G 207-9 and LP 133-144: light curve analysis and asteroseismology of two ZZ Ceti stars
Zs. Bognár, M. Paparó, L. Molnár, P. I. Pápics, E. Plachy, E. Verebélyi, Á. Sódor
aa r X i v : . [ a s t r o - ph . S R ] J un MNRAS , 1–13 (2016) Preprint 10 October 2018 Compiled using MNRAS L A TEX style file v3.0
G 207-9 and LP 133-144: light curve analysis andasteroseismology of two ZZ Ceti stars
Zs. Bogn´ar, ⋆ M. Papar´o, L. Moln´ar, P. I. P´apics, E. Plachy, E. Vereb´elyi, and ´A. S´odor Konkoly Observatory, MTA Research Centre for Astronomy and Earth Sciences, Konkoly Thege Mikl´os ´ut 15-17, H–1121 Budapest Instituut voor Sterrenkunde, KU Leuven, Celestijnenlaan 200D, B-3001 Leuven, Belgium
Accepted XXX. Received YYY; in original form ZZZ
ABSTRACT
G 207-9 and LP 133-144 are two rarely observed ZZ Ceti stars located in the middleand close to the blue edge of the ZZ Ceti instability domain, respectively. We aimed toobserve them at least during one observing season at Konkoly Observatory with thepurpose of extending the list of known pulsation modes for asteroseismic investigationsand detect any significant changes in their pulsational behaviour. We determined fiveand three new normal modes of G 207-9 and LP 133-144, respectively. In LP 133-144,our frequency analysis also revealed that at least at three modes there are actuallytriplets with frequency separations of ∼ m Hz. The rotational period of LP 133-144based on the triplets is ≃ h. The preliminary asteroseismic fits of G 207-9 predict T eff =
12 000 or
12 400
K and M ∗ = . − . M ⊙ values for the effective temperatureand mass of the star, depending on the assumptions on the spherical degree ( l ) valuesof the modes. These results are in agreement with the spectroscopic determinations.In the case of LP 133-144, the best-fitting models prefer T eff =
11 800
K in effectivetemperature and M ∗ ≥ . M ⊙ stellar masses, which are more than . M ⊙ largerthan the spectroscopic value. Key words: techniques: photometric – stars: individual: G 207-9, LP 133-144 – stars:interiors – stars: oscillations – white dwarfs
ZZ Ceti (or DAV) stars constitute the most populated groupof pulsating white dwarfs. Their light variations are the re-sults of local changes in their surface temperatures due tothe excitation of nonradial g -mode pulsations in their non-degenerate envelope. This envelope consists of an inner he-lium and an outer hydrogen layer, therefore, the hydrogenBalmer-lines dominate the spectra of ZZ Ceti stars. The pul-sations are driven by the so-called ‘convective driving’ mech-anism (Brickhill 1991; Goldreich & Wu 1999), as the drivingregion is associated with the base of the envelope convectionzone. Otherwise, pulsating white dwarfs are just like theirnon-pulsating counterparts, and the information we gain onwhite dwarf structures by asteroseismic investigations canbe essential to understand white dwarfs as a whole group.ZZ Ceti stars are short-period and low-amplitude pul-sators with − K effective temperatures andmodes excited typically in the − s period rangewith ∼ mmag amplitudes. However, within this period and ⋆ E-mail: bognar.zsofi[email protected] (Zs.B.) amplitude range the stars exhibit a large variety of pul-sational behaviour, from the star showing one rotationaltriplet only (G 226-29, Kepler et al. 1995) to the ‘rich’ DApulsators with more than a dozen normal modes known(see e.g. Bischoff-Kim 2009). Temporal variations in theirpulsational behaviour are also well documented in whitedwarfs, e.g. the case of GD 154 showing once a strongly non-sinusoidal light curve with one dominant mode and a seriesof its harmonic and near-subharmonic ( ∼ n / f i ) peaks in itsFourier transform, or just behaving as a simple multiperiodicpulsator another time (Robinson et al. 1978; Papar´o et al.2013). For comprehensive reviews of the observational andtheoretical aspects of pulsating white dwarf studies, seethe papers of Fontaine & Brassard (2008); Winget & Kepler(2008) and Althaus et al. (2010). We also refer to the work ofVan Grootel et al. (2013), in which the authors successfullyreconstructed the boundaries of the empirical ZZ Ceti insta-bility strip applying theoretical calculations, including itsextension to lower effective temperatures and surface gravi-ties, that is, further to the domain of the extremely low-massDA pulsators.White dwarf observations with the Kepler space tele- c (cid:13) Zs. Bogn´ar et al. scope revealed another new feature in ZZ Ceti stars, namelyrecurring increases in the stellar flux (‘outbursts’) in twocool DAVs being close to the red edge of the instability strip(Bell et al. 2015; Hermes et al. 2015b).G 207-9 and LP 133-144 were observed as part of ourproject aiming at least one-season-long local photometrictime series measurements of white dwarf pulsators. Our pur-poses are to examine the short-term variability of pulsationmodes in amplitude and phase, and to obtain precise periodsfor asteroseismic investigations. We have already publishedour findings on two cool ZZ Ceti stars (KUV 02464+3239,Bogn´ar et al. 2009 and GD 154, Papar´o et al. 2013), one ZZCeti located in the middle of the instability strip (GD 244;Bogn´ar et al. 2015), and the DBV type KUV 05134+2605(Bogn´ar et al. 2014). With the observations of G 207-9 andLP 133-144, we extended our scope of investigations tohigher effective temperatures in the DAV instability domain.
