) of the spectroscopicmass: M ∗ = 0 . ± . N p (cid:16) log NN (cid:17) + log σ where N p is the number of free parameters, and N the num-ber of observed periods. The BIC parameter estimates theabsolute quality of the period fit, by accounting for situa-tions in which there are different numbers of observed peri-ods and free parameters. In our case, N p = 2 (stellar massand effective temperature), and N = 3. The smaller thevalue of BIC, the better the quality of the fit. We obtainBIC = 0 .
11 for Case (1), BIC = − .
98 for Case (2), andBIC = 0 .
14 for Case (3). The period fit of Case (2) is excel-lent, as reflected by the corresponding BIC value. It couldbe compared with the BIC value of current asteroseismolog-ical period fits of pulsating white dwarfs [see, for instance,Bischoff-Kim & Østensen (2011)].
Time series imaging show SDSS J075415.12+085232.18 is anon-radial pulsator in the PG 1159 pre-white dwarf class,also called GW Vir. Its spectral effective temperature T eff = ( χ ) - ( χ ) - eff [K]00.511.5 ( χ ) - l= 1 χ = 0.05 χ = 0.62 l= 1, 2l= 2 χ = 0.67 Figure 3.
The inverse of the quality function of the period fitsin terms of the effective temperature. The vertical gray strip in-dicates the spectroscopic T eff and its uncertainties. Upper panelcorresponds to the case in which the three periods are associatedto ℓ = 1 modes, middle panel shows the situation in which thereis a mix of ℓ = 1 and ℓ = 2 modes, and lower panel displays thecase in which the three modes are ℓ = 2.
120 000 ±
10 000 K and C/He=0.33 by number is compara-ble to the prototype, and the main period at 525s is alsocomparable to the 516s main periodicity of PG 1159-035.Its low pulsation amplitude led to a small number of pe-riodicities detected, contrary to the prototype, which hasthe largest number of independent pulsations detected afterthe Sun. That PG 1159 stars probably have no atmosphericconvection layer might explain the absence of combinationfrequencies, even when large amplitudes are detected, as inPG 1159–035 itself (Costa et al. 2008a).These stars evolve fast, leading to substantial pe-riod change due to cooling and contraction, that shouldallow a detectable period change in a few years(Winget, Hansen, & van Horn 1983). Therefore the starshould be monitored at least yearly to allow evolutionarychanges determinations (Costa & Kepler 2008b).
REFERENCES
Althaus L. G., Serenelli A. M., Panei J. A., C´orsico A. H.,Garc´ıa-Berro E., Sc´occola C. G., 2005, A&A, 435, 631Althaus L. G., C´orsico A. H., Isern J., Garc´ıa-Berro E.,2010, A&ARv, 18, 471Bischoff-Kim A., Østensen R. H., 2011, ApJ, 742, L16Charpinet S., Fontaine G., Brassard P., 2009, Natur, 461,501Chote P., Sullivan D. J., Brown R., Harrold S. T., WingetD. E., Chandler D. W., 2014, MNRAS, 559 c (cid:13) , 1–5 iscovery of a new PG1159 (GW Vir) Pulsator Clemens J. C., Crain J. A., Anderson R., 2004, SPIE, 5492,331C´orsico A. H., Althaus L. G., 2006, A&A, 454, 863C´orsico A. H., Althaus L. G., Miller Bertolami M. M., 2006,A&A, 458, 259C´orsico A. H., Althaus L. G., Miller Bertolami M. M.,Garc´ıa-Berro E., 2009, A&A, 499, 257C´orsico A. H., Althaus L. G., Kepler S. O., Costa J. E. S.,Miller Bertolami M. M., 2008, A&A, 478, 869C´orsico A. H., Althaus L. G., Miller Bertolami M. M.,Werner K., 2007, A&A, 461, 1095C´orsico A. H., Miller Bertolami M. M., Althaus L. G., Vau-clair G., Werner K., 2007, A&A, 475, 619C´orsico A. H., Althaus L. G., Kawaler S. D., Miller Berto-lami M. M., Garc´ıa-Berro E., Kepler S. O., 2011, MNRAS,418, 2519Costa J. E. S., et al., 2008, A&A, 477, 627Costa J. E. S., Kepler S. O., 2008, A&A, 489, 1225Kanaan A., Kepler S. O., Winget D. E., 2002, A&A, 389,896Kawaler S. D., Bradley P. A., 1994, ApJ, 427, 415Kleinman S. J., et al., 2013, ApJS, 204, 5Koen C., Laney D., 2000, MNRAS, 311, 636Liebert J., Wesemael F., Husfeld D., Wehrse R., StarrfieldS. G., Sion E. M., 1989, AJ, 97, 1440McGraw J. T., Liebert J., Starrfield S. G., Green R., 1979,wdvd.coll, 377Metcalfe T. S., 2003, BaltA, 12, 247Miller Bertolami M. M., Althaus L. G., 2006, A&A, 454,845Quirion P.-O., 2009, CoAst, 159, 99Quirion P.-O., Fontaine G., Brassard P., 2009, JPhCS, 172,012077Schwarz H. E., et al., 2004, SPIE, 5492, 564Stetson P. B., 1991, ESOC, 38, 187Thompson S. E., Mullally F., 2009, Journal of Physics:Conference Series, 172, 012081Valdes F. G., 1998, ASPC, 145, 53Valdes F. G., Tody D., 1998, SPIE, 3355, 497Werner K., Heber U., Hunger K., 1989, LNP, 328, 194Werner K., Rauch T., & Kepler S.O., A&A, 564, A53Winget D. E., Kepler S. O., 2008, ARA&A, 46, 157Winget D. E., et al., 1991, ApJ, 378, 326Winget D. E., Hansen C. J., van Horn H. M., 1983, Natur,303, 781Woudt P. A., Warner B., Zietsman E., 2012, MNRAS, 426,2137This paper has been typeset from a TEX/ L A TEX file preparedby the author. c (cid:13)000