We collected photometric data both on G 207-9 ( B = . mag, a = h m s , d = + d m s ) andLP 133-144 ( B = . mag, a = h m s , d =+ d m s ) in the 2007 observing season. We used the 1-mRitchey-Chr´etien-Coud´e telescope at Piszk´estet˝o mountainstation of Konkoly Observatory. The detector was a Prince-ton Instruments VersArray:1300B back-illuminated CCDcamera. The measurements were made in white light andwith 10 or 30 s integration times, depending on the weatherconditions.We observed G 207-9 and LP 133-144 on 24 and 28nights, respectively. Tables 1 and 2 show the journals ofobservations. Altogether, 85 and 137 h of photometric datawere collected on G 207-9 and LP 133-144, respectively.We reduced the raw data frames following the standardprocedure: we applied bias, dark and flat corrections on theframes using iraf routines, and performed aperture pho-tometry of the variable and comparison stars with the irafdaophot package. We converted the observational times ofevery data point to barycentric Julian dates in barycentricdynamical time ( BJD
TDB ) using the applet of Eastman et al.(2010) . We then checked the comparison star candidates forvariability and instrumental effects. We selected three starsin the field of G 207-9 and two stars in the field of LP 133-144 and used the averages of these reference stars as com-parisons for the differential photometry of the two pulsators.The panels of Fig. 2 show the variable and the comparisonstars in the CCD fields. We applied low-order polynomial fitsto the light curves to correct for the instrumental trends andfor the atmospheric extinction. This method did not affectthe pulsation frequency domains. Figure 1 shows two illus-trative light curve segments of G 207-9 and LP 133-144. Allthe light curves obtained for both pulsators are presented inAppendix A and in Appendix B. iraf is distributed by the National Optical Astronomy Obser-vatories, which are operated by the Association of Universities forResearch in Astronomy, Inc., under cooperative agreement withthe National Science Foundation. http://astroutils.astronomy.ohio-state.edu/time/utc2bjd.html Table 1.
Journal of observations of G 207-9. ‘Exp.’ is the expo-sure time used.Run UT date Start time Exp. Points Lengthno. (2007) (BJD-2 450 000) (s) (h)01 Mar 26 4185.540 30 279 2.5702 Mar 27 4186.533 30 280 2.7003 Apr 02 4192.527 10 743 2.7004 Apr 03 4193.540 30 208 1.8805 Jun 15 4267.351 10 1001 4.4606 Jun 16 4268.374 30 427 4.1007 Jun 17 4269.396 10 1028 3.6808 Jun 18 4270.341 30 500 4.7509 Jun 19 4271.345 10 1173 4.8910 Jun 20 4272.454 30 198 2.2311 Jul 06 4288.389 30 184 2.4212 Jul 07 4289.347 10 1417 5.1813 Jul 08 4290.337 10 1449 5.6014 Jul 09 4291.351 30 34 0.3015 Jul 10 4292.434 30 227 2.2716 Jul 26 4308.434 30 349 3.1217 Jul 27 4309.337 10 1071 3.6918 Jul 30 4312.452 30 134 1.4219 Jul 31 4313.312 30 492 5.0320 Aug 01 4314.460 10 885 3.0521 Aug 10 4323.327 30 46 0.4022 Aug 13 4326.323 30 721 6.4323 Aug 14 4327.341 10 1646 5.7424 Aug 15 4328.315 10 1896 6.53Total: 16 388 85.16
We determined the frequency content of the datasets ondaily, weekly or monthly, and yearly time bases. We analysedthe daily observations with custom developed software tools,as the command-line light curve fitting program
LCfit (S´odor 2012).
LCfit has linear (amplitudes and phases)and nonlinear (amplitudes, phases and frequencies) least-squares fitting options, utilizing an implementation of theLevenberg-Marquardt least-squares fitting algorithm. Theprogram can handle unequally spaced and gapped datasets.
LCfit is scriptable easily, which made the analysis of therelatively large number of nightly datasets very effective.We performed the standard Fourier analyses of theweekly or monthly data subsets and the whole light curveswith the photometry modules of the Frequency Analysisand Mode Identification for Asteroseismology ( famias ) soft-ware package (Zima 2008). Following the traditional way,we accepted a frequency peak as significant if its amplitudereached the 4 signal-to-noise ratio (S/N). The noise level wascalculated as the average amplitude in a ± m Hz intervalaround the given frequency.
G 207-9 was announced as the 8th known member of pulsat-ing white dwarf stars in 1976 (Robinson & McGraw 1976).Four high ( F – F ) and one low ( F ) amplitude peaks weredetected at F = , F = , F = , F = and MNRAS , 1–13 (2016) he pulsations of G 207-9 and LP 133-144 -0.050.000.05 4289.35 4289.36 4289.37 4289.38 4289.39 4289.40 4289.41 4289.42 4289.43 4289.44 G 207-9 -0.050.000.05 4234.32 4234.33 4234.34 4234.35 4234.36 4234.37 4234.38 4234.39 4234.40 4234.41
LP 133-144 A m p li t ude [ m ag ] BJD
TDB -2450000
Figure 1.
Segments of the light curves of G 207-9 ( upper panel ) and LP 133-144 ( lower panel ) obtained at Piszk´estet˝o using 10 sintegration times.
Figure 2.
CCD frames with the variable and comparison stars marked. The field of view is ≈ ′ × ′ . F = m Hz. Even though G 207-9 is a relatively brighttarget, and has been known as a pulsator for decades, noother time series photometric observations and frequencyanalysis have been published on this star up to now.
The analyses of the daily datasets revealed one dominantand four low amplitude peaks in the FTs. The domi-nant frequency of all nights’ observations in 2007 was at m Hz. This was the 3rd highest amplitude mode in 1975( F = m Hz). Two lower amplitude frequencies at and m Hz reached the 4 S/N detection limit in 12 and14 of the daily datasets, respectively. Two additional low-amplitude frequencies were detected at and m Hzin 3 and 4 cases, respectively. These frequencies are medi-ans of the daily values. The and m Hz frequenciescould be determined separately only in the last three nights’datasets. We present the FT of one night’s dataset (the sec-ond longest run) in the first panel of Fig. 3.Considering the consecutive nights of observations,five weekly time base datasets can be formed: Week 1(JD 2 454 185–193), Week 2 (JD 2 454 267–272), Week 3(JD 2 454 288–292), Week 4 (JD 2 454 308–314) and Week 5 (JD 2 454 323–328). The Fourier analyses of these data ver-ified the five frequencies found by the daily observations.In three cases the first harmonic of the dominant frequencywas also detected. Additionally, the analysis of the Week 2and Week 5 data suggested that the peaks at ∼ and ∼ m Hz may be actually doublets or triplets and notsinglet frequencies. The separations of the frequency compo-nents were found to be between ∼ (close to the resolutionlimit) and m Hz, but we mark these findings uncertain be-cause of the effect of the 1 d − aliasing. The 2nd–6th panelsof Fig. 3 shows the FTs of the weekly datasets. We foundonly slight amplitude variations from one week to another.The amplitude of the dominant frequency varied between8.6 and 10.5 mmag.The standard pre-whitening of the whole dataset re-sulted 26 frequencies above the 4 S/N limit. Most of them areclustering around the frequencies already known by the anal-yses of the daily and weekly datasets. Generally, amplitudeand (or) phase variations during the observations can be re-sponsible for the emergence of such closely spaced peaks. Insuch cases, these features are just artefacts in the FT, as wefit the light curve with fixed amplitudes and frequencies dur-ing the standard pre-whitening process. Another possibilityis that some of the closely spaced peaks are rotationally split MNRAS , 1–13 (2016)
Zs. Bogn´ar et al.
Table 2.
Journal of observations of LP 133-144. ‘Exp.’ is theexposure time used.Run UT date Start time Exp. Points Lengthno. (2007) (BJD-2 450 000) (s) (h)01 Jan 15 4115.614 30 299 2.8102 Jan 17 4117.622 30 154 1.5403 Jan 26 4126.615 30 233 2.1904 Jan 28 4128.544 30 441 4.1705 Jan 30 4130.528 30 493 4.5806 Feb 17 4148.551 30 375 3.5307 Mar 15 4175.283 30 725 9.2708 Mar 16 4176.279 30 960 9.2909 Mar 22 4182.356 30 562 6.0410 Mar 24 4184.496 30 410 3.7611 Mar 25 4185.398 30 338 3.1212 Mar 26 4186.282 30 636 5.8813 Mar 27 4187.273 30 718 7.9414 Mar 30 4190.307 30 862 8.0515 Mar 31 4191.357 30 325 2.9816 Apr 01 4192.289 30 596 5.4717 Apr 03 4194.276 30 202 1.9218 Apr 12 4203.388 10 1483 5.5819 Apr 13 4204.337 10 1808 6.8020 Apr 14 4205.304 10 1946 7.6621 Apr 15 4206.309 10 1608 7.2722 Apr 16 4207.296 10 1422 5.5123 Apr 17 4208.378 10 1439 5.2624 May 10 4231.316 10 1026 4.0825 May 12 4233.358 30 259 2.3326 May 13 4234.315 10 1192 4.3227 May 14 4235.372 10 763 2.8228 May 16 4237.365 30 342 3.12Total: 21 617 137.27
Table 3.
G 207-9: frequency content of the 2007 dataset. Theerrors were calculated by Monte Carlo simulations. d f denotesthe frequency differences of the closely spaced frequencies to f , f or f . Frequency Period | d f | Ampl. S/N ± . ( m Hz) (s) ( m Hz) (mmag) f . ± . f . ± . f − ? 3414 . ± . f . ± . f − ? 1667 . ± . f . ± . f + ? 3437 . ± . f . ± . f − ? 1595 . ± . f . ± . f . ± . f . ± . frequencies. We can resolve such frequencies if the time baseof the observations is long enough. The Rayleigh frequencyresolution ( / D T ) of the whole dataset is . m Hz. We alsohave to consider the 1 d − alias problem of single-site obser-vations, which results uncertainties in the frequency deter-mination.We checked the frequency content of the whole dataset JD 2454326
Week 1
Week 2
Week 3
Week 4
Week 5 A m p li t ude [ m ag ] Frequency [ m Hz]
Figure 3.
G 207-9: amplitude spectra of one night’s observation( top panel ) and the weekly datasets ( lower panels ). by averaging three consecutive data points of the s mea-surements as a test. That is, we created a new, more homoge-neous dataset mimicking s exposure times. We then com-pared the frequency solutions of this s dataset with thefrequencies of the original mixed – s data. Finally, we ac-cepted as the frequencies characterizing the whole light curvethe frequencies that could be determined in both datasets,that is, without 1 d − differences. This resulted a reducedfrequency list of 12 frequencies. We list them in Table 3.There are still several closely spaced frequencies aroundthree of the main frequencies ( f , f and f ) remained withseparations between . and . m Hz. In the case of f and f , these separations are close to . m Hz (1 d − ). It ispossible that at least some of these frequency componentsare results of rotational splitting, but considering the uncer-tainties mentioned above, we do not accept them as rota-tionally split frequencies. In such cases when the frequencyseparations of the rotationally split components are around1 d − , multi-site or space-based observations are needed forreliable determination of the star’s rotational rate. MNRAS , 1–13 (2016) he pulsations of G 207-9 and LP 133-144 G 207-9 f
012 1600 1800 f f
01 5000 5200 f
01 7600 7800 f
01 6800 7000
01 3000 3200 f
01 3200 3400 f -200 0 200 A m p li t ude [ mm ag ] Frequency [ m Hz]
Figure 4.
G 207-9: FT of the whole dataset, the window function is given in the inset ( top panel ). We emphasized the frequency domainsof f − f and f , as seen in the original FT or during the pre-whitening process ( lower panels ). A m p li t ude [ mm ag ] Frequency [ m Hz] F F F =f F =f F f f f f f Figure 5.
G 207-9: comparison of the frequencies obtained in1975 ( red dashed lines ) and in 2007 ( black solid lines ). Theamplitudes of the 1975 observations are from the paper ofMukadam et al. (2006).
Summing it up: besides the five frequencies ( f – f )known also by the analyses of shorter (daily and weekly)data segments, we could detect two additional independentfrequencies ( f and f ) in the whole dataset. Frequency f at . m Hz was also detected in 1975 ( F = m Hz).Moreover, this was one of the dominant peaks at that time.Frequency f at . m Hz is a newly detected one. Notethat the frequency f is close to f + f = . m Hz, how-ever, the difference is . m Hz, which seems too large to claimthat f is the linear combination of these peaks consideringthe errors. Thus, we consider f as an independent mode.Fig. 4 shows the FT of the whole dataset and the frequencydomains of f − f on separate panels.Comparing the frequency content of the 1975 and 2007observations, we can conclude that three of the five frequen-cies found in the 1975 dataset did not appear in 2007 ( F , F and F ), while two stayed at an observable level ( F = f and F = f ). Figure 5 summarizes the frequencies of the twoepochs. It seems that even though there were no large am-plitude variations during our five-months observing seasonin 2007, on the time scale of years or decades, remarkable changes can happen in the pulsation of G 207-9: new fre-quencies can be excited to a significant level, while othermodes can disappear. The variability of LP 133-144 was discovered in 2003(Bergeron et al. 2004). Four pulsation frequencies were de-termined at that time, including two closely spaced peaks: F = . , F = . , F = . and F = . m Hz.Similarly to the case of G 207-9, no further results of timeseries photometric observations have been published up tonow.
We found four recurring frequencies in the daily datasetsat 3055, 3270, 3695 and 4780 m Hz (median values). Theiramplitudes varied from night to night, but the 4780 m Hzpeak was the dominant in almost all cases. One additionalpeak exceeded the 4 S/N limit at 5573 m Hz, but on one nightonly.We created four monthly datasets and analysed themindependently. These are Month 1 (JD 2 454 115–130),Month 2 (JD 2 454 175–194), Month 3 (JD 2 454 203–208)and Month 4 (JD 2 454 231–237). The analyses of themonthly data revealed that at the 3270, 3695 and 4780 m Hzfrequencies there are actually doublets or triplets with 2.6–4.7 m Hz frequency separations. This explains the differentamplitudes in the daily FTs. The 3055 m Hz frequency wasfound to be a singlet. In Month 3, the linear combination ofthe largest amplitude components of the 3270 and 4780 m Hzmultiplets also could be detected. The 5573 m Hz frequencywas significant in Month 2.
MNRAS000
MNRAS000 , 1–13 (2016)
Zs. Bogn´ar et al.
JD 2454234
Month 1
Month 2
Month 3
Month 4 A m p li t ude [ m ag ] Frequency [ m Hz]
Figure 6.
LP 133-144: amplitude spectra of one night’s observa-tion ( top panel ) and the monthly datasets ( lower panels ). The panels of Fig. 6 show the FT of one daily datasetand the monthly data. As in the case of G 207-9, there wereno remarkable amplitude variations from one month to an-other.The analysis of the whole 2007 dataset resulted inthe detection of 19 significant frequencies in the ∼ − m Hz frequency region. We also performed the test anal-ysis utilizing the averaged 30 s dataset, which confirmedthe presence of the 14 largest amplitude frequencies (theother five peaks remained slightly under the significancelevel). Thus we accepted them as the frequencies character-izing the pulsation of LP 133-144 and list them in Table 4.The Rayleigh frequency resolution of the whole dataset is . m Hz.The first eleven peaks in Table 4 are three triplets withfrequency separations of . − . m Hz ( f ), . m Hz ( f ) or . m Hz ( f ), and two singlet frequencies ( f and f ). In thecase of f , three peaks can be determined in the original10-30 s dataset with frequency separations of . − . m Hz.However, the low amplitude central peak of this triplet at f = . m Hz do not reach the 4 S/N significance limitin the test 30 s data. Still, to make the discussion of thetriplet structures clear, we added f to the list of Table 4in parentheses. Besides these, the first harmonic of f alsoappeared. Fig. 7 shows the FT of the whole dataset, the Table 4.
LP 133-144: frequency content of the 2007 dataset. Theerrors were calculated by Monte Carlo simulations. d f denotes thefrequency differences of the closely spaced frequencies to f , f f or f . We discuss the case of f in the text. The signal-to-noiseratios refer to the original 10-30 s dataset.Frequency Period | d f | Ampl. S/N ± . ( m Hz) (s) ( m Hz) (mmag) f ± f ± f ± f − ± f + ± f ± f + ± f − ± f + ± f − ± f ± f + ± f ± f − ± f ) 5569.618 ± consecutive pre-whitening steps at the multiplet frequenciesand at the frequency domains of f , f and f .We plot the frequencies of Bergeron et al. (2004) andthe frequencies found in the 2007 Konkoly observations to-gether in Fig. 8. Assuming that the closely spaced peaksat F and F are results of the not properly resolved compo-nents of the f triplet, we found, with similar amplitudes, allthe frequencies observed in 2003. Besides these, we detectedthree new frequencies: a relatively large amplitude mode at f , and two additional low-amplitude modes at f and f .That is, we doubled the number of modes can be used forthe asteroseismic fits.The schematic plot of the triplets can be seen in Fig. 9.It is clearly visible that the frequency separations of thecomponents are larger at higher frequencies. We discuss therotation of LP 133-144 based on the investigation of thesetriplets in Sect. 4.3.1. We built our model grid for the asteroseismic investi-gations of our targets utilizing the White Dwarf Evolu-tion Code ( wdec ; Lamb 1974; Kutter & Savedoff 1969;Lamb & van Horn 1975; Winget 1981; Kawaler 1986; Wood1990; Bradley 1993; Montgomery 1998; Bischoff-Kim et al.2008). The wdec evolves a hot polytrope model ( ∼ K)down to the requested temperature, and provides an equi-librium, thermally relaxed solution to the stellar structureequations. Then we are able to calculate the set of possi-ble zonal ( m = ) pulsation modes according to the adia-batic equations of non-radial stellar oscillations (Unno et al.1989). We utilized the integrated evolution/pulsation formof the wdec code created by Metcalfe (2001) to derivethe pulsation periods for the models with the given stel-lar parameters. More details on the physics applied in the wdec can be found with references in Bischoff-Kim et al. MNRAS , 1–13 (2016) he pulsations of G 207-9 and LP 133-144 LP 133-144 -200 0 2000123 f
01 3000 3100 01234 f f f
01 3200 3300 01234 f f f f f f
01 4700 4800 01 f f f
01 5500 5600 01 f
01 7100 720001
01 9500 9600 A m p li t ude [ mm ag ] Frequency [ m Hz]
Figure 7.
LP 133-144: FT of the whole dataset, the window function is given in the inset ( top panel ). We also plotted the FTs of theconsecutive pre-whitening steps at the multiplet frequencies f , f , f and f , the frequency domains of the singlets f and f , and thepeak at f ( lower panels ). The bottom panels show the residual spectra after pre-whitening with the denoted frequencies. (2008) and in our previous papers on two ZZ Ceti stars(Bogn´ar et al. 2009; Papar´o et al. 2013).Considering the limited visibility of high spherical de-gree ( l ) modes due to geometric cancellation effects, we cal-culated the periods of dipole ( l = ) and quadrupole ( l = )modes for the model stars only. The goodness of the fitbetween the observed ( P obs i ) and calculated ( P calc i ) periodswas characterized by the root mean square ( s rms ) value cal- culated for every model with the fitper program of Kim(2007): s rms = s (cid:229) Ni = ( P calc i − P obs i ) N (1)where N is the number of observed periods.We varied five main stellar parameters to build ourmodel grid: the effective temperature ( T eff ), the stellar mass( M ∗ ), the mass of the hydrogen layer ( M H ), the central oxy-gen abundance ( X O ) and the fractional mass point where MNRAS000
LP 133-144: FT of the whole dataset, the window function is given in the inset ( top panel ). We also plotted the FTs of theconsecutive pre-whitening steps at the multiplet frequencies f , f , f and f , the frequency domains of the singlets f and f , and thepeak at f ( lower panels ). The bottom panels show the residual spectra after pre-whitening with the denoted frequencies. (2008) and in our previous papers on two ZZ Ceti stars(Bogn´ar et al. 2009; Papar´o et al. 2013).Considering the limited visibility of high spherical de-gree ( l ) modes due to geometric cancellation effects, we cal-culated the periods of dipole ( l = ) and quadrupole ( l = )modes for the model stars only. The goodness of the fitbetween the observed ( P obs i ) and calculated ( P calc i ) periodswas characterized by the root mean square ( s rms ) value cal- culated for every model with the fitper program of Kim(2007): s rms = s (cid:229) Ni = ( P calc i − P obs i ) N (1)where N is the number of observed periods.We varied five main stellar parameters to build ourmodel grid: the effective temperature ( T eff ), the stellar mass( M ∗ ), the mass of the hydrogen layer ( M H ), the central oxy-gen abundance ( X O ) and the fractional mass point where MNRAS000 , 1–13 (2016)
Zs. Bogn´ar et al. A m p li t ude [ mm ag ] Frequency [ m Hz] F =f F F F =f f f f f Figure 8.
LP 133-144: comparison of the frequencies obtained in2003 ( red dashed lines ) and in 2007 ( black solid lines ). f f -2024681012 4776 4778 4780 4782 4784 4786 f
01 5566 5568 5570 5572 5574 f A m p li t ude [ mm ag ] Frequency [ m Hz]
Figure 9.
LP 133-144: schematic plot of the triplets found atdifferent frequency domains. The frequency errors are comparableto the width of the lines. the oxygen abundance starts dropping ( X fm ). We fixed themass of the helium layer ( M He ) at − M ∗ . The grid coversthe parameter range − K in T eff (the middle andhot part of the ZZ Ceti instability strip), . − . M ⊙ in stellar mass, − − − M ∗ in M H , . − . in X O and . − . in X fm . We used step sizes of K ( T eff ), . M ⊙ ( M ∗ ), . dex (log M H ) and 0.1 ( X O and X fm ). In the case of G 207-9, we could detect seven linearly inde-pendent pulsation frequencies by the 2007 Konkoly dataset( f − f ; see Table 3). The question is, if we could addmore frequencies to this list by the 1975 observations ofRobinson & McGraw (1976). As we mentioned already inSect. 3.1.2, two of the frequencies detected in 1975 were alsofound in the Konkoly data ( F = f and F = f ). The sta-tus of the remaining three 1975 frequencies is questionable.Assuming at least a couple of m Hz errors for the 1975 fre-
Table 5.
G 207-9 and LP 133-144: periods utilized for the modelfits. G 207-9 LP 133-144Period Period(s) (s) f f f f f f f f f f f f f F quencies, F = F − F (or F = F − F , or F = F + F ), thus,these three frequencies do not seem to be linearly indepen-dent. The fact that F and F are the two dominant peaks inthe FT of Robinson & McGraw (1976) suggests that F and F might be the parent modes and F is a combination peak.Furthermore, Robinson & McGraw (1976) pointed out that F − F ≈ F − F , thus, further combinations are possible. Wealso note that f of the Konkoly dataset is almost at twicethe value of F ( d f = . m Hz), however, there is no sign ofany pulsation frequency at . f in the 2007 data.We used two sets of observed periods to fit the calcu-lated ones. One set consists of the seven periods of f − f observed in 2007, while we complemented this list with theperiod of F detected in 1975 to create another set. We se-lected F because it was the second largest amplitude peakin 1975, which makes it a good candidate for an additionalnormal mode.In LP 133-144, we found all the previously observedfrequencies in our 2007 dataset, as we show in Sect. 3.2.2.Thus, we cannot add more frequencies to our findings, andperformed the model fits with six periods. We summarizedthe periods utilized for modelling in Table 5 for both stars. We determined the best-matching models considering sev-eral cases: at first, we let all modes to be either l = or l = . Then we assumed that the dominant peak is an l = ,considering the better visibility of l = modes over l = ones. At last, we searched for the best-fitting models assum-ing that at least four of the modes is l = , including thedominant frequency.We obtained the same model as the best-fitting astero-seismic solution both for the seven- and eight-period fits. Ithas T eff = K, M ∗ = . M ⊙ and M H = − M ∗ . Thismodel has the lowest s rms ( . − . s) both if we do notapply any restrictions on the l values of the modes, and as itgives l = solution to the dominant frequency, this model isalso the best-fit if we assume that the . s mode is l = .Note that in this model solution only this mode is an l = ,all the other six or seven modes are l = .In the case of four expected l = modes and seven peri-ods, the best-matching model has the same effective tem-perature ( T eff = K), a bit lower stellar mass ( M ∗ = . M ⊙ ), and thinner hydrogen layer ( M H = − M ∗ ).Assuming four l = modes and eight periods, the best- MNRAS , 1–13 (2016) he pulsations of G 207-9 and LP 133-144 matching model has T eff = K, M ∗ = . M ⊙ and M H = − . M ∗ . The second best-fit model is the same asfor four l = modes and seven periods. We denoted withopen circles these two latter models in Fig. 10 (left panel)on the T eff − M ∗ plane, together with the spectroscopic solu-tion. Both in the case of G 207-9 and LP 133-144, we uti-lized the T eff and surface gravity ( log g ) values provided byGianninas et al. (2011), and then corrected them accordingto the results of Tremblay et al. (2013) based on radiation-hydrodynamics 3D simulations of convective DA stellar at-mospheres. We accepted the resulting values as the best esti-mates for these atmospheric parameters. We converted thesurface gravities to stellar masses utilizing the theoreticalmasses determined for DA stars by Bradley (1996).Considering the mass of the hydrogen layer (see the leftpanel of Fig. 11), we found that most of the models up to s rms = . s are in the M H = − − − M ∗ range, while abouta dozen models predict thinner hydrogen layer down to − M ∗ . The best-fitting models favour the M H = − . M ∗ value.We summarize the results of the spectroscopic atmo-spheric parameter determinations, the former modelling re-sults based on the 1975 frequency list, the main stellar pa-rameters of the models mentioned above and the calculatedperiods fitted with our observed ones in Table 6. We alsolist the s rms values of the models. The T eff = K so-lutions are in agreement with the spectroscopic value. The T eff = K model seems somewhat too hot comparing tothe ∼ K spectroscopic temperature, but consideringthat the uncertainties of both values are estimated to bearound
K, this model still not contradicts to the obser-vations. The . − . M ∗ stellar masses are also closeto the value derived by spectroscopy, considering its uncer-tainty. Summing it up, we can find models with stellar pa-rameters and periods close to the observed values even if weassume that at least half of the modes is l = , including thedominant mode. The model with the lowest s rms ( . s) has T eff = K, M ∗ = . M ⊙ and M H = − . M ∗ if we do not apply anyrestrictions on the l values of modes. Generally, the best-matching models have masses around . M ⊙ , which are atleast . M ⊙ larger than the spectroscopic value. These mod-els provide − l = solutions to the observed modes.We searched for the best-matching models in a sec-ond run, assuming that the three largest amplitude modesshowing triplet structures at 209.2, 305.9 and 270.6 s areall l = modes. The best-matching model has the sameeffective temperature ( T eff = K), slightly larger mass( M ∗ = . M ⊙ ) and much thinner hydrogen layer ( M H = − . M ∗ ) than the previously selected model. The mass stillseems too large comparing to the spectroscopic value, butit gives l = solutions for all the four modes with tripletfrequencies, including the mode at 179.5 s. These modes areconsecutive radial overtones with k = − . We denoted thismodel with an open circle on the middle and right panels ofFig. 10. The hydrogen layer masses versus the s rms valuesof these models are plotted in the right panel of Fig. 11.This figure also shows that the best-fitting models have thinhydrogen layer with M H = − . M ∗ . Otherwise, two families of model solutions outlines: one with M H = − . − − . M ∗ and one with thinner, M H = − . − − . M ∗ hydrogen lay-ers. If we restrict our period fitting to the models with effec-tive temperatures and masses being in the range determinedby spectroscopy, the best-matching model has T eff = K, M ∗ = . M ⊙ and M H = − . M ∗ . However, the 179.5 smode is l = in this case, while all the other frequenciesare consecutive radial order l = modes.At last, we searched for models in this restricted param-eter space and assuming that all the four frequencies show-ing triplets are l = . Our finding with the lowest s rms has T eff = K, M ∗ = . M ⊙ and M H = − . M ∗ , however,its s rms is relatively large ( . s), which means that thereare major differences between the observed and calculatedperiods. Table 7 lists the stellar parameters and theoreticalperiods of the models mentioned above. For completeness,we included this last model solution, too.We concluded, that our models predict at least . M ⊙ larger stellar mass for LP 133-144 than the spectroscopicvalue. Nevertheless, it is possible to find models with lowerstellar masses, but in these cases not all the modes withtriplet frequency structures has l = solutions and (or)the corresponding s rms values are larger than for the largermass models. Considering the effective temperatures, the T eff = K solutions are in agreement with the spectro-scopic determination ( ∼ K) within its margin of error.As in the case of G 207-9, taking into account that the un-certainties for the grid parameters are of the order of thestep sizes in the grid, the T eff = K findings are stillacceptable.
A plausible explanation for the observed triplet structuresis that these are rotationally split frequency components of l = modes. We used this assumption previously in searchingfor model solutions for our observed periods. Knowing thefrequency differences of the triplet components ( d f ), we canestimate the rotation period of the pulsator.In the case of slow rotation, the frequency differences ofthe m = − , , rotationally split components can be calcu-lated (to first order) by the following relation: d f k ,ℓ, m = d m ( − C k ,ℓ ) W , (2)where the coefficient C k ,ℓ ≈ /ℓ ( ℓ + ) for high-overtone ( k ≫ ℓ ) g -modes and W is the (uniform) rotation frequency.In the case of LP 133-144, the presumed l = modesare low radial-order frequencies ( k = − ), but the C k ,ℓ val-ues of the fitted modes can be derived by the asteroseismicmodels. We used the average of the frequency separationswithin a triplet and calculated the stellar rotation rate sep-arately for f , f , f and f (see e.g. Hermes et al. 2015a). Weutilized the T eff = K, M ∗ = . M ⊙ model. The result-ing rotation periods are: P f = . d ( d f = . m Hz, C k ,ℓ = . ), P f = . d ( d f = . m Hz, C k ,ℓ = . ), P f = . d( d f = . m Hz, C k ,ℓ = . ) and P f = . d ( d f = . m Hz, C k ,ℓ = . ). The average rotation period thus . ± . d( ∼ h). This fits perfectly in the known rotation rates ofthe order of hours to days of ZZ Ceti stars (cf. Table 4 in MNRAS , 1–13 (2016) Zs. Bogn´ar et al. S t e ll a r m a ss [ M S un ] Figure 10.
Models on the T eff − M ∗ plane with their s rms values (colour coded). Left panel : G 207-9; models fitting with eight periodsand assuming that at least half of the modes is l = , including the dominant mode. The two models with the lowest s rms values (for moreexplanation, see the text) are denoted with open circles. Black dot marks the spectroscopic value (cf. Table 6). Middle panel : LP 133-144;models fitting with six periods and assuming that the three largest amplitude modes (also showing triplets) are l = . The best-fittingmodel is denoted with an open circle, the spectroscopic value presented in Table 7 is signed with a black dot. Right panel : magnifiedpart of the middle panel’s plot around the best-matching model. We used interpolation in the plots for better visibility. H y d r ogen l a y e r m a ss [ - l og M H / M * ] rms [s]G 207-9 4.05.06.07.08.01.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0rms [s]LP 133-144 Figure 11.
Mass of the hydrogen layer of models at different s rms values. Left panel : G 207-9, right panel : LP 133-144. We plotted thesame models that are presented in Fig. 10.
Fontaine & Brassard 2008). Note that the rotation periodscalculated by the different multiplet structures are stronglydepend on the actual values of observed frequency spacingsand also on the C k ,ℓ values, which vary from model to model.Thus the different rotation periods calculated for the differ-ent modes does not of necessarily mean that e.g. in this casewe detected differential rotation of the star, but we can pro-vide a reasonable estimation on the global rotation periodof LP 133-144. We have presented the results of the one-season-long photo-metric observations of the ZZ Ceti stars G 207-9 and LP 133-144. These rarely observed pulsators are located in the mid-dle and in the hot part of the instability strip, respectively. G 207-9 was found to be a massive object previously by spec-troscopic observations, comparing its predicted M ∗ > . M ⊙ mass to the average ∼ . M ⊙ value of DA stars (see e.g.Kleinman et al. 2013). In contrary, the mass of LP 133-144was expected to be around this average value.With our observations performed at Konkoly Observa-tory, we extended the number of known pulsation frequenciesin both stars. We found seven linearly independent modesin G 207-9, including five newly detected frequencies, com-paring to the literature data. We also detected the possiblesigns of additional frequencies around some of the G 207-9modes, but their separations being close to the 1 d − valuemakes their detection uncertain. Multi-site or space-basedobservations could verify or disprove their presence. In thecase of LP 133-144, we detected three new normal modesout of the six derived, and revealed that at least at three MNRAS , 1–13 (2016) he pulsations of G 207-9 and LP 133-144 Table 6.
G 207-9: results on the main stellar parameters obtained by spectroscopic observations (‘
Spectroscopy ’), previous model fits(‘
Modelling ’) and modelling presented in this paper (‘
This work ’). In the ‘
Modelling ’ part, we listed the observed periods used for thefits. In the case of our model fits, we listed the observed periods first and present the theoretical periods at the actual models. The s rms values are given at the effective temperatures in parentheses. In the case of model fits the identification of pulsation modes - sphericaldegree ( l ) and radial order ( k ) values - are also presented. T eff (K) M ∗ / M ⊙ -log M He -log M H Periods in seconds ( l , k ) Reference Spectroscopy:
12 078 ±
200 0.84 ± Modelling:
12 000 0.815 2.0 8.5 259.0, 292.0, 317.3, 557.3, 740.7, 787.5 ⋆ Castanheira & Kepler (2009)11 700 0.530 3.5 6.5 259.0, 292.0, 317.3, 557.3, 740.7, 787.5 ⋆ Castanheira & Kepler (2009)12 030 0.837 2.5 6–7 259.1 (1,4), 292.0 (2,10), 318.0 (1,5), Romero et al. (2012, 2013)557.4 (1,12), 740.4 (1,17)
This work: ⋆ The utilized periods were mean values of the periods of Robinson & McGraw (1976) and the periods of WD J0815+4437 showingsimilar pulsation modes.
Table 7.
Same as in Table 6 but for LP 133-144. T eff (K) M ∗ / M ⊙ -log M He -log M H Periods in seconds ( l , k ) Reference Spectroscopy:
12 152 ±
200 0.59 ± Modelling:
11 700 0.520 2.0 5.0 209.2 (1,2), 305.7 (2,7), 327.3 (2,8) Castanheira & Kepler (2009)12 210 0.609 1.6 ∼ This work: modes there are actually triplet frequencies with frequencyseparations of ∼ m Hz.All the pulsation modes of LP 133-144 and most of themodes of G 207-9 are found to be below 330 s, with ampli-tudes up to ∼ mmag. This fits to the well-known trendobserved at ZZ Ceti stars that at higher effective tempera-tures we see lower amplitude and shorter period light vari-ations than closer to the red edge of the instability strip(see e.g. Fontaine & Brassard 2008). We also found that onthe five-month time scale of our observations there were nosignificant amplitude variations in either stars. This suits to their location in the instability domain again, as shorttime scale large amplitude variations are characteristics ofZZ Cetis with lower effective temperatures. However, in thecase of G 207-9, the different frequency content of the 1975and 2007 observations shows that amplitude variations dooccur on decade-long time scale.In addition, similar pulsational feature of the two starsis that both show light variations with one dominant mode( A = − mmag) and several lower amplitude frequencies.The extended list of known modes allowed to performnew asteroseismic fits for both objects, in which we com- MNRAS , 1–13 (2016) Zs. Bogn´ar et al. pared the observed and calculated periods both with andwithout any restrictions on the l values of modes. The best-matching models of G 207-9 have found to be close to thespectroscopic effective temperature and stellar mass, pre-dicting T eff = or K and M ∗ = . − . M ⊙ .For LP 133-144, the best-fitting models prefer more than . M ⊙ larger stellar masses than the spectroscopic mea-surements and T eff = K effective temperatures. Themain sources of the differences in our model solutions andthe models presented by Castanheira & Kepler (2009), eventhough they also used the wdec , can arise from the differentperiods utilized for the fits, the different core compositionprofiles applied, and the different way they determined thebest-fitting models utilizing the amplitudes of observed pe-riods as weights to define the goodness of the fits. At last,we derived the rotational period of LP 133-144 based on theobserved triplets and obtained P rot ≃ h.Note that the results of the asteroseismic fits presentedin this manuscript are preliminary findings, and both ob-jects deserve more detailed seismic investigations utilizingthe extended period lists, similarly to the modelling pre-sented for other hot DAV stars, GD 165 and Ross 548(Giammichele et al. 2016). In the case of these objects, theauthors could identify models reproducing the observed peri-ods quite well while staying close to the spectroscopic stellarparameters, and also verified the credibility of the selectedmodels in many other ways, including the investigation ofrotationally split frequencies. ACKNOWLEDGEMENTS
The authors thank the anonymous referee for the construc-tive comments on the manuscript. The authors thank Agn`esBischoff-Kim for providing her version of the wdec and the fitper program. The authors also thank the contributionof E. Bokor, ´A. Gy˝orffy, Gy. Kerekes, A. M´ar and N. Sztan-k´o to the observations of the stars. The financial supportof the Hungarian National Research, Development and In-novation Office (NKFIH) grants K-115709 and PD-116175,and the LP2014-17 Program of the Hungarian Academy ofSciences are acknowledged. P.I.P. is a Postdoctoral Fellow ofthe The Research Foundation – Flanders (FWO), Belgium.L.M. and ´A.S. was supported by the J´anos Bolyai ResearchScholarship of the Hungarian Academy of Sciences.
REFERENCES
Althaus L. G., C´orsico A. H., Isern J., Garc´ıa-Berro E., 2010,A&ARv, 18, 471Bell K. J., Hermes J. J., Bischoff-Kim A., Moorhead S., Mont-gomery M. H., Østensen R., Castanheira B. G., Winget D. E.,2015, ApJ, 809, 14Bergeron P., Fontaine G., Bill`eres M., Boudreault S., Green E. M.,2004, ApJ, 600, 404Bischoff-Kim A., 2009, in Guzik J. A., Bradley P. A., eds,American Institute of Physics Conference Series Vol. 1170,American Institute of Physics Conference Series. pp 621–624,doi:10.1063/1.3246573Bischoff-Kim A., Montgomery M. H., Winget D. E., 2008, ApJ,675, 1512Bogn´ar Z., Papar´o M., Bradley P. A., Bischoff-Kim A., 2009,MNRAS, 399, 1954 Bogn´ar Z., Papar´o M., C´orsico A. H., Kepler S. O., Gy˝orffy ´A.,2014, A&A, 570, A116Bogn´ar Z., Papar´o M., Moln´ar L., Plachy E., S´odor ´A., 2015, inDufour P., Bergeron P., Fontaine G., eds, Astronomical Soci-ety of the Pacific Conference Series Vol. 493, 19th EuropeanWorkshop on White Dwarfs. p. 245 ( arXiv:1506.06960 )Bradley P. A., 1993, PhD thesis, Texas Univ., Austin.Bradley P. A., 1996, ApJ, 468, 350Brickhill A. J., 1991, MNRAS, 251, 673Castanheira B. G., Kepler S. O., 2009, MNRAS, 396, 1709Eastman J., Siverd R., Gaudi B. S., 2010, PASP, 122, 935Fontaine G., Brassard P., 2008, PASP, 120, 1043Giammichele N., Fontaine G., Brassard P., Charpinet S., 2016,ApJS, 223, 10Gianninas A., Bergeron P., Ruiz M. T., 2011, ApJ, 743, 138Goldreich P., Wu Y., 1999, ApJ, 511, 904Hermes J. J., et al., 2015a, MNRAS, 451, 1701Hermes J. J., et al., 2015b, ApJ, 810, L5Kawaler S. D., 1986, PhD thesis, Texas Univ., Austin.Kepler S. O., et al., 1995, ApJ, 447, 874Kim A., 2007, PhD thesis, The University of Texas at AustinKleinman S. J., et al., 2013, ApJS, 204, 5Kutter G. S., Savedoff M. P., 1969, ApJ, 156, 1021Lamb Jr. D. Q., 1974, PhD thesis, THE UNIVERSITY OFROCHESTER.Lamb D. Q., van Horn H. M., 1975, ApJ, 200, 306Metcalfe T. S., 2001, PhD thesis, The University of Texas atAustinMontgomery M. H., 1998, PhD thesis, The University of Texas atAustinMukadam A. S., Montgomery M. H., Winget D. E., Kepler S. O.,Clemens J. C., 2006, ApJ, 640, 956Papar´o M., Bogn´ar Z., Plachy E., Moln´ar L., Bradley P. A., 2013,MNRAS, 432, 598Robinson E. L., McGraw J. T., 1976, ApJ, 207, L37Robinson E. L., Stover R. J., Nather R. E., McGraw J. T., 1978,ApJ, 220, 614Romero A. D., C´orsico A. H., Althaus L. G., Kepler S. O., Castan-heira B. G., Miller Bertolami M. M., 2012, MNRAS, 420, 1462Romero A. D., Kepler S. O., C´orsico A. H., Althaus L. G., FragaL., 2013, ApJ, 779, 58S´odor ´A., 2012, Konkoly Observatory Occasional Technical Notes,15Tremblay P.-E., Ludwig H.-G., Steffen M., Freytag B., 2013,A&A, 559, A104Unno W., Osaki Y., Ando H., Saio H., Shibahashi H., 1989, Non-radial oscillations of starsVan Grootel V., Fontaine G., Brassard P., Dupret M.-A., 2013,ApJ, 762, 57Winget D. E., 1981, PhD thesis, Univ. RochesterWinget D. E., Kepler S. O., 2008, ARA&A, 46, 157Wood M. A., 1990, PhD thesis, Texas Univ., Austin.Zima W., 2008, Communications in Asteroseismology, 155, 17
APPENDIX A:
Normalized differential light curves of G 207-9 obtained in2007 at Piszk´estet˝o mountain station of Konkoly Observa-tory.
APPENDIX B:
Normalized differential light curves of LP 133-144 obtainedin 2007 at Piszk´estet˝o mountain station of Konkoly Obser-vatory.
MNRAS , 1–13 (2016) he pulsations of G 207-9 and LP 133-144 This paper has been typeset from a TEX/L A TEX file prepared bythe author.MNRAS000